AMD’s E-350 Fusion APU on the desktop

I’ll admit it. We are about to commit a terrible crime against the holy writ of product segmentation. We’re about to consider the merits of AMD’s Brazos and Intel’s Pine Trail—incredibly small, low-power, and inexpensive PC platforms—versus a whole range of full-grown desktop CPUs, a ridiculously unfair and inappropriate endeavor if ever there was one.

Before we dig into our heinous crime against carefully calibrated marketing, we should take a second to explain how we’ve arrived at this sorry state of affairs. First of all, you should know that we have some history of violating product segmentation lines when we’ve found it interesting to do so. After all, running a Pentium M on the desktop seemed at one point like a pretty decent idea. Since Brazos and Atom serve as the basis for a number of small-form-factor desktop systems, asking how they compare to larger, more expensive options felt like fair game to us.

Then there’s the fact that AMD made us do it. You see, back when the firm revealed the first details about the Bobcat CPU architecture used in Brazos, it predicted Bobcat would achieve an “estimated 90% of today’s mainstream performance in less than half the silicon area.” Now, such statements are famously slippery, but those words did conjure up some rather dreamy expectations inside of our heads. We likened Bobcat’s potential performance to 90% of the Athlon X2 255’s—then arguably a “mainstream” part as these terms tend to be used—and noted that “the X2 255 is more than up to the task of running modern games” and “should be plenty adequate for the vast majority of everyday computing tasks.” With prospects like that, a comparison seemed to be in order.

Also consider the fascinating fact that x86-compatible CPUs span the sheer range that they do, from tens of dollars to thousands, from a handful of watts to well over a hundred, from tiny “APUs” to behemoths comprised of over a billion transistors. Faced with this amazing dynamic range of cross-compatible processors, we couldn’t resist a little exploration. Particularly when we already had a fresh set of results for desktop CPUs spanning from $87 to $999 created for our Sandy Bridge review.

So yes, we have thrown AMD’s newly minted Brazos platform into the ring against Intel’s Pine Trail (the latest Atom platform) and forced the two of them to fight it out against a backdrop that includes the full range of today’s desktop processors. We think it’s an intriguing exercise. Of course, we’ve also taken time to investigate the question of adequacy for basic computing tasks, which will be critical success of these low-cost, lower-power platforms, regardless of how they match up to the big boys.

Brazos busts out

Let’s back up a second, for those who aren’t yet fully acquainted with AMD’s Brazos platform, and do some quick introductions. We’ve already previewed Brazos and taken a preliminary look at its performance, in case you missed those articles. In sum, Brazos is a low-cost, low-power, PC-compatible computing platform that aimed at a number of places where traditional PC processors won’t always easily fit: netbooks and ultraportables, compact home theatre PCs, ultra-slim desktops, embedded systems, and perhaps even tablets. You’ll grasp the basic concept behind Brazos if you compare it to the most common variants of Intel’s Atom, although AMD has promised superior integrated graphics and video decode acceleration for Brazos—key pain points for a lot of Atom-based systems.

The key pain point for Brazos, by contrast, may be all of the code names, brands, and acronyms being thrown about in reference to it. Perhaps we’re just a little dull after too many late nights with Infinity Blade lately, but it all seems rather bewildering to us. To help you keep track, here’s a quick Brazos-related decoder ring.

  • Bobcat — This code name refers to the architecture of the CPU cores that drive the main chip in the Brazos platform. Bobcat is an all-new, clean-sheet CPU architecture tailored for power-efficient operation. Bobcat supports many of the modern extensions to the x86 instruction set, including AMD64 and Supplemental SSE3. Unlike the Atom, the Bobcat core is capable of out-of-order instruction execution, potentially giving it the ability to execute and retire more instructions per clock.
  • Ontario — The code name given to the first chip to incorporate a Bobcat CPU core. Ontario has two Bobcat cores, DirectX 11-class Radeon integrated graphics, and an onboard memory controller. Ontario chips are rated to fit into a 9W thermal envelope.
  • Zacate — Somewhere along the way, AMD decided it needed a second code name to describe the variants of Ontario that will fit into an 18W power envelope, and Zacate was born. The bifurcation of marketing code names was likely deemed necessary because the 9W parts and the 18W parts will serve different markets. The Brazos system we’re reviewing today is Zacate-based.

  • Logical block diagram of the Brazos platform. Source: AMD.

  • Hudson — This is the I/O support chip that, together with Ontario/Zacate, forms the basis of the Brazos platform. Hudson handles most of the traditional duties of a core-logic chipset, although like many “chipsets” these days, it’s just a single chip. AMD calls Hudson an FCH for “Fusion controller hub,” a variation on the more widely used “I/O controller hub” or Intel’s “platform controller hub.”
  • Fusion — Back when AMD bought ATI and brought Radeon graphics into the fold, the top executives involved talked a lot about how CPUs and GPUs were on a collision course. They projected a future where the responsibility for the execution of programs could be shared between different types of resources. Branchy, sequential threads could run on traditional CPU cores, while problems ripe for parallelization could run on arrays of FLOPS-intensive execution units—in other words, highly evolved GPUs.

    Such talk got us computing enthusiasts all hot and bothered, so AMD decided to bottle the lightning. Soon, “fusion” came to be applied to a very different sort of animal, a product that incorporates very traditional CPU and graphics cores on the same chip for the sake of cost and power savings. This concept was much less exciting, but the fusion name granted the whole enterprise a halo of futuristic awesomeness.

    Finally, with the introduction of Ontario/Zacate, Fusion has become an official AMD brand name encompassing both the Brazos platform and future products based on an upcoming, much larger, Phenom-derived chip code-named Llano. Neither of them has any of the hallmarks of the fully realized fusion concept, such as a programming model with shared memory (or even shared cache) between the CPU and GPU cores.

  • APU — Stands for “accelerated processing unit,” AMD’s term to describe its CPUs with integrated graphics. Ontario/Zacate is billed as “the world’s first APU.” Intel calls its similar integrated chips CPUs, and given that the programming model hasn’t changed for either, we’re betting this new term won’t stick.
  • Vision — AMD’s all-encompassing brand for its PC platform solutions. Near as we can tell, Vision is AMD’s new flagship brand, supplanting Radeon and Phenom in importance, judged by the placement of Vision stickers on laptops and desktops.
  • Radeon HD 6310 — The name given to the integrated graphics processor (IGP) in Zacate. The lower-clocked version of this same silicon in Ontario has been dubbed the Radeon HD 6250.
  • Uh, none, really? — The official brand name of the Zacate/Ontario APUs. The Zacate model we’re reviewing today is known as the “AMD E-350 APU with Radeon HD 6310 graphics.” There’s no snappy name like Phenom or Athlon involved, just some alphanumeric designators. The E-series APUs are based on Zacate, while the 9W Ontario APUs are in the C series. The table below lays it all out for you.
Model CPU

cores

CPU

clock

speed

Radeon

graphics

IGP

clock

speed

TDP
E-350 2 1.6 GHz HD 6310 500 MHz 18 W
E-240 1 1.5 GHz HD 6310 500 MHz 18 W
C-50 2 1.0 GHz HD 6250 280 MHz 9 W
C-30 1 1.2 GHz HD 6250 280 MHz 9 W

Phew. So yeah, those are the Zacate models, and the handy list above will help you navigate the minefield of Brazos platform terminology for days and possibly even weeks before it all changes again.

The Hudson FCH (left) and Zacate APU (right)

Now that we’ve defined our terms, we can talk in a little more detail about the Zacate APU. The chip itself, shown on the right above, is a very compact 75 mm². It’s manufactured at TSMC on the same 40-nanometer fabrication process used to produce the latest Radeon and GeForce GPUs. Each of its two Bobcat-based cores has 32KB L1 data and instruction caches, along with 512KB of L2 cache.

The built-in Radeon graphics processor looks to be very similar to the discrete “Cedar” chip used in the Radeon HD 5450. In other words, it’s based on the same fundamental technology as other Radeon HD 5000-series GPUs, only scaled down to fit this application. The IGP has two SIMD engines, each with eight execution units that process five-wide vectors. That works out to 40 ALUs per SIMD and 80 ALUs total for the IGP overall. The texture units aligned with the SIMDs can sample and bilinearly filter up to eight texels per clock cycle, and the IGP’s lone ROP unit is capable of outputting four pixels per clock. At 500MHz on a Zacate APU, those specs don’t work out to a tremendous amount of graphics processing power. Because the IGP has to share a single DDR3 1066MHz memory channel with the two CPU cores, it probably won’t be as fast as a discrete Radeon HD 5450. Still, an integrated GPU anywhere in that neighborhood is a huge step up from what Intel has been fielding in its comparable Atom solutions.

Perhaps even more important than the graphics horsepower on offer is the full DirectX 11 feature set—this thing can do tessellation and high-precision texture formats, for gosh sakes—and the related fact that the IGP will be supported by AMD’s Catalyst video drivers. Even if the IGP is a little puny for some tasks, its basic compatibility should be quite good.

Aside from graphics, the IGP packs one of Zacate’s most notable features: the UVD3 video processing block, the same third-gen unit present in the latest Radeon HD 6000-series GPUs. Aboard a low-power system like this one, this logic’s ability to offload H.264 decoding chores from the CPU is absolutely critical to a good user experience, especially since the web is now so saturated with H.264-encoded videos. With proper support for UVD3, applications like Windows Media Center and Adobe Flash 10.2 can help Zacate play heavily compressed HD videos fluidly—something this class of CPU, even in dual-core form, may struggle to do. Witness the droves of Atom-based slim desktops and netbooks whose frustration factor has skyrocketed with YouTube’s rise. Dedicated video processing hardware should be more power-efficient than CPU-only software routines, as well.

Zacate also includes the necessary bits of glue to hold everything together, such as an internal bus, a memory controller, a four-lane PCI Express interface, and an interconnect to the Hudson I/O chip. All told, that’s an awful lot of capability packed into one little chip.

MSI’s Mini-ITX monster

Pictured above is the first Brazos-based motherboard to make its way into Damage Labs, MSI’s E350IA-E45. This thing is bristling with connectivity options, befitting Brazos’ mission as a compact but competent version of the do-everything PC platform. Like many Brazos boards, this one fits the 17-cm-by-17-cm dimensions of the Mini-ITX form factor.

Also like all Brazos-based mobos, the APU CPU processor is soldered in place, with a heatsink/fan combo already mounted. If you’re like me, you’ll take one look at that tiny fan and get flashbacks to whiny chipset fans of old. Nearly every small fan like that, no matter how quiet it starts out, eventually ends up rattling and buzzing like Charlie Sheen halfway into a briefcase. MSI tells us the cooler on our pre-release sample differs from the final one, which will be larger and bluer than what you see above. In fact, we got a look at what appears to be the production cooler at CES:

Happily, that’s a nicer looking deal. Less happily, it still includes a teeny little fan. We did find that MSI had implemented some very decent fan speed controls in this board’s BIOS, and the fan on our cooler was pretty quiet with the right settings enabled. Still, we can’t help but worry about how it will sound over time.

There are many possible applications for a Zacate-based mobo like this one, but when we lay eyes on that HDMI output port, the eight-channel analog outputs, and two flavors of SPDIF ports, we think about the potential to build a small, quiet home theater PC around it. Don’t expect anything too special on the analog audio front, though. The board has a Realtek ALC887 audio codec onboard. They’re cheap and competent, but you’d want to use one of the digital audio outputs for a home theater system.

Notice the one PCIe x16 slot on the board, suggesting the possibility of hosting a TV tuner or even—crucially for the evil experiment we’re about to conduct—a discrete graphics card. Although that’s a real x16 slot physically, it has only four lanes connected electrically, the maximum Zacate allows.

We should pause to note that this is not an enthusiast-class motherboard in the usual sense. The board has a UEFI BIOS with a very traditional (and quite snappy) text-based menu interface on top, and there is an “overclocking” section present. However, the only setting that might be construed as overclocking is the ability to run the DDR3 memory at 1333MHz, higher than AMD’s officially supported max of 1066MHz. The rest of the settings control things like DRAM timings and enabling Cool’n’Quiet. There’s no option for raising the CPU clock speed at all. Before you think that’s horribly weak sauce, consider for a moment that many Atom boards in this class don’t even offer DRAM timing controls, let alone a temperature-based threshold for the CPU fan speed like MSI provides here. This board won’t win any medals for tweakability, but MSI has acquitted itself rather well overall, in our view.

MSI plans to offer two models of this board. The one with the big, blue heatsink and USB 3.0 support should cost about $139.99 at online vendors, while a lower-end version with a smaller cooler and no USB 3 will go for 10 bucks less. Although we’ve not yet seen either one listed at Newegg, MSI tells us they should be showing up for sale any day now.

We were a little surprised when we first got those prices from MSI. AMD had told us to expect Brazos boards to start at “somewhere under $100.” Clearly, this is more of a premium product than the cheaper options apparently also coming. Yes, we’re talking about a motherboard, CPU, and cooler combo, but you can pay less for a combo based on a low-end desktop processor, like this Biostar mobo, AMD Sempron, and cooler combo for 80 bucks. You’ll also find older dual-core Atom boards like this one for $75. Obviously, at $140, you’re paying a premium for miniaturization, for low power draw, for Brazos’ expanded graphics and video capabilities versus Atom, and perhaps for protection from rogue elements of society. Systems in this class are meant to be dirt cheap, like a jumbo pack of Twinkies from Costco. Feels like we’re initially being asked to pay Whole Foods Organic Cream-Filled Cakes prices here, to me. Perhaps these boards will settle closer to $100 over time.

Atom takes the Pine Trail

As you may have gathered, Brazos’ most natural competitor is Intel’s Atom processor, specifically the version aimed at netbooks and compact desktop systems (which Intel calls “nettops.”) The latest Atom platform in that category is code-named Pine Trail, and the CPU at the center of it was developed under the name Pineview, just to keep the pine-scented confusion swirling. The Pineview Atom is remarkably similar to Zacate, incorporating one or two low-power CPU cores, an IGP, a memory controller, and an interconnect to a separate I/O chip. (We first covered the platform in some depth here.) At 87 mm² and 66 mm², respectively, the dual- and single-core variants of Pineview bracket the 75 mm² Zacate in size.

The NM10 PCH (left) and a dual-core Pineview CPU (right)

Pineview’s CPU cores use a simpler in-order execution scheme, but unlike Bobcat, each core can track and execute two threads simultaneously. The Atom’s architects chose symmetric multithreading rather than out-of-order execution as a means of achieving higher per-clock performance, and so far, that’s proven to be a pretty good bet. SMT, better known at Hyper-Threading in Intel’s marketing lexicon, is most helpful at some of the toughest points of everyday PC use, such as periods of heavy multitasking or when dealing with media playback and manipulation.

The integrated GMA 3150 graphics core looks to be less formidable, based as it is on a rather old Intel chipset IGP. Raw graphics power is one of its weaknesses, but more critically, the Pineview IGP supports only the most rudimentary version of DirectX 9, known as Shader Model 2.0. Even there, the IGP lacks support for some texture formats available in the first DX9/SM2.0 GPUs like the Radeon 9700. Thus, even if Intel’s graphics drivers were a model of quality and compatibility, the GMA 3150 would fall short of the minimum requirements for great many games.

That’s a shame, but the worst part of the story, as we’ve noted, is that fact that Pineview’s video processing logic can’t accelerate the decoding of H.264 video, so that burden will fall entirely on the Atom CPU.

One other difference between Zacate and Pineview is the memory controllers. Although both support a single DRAM channel, Pineview’s support is limited to DDR2 memory at a peak of 800MHz. Performance-wise, that’s not likely to amount to much of a difference versus Zacate’s 1066MHz DDR3 channel, but it may mean more on the power front due to DDR3’s lower operating voltages, particularly because Zacate supports low-voltage variants of DDR3 memory.

While searching through the Newegg listings for an appropriate product to compare against the MSI Brazos board, we came across this unassuming offering from Jetway for $130. This board has several things we wanted. It’s based on the Pine Trail platform with an Atom D525 processor, the fastest desktop/nettop version of Pineview. The D525’s two CPU cores run at 1.8GHz, and the whole chip has a TDP of just 13W, a little lower than Zacate’s 18W. Like the MSI Brazos board, the Jetway has a PCIe x16 slot, raising the threat of a discrete graphics card. Although many folks would likely cite Nvidia’s second-gen Ion platform—essentially Pine Trail plus a small discrete GPU—as Brazos’ most appropriate foil, we wanted to see how AMD’s latest stacks up directly against Intel’s total platform solution, which is more widely used than Ion. Of course, this Pine Trail setup can easily be augmented with any discrete GPU we might choose to drop into its PCIe x16 slot, as well.

Finding a Mini-ITX board with an Atom D525 and a PCIe x16 slot isn’t easy, and we had to make a few compromises. We’re not so pleased with the Jetway’s lack of digital audio and video outputs, for instance. Still, the totally passive cooling on the board is nifty, and the expansion slot allows us to overcome both of those output shortcomings via a single HDMI port.

The final implements of destruction

Our nefarious plan to compare these low-power platforms to one another and, scandalously, to standard desktop systems requires a few more tools.

One of those is a low-end discrete graphics card, with which we can test AMD’s claim that Zacate includes “discrete level” graphics. Pictured above is a Radeon HD 5450, shamelessly stuffed into the PCIe slot on our MSI Brazos board. Although a great many Brazos systems will no doubt use the IGP, this card will allow us a nice basis of comparsion for the graphics component alone. We can also see how the Atom D525 might fare if it were granted the assistance of a Radeon GPU.

More questionably, we populated both boards’ PCIe slots with a more powerful graphics card—a GeForce GTX 460 1GB—for the portion of our comparison that includes full-scale desktop processors. Doing so meant all of the CPUs involved would be working with the same GPU, although we should point out that the Atom and Brazos systems have only four lanes connected to their PCIe x16 slots. We’ll probably lose our license over this.

Both of our Mini-ITX contestants proved to be perfectly happy when connected to the 610W Silencer PSU that’s standard issue for our CPU test rigs, but we couldn’t really address the question of power efficiency when using a power supply intended to deliver well over ten times the juice that our itty-bitty integrated systems require at peak. That’s why we ordered up the power supply you see pictured above, the nMediaPC HTPC 1080iP. This puppy has a max power rating of 85W, supplied by a completely passive, laptop-style power brick that plugs into a break-out PCB with standard ATX-style connectors.

Using it with the passively cooled Jetway Atom board, the Radeon HD 5450 shown above, and a Corsair SSD produced some freaky, “Is it on?” moments, because there was virtually no sound whatsoever emanating from the computer as it POSTed, booted, and ran programs. We achieved a similar effect, at least part of the time, by setting the MSI board’s minimum CPU fan speed to 0% and giving it a relatively high 65° C target temperature.

Our testing methods

As ever, we did our best to deliver clean benchmark numbers. Tests were run at least three times, and we reported the median of the scores produced.

Our test systems were configured like so:

Processor
Athlon II X3 455 3.3GHz

Phenom II X2 565 3.4GHz

Phenom II X4 840 3.2GHz

Phenom II X4 975 3.6GHz

Phenom II X4 1075T 3.0GHz

Phenom II X4 1100T 3.3GHz

Pentium
Extreme Edition 840 3.2GHz
Pentium
G6950 2.8GHz
Core
2 Duo E6400 2.13GHz
Core
i3-560 3.33 GHz

Core i5-655K 3.2GHz

Core i5-760 2.8GHz

Core i7-875K 2.93GHz

Core
2 Quad Q9400 2.67GHz
Motherboard Gigabyte
890GPA-UD3H
Asus
P5E3 Premium
Asus
P7P55D-E Pro
North bridge 890GX X48 P55
South bridge SB850 ICH9R
Memory size 8GB
(4 DIMMs)
8GB
(4 DIMMs)
8GB
(4 DIMMs)
Memory type Corsair

CMD8GX3M4A1333C7

DDR3 SDRAM

Corsair

CMD8GX3M4A1600C8

DDR3 SDRAM

Corsair

CMD8GX3M4A1600C8

DDR3 SDRAM

Memory speed 1333 MHz 800
MHz
1066 MHz
1066 MHz
1333 MHz
1333 MHz
Memory timings 8-8-8-20 2T 7-7-7-20 2T 7-7-7-20 2T
7-7-7-20 2T
8-8-8-20 2T
8-8-8-20 2T
Chipset

drivers

AMD
AHCI 1.2.1.263
INF
update 9.1.1.1025

Rapid Storage Technology 9.6.0.1014

INF
update 9.1.1.1025

Rapid Storage Technology 9.6.0.1014

Audio Integrated

SB850/ALC892 with

Realtek 6.0.1.6235 drivers

Integrated

ICH9R/AD1988B with 

Microsoft drivers

Integrated

P55/RTL8111B with

Realtek 6.0.1.6235 drivers

Processor Core
i7-950 3.06 GHz

Core i7-970 3.2 GHz

Core i7-980X Extreme 3.3 GHz

Core
i3-2100 2.93 GHz

Core i5-2400 3.1 GHz

Core i5-2500K 3.3 GHz

Core i7-2600K 3.4 GHz

Atom
D525 1.8 GHz
AMD
E-350 1.6GHz
Motherboard Gigabyte
X58A-UD5
Asus
P8P67 Deluxe
Jetway
NC94FL-525-LF
MSI
E350IA-E45
North bridge X58 P67 NM10 Hudson
M1
South bridge ICH10R
Memory size 12GB
(6 DIMMs)
8GB
(4 DIMMs)
4GB (2 DIMMs) 4GB (2 DIMMs)
Memory type Corsair

CMP12GX3M6A1600C8

DDR3 SDRAM

Corsair

CMD8GX3M4A1600C8

DDR3 SDRAM

Corsair

CM2X2048-8500C5D

DDR2 SDRAM

Corsair

CMD8GX3M4A1333C7

DDR3 SDRAM

Memory speed 1333 MHz 1333 MHz 800
MHz
1066 MHz
Memory timings 8-8-8-20 2T 8-8-8-20 2T 5-5-5-18
2T
7-7-7-20 2T
Chipset

drivers

INF update 9.1.1.1020

Rapid Storage Technology 9.5.0.1037

INF update
9.2.0.1016

Rapid Storage Technology 10.0.0.1046

INF update 9.1.1.1020

Rapid Storage Technology 9.5.0.1037

AMD
AHCI 1.2.1.275
Audio Integrated

ICH10R/ALC889 with

Realtek 6.0.1.6235 drivers

Integrated

P67/ALC889 with

Microsoft drivers

Integrated

NM10/ALC662 with

Realtek 6.0.1.6235 drivers

Integrated

Hudson M1/ALC88 with

Realtek 6.0.1.6235 drivers

They all shared the following common elements:

Hard drive Corsair
Nova V128 SATA SSD
Discrete graphics Asus
ENGTX460 TOP 1GB (GeForce GTX 460) with ForceWare 260.99 drivers
OS Windows 7 Ultimate x64 Edition
Power supply PC Power & Cooling Silencer 610 Watt

Our test systems for integrated graphics looked a little bit different. They were configured like this:

Processor
Phenom II X4 1075T 3.0GHz
Core
i3-2100 2.93 GHz

Core i5-2500K 3.3 GHz


Core i5-655K 3.2GHz
Atom
D525 1.8 GHz
AMD
E-350 1.6GHz
Motherboard Gigabyte
890GPA-UD3H
Intel
DH67BL
Gigabyte
H57M-USB3
Jetway
NC94FL-525-LF
MSI
E350IA-E45
North bridge 890GX H67 H57 NM10 Hudson
M1
South bridge SB850
Memory size 8GB
(4 DIMMs)
8GB (4 DIMMs) 8GB
(4 DIMMs)
4GB (2 DIMMs) 4GB (2 DIMMs)
Memory type Corsair

CMD8GX3M4A1333C7

DDR3 SDRAM

Corsair

CMD8GX3M4A1600C8

DDR3 SDRAM

Corsair

CMD8GX3M4A1600C8

DDR3 SDRAM

Corsair

CM2X2048-8500C5D

DDR2 SDRAM

Corsair

CMD8GX3M4A1333C7

DDR3 SDRAM

Memory speed 1333 MHz 1333 MHz 1333 MHz 800
MHz
1066 MHz
Memory timings 8-8-8-20 2T 9-9-9-24
2T
8-8-8-20 2T 5-5-5-18
2T
7-7-7-20 2T
Chipset

drivers

AMD
AHCI 1.2.1.263
INF update
9.2.0.1016

Rapid Storage Technology 10.0.0.1046

INF
update 9.1.1.1025

Rapid Storage Technology 9.6.0.1014

INF update 9.1.1.1020

Rapid Storage Technology 9.5.0.1037

AMD
AHCI 1.2.1.275
Audio Integrated

SB850/ALC892 with

Realtek 6.0.1.6235 drivers

Integrated

P67/ALC892 with

Microsoft drivers

Integrated

P55/ALC889 with

Realtek 6.0.1.6235 drivers

Integrated

NM10/ALC662 with

Realtek 6.0.1.6235 drivers

Integrated

Hudson M1/ALC88 with

Realtek 6.0.1.6235 drivers

Graphics Integrated

Radeon HD 4290 with

Catalyst 10.12 drivers

Integrated

Intel HD Graphics with

8.15.10.2266 drivers

Integrated

Intel HD Graphics with

8.15.10.2246 drivers

Integrated

Intel  GMA 3150 with

8.14.10.2230 drivers

Integrated

Radeon HD 6310 with

Catalyst 11.1 drivers

They shared the following common elements:

Hard drive Corsair
Nova V128 SATA SSD
OS Windows 7 Ultimate x64 Edition
Power supply PC Power & Cooling Silencer 610 Watt

We’d like to thank Asus, Corsair, Gigabyte, and OCZ for helping to outfit our test rigs with some of the finest hardware available. Thanks to Intel and AMD for providing the processors, as well, of course.

The test systems’ Windows desktops were set at 1900×1200 in 32-bit color. Vertical refresh sync (vsync) was disabled in the graphics driver control panel.

We used the following versions of our test applications:

The tests and methods we employ are usually publicly available and reproducible. If you have questions about our methods, hit our forums to talk with us about them.

Power consumption and efficiency

We used a Yokogawa WT210 digital power meter to capture power use over a span of time. The meter reads power use at the wall socket, so it incorporates power use from the entire system—the CPU, motherboard, memory, graphics solution, hard drives, and anything else plugged into the power supply unit. (The monitor was plugged into a separate outlet.) We measured how each of our test systems used power across a set time period, during which time we ran Cinebench’s multithreaded rendering test.

We’ll start with the show-your-work stuff, plots of the raw power consumption readings. We’ve broken things down by socket type in order to keep them manageable.

For space reasons, we’ve only included the most relevant portion of the CPUs we tested in the plot below. For the full slate of power-over-time plots, see our Sandy Bridge review. Also, you’ll see that we have two sets of results for the Atom- and E-350-based systems. The primary set comes from a configuration with a discrete GTX 460 graphics card and our 610W standard PSU connected. The second set was captured with only integrated graphics in use and our laptop-style power brick supplying the juice.

This plot gives us our first indication of the E-350 APU’s power draw and performance both. As you can see, the Atom and the E-350 setups use fewer watts under load than any of the conventional desktop CPU-based systems do at idle. Switching our PSUs and graphics solutions drops power consumption ever further, well below 30W for both the Atom and Zacate systems. The E-350 and the Atom D525 track very closely together in power draw overall.

As for performance, we’ve had to extend our time window for completion of the rendering task substantially in order to accommodate the low-power processors. Even the Core 2 Duo E6400 takes roughly half the time to finish the job. The E-350 finishes ahead of the Atom D525, but not by a lot.

We can slice up these raw data in various ways in order to better understand them. We’ll start with a look at idle power, taken from the trailing edge of our test period, after each CPU has completed the render. Next, we can look at peak power draw by taking an average from the ten-second span from 15 to 25 seconds into our test period, when the processors were rendering.

Here’s a clearer view of idle and peak power consumption. Check out the E-350 system pulling only 14W at idle on the brick PSU. Whoa. The E-350 is a pretty close match to the Atom D525 under load, which is no bad thing. Amusingly, the Pentium EE 840 system draws nearly ten times as much power under load as the E-350 one does. Atom and Zacate are unquestionably in a different weight class than the desktop CPUs, a fact we’ll want to remember so we can keep the forthcoming performance results in perspective.

We can pinpoint efficiency more effectively by considering the amount of energy used for the task alone. Since the different systems completed the render at different speeds, we’ve isolated the render period for each system. We’ve then computed the amount of energy used by each system to render the scene. This method should account for both power use and, to some degree, performance, because shorter render times may lead to less energy consumption.

If you think of energy efficiency in terms of power draw at the wall socket alone, the E-350 and Atom are our obvious champions. However, once you consider more targeted measures like the energy required to complete a task, today’s regular desktop processors can be surprisingly efficient. The long completion times of the Atom and Zacate systems put them at a disadvantage. Nevertheless, the E-350 on the brick PSU claws its way to the middle of the pack, several notches above the Atom D525 system.

Memory subsystem performance

These results don’t always track directly with real-world performance, but they do give us some insights into the CPU and system architectures involved. For this first test, the graph is pretty crowded. We’ve tried to be selective, only choosing a subset of the CPUs tested. This test is multithreaded, so more cores—with associated L1 and L2 caches—can lead to higher throughput.

These results nicely serve to orient our expectations for the Zacate-versus-Atom showdown ahead by illustrating how closely matched the two are in terms of cache size and speed. This test does benefit from the presence of additional cores (and thus caches), but even an older dual-core CPU like the Pentium Extreme Edition 840—essentially two Pentium 4s glued together—roughly doubles the cache bandwidth of the little guys.

The E-350’s single channel of DDR3 memory at 1066MHz doesn’t confer it any noticeable advantage over the Atom’s 800MHz DDR2 channel. Some of the older desktop CPUs saddled with a front-side bus bottleneck achieve comparable bandwidth or latency, but only the Pentium EE 840 has a combination that’s arguably less attractive. Of course, the fact that Atom and Zacate are only driving a single memory channel contributes to their miserly platform power draw.

Productivity

SunSpider JavaScript performance

Here’s our first look at application performance, courtesy of this JavaScript browser benchmark, and it turns out the foreshadowing we saw on the last couple of pages wasn’t a fluke. Even today’s slowest $87 desktop CPUs, the Athlon X3 455 and the Pentium G6950, very much outperform our two lightweights.

There are two sides of this unfolding performance story. The E-350 APU is undoubtedly in the same overall performance class as the Atom D525—it’s not delivering anything close to what we’d consider “90% of today’s mainstream performance” like AMD had led us to expect—but it is about 30% faster than the Atom.

7-Zip file compression and decompression

7-Zip is nicely multithreaded and allows the Atom to demonstrate how Hyper-Threading can sometimes grant it higher performance than Zacate’s out-of-order execution scheme. Although this looks like another case of domination by the desktop CPUs, we’re actually somewhat impressed by the decompression results, where the little guys deliver a high proportion of the throughput of the Conroe-based Core 2 Duo E6400, a pretty potent CPU in its day.

TrueCrypt disk encryption

This full-disk encryption suite includes a performance test, for obvious reasons. We tested with a 500MB buffer size and, because the benchmark spits out a lot of data, averaged and summarized the results.

TrueCrypt has added support for Intel’s custom-tailored AES-NI instructions since we last visited it, so the encoding of the AES algorithm, in particular, should be very fast on the Intel CPUs that support those instructions. Those CPUs include the six-core Gulftowns, the dual-core Clarkdales, and Sandy Bridge.

The big boys at the top of the charts here are helped mightily by their support hardware AES acceleration, which makes this something less than a fair fight. The take home message for our Mini-ITX contenders is a little more practical than raw comparative numbers, though. At 40-45 MB/s, their throughput isn’t sufficient to sustain real-time disk encryption without slowing down the storage subsystem. Even slower mobile hard drives can sustain transfers at nearly twice those rates.

Image processing

The Panorama Factory photo stitching
The Panorama Factory handles an increasingly popular image processing task: joining together multiple images to create a wide-aspect panorama. This task can require lots of memory and can be computationally intensive, so The Panorama Factory comes in a 64-bit version that’s widely multithreaded. I asked it to join four pictures, each eight megapixels, into a glorious panorama of the interior of Damage Labs.

In the past, we’ve added up the time taken by all of the different elements of the panorama creation wizard and reported that number, along with detailed results for each operation. However, doing so is incredibly data-input-intensive, and the process tends to be dominated by a single, long operation: the stitch. Thus, we’ve simply decided to report the stitch time, which saves us a lot of work and still gets at the heart of the matter.

Because it measures completion time in seconds, this test gives us an easily accessible sense of how going with a lightweight CPU could affect the user experience. You’d be waiting roughly a minute longer for the panorama to be stitched together on an E-350 than on an Athlon II X3 455. With an Atom D525, you’d be waiting another 19 seconds beyond that.

picCOLOR image processing and analysis

picCOLOR was created by Dr. Reinert H. G. Müller of the FIBUS Institute. This isn’t Photoshop; picCOLOR’s image analysis capabilities can be used for scientific applications like particle flow analysis. Dr. Müller has supplied us with new revisions of his program for some time now, all the while optimizing picCOLOR for new advances in CPU technology, including SSE extensions, multiple cores, and Hyper-Threading. Many of its individual functions are multithreaded.

At our request, Dr. Müller graciously agreed to re-tool his picCOLOR benchmark to incorporate some real-world usage scenarios. Thus, we now have four tests that employ picCOLOR for image analysis. We’ve summarized the results of those tests below, alongside the results from picCOLOR’s synthetic tests of image processing functions.

The baseline for all of these results is a Pentium III 1GHz, interestingly enough. That chip would score 1.0 on these tests, so the E-350 APU is 3-4 times as fast as the PIII. Then again, the E-350 is only about half as quick as the Core 2 Duo E6400. The good news for AMD’s new APU? It’s again a little quicker than the Atom D525.

Video encoding

x264 HD benchmark

This benchmark tests one of the most popular H.264 video encoders, the open-source x264. The results come in two parts, for the two passes the encoder makes through the video file. I’ve chosen to report them separately, since that’s typically how the results are reported in the public database of results for this benchmark.

These low-power processors tend to struggle when asked to decode an H.264 video, so we’re not surprised to see them churning out low frame rates when encoding a video into the same format. The parity between the Atom D525 and the AMD E-350 is striking, though.

Windows Live Movie Maker 14 video encoding

For this test, we used Windows Live Movie Maker to transcode a 30-minute TV show, recorded in 720p .wtv format on my Windows 7 Media Center system, into a 320×240 WMV-format video format appropriate for mobile devices.

The $87 Pentium G6950 transcodes this video in under 10 minutes, while the two lightweight are left cranking away for nearly half an hour. The E-350 finishes just shy of two minutes before the Atom D525 does, but neither one is well suited to this sort of media manipulation.

3D modeling and rendering

Cinebench rendering

The Cinebench benchmark is based on Maxon’s Cinema 4D rendering engine. It’s multithreaded and comes with a 64-bit executable. This test runs with just a single thread and then with as many threads as CPU cores (or threads, in CPUs with multiple hardware threads per core) are available.

POV-Ray rendering

We’re using the latest beta version of POV-Ray 3.7 that includes native multithreading and 64-bit support.

Valve VRAD map compilation

This next test processes a map from Half-Life 2 using Valve’s VRAD lighting tool. Valve uses VRAD to pre-compute lighting that goes into games like Half-Life 2.

These CPUs aren’t intended for this sort of work, as they obligingly demonstrate for us.

Scientific computing

MyriMatch proteomics

Our benchmarks sometimes come from unexpected places, and such is the case with this one. David Tabb is a friend of mine from high school and a long-time TR reader. He has provided us with an intriguing new benchmark based on an application he’s developed for use in his research work. The application is called MyriMatch, and it’s intended for use in proteomics, or the large-scale study of protein. I’ll stop right here and let him explain what MyriMatch does:

In shotgun proteomics, researchers digest complex mixtures of proteins into peptides, separate them by liquid chromatography, and analyze them by tandem mass spectrometers. This creates data sets containing tens of thousands of spectra that can be identified to peptide sequences drawn from the known genomes for most lab organisms. The first software for this purpose was Sequest, created by John Yates and Jimmy Eng at the University of Washington. Recently, David Tabb and Matthew Chambers at Vanderbilt University developed MyriMatch, an algorithm that can exploit multiple cores and multiple computers for this matching. Source code and binaries of MyriMatch are publicly available.
In this test, 5555 tandem mass spectra from a Thermo LTQ mass spectrometer are identified to peptides generated from the 6714 proteins of S. cerevisiae (baker’s yeast). The data set was provided by Andy Link at Vanderbilt University. The FASTA protein sequence database was provided by the Saccharomyces Genome Database.

MyriMatch uses threading to accelerate the handling of protein sequences. The database (read into memory) is separated into a number of jobs, typically the number of threads multiplied by 10. If four threads are used in the above database, for example, each job consists of 168 protein sequences (1/40th of the database). When a thread finishes handling all proteins in the current job, it accepts another job from the queue. This technique is intended to minimize synchronization overhead between threads and minimize CPU idle time.

The most important news for us is that MyriMatch is a widely multithreaded real-world application that we can use with a relevant data set. I should mention that performance scaling in MyriMatch tends to be limited by several factors, including memory bandwidth, as David explains:

Inefficiencies in scaling occur from a variety of sources. First, each thread is comparing to a common collection of tandem mass spectra in memory. Although most peptides will be compared to different spectra within the collection, sometimes multiple threads attempt to compare to the same spectra simultaneously, necessitating a mutex mechanism for each spectrum. Second, the number of spectra in memory far exceeds the capacity of processor caches, and so the memory controller gets a fair workout during execution.

Here’s how the processors performed.

STARS Euler3d computational fluid dynamics

Charles O’Neill works in the Computational Aeroservoelasticity Laboratory at Oklahoma State University, and he contacted us to suggest we try the computational fluid dynamics (CFD) benchmark based on the STARS Euler3D structural analysis routines developed at CASELab. This benchmark has been available to the public for some time in single-threaded form, but Charles was kind enough to put together a multithreaded version of the benchmark for us with a larger data set. He has also put a web page online with a downloadable version of the multithreaded benchmark, a description, and some results here.

In this test, the application is basically doing analysis of airflow over an aircraft wing. I will step out of the way and let Charles explain the rest:

The benchmark testcase is the AGARD 445.6 aeroelastic test wing. The wing uses a NACA 65A004 airfoil section and has a panel aspect ratio of 1.65, taper ratio of 0.66, and a quarter-chord sweep angle of 45º. This AGARD wing was tested at the NASA Langley Research Center in the 16-foot Transonic Dynamics Tunnel and is a standard aeroelastic test case used for validation of unsteady, compressible CFD codes.
The CFD grid contains 1.23 million tetrahedral elements and 223 thousand nodes . . . . The benchmark executable advances the Mach 0.50 AGARD flow solution. A benchmark score is reported as a CFD cycle frequency in Hertz.

So the higher the score, the faster the computer. Charles tells me these CFD solvers are very floating-point intensive, but they’re oftentimes limited primarily by memory bandwidth. He has modified the benchmark for us in order to enable control over the number of threads used. Here’s how our contenders handled the test with optimal thread counts for each processor.

If you have scientific computing apps to run, you might want to consider stepping up to at least an $87 CPU. Just sayin’.

StarCraft II

We tested StarCraft II by playing back a recording of an epic 30-minute, eight-player match that we found online and capturing frame rates with Fraps. Thanks to the relatively long time window involved, we decided not to repeat this test multiple times, like we usually do when testing games in this fashion.

Interesting. Even with the aid of a GeForce GTX 460 (albeit via only four PCIe lanes), the E-350 and Atom D525 can’t handle SC2 at these detail levels. We have several crackpot theories about why, involving things like the number of PCIe lanes going to the graphics card and the CPU overhead required just to drive a fast GPU. Still, we can’t avoid the obvious interpretation. The Pentium EE 840 also struggles in this game, and it has a full 16 PCIe lanes available to it. The Atom and E-350 are typically slower than the Pentium EE 840. Ergo, ouch.

Battlefield: Bad Company 2

Most of our performance tests are scripted and repeatable, but Battlefield: Bad Company 2, we used the Fraps utility to record frame rates while playing a 60-second sequence from the game. Although capturing frame rates while playing isn’t precisely repeatable, we tried to make each run as similar as possible to all of the others. We raised our sample size, testing each Fraps sequence five times per CPU in order to counteract any variability. We’ve included second-by-second frame rate results from Fraps, and in that case, you’re seeing the results from a single, representative pass through the test sequence.

These results follow the same cruel logic the StarCraft II ones did, only frame rates in this game suffer even more on the lower-end desktop CPUs, such as the Pentium G6950, and on the older processors like the E6400. The Atom D525 is faster than the E-350, but neither is anywhere close to competent here.

Civilization V

The developers of Civ V have cooked up a number of interesting benchmarks, two of which we used here. The first one tests a late-game scenario where the map is richly populated and there’s lots happening at once. As you can see by the setting screen below, we didn’t skimp on our the image quality settings for graphics, either. Doing so wasn’t necessary to tease out clear differences between the CPUs.

Civ V also runs the same test without updating the screen, so we can eliminate any overhead or bottlenecks introduced by the video card and its driver software.

None of the results are encouraging, but the “no render” tests mean there’s no hiding from it. Even without the burden of driving the GPU, our two lightweight processors aren’t up to running this game smoothly, as frame rates in the teens and lower indicate unambiguously.

F1 2010

CodeMasters has done a nice job of building benchmarks into its recent games, and F1 2010 is no exception. We’ve documented the image quality settings we used for this game here.

Metro 2033
Metro 2033 also offers a nicely scriptable benchmark.

Given what we’ve seen on the preceding pages, the results from these two games are not a big surprise.

Integrated graphics

Our round of gaming tests with the GeForce GTX 460 left us with some lingering questions. How much of the poor performance from the Atom and E-350 could be attributed to their four-lane PCIe connections to the graphics card? How much of it was attributable to the relatively high image quality settings we used? Could these CPUs run these games acceptably at lower settings?

We also had some natural questions about how the much-hyped IGP in Zacate stacks up—not only to the known-quantity Pineview IGP, but also to the IGP in AMD’s own 890GX chipset for Athlon and Phenom processors. We wanted to see how Zacate’s IGP fares against its close cousin in the discrete world, the Radeon HD 5450, as well.

To answer those questions, we conducted a new round of tests using the integrated graphics processors from Zacate, Atom, the 890GX, and a couple of Sandy Bridge chips. We threw in the Radeon HD 5450 on the two lightweight CPUs for comparison, and we also included a GeForce GTX 460, to see if it could perform better in those four-lane PCIe slots at lower image quality settings.

We tested Bad Company 2 at the lowest IQ settings across the board with a 1280×800 display resolution. The Atom’s IGP couldn’t run this game, so it had to sit out.

Funny, isn’t it, how some of the faster CPUs are brought to parity with the Atom and E-350 when they’re forced to rely on their integrated graphics processors? Still, neither the Zacate nor the Atom can really muster acceptable frame rates, regardless of the graphics solution used.

We’re not quite sure what to make of the fact that the E-350 is measurably faster with the Radeon HD 5450 than with the GeForce GTX 460. The 5450 is a much slower GPU, generally speaking.

Next, we tested StarCraft II at the game’s “Medium” quality presets, F1 2010 with its “Ultra low quality” settings, and Civ V with absolutely everything turned to its lowest possible level. We used a 1280×800 display res in all cases except one: the Atom IGP would only run SC2 at 1280×720. That same IGP required some lower IQ settings in SC2 for compatibility’s sake, too. As you’ll see, the lower settings didn’t help it much.

I think we have some answers. No, dropping to lower quality settings doesn’t unburden the four-lane PCIe connections on the Brazos and Pine Trail systems and magically produce playable frame rates. The CPUs themselves are just too slow to handle these brand-new, big-name games.

I’m not sure we can divine whether Zacate’s IGP is superior or inferior to, say, the 890GX’s or the Radeon HD 5450, given these results. We can surmise that the practical difference for these top-flight games is immaterial, though.

We think the E-350’s IGP deserves closer attention in a different context, so let’s move on.

Some more, uh, age-appropriate games

We could test top-flight games and report unplayable frame rates in the teens until we’re blue in the face, but you really shouldn’t hold your breath while benchmarking. Makes the long tests uncomfortable.

Instead, we decided to breathe freely, focus on the Zacate-versus-Atom angle, and try out some older games and indie titles with less stringent hardware requirements.

Our first stop was Shank, a side-scrolling beat-’em-up that’s loads of fun.
My interest here was intensified because I had already tried running this game on my Phenom II-based home theater PC with an AMD 785G chipset, and the IGP on that system was too slow to deliver playable frame rates at 1920×1080. By contrast, the Zacate IGP was able to churn out an acceptable stream of frames at 1920×1200 using Shank‘s image quality defaults. The CPU didn’t seem to be a big hindrance, either, since gameplay felt fairly smooth. I was surprised when I fired up Fraps and found out that frame rates were only averaging around 20 FPS, but some games don’t need 60Hz to give a good sense of motion.

The Atom IGP, meanwhile, couldn’t run Shank. I believe the game requires support for DX9’s Shader Model 3.0, which rules out the GMA 3150.

In the indie action RPG Torchlight, at the quite decent settings shown above, the E-350 APU managed nicely, averaging about 35 FPS on the frame-rate counter. Popping in a Radeon HD 5450 raised frame rates to about 43 FPS, just a little faster than the integrated graphics. The Atom IGP required some concessions, such as dropping down to 1440×1050 resolution, just to run the game correctly. Even then, it could only manage 6 FPS. You’d probably need to use Torchlight‘s special ultra-low-quality netbook mode to get a usable frame rate out of the Intel IGP.

Lego Batman was one the few places where we ran into a compatibility problem with the E-350’s IGP. Using the default image quality options, we’d see only a black screen with a few indicators on it after the game started. Some googling around revealed that many folks have run into this same problem on discrete Radeons. We were able to work around it by disabling the game’s enhanced graphics, reducing image quality. So configured, at 1920×1200 with high texture quality, the E-350 APU produced frame rates of 30-40 FPS.

On the Atom IGP, we didn’t run into any black screen issues, perhaps because the enhanced graphics settings were completely unavailable. We had to bump the resolution down to 1280×768, and even then, the game felt sluggish, with frame rates ticking away at 15-20 FPS.

Emboldened by our successes with those other games, we decided to try a newer shooter based on the Unreal Engine that undergirds so many console ports these days. Borderlands seemed like just the ticket. Sadly, the E-350 wasn’t quite up to it. Frame rates at the dock area you see pictured above averaged about 25 FPS at 1024×600 with the lowest possible image quality options. The Radeon HD 5450 was faster, averaging 35 FPS, even though we had to run at 1024×768, since 1024×600 wasn’t available with it. (You can probably guess by now, but no, the Atom IGP couldn’t run this one at all, since Shader Model 3.0 is required.) Playing Borderlands on the E-350 IGP isn’t a fun experience; it’s just too slow.

The E-350’s stumble in Borderlands kicked off a bargaining process in which we started seeking older shooters that would run acceptably on the IGP. Call of Duty 4: Modern Warfare was an obvious target, but the E-350 couldn’t muster enough performance to make playing the game anything but painful. Our next stop was Unreal Tournament III, but alas, the frame rates again weren’t consistently high enough to satisfy us. Switching over to the Radeon HD 5450 helped some, but not enough. At the end of the day, even older shooters and console ports aren’t a good bet to run well on the E-350. The overall experience is very much like what we’ve experienced with Nvidia’s Ion platform: better than the dreadful Atom IGP, but still not good enough handle yesterday’s more demanding PC games, let alone today’s.

Video playback

Our last round of tests covers one of the E-350’s most notable features, accelerated playback of H.264 video. To see what the practical difference would be between an unaccelerated dual-core Atom and an E-350, we played the trailer for Iron Man 2 in various formats: three QuickTime .mov files with H.264 encoding at the three most commmon resolutions, and then in the Flash video format via YouTube. We used Window Media Player to view the QuickTime files and Firefox 4 beta 11 for the bulk of the YouTube videos.

The CPU utilization numbers you’ll see below are what we observed via Windows Task Manager while playing the video. They’re not intended to be super-precise, and heck, we have reservations about how accurate they really are on the Atom due to Hyper-Threading. They should give you some sense of how close the CPU was to the limits of its capability in each case, at least.

  Atom
D525
AMD
E-350
  CPU utilization Result CPU utilization Result
H.264 480p 1-16% Perfect 2-10% Perfect
H.264 720p 6-49% Perfect 5-18% Perfect
H.264 1080p 34-64% Perfect 3-64% Perfect
YouTube 720p windowed 46-64% Smooth 36-49% Smooth
YouTube 1080p windowed 79-90% Choppy 37-50% Smooth
YouTube 1080p
full-screen
Choppy Smooth
YouTube 1080p windowed
– IE9 beta
72-86% Choppy 11-37% Smooth

The QuickTime H.264 videos are no problem for either system. Because these are dual-core CPUs, the unaccelerated Atom D525 isn’t quite the basket case that, say, a single-core Atom in a netbook might be when asked to handle these video formats. The D525 even deals with a 720p Flash video in an expanded windowed view pretty well. It hits a wall at 1080p, however, and just can’t cope.

The E-350 copes well with our higher-definition Flash videos, all the way up to full-screen 1080p.

That’s not to say that Flash video is always a breeze on Brazos. The scores in the table above come from a release candidate version of Flash 10.2 that was supplied to us by AMD. The final version of Flash 10.2 has since been released by Adobe, and it doesn’t properly support UVD3 acceleration on the E-350. Without help from the video decode hardware, CPU utilization is substantially higher, and 720p Flash videos can be choppy in full-screen mode.

AMD has been telling us for months that it’s working with Adobe on Flash acceleration for the Brazos platform, so we were taken aback by the lack of support in the public 10.2 release. The latest official word from AMD on the issue reads like so: “AMD is working closely with Adobe to resolve this issue and we expect an updated 10.2 in the near future.”

We’re hopeful AMD and Adobe will get this issue resolved, but we’re not entirely comfortable recommending Brazos as a platform for web video playback at present. We need to see a working, public version of Flash with proper acceleration, and ideally, we’d like to see acceleration working properly through several generations of Flash releases. Brazos’ basic competency at web video playback relies on successful collaboration between AMD and Adobe—not just once, but consistently. The problems with Flash 10.2 serve to illustrate how fragile that competency can be. If you want to be absolutely certain your system will be up to the task of handling HD Flash video, the only sure-fire option is getting a faster CPU.

Then again, good things are possible with really robust GPU support, as our foray into testing with the IE9 beta shows. IE9 includes GPU acceleration for all of the browser’s visuals, and Flash 10.2 makes use of it. On Brazos, that combination yields markedly lower CPU utilization. Further good things look to be possible once Adobe’s grand plan for its Stage Video API unfolds, although we understand that sites like YouTube will have to modify their player programs to take advantage.

Curious about its potential to serve as the engine for a home theater PC, I spent some time trying out our Brazos system with Windows Media Center. Playback of a recorded HDTV program from a local drive was seamless, and skipping forward and back around commericals and such was quick and stutter-free. I was even able to access 720p recorded videos on my home theater PC and use the Brazos system to play those. Quick jumps ahead or back were occasionally hindered by networking speeds—my HTPC is on an 802.11n Wi-Fi connection—but the E-350 looked to be quite happy in the Media Center extender role.

Conclusions

In terms of size, power draw, CPU performance, and price, AMD’s E-350 APU is the same class of solution as Intel’s dual-core Pine Trail Atom. Everything we’ve seen in the preceding pages has made that fact abundantly clear. Full-sized desktop processors are quite a bit faster than either the Atom D525 or the E-350. AMD’s bravado last summer in declaring that the Bobcat core would reach “90% of today’s mainstream performance in less than half the silicon area” was fun to witness, but it was a bit too optimistic, in our view. Zacate’s appeal is in its small size, blessedly low power draw, and potentially low overall system costs.

Once you get past that realization, the next one follows almost immediately: the E-350 APU is the new champ in its weight class. With dual Bobcat cores at 1.6GHz, the E-350 is generally about 10-20% quicker than an Atom D525 at 1.8GHz. The two processors sometimes trade places in the standings depending on the workload, but the newcomer from AMD comes out ahead overall. The E-350 reaches this level of performance without drawing any more power than the Atom, either. In fact, our Brazos system drew measurably less power than our Atom D525 test rig, especially when idling. 14W for an entire desktop system is pretty darned remarkable.

What really sets the E-350 apart from the Atom, though, comes through its Radeon heritage. The ability to play H.264-encoded videos is practically indispensable these days, and the E-350’s UVD3 video decoder handles that chore with ease. That’s a concrete, practical improvement in usability over the Atom, even in its dual-core form. We still have some trepidation about the state of video acceleration in Adobe Flash, but assuming Adobe and AMD can iron out those software problems, the hardware is certainly capable.

Much like its CPU counterpart, the E-350’s integrated GPU won’t keep pace with very latest PC games, but it’s good enough to run a host of indie and casual titles like Shank and Torchlight quite well. Frustratingly, the Atom’s weak-sauce IGP makes even modest games like these seem hopelessly out of reach. We’re much more pleased with Zacate’s balance of resources. Over the next six months or so, we expect Zacate’s superior IGP to matter even more, as GPU acceleration of web browsers and Flash animation becomes nearly widespread.

The E-350 Fusion APU is a wonder of integration, and AMD has set a new standard for basic computing platforms with the Brazos platform. For users whose needs are confined to simple office productivity, communications, and media consumption, the E-350 may well be sufficient. For those places where you might have considered an Atom- or Ion-based slim desktop system before, you’d now do better to consider a Brazos-based offering. In fact, we’re left wondering how Intel could possibly continue its long-standing (and seemingly intentional) neglect of the graphics and video capabilities of this Atom platform for another generation. AMD has forced the issue. Without some help, the Atom will deserve to lose badly, both in “nettops” and netbooks, from here forward.

Still, to address the title of this article quite directly, we think the appeal of the E-350 APU on the desktop will be limited. Processors like Sandy Bridge (and probably AMD’s own upcoming Llano) are also wonders of integration, on a larger scale. They’re too good to be ignored and, if time is money, they’re worth every extra penny for most of us. The Brazos platform is best suited to a host of other locations: in the living room, inside of specialized embedded systems, tucked into a slim ultraportable, or perhaps even beneath the touchscreen of a tablet PC.

Comments closed
    • Sam125
    • 9 years ago

    Thanks for the review Scott! The E-350 looks like a good Atom replacement which I’ve seen used as a terminal or even an entry level general purpose machine. I’m personally waiting for llano reviews and November to see if Apple ends up adopting AMD APUs before I take the plunge but this certainly seems like a pretty good first step.

    • burntham77
    • 9 years ago

    This is looking like a good HTPC option for some seriously low power draw. My only concern is that it has just the one PCIe slot. Good for popping in a low end GPU, but I’d love to see a motherboard with a PCIex1 or PCI slot for a soundcard (like the Xonar). That would make for a perfect HTPC foundation.

      • UberGerbil
      • 9 years ago

      Well, then you’re looking at an mATX motherboard and 3rd party passive heatsinks, I’d imagine. But frankly, I don’t see anything wrong with the integrated graphics for HTPC use (unless you’re really talking about a gaming machine that also plays media). For that matter, I don’t see a huge problem with using HDMI or the SPDIF outs for audio, either, but then I’m not a hardcore audiophile. And since I’d probably use an HDHomeRun to supply the video (rather than a tuner card), I could do without the PCIe slot altogether — and use a very low-profile case (provided it had some kind of quiet cooling). I’m seriously considering such a build, in fact.

    • raddude9
    • 9 years ago

    How about some Browser benchmarks next time.
    Apart from a javascript benchmark or two I’d like to see if the built in GPU can accelerate the graphics in modern browsers.

      • hoohoo
      • 9 years ago

      Yes it does.

    • BeowulfSchaeffer
    • 9 years ago

    I’d like to see a Distributed Computing test included, with both a cycle per hour and cycle per watt.

    • evilpaul
    • 9 years ago

    But can Brazos play Super Meat Boy? Inquiring minds want to know (and got it for like $3 on Steam :-P).

    • Dan37w
    • 9 years ago

    Hi, i see there are a lot of technical people here.
    Extremely detailed review, maybe too detailed for me 🙂
    I’ll probably make a future purchase of a zacate 350 nettop when released.

    I have 2 questions:
    – Is it slower in web browsing than a regular desktop ? (I’m on athlon X4 right now)
    I mean if it’s a little slower that’s ok, but hopefully it won’t be in slow motion and everything moving sluggish 😛
    – And will it play max payne 2 on 1080p all-high, as i can play on my integrated HD4200 ?
    I know that the 6310 is better than HD 4200, but i’m asking regarding cpu performance…
    thank you

      • khands
      • 9 years ago

      My guess to both of those questions would be it should be capable of doing so (max payne 2 being 8 years old now). Your Athlon (2 I’m assuming) x4 would beat the snot out of it but Zacate should be capable of what you’re asking of it there. For comparison’s sake, here was its sys reqs at release.

      * Pentium III 1.0 GHz CPU,
      * 256 MB RAM,
      * 32 MB video card RAM,
      * DirectX 9.0,
      * Windows 98 (WIN)

        • Dan37w
        • 9 years ago

        Thanks for the reply,
        I’m looking into a low power desktop replacement since i’m no gamer (of newer games atleast) and I’m tired of seeing my cpu at 1-5 % usage 95% of the time…just paying electricity for nothing…and the constant humming of all the fans and two 7200 rpm drives iritates me.
        Ok, thanks, guess I’ll have to wait for a nettop (preferably z-box), hope it won’t be long until they release them…

          • swaaye
          • 9 years ago

          You won’t be saving any money unless you can sell off your current box for a nice amount. And yea this Brazos hardware is much slower than what you have. Its CPU performance for most apps is going to be lower than the original Athlon 64 lineup from 2003-4, although it’s a dual core so that will offset that a little bit.

      • hoohoo
      • 9 years ago

      Speaking from using the ASUS micro-ATX E350 and also the Sapphire mini-ITX E350, running Win7 and OpenSuSE (initially 11.3 but moved to 11.4): web browsing is perfectly good, video playback 1920×1200 on Win7 is perfectly good but on Linux the video suffers from a bit of tearing.

      Can’t comment on Max Payne2. FWIW L4D at 1920×1200 managed between 6FPS and roughly 30FPS. But, heheh, L4D is just as much fun in the middle of a zombie swarm at 6FPS as at 60FPS!

      I returned the ASUS as it had memory issues running Linux (fixed by the kernel supplied in OpenSUSE 11.4, but still). Have used the Sapphire for two weeks now, and unless I was playing a game I used the E350 and left my Core2Quad+GTX560Ti box turned off.

    • jpostel
    • 9 years ago

    About the Starcraft II results:

    I was reading an article on Anandtech (can’t find the one I was reading, but it’s mentioned in [url=http://www.anandtech.com/show/3987/amds-radeon-6870-6850-renewing-competition-in-the-midrange-market/10<]others[/url<]) where they mentioned Starcraft II was mostly CPU limited rather than GPU limited. There are only certain scenarios where GPUs make a big difference. That might explain the results.

    • DPete27
    • 9 years ago

    I find it amusing that you can write a conclusion saying that the Fusion APU outperforms an Ion platform when you don’t show any proof in the article. I think the point was missed in this article as far as comparing competing products which would be the Fusion vs. Atom/Ion systems. It’s pretty well known that the Fusion chips’ integrated graphics are far superior to Atom’s as has been mentioned in a couple other previous articles regarding Fusion. What I really want is to know if the E-350 can outperform an Atom D-525/ION2 system (which I would say is the REAL competition for Fusion and quite possibly the design goal)…. Strapping a GTX460 to the Fusion and Atom systems was hilarious though.

      • DancinJack
      • 9 years ago

      [quote<]...nobody that reads tech sites would buy an Atom system without Ion.[/quote<] Just because someone reads a tech site does not mean they are a hardware enthusiast or even know what the heck Ion is. Don't assume everyone is as informed as you are. When you write the article, you can chose what to compare it against. Don't get pissy becuase you chose to read an article that didn't contain comparisons you were looking for. Thanks for wasting my time with your content-less comment.

      • jpostel
      • 9 years ago

      Not a direct comparison, but the Integrated Graphics section slapped a GTX 460 on them, and thats a heck of a lot more GPU than ION2.

      • Cyril
      • 9 years ago

      We compared Zacate to an Ion-powered CULV nettop in this article:

      [url<]https://techreport.com/articles.x/19981[/url<]

      • eitje
      • 9 years ago

      [quote<]I mean, come on, nobody that reads tech sites would buy an Atom system without Ion.[/quote<] I did.

        • tfp
        • 9 years ago

        So did I. It makes a great headless storage and print server. Uses very little power vs my old setup.

    • Xylker
    • 9 years ago

    I just bought one of these. The Gigabyte one, and I really like it.

    So, I spent $192 on the MB/CPU/GPU/Sound card/Memory and I paired it with a $170 SSD… I am still working out how to best get the PSU and case assembled for my particular purpose. But, all things considered, I am very pleased with my purchase.

      • DancinJack
      • 9 years ago

      How is the SATA controller?

        • Xylker
        • 9 years ago

        What sort of answer would you like? That is to say, it worked fine for OS install and light usage to date. But if you have a free benchmark you’d like me to run, I would be happy to oblige.

          • DancinJack
          • 9 years ago

          HD Tach maybe…so long as you have a HDD and not a SSD.

          [url<]http://www.simplisoftware.com/Public/index.php?request=HdTach[/url<] sequential reads and writes puhlease?

            • Xylker
            • 9 years ago

            I linked some graphs in the forum:
            [url<]https://techreport.com/forums/viewtopic.php?f=22&t=75609[/url<]

      • Dan37w
      • 9 years ago

      Hi Xylker,
      How is the zacate in everyday experience ?
      I mean web browsing, some office work, some youtube viewing….regular stuff
      Is it sluggish compared to a normal desktop or not ?
      I’m kinda set on replacing my desktop with a nettop…but wouldn’t like to experience constant slo-mo in everything 🙂
      Thank you

        • Xylker
        • 9 years ago

        I am at work now, but this weekend I can take some time to do a few benchies with a 7200 rpm disk and a 10k raptor.

        My daily use is with an SSD, so it will take some time to do the testing.

    • wiak
    • 9 years ago

    i miss SATA Controller test, and a 1080p H.264/VC-1 40mbit Blu-Ray Test (this is different from a low bitrate 10mbit 1080p h264 trailer from trailers.apple.com as it uses different profile and higher bitrates that can be alot harder to decode)

    am pretty sure the atom system will be murdered by running a 40mbit blu-ray on it, brazos wont have any problems because it got a hardware decoder built into the apu

    lets hope the gigabyte review will have these, btw other than that nice review, gives you a wide range of cpus to compare it to and how the power usage compares

    you should also check out Trine (with PhysX) and Zero Gear (uses Open Source Ogre 3D and BulletPhysics) for the indie test
    [url<]http://trine-thegame.com[/url<] [url<]http://myzerogear.com[/url<] regarding flash, did you guys test it on IE8/IE9? not only firefox?, i belive i read some place that amd test it on IE, i know ie is shit bit still might want to test it with the activex plugin best regards wiak

    • MadManOriginal
    • 9 years ago

    Nice review. What I took away from this most review which hasn’t been commented upon explicitly already is how important PSU matching is with the ultra-low power draw systems. A 50%/25-30W drop is nothing to sneeze at especially for 24/7 systems. My only concern would be for certain applications like multi-spinning-platter-NASes use where the power draw spike upon the (hopefully rare) reboot could overwhelm the powerbrick PSU. Do people know if ‘by default’ there is a staggered disk spin-up, or if it’s available by a setting, or if not could TR do a quick article or has another reputable site already done so?

      • insulin_junkie72
      • 9 years ago

      Yeah, using the SPCR numbers on their appropriate reviews and doing the AC-DC conversions based on the used power supplies (PicoPSU, Seasonic 400ET), it turns out there’s about a 4 watt (using DC) difference between the 350 and any Sandy Bridge* under both idle and HD MKV playback.

      Power supply used makes a HUGE difference.

      * yes, the price difference is pretty big, so not apples-to-apples. Does further illustrate how power-efficent SB is, though, for a CPU of it’s performance class if it gets that close to Zactate.

      • flip-mode
      • 9 years ago

      Definitely. Good call Scott!

      • Cyril
      • 9 years ago

      I believe Scott measured the power draw of the E-350 system [i<]without[/i<] the GeForce GTX 460 when using the power brick, which would explain the big drop in power consumption.

      • ermo
      • 9 years ago

      I belive staggered spin-up is part of the SATA spec. However, it looks as if it is optional for consumer level SATA hardware, so it is up to the vendor to actually implement support for it in firmware.

      • esterhasz
      • 9 years ago

      There are plenty of picoPSU users in the SPCR forums. Myself, I have my HTPC/home server running on picoPSU nor for nearly five years, with a 120W brick that has a fan come on at 90W. On my old X2@1.2Ghz the fan sometimes came on on reboot with four drives but now with Zacate and 2 spinning drives it doesn’t ever get even near 50W… funny thing is, when I bought the PSU+brick combo, it was like $70 and with 90%+ efficiency it has certainly saved me and the planet quite a bunch over the years running 24/7…

      With a 120W or 150W brick you could probably run a SB system!

        • NeelyCam
        • 9 years ago

        PicoPSU DC/DC converter efficiency should be around 95%, but the power brick is lower – I don’t know what it is, but looking at AC adapters in general, the 100-150W ones I’ve found are somewhere around 90%.

        0.9*0.95=0.855.

    • I.S.T.
    • 9 years ago

    I’m really not surprised it couldn’t handle the console ports(Especially Borderlands, which is unusually slow for one anyway). Zacate’s GPU is weaker than the PS3 or X360’s GPU by far.

    I’m very interested to see how Llano does on these games. Given the specs of the GPU, it should fare nicely, but… It’s not guaranteed. It will have far less bandwidth than a dedicated card, after all.

      • stmok
      • 9 years ago

      Llano or AMD A-series’s GPU (Beavercreek) is spec’ed like a Radeon HD 5570.

      AMD’s own numbers indicate that, (due to shared memory bandwidth with the x86 “Husky” cores); it’ll perform more like a Radeon HD 5550…Well, its a little bit faster.

      The A-series will use dual channel DDR3-1600 memory controller.

      Performance wise, the quad-core version will perform in the same class as an Athlon II X4 640 or Core i3 530 with a Radeon HD 5550 discrete video card…This isn’t surprising as the A-series are replacing the Athlon II line.

      It’ll be much better than the E-350’s GPU, but don’t expect miracles of GPU awesomeness. (Give you good frame-rate with all quality features disabled in games.)

    • ET3D
    • 9 years ago

    Would have liked that Sempron benchmarked, the one mentioned on the first page. Would have been interesting how that $80 board+CPU fare in the comparison

    Regarding the low GTX 460 performance, the obvious explanation is that its driver is CPU heavy, in particular with the workload associated with Bad Company 2.

    Anyway, great review. I appreciate detailed benchmarks and comparisons, and it was nice to see power draw with two types of PSU, something that people complained about in other E-350 reviews. A little overboard with the comparison to other CPU’s (since they’re all way faster, and using a smaller sample would have been enough to illustrate that), but that’s a minor niggle.

      • Palek
      • 9 years ago

      [quote<]Regarding the low GTX 460 performance, the obvious explanation is that its driver is CPU heavy, in particular with the workload associated with Bad Company 2.[/quote<] I have close to zero knowledge of driver programming so my comments are pure guesswork, but I thought that, rather than the driver being CPU heavy, it's more likely that the wide parallelism of the GTX 460 is the culprit. The E-350 is a slow CPU by modern standards and it is probably struggling to keep feeding the GTX 460's gajillion shaders and whatnot with instructions and data. The associated overhead actually becomes so heavy as to overwhelm real work. The Radeon 5450 or the E-350 IGP, on the other hand, are much-much narrower and therefore better matched to the processor. Does this make any sense?

        • djgandy
        • 9 years ago

        Kinda, but the way the CPU feeds the GPU is through the driver :-).

        The whole point of a GPU is that your CPU doesn’t need to feed it too much. However a GPU like the 460 would usually be used in something with at least a Core 2 Duo which has significantly more processing power than either of these chips.

        It shouldn’t make too much difference how many cores the GPU has. The same basic work has to be done, and the same data is sent to the GPU. The extra GPU grunt just means the GPU can do its processing faster. A big issue for these low end processors is shader compilation. Modern games have huge shaders and compiling these down to GPU instructions can be quite costly, this of course has to be done on the CPU.

        ET3D is probably right, and with a profiler this is easily verifiable. Those games may have heavy CPU logic anyway, and the driver overhead could be significant for such low performance CPU.

          • Palek
          • 9 years ago

          It would be interesting to see CPU load figures for a game benchmark on two configurations that use two different ranks of GPUs from the same architecture generation, such as the Radeon HD 5670 versus the Radeon 5870, but are otherwise identical. If what you say is correct CPU load should be close in these two scenarios.

            • djgandy
            • 9 years ago

            Well it is pretty easy to do actually for anyone with 2 cards. Download AMD codeanalyst and run the game. It’ll give you a full profile of what the system is doing and where it is spending its time.

    • swaaye
    • 9 years ago

    Up until recently, I had a 12.1″ subnote with the Turion Neo X2, which is a 1.6 GHz K8 dual core. It also had a discrete Radeon 3450. I would prefer not to have that level of performance again in a general purpose PC. 🙂 That sounds like this Zacate platform essentially.

    It was quite sluggish in Vista/7 and gaming was not really worthwhile. It struggled with games like Sins of a Solar Empire and even KOTOR a bit. UT3 was barely playable. Let’s not sit around and pretend that this kind of hardware is a desirable game machine or even offers desktop performance that’s above average. What it feels like is a PC from 2004 or so.

    As an Atom replacement, maybe it’s ok. Atom sucks. This sucks a little less. Making a highly portable machine or a low power server of some sort are about the only interesting applications IMO. I’d buy a 9″ netbook with this inside, in a blink, because my Eee900 is definitely getting a bit worn out after 3 years.

      • dpaus
      • 9 years ago

      I had a Dell Inspiron with the same CPU (except I think I splurged and got the 1.8 GHz model), but, since I didn’t game on it, I was quite happy with its day-to-day performance with Vista.

      I’d like to see what this platform can do when clocked to about 2.5 GHz (TDP be damned, I say!) and/or with an extra core or two. I think with a little work, this could be AMD’s next mainstream platform, and good one at that.

      That doesn’t change the fact that I’d love to find a 13″ laptop with a Phenom II X6 1100T in it and DisplayPort video out 🙂

      • UberGerbil
      • 9 years ago

      I have to disagree. I was stuck using a 1.8GHz Neo X2 w/4330 (Dell Zino HD 400) for a month, and it was no hardship at all. Under Win7 it was in no way sluggish as a general purpose PC. It would’ve been fine for Sims and probably WoW, which is all the gaming I’d expect from a “general purpose” PC; I probably would’ve been disappointed trying to run non-Source-based FPSs on it or other non-casual games, but I didn’t attempt that and I don’t expect a “general purpose” PC to run those: that’s what a gaming PC is for.

      Now, on a subnote with a sluggish mobile hard drive — and especially if inadequate RAM led to paging — I could imagine the experience might be different. But on a well-balanced system with 4GB of RAM and a 7200rpm HD, it was responsive and completely adequate for browsing (including Flash) and office apps and 12MP image editing and every other “general purpose” task I threw at it (even some light Visual Studio work). Sure, it feels “like a PC from 2004” on those tasks — because a PC from 2004 ran those completely adequately too: on “general purpose” tasks, performance has mostly ceased to matter, and you can’t tell the difference in daily use. About the only “general purpose” task that has continued to demand more hardware is ever-higher-bandwidth video, but how many 1080p60+ videos do you really encounter every day?

        • swaaye
        • 9 years ago

        If you’re happy with it, by all means enjoy it. I’m not interested in paying for that again unless it is used in an application where its low heat output and power usage is truly leveraged and no better alternative is available.

          • UberGerbil
          • 9 years ago

          But for those applications, it appears to be the best alternative currently available. I was just suggesting that if you did employ one in such an application, using it for more general purpose PC tasks wouldn’t be a particular hardship.

            • swaaye
            • 9 years ago

            That’s very true. I would kill to see this in a 9″ netbook. It would use less power than my EeePC 900 with its Celeron 900 + 910 + ICH7 setup and blow it out of the water in performance. But nobody seems to want to go below 10″ anymore and those 11.6″ things should really just have a Core i3 inside.

            • UberGerbil
            • 9 years ago

            Well, I think the mfrs mostly followed the market as given the choice people seemed to migrate towards machines with screens larger than 10″ (they wanted ultra-cheap ultra-portables, not cheap ultra-ultra-portables). But I do wonder how much of that was also a result of dissatisfaction with the limited [i<]resolution[/i<] in those 10" screens, and not the limited dimensions; Windows is pretty painful at 1024x600, after all. And that resolution limit was enforced by Intel's Atom licensing, not by any technical requirement. I agree, it would be extremely interesting for someone to attempt another 9" or 10" machine with at least full 720p resolution and Brazos guts. That said, I for one am very intrigued (to the point of considering purchase) by the Lenovo X120e, which is one of "those 11.6" things" -- it's the right size (and more importantly, has a matte screen and decent keyboard), and while I"d like to be able to compare it to something similar with a Sandy Bridge I3 inside, the Brazos guts are fine with me for that application (really the only thing I'd expect to gain from SB is better battery life). But the real combination that intrigues me would be a step up from that: the U260, with SB guts for better battery life (though I suspect that machine really just needs a bigger battery no matter what it has inside) .

    • jensend
    • 9 years ago

    Add me to those who think TR should put a disclaimer about the 90% per-clock bit. I’m not saying that TR did anything wrong in faulting AMD for not living up to the most straightforward interpretation of their statement. However, I think their level of success at meeting the qualified statement is an important part of the story- and perhaps some Zacate boards may allow significant overclocks.

    It’s true that for many of us a Zacate desktop probably doesn’t make sense except possibly as a media center. However, for those who have systems around (secondary boxes, your relatives’ boxes, etc) which haven’t been upgraded since before Conroe, Zacate might be worth looking at. For lots of people, the P4 Northwood was the advent of “good enough” computers. Given how Zacate compares to the Pentium Extreme Edition, I think it’d outperform single-core Northwood, Prescott, etc comfortably, also giving the ability to watch H.264 video comfortably and run recent casual games (which their previous systems couldn’t do), while costing very little and always using less than a quarter of the power their old systems used.

    I think the next generation of Bobcat- with a die shrink- should be quite interesting.

    • jensend
    • 9 years ago

    How on earth did my comment get doubled? I only hit the button once.

      • NeelyCam
      • 9 years ago

      Looks like you pressed it twice real fast.

      I consider this to be a bug in TechReport forum code – one I requested to be fixed to no avail.

    • Bauxite
    • 9 years ago

    I think asus has the only itx board I’d consider for a mini htpc setup, thanks to its decently sized passive heatsink. All the fan coolers I’ve seen so far give me high pitched nightmares. (They also have a matx board which is sort of amusing)

    Using a minicard for upgradable wifi is a nice bonus (or possibly tuners etc in the future) since slots are so hard to come by in this form factor.

    I wonder if e350 has enough muscle to keep something like a ceton tuner recording 4 or 6 streams with no issues, that would be a fun setup to test. The help from uvd should keep playback working even with heavy recording going on.

      • msi-rajiv
      • 9 years ago

      Hi Bauxite – the MSI retail version has a slightly larger heatsink on it. These were early sample models so they do not accurately reflect the retail heatsink.

      You can see the actual one here:

      [url<]http://www.msi.com/product/mb/E350IA-E45.html[/url<]

      • mutarasector
      • 9 years ago

      I was wondering this myself, but I seriously doubt it. While the GPU could certainly handle the streamed content, I suspect the X86 cores would be lacking since Ceton’s recommendation is a dual core X86 clocked at 2.4GHz.

      What Zacate might be able to handle is a possible 2-tuner USB variant of Ceton’s tuner when they bring it out, or perhaps a 3 tuner HomeRun HD tuner by Silicon Dust, but by then, Krishna and Wichita will be out as well.

    • flip-mode
    • 9 years ago

    I find it hard to criticize AMD for the “90% of mainstream” comment because it was so vague to begin with – at least in terms of criticizing on performance grounds. As a marketing line, I think it’s crummy. Honestly, it’s a pretty crummy marketing line any way you look at it. Crummy because it’s vague. Crummy because 90% is essentially saying it’s a downgrade. It somehow simultaneously sets expectations both high and low.

    So, honestly, AMD delivered a marketing line that’s no better or no worse than most other marketing lines.

    As for performance, well, it’s good enough for certain uses and not nearly good enough for other uses. It’s 20 to 30 percent faster than Atom at a lower clockspeed.

    So, Zacate/Ontario is to Atom what Core is to Phenom. And if you’re looking for something to do low power tasks at low power consumption levels, that’s probably all that matters.

      • dpaus
      • 9 years ago

      [i<]"Zacate/Ontario is to Atom what Core is to Phenom"[/i<] -- don't y'all just [i<]hate[/i<] it when flip-mode is [i<]so[/i<] on-point?

      • hoohoo
      • 9 years ago

      The strength of the E350 is video. I play MP4/MKV/DIVX on my E350 at 1920×1200 or 1920×1080 and it works quite well. Atom cannot touch that.

      Atom-only solutions are poor at video. Atom+ION compare well to E350. For manufacturers though E350 has lower BOM than Atom+ION. Do the math and you will see what AMD was shooting for.

    • donkeycrock
    • 9 years ago

    Would you recommend this platform for a home server?

      • flip-mode
      • 9 years ago

      Yes, but due to the power consumption rather than the performance. The performance of Atom is good enough for that, so this is even better, but with lower power consumption than Atom. Kind of a dead ringer.

        • Bauxite
        • 9 years ago

        It looks like 4+ sata ports is the norm as well, quite a few atom boards only come with 2.

          • flip-mode
          • 9 years ago

          Which is fine for a home server. The 4x pcie means you could plug in a pretty decent RAID card too.

            • Bauxite
            • 9 years ago

            2 ports is not fine for a DIY home server, though the temptation to cut plastic and traces in the atom price land is high.

            Thankfully many of the prebuilt home servers with atom have 4 ports/4 drive bays, but you have to shop carefully in DIY land.

            Anyways it looks like this platform can go up to 6 ports native off the “southbridge”, 6gbps is just a bonus.

        • swaaye
        • 9 years ago

        When do you start saving though? You have the initial money outlay to recover and what about the energy used to build and ship the new hardware? I think it makes more sense to retask old hardware as a home server even if it’s fairly power hungry.

          • flip-mode
          • 9 years ago

          Depends on how old. Depends on features. Sometimes old hardware is absolutely obsolete because it is simply not doing a good enough job or not doing the job at all. Maybe and add in card can fix it, but at what cost? If an add in card is $40 or you could pick up an essentially new system for $100 that has more features and better power consumption and maybe even better performance, and how bout that form factor, then… well, I’d try to go for the new system.

            • swaaye
            • 9 years ago

            That’s true. It depends on what you need and what you have lying around. I’m starting to come across a lot of old stuff with SATA built-in these days though. And I already have a SATA PCI card. It’s enough to work as a gigabit NAS really.

          • NeelyCam
          • 9 years ago

          [quote<]I think it makes more sense to retask old hardware as a home server even if it's fairly power hungry.[/quote<] Too loud.

            • swaaye
            • 9 years ago

            You mean fans? I can fix that. Unless you’re thinking of using a 3.2 GHz Prescott or something! 😉

            • NeelyCam
            • 9 years ago

            Touche. 🙂

            No, I meant old stuff was generally higher power, with loud fans not only on the CPU cooler, but also inside the PSU. I don’t miss the whiny days…

            • hoohoo
            • 9 years ago

            My home NFS/SAMBA/LAMP server is a Phenom 9500. It sits in the kitchen beside the fridge. The fridge is louder LOL. But I understand your point.

      • wibeasley
      • 9 years ago

      A related question: ANy reason to think its disk IO wouldn’t match the performance of a, say, 4 year-old Athlon?

        • swaaye
        • 9 years ago

        I don’t see why disk performance would be a problem. That’s done with DMA and should not have a large CPU impact.

        It’s possible that it might not be able to handle full gigabit network throughput though. Not sure there.

      • hoohoo
      • 9 years ago

      Maybe OT. My home server uses a Dell PERC 5i RAID card. I tried that card in the ASUS micro-ATX E350 board and the board refused to talk to it. Linux reported errors related to interrupt 19, Win7 said nothing at all. From my limited understanding of the problem it seems like ASUS assumed the only thing ever plugged into the PCIe slot would be a video card and bought the BIOS code accordingly.

      Sapphire’s PURE FUSION Mini E350 has a BIOS setting to control how int19 is handled but I’ve not tried the PERC in the Sapphire yet.

      If an E350 board could talk to a RAID card then I suspect an E350 would make a fine low-load server.

    • mczak
    • 9 years ago

    No DDR3-1333 benchmarks?
    I’d be interested in how it performs in some of the game benchmarks with the memory overclocked – looks to me this could help a bit (also would be nice to know if this makes a difference in power draw).

      • Palek
      • 9 years ago

      Seconded!

      • hoohoo
      • 9 years ago

      Brazos has a single channel memory controller. It is never going to produce fantastic bandwidth as a result. This chip is targeted at a specific market: netbook and thin/light notebook.

      This article was interesting but as the author said in the introduction (paraphrase) it compares apples and oranges.

      I really would have liked to see some numbers for NFS or SAMBA performance – I think E350 may fare well in smallish servers.

    • vvas
    • 9 years ago

    Somehow I don’t think that the limitation of the Brazos and Pineview systems is the 4x PCIe speed. My guess is that, even if they had the full set of 16 PCIe lanes to connect a discrete GPU, they would still struggle with the benchmarks that have been used for this article. Instead it looks like it’s the CPUs that are not fast enough to feed the GPUs with the data they need.

    • esterhasz
    • 9 years ago

    I bought an ASRock E350M1 a couple of weeks ago to replace an undervolted X2 3600+@1.2Ghz HTPC/home server that was showing signs of old age. I was pleasantly surprised by a) the low noise of the 40mm fan after activating fan control and b) how much more reactive the system feels compared to the old X2 and c) the drop in power consumption from 50W (with an SSD, 2 HDDs and a Xonar sound card, picoPSU, 8GB Ram) to about 32W. The thing is really perfectly fitted for the types of tasks I use it for.

    The real problem for AMD is that I don’t see TMSC spitting out the quantities necessary to feed the market after the rather positive reviews…

    • dragmor
    • 9 years ago

    [b<]Computer Enthusiast "mainstream" is way over the top[/b<] I'd argue that AMD lived up to their 90% performance of mainstream ideal. Its our general view of "mainstream" that is skewed. Mainstream means the PC the average person buys. Even if you looked at Today using the Dell website "mainstream" is likely to be an Inspiron 560 ($650 Oz, E5800 Dual 3.2ghz 800mhz FSB with X4500 integrated graphics) or Inspiron 15 laptop ($600 T4500 Dual 2.3ghz 800mhz FSB with 4500HMD integrated graphics). These would be today's most common machines (after balancing out the netbooks and i3). Nine months ago you were looking at 2.8ghz in desktops and single core 2.0ghz celerons in Laptops. Sure these CPU's are faster than the E350, but the graphics and video performance is way behind. Is the E350 APU (CPU and graphics performance) 90% of the Inspiron PC in mainstream tasks i.e. email, web browsing, SIM3, WoW, youtube, etc. I'd say YES. JPR lists Intel has having 52.5% of the graphics market in Q4 2010. That means mainstream graphics is still Intel integrated and most likely 3500HD or 4500HD level. Why do you think Valve was so happy with Sandybridge graphics? The following article lists 775 (C2D) as 65% of Intel shipments in Q4 2010. ATOM would be at least another 5%. [url<]http://news.softpedia.com/news/LGA-775-Accounts-for-65-Percent-of-Intel-CPU-Shipments-in-Q4-2010-168686.shtml[/url<]

    • tejas84
    • 9 years ago

    “AMD’s bravado last summer in declaring that the Bobcat core would reach “90% of today’s mainstream performance in less than half the silicon area” was fun to witness, but it was a bit too optimistic, in our view.”

    Quote from above article by Scott Wasson

    Enough said. Typical AMD bs with nothing to back it up.

      • Hattig
      • 9 years ago

      It was always clear that this was per-clock. This review was unnecessarily harsh regarding this, in my opinion.

      This review should have thrown in a MacBook Air so we could have seen how a 1.4GHz Core 2 Duo with GeForce 320M chipset/graphics compared.

        • djgandy
        • 9 years ago

        I have a Core Duo U2500 (1.2GHz) that can beat Bobcat and Atom in nearly all of the CPU benchmarks. I’d rather have a cut down full super scalar processor at a low voltage than either of these processors.

        Atom is surely due for a refresh in the next year or two? The desktop processors have jumped so far ahead, and the technology within them will surely be used in the low power segment.

        Sandy bridge looks like it is pushing even nearer to the Atom/Bobcat power usage though. The CULVs were very good anyway, so SB ULVs are going to basically render Atom useless in terms of performance per watt. Sure atom may idle lower, but SB can complete the task in 1/10th of the time.

        The only issue is it is hard to find a configuration where you can have a decent processor without 20 usb ports and a huge chipset that burns power for nothing. That is where these Atom/Bobcat type systems excel. They have lightweight chipsets, although in Atoms case the chipset is still a powerhungry beast.

          • Anonymous Coward
          • 9 years ago

          Atom and Bobcat will always have platform cost advantages, that is important as well.

            • djgandy
            • 9 years ago

            Of course hence SB only rendering atom useless in terms of performance per watt.

          • vext
          • 9 years ago

          No, U2500 performance is about 68% of the E-350. You’re probably thinking of the SU7300. Here’s an interesting comparison from CpuMark benchmarks: [url<]http://www.cpubenchmark.net/midlow_range_cpus.html[/url<] U2500: 529 Atom D525: 711 E-350: 770 Neo II L325 (1.5Ghz): 818 Neo II K625 (1.5Ghz): 948 SU7300: 979 Athlon X2 3800+: 1047 SB ULVs have much higher performance, but they also cost a lot more. Zacate is built to be extremely cheap to manufacture. I wouldn't be surprised if Zacate ends up in small tablets. I don't think we'll see SB in tablets.

            • OneArmedScissor
            • 9 years ago

            How cute, the E-350 has almost the exact same score as the K325 and K345. Good thing nobody actually bothered comparing the E-350 to any equivalent laptop Athlon II in any review I’ve seen before they busted out the pitchforks.

      • ermo
      • 9 years ago

      Just to clarify, “90% of today’s mainstream performance” refers to this:

      [quote<] [i<]"AMD's performance target for Bobcat was 90% of the performance of K8 at the same clock speed and our Photoshop CS4 benchmark shows that AMD can definitely say that it has met that goal. At 1.6GHz the E-350 manages to outperform a pair of K8s running at 1.5GHz in the Athlon X2 3250e." [/i<] [/quote<] [url=http://www.anandtech.com/show/4023/the-brazos-performance-preview-amd-e350-benchmarked/3<]Link to quote[/url<] [s<]I'm personally chalking this up to "creative license" on Scott's behalf ;)[/s<]

        • StuG
        • 9 years ago

        Hmm, thats just a little different than “today’s mainstream performance”.

        • flip-mode
        • 9 years ago

        Wow, nice find. In that case, AMD delivered in full.

        • Damage
        • 9 years ago

        Chalk it up instead to the reps from AMD to talked with us about Bobcat. They didn’t offer such careful contextual qualifications. Just said Bobcat would reach an “estimated 90% of today’s mainstream performance in less than half the silicon area.”

        Look, I wouldn’t make too much of that quotation. I said up front such statements are famously slippery. We wanted to see how Zacate/Atom compared to the big-boy CPUs regardless. But if you’re looking for who took creative license with its performance estimates, look toward AMD.

          • ermo
          • 9 years ago

          @Damage:

          Just thought that the technical basis, which (as you point out) marketing then proceeded to distort, should be given at least a little exposure given the mud-slinging in tejas84’s comment.

          It doesn’t change the fact that the E-350 setup in your piece is nowhere near the point, performance-wise, where it can replace a cheap MATX build costing roughly the same, so I can’t fault you for the conclusion.

      • Anonymous Coward
      • 9 years ago

      How clever of you.

    • codedivine
    • 9 years ago

    Scott, can I post a special request ?
    Can you run the PCIeSpeedTest from AMD?
    [url<]http://developer.amd.com/gpu/amdapppowertoy/pages/default.aspx[/url<] Despite its name, in this case (assuming no discrete GPU installed), it will test the speed of data copy from CPU to GPU (and vice versa). For GPGPU purposes, it should be quite interesting to see how fast data can be copied from CPU/GPU. For discrete cards, often for GPGPU the bottleneck is PCIe and I wanted to see how do Fusion chips perform in this regard. It will only take a minute or two to run. Thanks in advance!

    • sircharles32
    • 9 years ago

    I think AMD’s comments about performance might hold a little more weight, if you had taken clock speed into consideration.
    If you had taken an Athlon II X2 and under clocked it to 1.6 GHz, then perhaps Brazos would have performed closer to the presumed 90% mark.

    Just a thought.

      • maroon1
      • 9 years ago

      No it won’t. It would be closer to the truth if you say that Zacate is 90% performance of the old K8 based Athlon

      E-350 @ 1.6GHz performs worse than the old K8 based Athlon X2 @ 1.5GHz
      [url<]http://www.anandtech.com/show/4023/the-brazos-performance-preview-amd-e350-benchmarked/3[/url<] It is not slower in every case, but on average it slower than lower clocked Athlon And we are talking about the old Athlon (K8 based), not Athlon II which is based on K10 architicture.

        • OneArmedScissor
        • 9 years ago

        Phenom’s big difference is the fancy 48 way shared L3 cache – if it’s there, but it’s not in the “mainstream.”

        Clock to clock difference in the “real world” between any Athlon 64, Athlon II, or Bobcat dual-core comes down largely to the cache of the particular model, but it’s just barely even quantifiable, much less noticeable.

        The only things the E-350 did “worse” than the very equivalent 3250e in your link are pretty much bogus. The 1st pass of the x264 encode is slightly slower – but then it’s equal on the 2nd pass, the part that’s actually CPU intensive and slow. Cinebench is slightly slower, but who does 3D rendering on a low power laptop?

        Yes, it’s slower outright at WinRAR, and does tell us one little detail, as that’s the only thing such a computer might encounter that’s highly bandwidth dependent. Bobcat chips allegedly run their cache at half the CPU clock, but what it tells us is that it likely helps more than it hurts. It’s the corner everyone could have been cutting with low power laptops to begin with. Nobody will notice!

        Why nitpick these things? Would you really be able to tell even an Athlon II with more cache or Brazos computer apart in a blind test?

          • swaaye
          • 9 years ago

          Well sure it’s similar in some tests, but in others Brazos has a >20% disadvantage per clock to that old Athlon 64 X2. That means Brazos’ performance varies considerably. That Athlon X2 @ 1.5 GHz is frankly already a dog.

          You can’t put an Athlon X2 in a netbook though which is the only place Brazos should be used.

    • Meadows
    • 9 years ago

    “[i<]On the Atom IGP, we didn't run into any black screen issues, perhaps because the enhanced graphics settings were completely unavailable.[/i<]" I did giggle. In all honesty though, this is supremely disappointing. The review's target is garbage. I really did expect more.

    • Bensam123
    • 9 years ago

    One vote for that PSU facing off against the beast!

    I would say AMD and NVidia heavily optimize their graphics cards for having all the lane available and if the aren’t then it is forced to cope. I doubt the same results would be reproducible using a PCI-E 1.0 x8 slot compared to the PCI-E 2.0 x4 slot.

    That is almost more interesting a find then the Brazos itself, since a lot of hardware vendors substitute in x8 slots or they have a x4 slot that someone plugs their graphics card into unknowingly, it may garner further attention.

    • kizzmequik_74
    • 9 years ago

    Nice review. I’ve been looking at a Zacate system to serve as an always-on media center/file server/torrent rig, so I was a bit disappointed not to see tests on the storage controller’s performance.

    Still, an E-350 board of some sort looks like my next tech purchase.

    • Crayon Shin Chan
    • 9 years ago

    How does it perform in EVE Online? Might get one to replace the desktop at home, an aging Athlon XP.

    EDIT: I really appreciate your including more powerful processors in the review. Puts everything in perspective.

      • DancinJack
      • 9 years ago

      Not too long ago (can’t remember which EVE version) my Athlon X2 4400 and 8800GTS320 ran it fine. This may have been up to 2 years ago mind you…

    • DrDillyBar
    • 9 years ago

    Slap on a second monitor and sounds like a perfect candidate for a thin clint with a RDP window.

      • mnecaise
      • 9 years ago

      “I do not think that word means what you think it means.”

      Your idea of a thin client is a bit on the heavy side.

Pin It on Pinterest

Share This