Intel’s X48 Express chipset

Intel’s Core 2 Extreme QX9770 is due out this quarter, bringing with it not only four Penryn cores clocked at 3.2GHz, but also a front-side bus running at 1600MHz. That faster FSB presents a bit of a problem, though, because Intel’s current P35 and X38 Express chipsets only support front-side bus speeds up to 1333MHz. So the QX9770 needs a new chipset, or at least a new north bridge, which is where the X48 Express comes in.

Based on Intel’s existing X38 Express, the X48 brings with it official support for 1600MHz front-side bus speeds. With this faster bus, the X48 looks poised to supplant its predecessor atop Intel’s chipset lineup, which means a new wave of motherboards is coming from all the usual suspects.

Gigabyte’s X48T-DQ6 is the first X48 offering to hit our labs, and in addition to packing Intel’s latest chipset, it comes with an innovative Dynamic Energy Saver mechanism designed to lower power consumption. Join us as we put this power saving scheme to the test and run the X48 Express through it paces against the rest of Intel’s chipset lineup.

A new north bridge

The X48 Express’ defining feature is its new north bridge component—a chip that might not actually be new at all. Intel engages in a process called binning with its processors, sorting what are essentially identical chips based on the clock speeds at which they’re comfortable running. The same approach could be applied to north bridge chips, making the X48 a bin sort of X38 chips comfortable running a 1600MHz front-side bus. Based on our own experiences overclocking a couple of X38 boards to front-side bus speeds of 2000MHz and higher, we’d wager there are plenty of X38 chips up to the task of a mere 1600MHz.

X48
Express
X38
Express
P35 Express

Front-side bus
1600/1333/1066MHz 1333/1066MHz 1333/1066MHz

Memory controller
DDR2-800
DDR3-1600
DDR2-800
DDR3-1333
DDR2-800
DDR3-1066

PCI Express 1.1 lanes
0 0 16

PCI Express 2.0 lanes
32 32 0

Multi-GPU support
CrossFire CrossFire CrossFire*

Chipset interconnect
DMI DMI DMI

Peak interconnect bandwidth
2GB/s 2GB/s 2GB/s

Apart from its faster system bus, the X48’s features mirror those of the X38 Express. Both feature 32 second-generation PCI Express lanes, providing substantial bandwidth to not only single graphics cards, but also CrossFire configurations. However, despite working with Nvidia to bring SLI support to its Skulltrail platform, Intel’s desktop chipsets remain locked out of the green team’s multi-GPU rendering scheme. That strikes us as particularly petty on Nvidia’s part, and while it blunts some of the X48’s appeal, CrossFire has matured into a viable and in some cases preferable alternative to SLI.

The X48 Express north bridge chip cap

Falling prices have kept DDR2 memory popular among enthusiasts, so it’s a good thing the X48 retains support for the older memory type. Expect the first X48 boards to have DDR3 memory slots, though. Based on our conversations with motherboard makers, it looks like most are aiming for DDR3 first, with DDR2 offerings to follow.

On the DDR3 front, the X48 should support memory bus speeds up to 1600MHz. However, depending on your motherboard, that top memory speed may only be available if you’re also running a 1600MHz front-side bus. Our Gigabyte X48T-DQ6 lacks the dividers necessary to run 1600MHz memory with a 1333MHz FSB, for example, and others are also likely to face this limitation. Motherboard makers have some leeway when it comes to defining bus dividers, which is why you’ll find plenty of P35 Express boards capable of clocking DDR3 memory at 1333MHz, despite the fact that the chipset only officially supports memory bus speeds up to 1066MHz.

Hardly a secret, the ICH9R shows up at the south bridge

While it’s unclear whether the X48 north bridge chip is merely a carefully selected and further validated X38, the chipset’s south bridge component is Intel’s existing ICH9R. Don’t let the “Secret” chip marking fool you; this is the same silicon Intel launched last summer with the P35 Express. Nearly a year after its initial launch, the ICH9R still packs a competitive feature set, including six SATA RAID ports, 12 USB ports, and six gen-one PCI Express lanes. Intel also says the ICH9R has an embedded Gigabit Ethernet controller, although we’ve yet to see a motherboard actually make use of it.

DDR3 rises to the occasion

Intel first cracked the seal on DDR3 memory with its P35 Express chipset, and at the time, the new memory type hardly looked poised to take over from DDR2. Prices were too high, for one, and DDR3 hadn’t yet ramped to significantly higher clock speeds than existing DDR2 modules. But times have changed, speeds have scaled, and prices have fallen.

Today, DDR3 memory is available at speeds as high as 2000MHz. That’s an impressive milestone considering that DDR3 started out at roughly half that speed nearly a year ago. Those of us who have been around for a while will remember that DDR2 certainly didn’t ramp this quickly after its initial launch.

Unfortunately, there’s still an inescapable price premium associated with DDR3. Modules like Corsair’s CM3X1024-1600C7DHX, which are rated for 7-7-7-20 timings at 1600MHz, run about $400 for a 2GB dual-channel kit. Believe it or not that’s an improvement over what was available last summer, when $400 would have only bought you a 2GB kit of DDR3-1066 memory rated for 9-9-9-24 timings. The price of 2GB DDR3-1066 kits has fallen to around $220.

Despite falling DDR3 prices, the fact that the bottom has essentially dropped out of the DDR2 memory market makes for a tough sell. DDR3 is undoubtedly the future, but when you can get 2GB of DDR2 for less than $50, paying more than four times the price for DDR3 is difficult—if not impossible—to justify rationally.

Gigabyte’s X48T-DQ6
Remixed for X48

Manufacturer Gigabyte
Model X48T-DQ6
Price (MSRP) $300
Availability Soon

While the X48 Express may not offer a whole lot in terms of new features and capabilities, at least it’s spawned a wave of exciting new motherboard designs, right? Well, not exactly. Because the X48 and X38 chipsets are so similar, motherboard makers have been able to dip into their existing product lines for more than just inspiration. Gigabyte’s X48T-DQ6, for example, bears an uncanny resemblance to the company’s X38-DQ6. There are a few minor differences here and there, such as the size of the north bridge heatsink and the placement of a few onboard components, but the overall design appears unchanged.

Not that there’s anything wrong with that. We quite liked the X38-DQ6 and even gave it a TR Recommended award, so there’s something to be said for not messing with the formula.

For Gigabyte, part of that formula includes “Ultra Durable 2” branding that denotes the use of higher quality electrical components, including solid-state capacitors, ferrite core chokes, and low-resistance MOSFETs. High quality components should be a given for motherboards in this price range, but that’s sadly not always the case, which I suppose is why Gigabyte makes such a fuss over its Ultra Durable brand. Never mind that the same Ultra Durable components can be found on a number of the company’s more affordable mid-range boards, as well.

Another element to Gigabyte’s formula is a turquoisy blue board color that has persisted for as long as I can remember. As other manufacturers plunder every color in the rainbow, and even a few that aren’t, Gigabyte has remained faithful to a single unique hue—one that neatly matches the company’s graphics cards, too.

I have to admit that the combination of a blue board with copper heatsinks looks fantastic. Sure, the board will be hidden away in an enclosure and perhaps only occasionally observed through a case window, but if you’re going to build up a high performance hot rod you might as well make sure the engine bay is easy on the eyes.

The engine, in this case, would be a processor to slide into the X48T’s LGA775 socket. Gigabyte keeps the socket area relatively free of obstructions, using low profile capacitors in areas that might interfere with standard heatsink designs. Larger aftermarket coolers that fan out from the CPU socket might conflict with the board’s beefy north bridge cooler, though.

Passive chipset and VRM cooling is provided by a set of copper heatsinks tied together by a modest (at least by today’s motherboard standards) heatpipe network. This relatively simple arrangement puts surface area before flashy excess, and unlike some passively cooled motherboards, you don’t need to strap on a noisy chipset fan to overclock the front-side bus.

The X48T’s cooling apparatus extends to the underside of the board, as well. Here, we find finned metal back plates for the south bridge, north bridge, and the CPU socket itself. The socket back plate can also be removed if you want to use an aftermarket processor heatsink with a back plate of its own.

On the flip side, Gigabyte neatly arranges the X48T’s storage-related ports in the bottom right-hand corner of the board. Here we find six orange SATA ports hanging off the ICH9R south bridge with an additional two purple ports making an appearance courtesy of a JMicron controller rebranded as a Gigabyte GSATA chip. This auxiliary storage controller is also responsible for the board’s IDE port.

More interesting than the origins of the X48T’s SATA ports is their placement. Gigabyte shows particular attention to detail here, ensuring that longer double-wide graphics cards won’t block access to any of the onboard SATA ports. Longer expansion cards won’t interfere with the board’s low-profile south bridge heatsink, either.

Gigabyte’s board designers deserve a gold star for the X48T’s expansion slot layout, too. The board features a good mix of PCI Express x16, x1, and standard PCI slots. Even with a double-wide CrossFire config, you’ll still have two PCIe x1s and a standard PCI slot to play with.

A year ago, we would have rather seen Gigabyte bias the slot X48T’s configuration towards standard PCI slots. However, with sound cards like Asus’ Xonar D2X making their way to PCI Express, we’re running out of reasons to keep standard PCI slots around.

The X48T certainly packs plenty of expansion capacity without them. Just look at the port cluster, which houses eight USB ports and two flavors of Firewire. You get a pair of S/PDIF digital audio outputs, too: one TOS-Link and one coaxial. And because the board uses Realtek’s swanky ALC889A codec chip, it’s possible to encode multi-channel audio into a DTS Connect stream that can be passed through the board’s digital output ports.

External Serial ATA connectivity is conspicuously missing from the X48T’s backplane, but that’s because Gigabyte has a more flexible solution bundled in the box. Rather than tying eSATA to port cluster connectors arbitrarily assigned to an onboard storage controller, Gigabyte lets you roll your own eSATA connections using any internal Serial ATA port. A PCI back plate provides the external Serial ATA interface, with additional cabling—including power if you want to quickly hook up a standard internal hard drive outside your system—to make it all work. This solution may not be as clean of an implementation as a backplane eSATA connector, but I’ll gladly swallow a little cable clutter for the ability to pick and choose which SATA ports get routed to the outside world.

BIOS options galore

Official support for a 1600MHz front-side bus makes the X48 Express chipset a prime candidate for more extreme overclocking exploits, and Gigabyte has ensured that the X48T-DQ6’s BIOS is up to the task.


Bus speeds
FSB: 100-700MHz in
1MHz increments

PCIe: 90-150MHz in 1MHz increments


Bus multipliers
CPU: 6x-8x (Core
2 Duo E6750)
DRAM: 2, 2.4, 2.5, 2.66, 3, 3.2, 3.33, 4
Voltages CPU:0.5-2.35V in 0.00625V increments

DRAM: +0.05-1.55V in 0.05V increments

PCIe: +0.05-0.75V in 0.05V increments
FSB: +0.05-0.035V in
0.05V increments
MCH:
+0.025-0.775V in 0.025V increments
CPU GTLREF2:
-3-9% in 3% increments
CPU/PCIe driving control: 0.7-1V in 0.1V increments


Monitoring
Voltage, fan
status, and temperature monitoring

Fan speed control
CPU

Front-side bus speed options are actually available up to 700MHz (2800MHz quad-pumped), giving potential overclockers more than enough of headroom to play with. A total of eight multipliers yields plenty of memory bus speed options, as well.

The BIOS really shines in the voltage department, where users have control over seven different system voltage settings. Extreme overclockers will be pleased to note that CPU core voltage options are available up to a whopping 2.35V in ultra-fine 0.00625V increments. If you prefer to underclock, you can lower the core voltage to as little as 0.5V with the same granularity. Memory overvolting options are also impressive, allowing users to push an additional 1.55V on top of DDR3’s default voltage of 1.5V.

As one might expect, the BIOS also includes all the usual memory timings. Temperature-based fan speed control makes an appearance, as well, but only for the CPU fan, and without arbitrary control over actual fan speeds and target temperatures. It’s a little disappointing to see motherboard makers dedicating so little attention to an important feature like temperature-based fan speed control. Just about every enthusiast-oriented motherboard that has passed through our labs in the last couple of years has offered more than enough options on the overclocking front, but precious few give users meaningful control over fan speed profiles.

A new Dynamic Energy Saver feature that Gigabyte has rolled out in its latest motherboards suggests that the company is at least serious about reducing power consumption. This functionality isn’t exposed through the BIOS (at least not yet, although it’s something that Gigabyte is working on), but rather through Windows software bundled with the X48T-DQ6.

When enabled, DES dynamically scales the number of power phases used to feed the processor based on system load. A minimum of four power phases are active at all times, with additional phases activated in pairs. Since our X48T has 12-phase power, DES can choose from five different phase configurations.

Gigabyte’s DES Windows app lets users keep tabs on how many power phases are in use at a given time. More interestingly, it also calculates how many watts a system’s processor is using, feeding that information into a counter that tracks overall power savings. We’ll see just how well DES works when we probe power consumption a little later in this review.

Specifics on specifications

If you prefer your motherboard specifications carefully compiled into a single chart, the one below details Gigabyte’s X48T-DQ6. Knock yourself out.


CPU support
LGA775-based
Celeron, Pentium 4/D, Core 2 processors

North bridge
Intel X48 Express

South bridge
Intel ICH9R

Interconnect
DMI (2GB/s)

Expansion slots
2 PCI Express x16

3 PCI Express x1
2 32-bit/33MHz PCI


Memory
4 240-pin DIMM
sockets

Maximum of 8GB of DDR3-1066/1333/1600 SDRAM


Storage I/O
Floppy disk

1 channel ATA/133 via JMicron JMB363

6 channels 300MB/s Serial ATA with RAID 0, 1, 10, 5 support
2 channels
300MB/s Serial ATA with RAID 0, 1 support via JMicron JMB363

Audio 8-channel HD audio via Realtek
ALC889A codec
Ports 1 PS/2 keyboard
1 PS/2 mouse

8 USB
2.0 with headers for 4 more

2 RJ45 10/100/1000 via Realtek RTL8111B
2 1394a Firewire via
Texas Instruments TSB43AB23 with header for 1 more

1 front out
1 analog bass/center out

1 analog rear out

1 analog surround out

1 analog line in

1 analog mic in
1 digital coaxial S/PDIF out
1 digital TOS-Link S/PDIF
out

Our testing methods

Since the bulk of initial X48 boards appear to be of the DDR3 variety, we’ve evened the playing field and used DDR3 all around. This restricts our comparative platforms to Intel chipsets, notably DDR3 boards based on the X38 and P35 Express. Nvidia doesn’t yet have a DDR3 chipset for Core 2 processors, so the nForce platform will have to ride the pine this time around. Check our nForce 780i SLI review for the latest on the green team’s Core 2 chipsets.

All tests were run three times, and their results were averaged.

Processor

Core 2 Duo E6750 2.67GHz
System bus 1333MHz (333MHz
quad-pumped)

Motherboard


Asus P5K3 Deluxe


Asus P5E3 Deluxe
Gigabyte X48T-DQ6
Bios revision 0604 0201 F4D

North bridge
Intel P35 Express Intel X38 Express Intel X48 Express

South bridge
Intel ICH9R Intel ICH9R Intel ICH9R
Chipset drivers Chipset 8.3.1.1009

AHCI 7.8.0.1012

Chipset 8.3.1.1009

AHCI 7.8.0.1012

Chipset 8.3.1.1009

AHCI 7.8.0.1012

Memory size 2GB (2 DIMMs) 2GB (2 DIMMs) 2GB (2 DIMMs)

Memory type


Corsair CM3X1024-1600C7DHX DDR3 SDRAM
at 1333MHz
CAS latency
(CL)
7 7 7
RAS to CAS
delay (tRCD)
7 7 7
RAS precharge
(tRP)
7 7 7
Cycle time
(tRAS)
21 21 21
Command rate 1T 1T 1T

Audio codec
Integrated

AD1988B with 5.10.1.6110 drivers

Integrated AD1988B with
7.0.0.0 drivers
Integrated ALC889A with 1.88
drivers
Graphics

GeForce 8800 GT 1GB PCIe
with ForceWare 169.25 drivers
Hard drive
Western Raptor X 150GB
OS

Windows Vista Ultimate x86
with KB936710, KB938194, KB938979, KB940105
updates

Thanks to Corsair for providing us with memory for our testing. 2GB of RAM seems to be the new standard for most folks, and Corsair hooked us up with some of its 1GB DIMMs for testing.

All of our test systems were powered by OCZ GameXStream 700W power supply units. Thanks to OCZ for providing these units for our use in testing.

Finally, we’d like to thank Western Digital for sending Raptor WD1500ADFD hard drives for our test rigs. The Raptor’s still the fastest all-around drive on the market, and the only 10K-RPM Serial ATA drive you can buy.

We used the following versions of our test applications:

The test systems’ Windows desktop was set at 1280×1024 in 32-bit color at an 85Hz screen refresh rate. Vertical refresh sync (vsync) was disabled for all tests.

All the tests and methods we employed are publicly available and reproducible. If you have questions about our methods, hit our forums to talk with us about them.

Memory performance

With the same DDR3 DIMMs running at 1333MHz with 7-7-7-20 timings, we see little difference in memory performance between Intel’s top three chipsets. The X48 and X38 are all but even here, with the P35 only pulling up a little short in the bandwidth test.

Memory controllers don’t always handle four DIMMs gracefully, so we popped an additional two memory modules into each system for another round of tests.

Four DIMMs certainly don’t favor the X48. Not only does Intel’s flagship deliver the lowest bandwidth scores and highest access latencies here, the budget P35 Express vaults into the lead. Gigabyte is still tweaking the X48’s BIOS, and this looks like one area where performance could definitely be improved.

The following latency graphs are a little indulgent, so I won’t be offended if you skip them. They show access latencies across multiple block and step sizes, painting a fuller picture of memory controller performance with each chipset. I’ve arranged the graphs in order of highest latency to lowest, taking results from testing with two DIMMs. Yellow represents L1 cache, light orange is L2, and dark orange is main memory.

While the P35 is clearly slower than its high-end counterparts when accessing larger blocks with a step size of 128, there’s little difference between the X38 and X48 here.

STARS Euler3d computational fluid dynamics

Few folks run fluid dynamics simulations on their desktops, but we’ve found this multi-threaded test to be particularly demanding of memory subsystems, making it a good link between our memory and application performance tests.

Score one for the X48, which narrowly edges out the P35 Express in this test. The X38 is a little slower still, although scores for all three are quite close.

WorldBench

WorldBench uses scripting to step through a series of tasks in common Windows applications. It then produces an overall score. WorldBench also spits out individual results for its component application tests, allowing us to compare performance in each. We’ll look at the overall score, and then we’ll show individual application results alongside the results from some of our own application tests.

Only three points separate our chipsets in WorldBench’s overall score, but the X48 Express finds itself at the bottom of the pile. Let’s see if we can find out why by breaking down the overall score into bite-sized application tests.

The X48 actually comes out ahead in two of WorldBench’s three multimedia editing and encoding tests, although never by more than a couple of seconds.

Scores are also tight through WorldBench’s office and multitasking tests.

WorldBench’s 3dsmax tests can’t coax much of a performance difference out of the chipsets, either.

Nero and WinZip do, however. In the former, the X48 finds itself at the back of the pack by a good 30 seconds. Things only get worse in WinZip, where the X48 is a full 50 seconds off the pace set by the X38 Express.

We’ve seen Vista’s (supposedly) intelligent disk caching mechanism affect WorldBench performance in these last two application tests, so I wouldn’t put much stock in the X48 actually being slower than the X38 and P35 here.

Gaming

Even at relatively low resolutions, we see little difference in game performance between the X48 and Intel’s other chipsets. The X48 does come out on top across the board, but only by the slimmest of margins in most tests.

Power consumption

We measured system power consumption, sans monitor and speakers, at the wall outlet using a Watts Up Pro power meter. Power consumption was measured at idle and under a load consisting of a multi-threaded Cinebench 10 render running in parallel with the “rthdribl” high dynamic range lighting demo. We also ran a set of tests on the Gigabyte X48 board with its Dynamic Energy Saver software activated.

The X48 Express isn’t much more power-hungry than its predecessor. Gigabyte’s Dynamic Energy Saver helps a little, too, although its effects are more pronounced under load than they are at idle. Interestingly, the DES application indicated that our test system was only making use of eight of 12 power phases when under load.

Don’t pay too much attention to the P35 Express numbers here. The Asus P5K3 Deluxe motherboard we used for testing is known for having higher power consumption than other P35 boards.

Overclocking

For our overclocking tests, we dropped our CPU multiplier to 6X—its lowest possible value. Next, we dropped the memory bus multiplier to 2X to keep our DDR3 modules from limiting front-side bus speeds. With that set, our attention turned to cranking the front-side bus, which we did in 10MHz increments, testing stability with a combined load of Prime95 and the rthdribl HDR lighting demo along the way.

Overclocking the X48T-DQ6 proved easy until we hit around 440MHz. For whatever reason, the board just wasn’t stable between that speed and 470MHz, where no amount of fiddling with chipset, front-side bus, or even processor voltages would get it through our stability test. However, the X48T was rock solid with default voltages when we pushed the FSB to 480MHz and then 500MHz, where we finally hit the wall.

A 500MHz (2000MHz with quad-pumping taken into account) front-side bus is hardly a new milestone for Core 2 overclocking—X38, P35, and even nForce boards have been playing in that range for a while now—but it’s still fast enough for all but the most extreme overclocking exploits.

As is always the case with overclocking, your mileage may vary.

Motherboard peripheral performance

Core logic chipsets integrate a wealth of peripherals, but they don’t handle everything. Firewire and audio are farmed out to auxiliary chips, for example. Intel chipsets also rely on third-party silicon for networking, and many motherboards feature additional SATA controllers to complement south bridge Serial ATA offerings.

To provide a closer look at the peripheral performance you can expect from the motherboards we’ve tested today, we’ve complied Ethernet, USB, Firewire, Serial ATA, and Audio performance results below. We’ve even thrown in some PCI and PCIe performance tests. However, since the P35, X38, and X48 chipsets all use the same ICH9R south bridge, don’t expect meaningful performance differences between the boards in tests that probe the south bridge alone.

NTttcp Ethernet
performance
Throughput (Mbps) CPU utilization (%)

Asus P5E3 Deluxe (88E8056)
938.775 14.60

Asus P5E3 Deluxe (RTL8169)
378.837 39.58

Asus P5K3 Deluxe (88E8056)
937.432 15.21

Asus P5K3 Deluxe (RTL8169)
361.517 37.80

Gigabyte X48T-DQ6 (1)
930.312 46.65

Gigabyte X48T-DQ6 (2)
931.402 46.61

The X48T-DQ6’s onboard Gigabit Ethernet controllers provide plenty of throughput, but CPU utilization is higher than we’d like to see. Note that Asus boards have particularly low throughput with their PCI-based RTL8169 GigE chipsets. We’ve seen higher scores from those boards before, suggesting that something in our latest test system configuration is holding back performance here.

PCI Ethernet
performance
Throughput (Mbps) CPU utilization (%)

Asus P5E3 Deluxe
89.342 2.90

Asus P5K3 Deluxe
85.299 3.28

Gigabyte X48T-DQ6
78.194 3.28

Slow PCI Gigabit Ethernet performance persists when we toss a discrete networking card into each system. Throughput is positively dismal here, and we observed similarly low scores when recently testing Intel’s G35 chipset. We’ve also seen much higher PCI throughput with the Asus boards in previous reviews.

Since lower PCI Ethernet throughput seems to be tied to something in our latest system configurations, I wouldn’t count it against any of the boards here. We’ve contacted Intel about the issue and they’re looking into it.

PCI Express Ethernet
performance
Throughput (Mbps) CPU utilization (%)

Asus P5E3 Deluxe
938.953 19.57

Asus P5K3 Deluxe
937.762 18.61

Gigabyte X48T-DQ6
927.387 18.16

There’s certainly nothing wrong with PCI Express throughput.

HD Tach USB
performance

Read burst

speed (MB/s)


Average read

speed (MB/s)


Average write

speed (MB/s)


CPU utilization

(%)


Asus P5E3 Deluxe
33.9 32.5 29.9 4.3

Asus P5K3 Deluxe
33.9 32.6 25.9 4.0

Gigabyte X48T-DQ6
33.9 32.6 30.1 5.7

USB performance looks pretty good, too. Curiously, though, the P5K3 Deluxe is a little slower than the others here, despite the fact that it shares the same ICH9R USB controller.

HD Tach
Firewire performance

Read burst

speed (MB/s)


Average read

speed (MB/s)


Average write

speed (MB/s)


CPU utilization

(%)


Asus P5E3 Deluxe
42.3 37.5 28.8 1.7

Asus P5K3 Deluxe
42.3 37.4 28.7 0.7

Gigabyte X48T-DQ6
42.1 37.5 28.7 2.3

None of these boards skimp in the Firewire department, so we have excellent performance across the board.

HD Tach Serial
ATA performance

Read

burst

speed

(MB/s)


Average

read

speed

(MB/s)


Average

write speed

(MB/s)


CPU

utilization

(%)


Random

access

time

(ms)


Asus P5E3 Deluxe
133.5 75.2 101.0 2.0 8.4

Asus P5K3 Deluxe
133.7 75.2 98.7 3.7 8.3

Gigabyte X48T-DQ6 (ICH9R)
133.2 75.2 94.5 2.7 8.3

Gigabyte X48T-DQ6 (JMB363)
129.5 75.2 47.4 5.3 8.1

Write speeds are a little slow for the X48T’s JMicron storage controller, but that may be because it doesn’t employ Native Command Queuing. The relatively high scores we see in HD Tach’s write speed test seem to only occur with storage controllers that implement NCQ.

RightMark Audio
Analyzer audio quality

Overall score

Frequency response

Noise level

Dynamic range

THD

THD + Noise

IMD + Noise

Stereo Crosstalk

IMD at 10kHz

Asus P5E3 Deluxe
4 5 3 3 3 1 3 4 3

Asus P5K3 Deluxe
4 5 3 1 3 1 3 4 3

Gigabyte X48T-DQ6
4 5 3 3 3 1 3 4 3

Nothing to see here. Move along.

Conclusions

The X48 Express is a solid addition to Intel’s chipset lineup, and a necessary one given the upcoming Core 2 Extreme QX9770’s 1600MHz front-side bus. However, apart from the faster system bus, the X48 does little to differentiate itself from Intel’s existing X38 Express. Unless you’re looking to run a processor that demands a 1600MHz FSB or want a guaranteed front-side bus overclock to that speed, we see little incentive to opt for the X48 over Intel’s other chipsets. For all practical purposes, the X38 and P35 Express are every bit as fast as the X48, at least when running on a 1333MHz front-side bus.

One thing that the X48 will bring to the table, or rather the market, is a new batch of high-end LGA775 motherboards. Don’t bet on seeing many designs that differ radically from current X38 boards, though. Gigabyte’s X48T-DQ6, for example, is quite similar to the company’s X38 offerings. That’s a great foundation to work with, and the result is a quality motherboard rich in features and BIOS options, but there isn’t much in the way of new hotness.

At least Gigabyte has rolled out something innovative with its Dynamic Energy Saver tech, which through the magic of dynamically processor power phase scaling, is capable of lowering power consumption at idle and particularly under load. DES isn’t exclusive to the X48T, though; you’ll be able to get it on a whole slew of Gigabyte boards, including less expensive mid-range models.

It’s those affordable mid-range models that really ruin things for not only the X48 Express, but also the X48T-DQ6. Gigabyte’s suggested retail price for the board is $300, and we’d expect most X48 boards to ring up in the same range. That’s double the cost of P35 offerings that boast comparable performance, and based on our own experiences, have no problems overclocking to 1600MHz front-side bus speeds and beyond.

For enthusiasts in search of the sweet spot, the X48 Express misses the mark. However, it’s a fine chipset and a fitting flagship for upcoming processors that require a 1600MHz front-side bus.

Comments closed
    • lucas1985
    • 12 years ago

    Core logic development has been stagnant since the .. umm … nForce4 (PCI-E for AMD, single-chip design, SLI, 10 USB ports, integrated Gigabit NIC with acceleration -total failure-, etc) with some exceptions (the latest IGP from AMD and the embedded/low-power market)

    • Hattig
    • 12 years ago

    Nice photos.

    • Usacomp2k3
    • 12 years ago

    Have you guys gotten SP1 yet?

      • oldDummy
      • 12 years ago

      Yes and it improves file transfers quite a bit.

      Regarding X48: I’m running my Q6700 at 1600 with a G33 Chipset.

      Whats the big deal?

        • Forge
        • 12 years ago

        Reading comprehension is such a dying art.

        X48 isn’t the first chipset ever to do 1600 FSB. It’s not even the first Intel chipset to do it.

        It *is* the first chipset to do it officially and at a low/stock voltage, and with tight/stock timings.

        Really, is this so hard to understand? An e6600 at 1333FSB is not an E6850. The FSB/clocks are the same, but a hypothetical average E6600 will need an above-stock cooler or above-stock voltage to do it, sometimes both. Often the E6850 in this example would have more headroom due to the lower starting temps/volts, but that’s not always the case.

        P.S. Don’t start talking about your E6600 doing 400FSB and stock volts, this is a METAPHOR, not an EXAMPLE.

          • d0g_p00p
          • 12 years ago

          Thank you.

          • Krogoth
          • 12 years ago

          Nope, X48 is just a revision of X38. It is a lot like how Nforce 2 Ultra (officially handled 400Mhz FSB) was to Nforce 2 (could unofficially do 400Mhz FSB).

          A lot of the X38 motherboards on the market already offer “unofficial” 1600Mhz FSB support without requiring voltage boosts.

          X48 is purely a marketing product to concise with launch of XE Yorkfields.

            • flip-mode
            • 12 years ago

            “coincide”

    • Hance
    • 12 years ago

    Basically another Ho Hum chipset. My old Gigabyte GA965P-DS3 will do 1600 FSB so the X48 has basically nothing to offer me.

      • toxent
      • 12 years ago

      Lucky, i still can’t get mine past 350 MHz.

      I even bought a new cooler for it. 🙁

      • rosselt
      • 12 years ago

      hell my old 975x did 400fsb (1600)
      it did have trouble with all four DIMM slots full though and i doubt it could
      of pushed a quad that high ethier.
      my maximus formula (x38) handles all four DIMM slots at either
      ddr2-800 3-3-3-8 or ddr2-1066 4-4-4-12 like a champ
      i found out yesterday i can just flash a rampage formula (x48) BIOS
      onto my MF and get all the new nice MCH timing options too! suffer in your jocks ASUS!

      the only thing x48 has over x38 though is its validated for 400fsb @ 1.25v
      and it might handle tighter MCH timings over x38 at lower voltage levels
      a Anadtech Article shows thier x48 sample scaling higher with less volts.
      really not worth going out and getting another board for it all though

    • jwb
    • 12 years ago

    When are you guys going to start measuring CPU utilization normalized by performance? It should be in CPU per Gbps for ethernet, CPU per MBps for SATA and FireWire, etc. I don’t care if chip X has triple the CPU usage while going ten times faster. In fact, that seems desirable. The way you present it now, I have to do the math in my head.

      • odizzido
      • 12 years ago

      I have been thinking the same thing. It’s been a problem for a good while now.

      • Usacomp2k3
      • 12 years ago

      Meh. I enjoy the fact that I can read and don’t need them to do it for me 😉

      • Saber Cherry
      • 12 years ago

      /agree

      I would like to see all throughput numbers in CPU-nanoseconds per byte: (0.000000001*bytes)/(cpu time used). The raw average CPU usage is usually irrelevant.

    • bdwilcox
    • 12 years ago

    $300 just for a 1600MHz FSB? Looks like Intel is acting like AMD doesn’t exist anymore. Can’t wait till AMD ramps the Phenom’s speeds and combines them with cheap AMD chipset based boards. Hopefully it catches Intel with their pants down, like the Athlon64 did once before. Like Microsoft, Intel’s biggest competitor is it’s own arrogance.

      • BoBzeBuilder
      • 12 years ago

      Almost everyone is acting like AMD doesn’t exist. Intel, nVidia, even the consumers. HA.

      • dragmor
      • 12 years ago

      The DQ6 has always been expensive and your dollar is dropping. The 965DQ6 was $300 Aus vs $200 Aus for the DS3. The new X48 board is priced the same Aus, its just that your dollar has dropped from 1US = 0.8 Aus to 1US = 0.95 Aus.

    • Fastfreak39
    • 12 years ago

    wow the TR crew is really cranking out those articles lately, loving it of course.

    • UberGerbil
    • 12 years ago

    The combination of a blue board with copper heatsinks may look fantastic, but the combination of hot pink and electric green on the DIMM slots is anything but “easy on the eyes.”

    Are those really little piston icons in their monitoring app?

      • Mithent
      • 12 years ago

      Look like it to me. I was wondering if they were animated.

    • flip-mode
    • 12 years ago

    I wonder what’s up with ICH9R’s gig-e going completely unused by manufacturers. Is there some untold story there?

      • deruberhanyok
      • 12 years ago

      I always scratch my head when I see that, too. It’s like… hello, it’s there already, is it really cheaper to put an entirely seperate NIC on the motherboard than it is to utilize that?

        • Saber Cherry
        • 12 years ago

        Similarly, why is an additional Realtek chip used for audio instead of the on-die version? Especially considering Intel *[

          • crazybus
          • 12 years ago

          The onboard HDA audio controller is used, in both cases. The Realtek chips on most motherboards are CODECs, which as the name suggests, handle the digital to analog conversion and vice versa, along with s/pdif ouput and basically anything on the I/O side of the audio setup. They function in conjunction with the audio controller on the chipset, whether it be Intel, nVidia, or AMD. The driver interface is usually implemented by the CODEC provider, although it is sometimes possibly to use generic drivers sourced from the chipset manufacturer (e.g. nVidia has provided AC’97 drivers which work with their AC’97 supporting chipsets regardless of codec manufacturer).

            • Saber Cherry
            • 12 years ago

            Thanks for the links. I have not had a chance to read them yet, though I intend to. However, your response has left me even more confused.

            1) CODEC == Digital to Analog Converter? Where I come from, that’s called a DAC, and it is too small, cheap, and simple to actually have multiple versions that cost nontrivially different amounts of money and need drivers.

            2) If the additional (off-Southbridge) sound chip was actually just a DAC, it would not be possible for it to incorrectly handle EAX, which is (a very limited form of) 3D-aware sound modeling done at least partly in the digital domain, and therefore must be aware of things other than just an incoming audio data stream… A DAC would never deal with EAX in the first place.

            On the other hand, I went here:

            §[<http://en.wikipedia.org/wiki/Digital-to-analogue_converter<]§ ..and nowhere on the page do they call a DAC a codec. But the picture in the upper-right shows a Cirrus Logic DAC integrated onto a Creative X-Fi, and it looks just like the sort of chip on a motherboard that is called the "audio codec". So as I said, I'm still confused about exactly how the responsibilities are split between the Southbridge and additional chip, how it is possible to screw one up if it is mainly just a DAC, and why they need drivers, though again, I still need to read those articles. Last night, after reading your post, I tried to install Via's Vinyl Audio drivers that supposedly support my Southbridge instead of the codec provider's drivers. Well... the setup program hung for a couple minutes, then gave an error message with a error code return value of -2 billion. So I installed them manually from the inf file, which seemed to be working well, until... I got my first ever blue-screen-of-death in Windows XP that was not a result of catastrophic hardware failure! That really messed up my computer (worse than expected of Via), disabling my keyboard and both optical drives, and causing HDD data corruption in over 20 Windows system32 files, which was fortunately caught by the automatic scandisk on reboot that I could not skip because my keyboard was disabled. But I will take your word for it that SOMETIMES the chipset provider's generic drivers will work :)

            • crazybus
            • 12 years ago

            Hmm….I guess generic drivers don’t always work. The last two times I’ve tried them they have (nVidia and VIA chipsets with Realtek codecs), but with limited functionality.

            Regarding the whole DAC = codec thing, well of course a codec is more than just a DAC, as they also feature an ADC, a low power amplifier and the hardware for digital audio I/O. Besides that, they have very little in the way of logical capability. Realtek has their codec’s datasheets available on their website with documentation of all the commands the codec can receive. It’s not all that extensive. All of the audio processing such as EAX, AC3/DTS encoding, etc. is done in software so the actual codec hardware has little to do with that.

            I imagine the only reason codec providers differentiate between the version of their products that offer real-time AC3/DTS encoding is because of licensing fees.

            If you’re interested, here’s the link to Intel’s HD Audio specs:
            §[<http://download.intel.com/standards/hdaudio/pdf/HDAudio_03.pdf<]§ I'll warn you that it's a bit of a dry read.

      • ssidbroadcast
      • 12 years ago

      Intel”Secret”Inside™ !

    • donkeycrock
    • 12 years ago

    blaa blaa blaa, hoe hum. NEXT

    • Jive
    • 12 years ago

    Im glad to see high quality components in a mobo but don’t think i would pay more for such little difference in power consumption.

    • Krogoth
    • 12 years ago

    Didn’t read the article yet.

    In a nutshell, X48 = X38 Revision 1.1 with “official” 1600Mhz FSB support.

    Edit: Read article and it is confirms what I knew about X48.

      • d0g_p00p
      • 12 years ago

      Look at the big brain on Krogoth.

Pin It on Pinterest

Share This