Then Shuttle shifted its attention to selling complete systems, reducing what was once a torrent of new barebones designs to little more than a trickle. Shuttle’s competitors didn’t exactly pick up the slack, either. In fact, many retreated from the small form factor barebones market altogether, leaving enthusiasts with few reasonable alternatives.
Fortunately, we’re still treated to a new barebones XPC every so often. The latest such box is the SD39P2, which supports Intel’s latest quad-core processors, up to 8GB of memory, two hard drives, and both PCI Express and PCI expansion cardsin a form factor the size of a toaster. The question, of course, is whether this latest toaster has the performance, features, and attention to detail that made Shuttle king of the small form factor. Read on to see what we found.
There’s only so much one can squeeze into a form factor that measures just 325mm long, 210mm wide, and 220mm tall (12.8″ long, 8.3″ wide, and 8.7″ tall, if 32 degrees seems to you like a good temperature for water to freeze), but the SD39P2’s spec sheet is surprisingly loaded.
|CPU support||LGA775-based Celeron, Pentium 4/D, Core 2 processors|
|North bridge||Intel 975X Express|
|South bridge||Intel ICH7R|
|Expansion slots||1 PCI Express x16
1 32-bit/33MHz PCI
|Memory||4 240-pin DIMM sockets
Maximum of 8GB of DDR2-533/667/800 SDRAM
|Storage I/O||Floppy disk
1 channels ATA/133
3 channels Serial ATA with RAID 0, 1, 10, 5 support
|Audio||8-channel HD audio via ICH7R and Realtek ALC888 codec|
6 USB 2.0 (rear) with headers for 2 more
2 USB 2.0 (front)
1 1394a Firewire via VIA VT6308 (front)
1 1394a Firewire via VIA VT6308 (rear)
1 RJ45 10/100/1000 via Broadcom BCM5789
1 eSATA 1 analog front out
1 analog bass/center out
1 analog rear out
1 analog surround out
1 analog line in
1 analog mic in (front)
1 analog headphone out (front)
1 coaxial digital S/PDIF output
1 TOS-Link digital S/PDIF output
1 TOS-Link digital S/PDIF input
|Bus speeds||FSB: 133-400MHz in 1MHz increments
DRAM: 533, 667, 800MHz
PCIe: sync, locked at 100MHz
CPU: +0.025-0.8V in 0.025V increments
DRAM: 2.0-2.4V in 0.2V increments
|Monitoring||Voltage, fan status, and temperature monitoring|
|Fan speed control||CPU, system|
You might be surprised to see Intel’s 975X Express chipset in Shuttle’s latest XPC. After all, the 975X is far from Intel’s newest core logic chipset. The 975X is actually a pretty good match for the SD39P2, thoughperhaps even a better match than the P965.
First, the 975X is every bit as fast as the P965. The P965 may have a fancy new memory controller, but it doesn’t actually deliver greater bandwidth or lower latency than the 975X. There is virtually no difference in application performance between the two chipsets, either.
The P965’s real advantages over the 975X don’t become apparent until we consider Intel’s new ICH8R south bridge, whose six Serial ATA and 10 USB ports trump the ICH7R’s four and eight. Because the SD39P2 uses the older ICH7R, its port selection is a little more limited. A couple of extra USB ports would be nice, I’ll admit, but eight should be plenty for most folks. There wouldn’t be much point in equipping an XPC limited to just three internal drive bays with six Serial ATA ports, either; the ICH7R’s four SATA ports are more than up to the task. The fact that the ICH7R retains an ATA controller is a bonus for the SD39P2, as well.
The SD39P2 does need a few third-party chips to augment the chipset. Networking, audio, and Firewire are handled by auxiliary silicon from Broadcom, Realtek, and VIA, respectively.
Shuttle has been refining its P-series chassis for years now, and it’s dangerously close to perfection for a form factor of this size.
The faceplate just doesn’t do it for me, though. Yes, it’s sleek, and even stylish. However, mixing flat, brushed, and polished plastics looks a little muddled to me, especially when nothing exactly matches the aluminum skin that covers the rest of the enclosure.
At least the design isn’t vulgar, so it should have pretty broad appeal. Shuttle’s even exercised a measure of restraint with the polished bits so that the entire face doesn’t turn into a giant fingerprint magnet.
Since end users can’t be relied on not to break the XPC’s all-black dress code, Shuttle hides the system’s external 3.5″ and 5.25″ drive bays behind a couple of stealthy doors. The 5.25″ drive bay door is spring-loaded, opening and closing automatically with tray-loading optical drives. You’ll actually have to open the 3.5″ drive bay door manually if you want to get at a floppy drive or memory card reader, though.
Along the bottom edge of the XPC we find another hinged door that hides the system’s front-mounted port array. You get headphone and microphone jacks up front in addition to one Firewire and two USB ports. The reset button is also tucked away over to the left, although it’s recessed enough that you’ll probably need a pin, pen, or mangled paper clip to press it.
The rest of the SD39P2’s ports can be found at the rear, where there’s no shortage of cutting-edge connectivity. In addition to the requisite array of USB, Ethernet, and Firewire ports, you also get an eSATA port for external Serial ATA devices. Shuttle has gone all out in the audio department, too, outfitting the SD39P2 with a full array of analog input and output ports in addition to TOS-Link digital S/PDIF input and output ports, and an additional coaxial S/PDIF output.
Along with the regulars, the SD39P2 also features a handy recessed CMOS reset button at the rear. This will save you from having to pop the case open and to reset the CMOS jumper manually in the event that BIOS fiddling or overclocking renders the system unable to post.
With all these goodies at the rear, it’s almost easy to forget that the SD39P2 is devoid of “legacy” PS/2, serial, and parallel ports. The absence of the latter two isn’t likely to cause a stir, but there are some who stubbornly cling to old-school PS/2 peripherals or KVM switches. Perhaps it’s time to upgrade, folks.
Pulling off a sleek form factor exterior is relatively easy, but making everything work inside the chassis is considerably more difficult. This is where Shuttle’s experience really shines, and so it should, since Shuttle has now made a business of building complete systems with its chassis.
A handful of thumbscrews is all that holds the SD39P2’s aluminum skin in place, making it relatively easy to get at the guts of the beast. Even with just the outer skin removed, the system’s internals are remarkably accessible.
Up top, we find a couple of 3.5″ hard drive bays held in place by standard screws. Shuttle’s older P-series designs had largely tool-free internals that made liberal use of plastic struts and clips to hold everything in place. It was a pretty slick setup, but one that Shuttle ultimately abandoned in favor of a more traditionaland probably more durabledesign. I’m not even sure I prefer one configuration over the other; the older plastic clips made installation a snap, but they certainly didn’t hold drives as tightly as the SD39P2’s metal cages and screws.
Around the left side of the system, we have relatively easy access to all four of the SD39P2’s DIMM slots. Amazingly, there’s quite a bit of clearance for taller memory modules. Corsair’s XMS Pro DIMMs fit easily, and you can even squeeze the company’s taller Dominator modules into the system without removing additional components.
From this angle we can also see that Shuttle has gone with passive chipset cooling for the 975X Express. The P2 chassis has a generous amount of internal airflow, so there’s little need to worry about the chipset overheating.
Turning our attention to the other side of the system, we find the SD39P2’s PCI and PCI Express expansion slots. Shuttle’s first stab at a Core 2-compatible cube featured dual PCIe x16 slots and support for CrossFire multi-GPU configurations, but this latest attempt opts for a more sensible combination of x16 and standard PCI slots. The XPC chassis only has room for two expansion slots, and you’re generally better off with a single double-wide graphics card than a pair of single-slot cards in CrossFire. It’s nice to have a PCI expansion option, too, if only because sound cards remain wedded, tragically, to the PCI bus.
In case you’re wondering, it is possible to squeeze a GeForce 8800 GTS into the SD39P2 with relatively little drama. Shoehorning the larger GTX into the, er, shoebox is a little more complicated, particularly because the system only has one six-pin PCIe power connector; the GTX requires two. Incidentally, the SD39P2’s power supply is a 400W unit rated for a peak output of 450W.
Sliding any double-wide graphics card into an XPC can be a little tricky, and it’s common for the bottom tips of the PCI back plate to make contact with the motherboard. Shuttle has you covered, though; a plastic cover protects sensitive motherboard components between the expansion slots and PCI backplane.
Getting at the SD39P2’s DIMM and expansion slots is relatively easy with just the outer skin removed, but to access the CPU socket and internal drive bays, we have to pop the XPC’s top. Out come the hard drive trays, which hold the entire upper drive cage in place, giving us a much better look at the system’s LGA775 socket.
With the SD39P2’s drive bays removed, it’s easy to access the system’s CPU socket. Note that Shuttle has equipped the system’s VRMs with a nice, chunky heatsink. Air flows from left to right in P-series chassis (top to bottom in the picture), so that VRM heatsink will get a little extra airflow, as well.
Cooling components in the cramped confines of a small form factor chassis isn’t easy, but Shuttle has nailed it with the P-series. The system has three distinct cooling zones; one for the hard drives, one for the processor, and another for the power supply and system internals. All fans are variable speed units, and with the Core 2 Duo’s modest heat output, that makes the SD39P2 one of the quietest small form factor systems we’ve ever tested. Our XPC test rig was every bit as quiet as my personal workstationa system I’m pretty anal about keeping as silent as possible.
An intricate heatpipe cooler is at least partially responsible for the SD39P2’s low noise levels. Shuttle has been using a similar design for years now, and it’s proven quite effective. In fact, this latest version actually shaves off some of the radiator bulk, presumably because it just isn’t needed to keep a Core 2 processor cool.
As you can see, the cooler comes equipped with a four-pin fan header that allows smooth fan speed ramping based on CPU temperaturesno annoying oscillations between high and low fan speeds here. The cooler screws directly into the motherboard, and screws are of fixed length, so it’s impossible to over-tighten them.
Turning the cooler on its side reveals a copper base plate that leads directly to four heatpipes. The basic shape of this cooler can actually be traced to Shuttle’s original G-series designs, which were introduced way back in 2002.
After gushing over the SD39P2’s internals for a while, you’ll eventually want to actually assemble the system. Installation is a breeze, in part because you’ve already had to remove the drive bays to get at the CPU socket.
Screwing optical drives and hard disks into the removable drive takes just minutes, and everything falls into place from there. Installation isn’t quite as quick as with Shuttle’s older tool-free internals, but since you’ve always had to bust out a screwdriver to remove the CPU heatsink and expansion cards, it’s not that inconvenient to have to use one on the drive bays.
Shuttle has all but perfected the small form factor chassis with the SD39P2, but can the BIOS deliver as well?
Not so much.
The BIOS is certainly well organized, with all the overclocking and tweaking options available on a single page. But they only all fit on one page because there’s relatively few of them.
Front-side bus speed options are only available up to 400MHz, which isn’t a good sign for those looking to really push a Core 2 Duo. Memory bus speed options are limited to 533, 667, and 800MHz, as well. At least you’re able to lock the PCI Express clock to 100MHz.
Things don’t get any better on the voltage front, where you can only increase the CPU voltage by up to 0.8V. DRAM voltages are available up to 2.4V, but only in 0.2V increments. Chipset voltages, on the other hand, are nowhere to be found.
So the SD39P2’s BIOS isn’t well-equipped for overclocking. Shuttle does, however, provide a decent amount of control over the system fan speeds and even the brightness of the blue power-on LED.
Users are presented with four different system fan profiles and a whopping six fan profiles for the CPU cooler, including an unfortunately-named “eXtreme PC” mode. However, the BIOS doesn’t allow users to define fan speed profiles and target temperatures on their own. That functionality is only available through Shuttle’s XPC Tools Windows software, but it doesn’t yet support the SD39P2. Shuttle says a new version of XPC Tools with SD39P2 support should be ready in a few weeks.
We’re comparing the performance of the XPC SD39P2 to that of a collection of Core 2 motherboards, including Asus’ P5N-E SLI and Striker Extreme, DFI’s LANParty UT ICFX3200-T2R/G, and EVGA motherboards based on Nvidia’s nForce 680i SLI and 680i LT SLI reference designs.
Although we had no problems running a 1T command rate on the Asus and EVGA boards, the DFI wasn’t stable at 1T, so we had to back it off to a 2T command rate for testing. Command rate control isn’t available on the XPC SD39P2. In fact, we’ve yet to see command rate control exposed on motherboards with Intel chipsets.
All tests were run at least twice, and their results were averaged, using the following test systems.
|Processor||Core 2 Duo E6700 2.67GHz|
|System bus||1066MHz (266MHz quad-pumped)|
|Motherboard||Asus P5N-E SLI||Asus Striker Extreme||EVGA 122-CK-NF68||EVGA 122-CK-NF67||DFI LANParty UT ICFX320-T2R/G||Shuttle XPC SD39P2|
|North bridge||Nvidia nForce 650i SLI SPP||Nvidia nForce 680i SLI SPP||Nvidia nForce 680i SLI SPP||Nvidia nForce 680i LT SLI SPP||AMD RD600||Intel 975X Express|
|South bridge||Nvidia nForce 430i SLI MCP||Nvidia nForce 680i SLI MCP||Nvidia nForce 680i SLI MCP||Nvidia nForce 680i LT SLI MCP||AMD SB600||Intel ICH7R|
|Chipset drivers||ForceWare 8.26||ForceWare 9.53||ForceWare 9.53||ForceWare 9.53||Catalyst 7.2||Chipset 188.8.131.520
|Memory size||2GB (2 DIMMs)||2GB (2 DIMMs)||2GB (2 DIMMs)||2GB (2 DIMMs)||2GB (2 DIMMs)||2GB (2 DIMMs)|
|Memory type||Corsair TWIN2X2048-8500C5 DDR2 SDRAM at 800MHz|
|CAS latency (CL)||4||4||4||4||4||4|
|RAS to CAS delay (tRCD)||4||4||4||4||4||4|
|RAS precharge (tRP)||4||4||4||4||4||4|
|Cycle time (tRAS)||12||12||12||12||12||12|
|Audio codec||Integrated nForce 430i/ALC883 with Realtek HD 1.57 drivers||Integrated nForce 680i/AD1988B with 184.108.40.20670 drivers||Integrated nForce 680i/ALC885 with Realtek HD 1.57 drivers||Integrated nForce 680i LT/ALC885 with Realtek HD 1.57 drivers||Integrated SB600/ALC885 with Realtek HD 1.57 drivers||Integrated ICH7R/ALC888 with Realtek HD 1.57 drivers|
|Graphics||GeForce 7900 GTX 512MB PCI-E with ForceWare 93.71 drivers|
|Hard drive||Western Digital Caviar RE2 400GB|
|OS||Windows XP Professional|
|OS updates||Service Pack 2|
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.
With the exception of the SD39P2, 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.
We used the following versions of our test applications:
- SiSoft Sandra Standard XI SP1
- WorldBench 5.0
- TCD Labs HD Tach v3.01
- Futuremark 3DMark06 Build 1.02
- Splinter Cell Chaos Theory v1.05
- RightMark Audio Analyzer 5.5
- RightMark 3D Sound 2.3
- Cinebench 9.5
- Sphinx 3.3
- Quake 4 1.30 with trq4demo1 demo
- F.E.A.R. 1.08
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.
The SD39P2 sits in the middle of the pack in our first round of memory subsystem tests, but things don’t get interesting until we pop a couple of extra DIMMs into the XPC.
Motherboards don’t always handle four-DIMM configurations well, so we also test motherboards and small form factor systems with four DIMMs installed. Unfortunately, the SD39P2 refused to post with four DIMMs at 800MHz, regardless of memory voltage, timings, or the DIMMs we used. Four DIMMs at 667MHz were perfectly stable, though, even with aggressive timings. We asked Shuttle about this issue, and they acknowledged that it’s a problem. They say they’re working on it, but don’t yet have an ETA for a fix.
The SD39P2 wasn’t the only platform that has trouble with four-DIMM configurations. We also had to slow the P5N-E’s DRAM command rate to 2T to get four DIMMs running stable on that board.
Scaling the SD39P2’s memory speed back to 667MHz puts the XPC at a disadvantage, and it has a hard time keeping up in our memory bandwidth and latency tests.
Of course, memory subsystem performance isn’t always indicative of what happens with real-world applications, as WorldBench shows. The SD39P2 is out ahead here, although only by a single point over EVGA’s nForce 680i SLI motherboard.
Scores are all but tied in our first round of gaming tests, which use relatively low resolutions and detail levels.
Our first round of gaming tests was conducted with lower in-game detail levels and display resolutions, but we’ve cranked things up for a second round. These tests use high resolutions, high detail levels, and anisotropic filtering and antialiasing, and we normally use them to tax a motherboard’s multi-GPU support. The SD39P2 only supports a single graphics card, but we’ve thrown in these results to illustrate what you lose by not being able to add a second card.
We’ve tested each board with a single GeForce 7900 GTX. The Asus and EVGA boards were also tested with two 7900 GTXs running in SLI. Since the ICFX3200 doesn’t support SLI, we tested it with a Radeon X1900 XTX in single-card and CrossFire configurations.
As one might expect, the XPC can’t keep up with multi-GPU CrossFire and SLI configurations. However, if we look only at results with a single graphics card, the SD39P2 is right in the thick of things.
Sphinx speech recognition
Cinebench scores are too close to call, but the SD39P2 scores a win in Sphinx. The XPC is just a hair faster than Asus’ Striker Extreme and P5N-E, but it easily distances itself from the EVGA and DFI boards.
Positional audio performance is largely dictated by audio codec drivers, and all but one of systems we tested has drivers provided by Realtek. Fear the crab. Or rather, fear Analog Devices. The Asus Striker Extreme is the only board to really differentiate itself here, and it’s sporting a codec chip from Analog Devices.
We should also note that some of Realtek’s HD audio drivers fail to correctly implement EAX occlusions and obstructions, rendering some games virtually unplayable with EAX effects enabled. That’s particularly damning for the SD39P2, whose tight confines make it impossible to run a discrete sound card alongside a double-wide graphics card.
We used an M-Audio Revolution 7.1 sound card for recording in RightMark’s audio quality tests. Analog output ports were used on all systems. To keep things simple, I’ve translated RightMark’s word-based quality scale to numbers. Higher scores reflect better audio quality, and the scale tops out at 6, which corresponds to an “Excellent” rating in RightMark.
The SD39P2’s performance in RightMark Audio Analyzer is unremarkable, but at least audio quality doesn’t vary much between the front and rear analog outputs. Thankfully, the XPC’s S/PDIF outputs make it easy to pass pristine audio bitstreams to compatible receivers or digital speakers.
ATA performance was tested with a Seagate Barracuda 7200.7 ATA/133 hard drive using HD Tach 3.01’s 8MB zone setting.
The SD39P2 is a hair slower in our ATA write speed test, but otherwise competitive with the rest of the field.
Serial ATA performance
Moving to Serial ATA, we tested performance with a Western Digital Raptor WD360GD SATA hard drive. Again, we used HD Tach 3.01’s 8MB zone test.
Our USB transfer speed tests were conducted with a USB 2.0/Firewire external hard drive enclosure connected to a Seagate Barracuda 7200.7 hard drive. We tested with HD Tach 3.01’s 8MB zone setting.
The SD39P2’s USB write speeds are disappointing, but read speeds and CPU utilization lead the pack.
Our Firewire transfer speed tests were conducted with the same external enclosure and hard drive as our USB transfer speed tests.
Write performance again proves problematic for the SD39P2. Firewire read speeds are competitive, though.
We evaluated Ethernet performance using the NTttcp tool from Microsoft’s Windows DDK. The docs say this program “provides the customer with a multi-threaded, asynchronous performance benchmark for measuring achievable data transfer rate.”
We used the following command line options on the server machine:
ntttcps -m 4,0,192.168.1.25 -a
..and the same basic thing on each of our test systems acting as clients:
ntttcpr -m 4,0,192.168.1.25 -a
Our server was a Windows XP Pro system based on Asus’ P5WD2 Premium motherboard with a Pentium 4 3.4GHz Extreme Edition (800MHz front-side bus, Hyper-Threading enabled) and PCI Express-attached Gigabit Ethernet. A crossover CAT6 cable was used to connect the server to each system. The boards were tested with jumbo frames disabled.
Broadcom is the business, apparently. The SD39P2’s BCM5789 Gigabit Ethernet controller has plenty of throughput and lower CPU utilization than GigE solutions available on competing enthusiast motherboards.
We measured system power consumption, sans monitor and speakers, at the wall outlet using a Watts Up power meter. Power consumption was measured at idle and under a load consisting of a multi-threaded Cinebench 9.5 render running in parallel with the “rthdribl” high dynamic range lighting demo.
Unfortunately, we couldn’t get the Core 2 Duo’s C1E enhanced halt state working properly on the SD39P2. There’s no C1E option in the BIOS, and the system refused to lower CPU clock speeds at idle unless we invoked SpeedStep via Windows’ minimal power management profile. We contacted Shuttle about the issue, and they’re looking into it, but it’s yet to be resolved. DFI’s ICFX3200-T2R/G experienced similar problems with the C1E enhanced halt state.
Since the SD39P2 uses a different power supply than the other systems we tested, these results aren’t entirely comparable. However, it’s worth noting that the XPC’s power draw is quite minimal when SpeedStep is enabled; you just have to invoke an OS power management profile that uses it.
For our overclocking tests, we swapped our Core 2 Duo E6700 engineering sample for a retail E6300 we’ve had up to 3.3GHz without extra voltage or cooling. System stability was tested with Prime95 and the rthdribl HDR lighting demo.
The fact that the SD39P2’s BIOS is limited to front-side bus speed options of only 400MHz didn’t prove to be a limiting factor for us. Our system was only stable up to front-side bus speeds of 360MHz, no matter how much we fiddled with CPU voltages or fan speeds. Since the XPC’s BIOS lacks control over the chipset voltage, there wasn’t much else we could do to coax higher speeds out of the XPC.
Even getting to 360MHz wasn’t easy. The SD39P2 apparently runs out of memory bus dividers, because with a 360MHz front-side bus, the memory was running at a whopping 451MHz (902MHz with DDR’s clock-doubling properties taken into account). Trying to force lower memory speeds in the BIOS didn’t work, and we actually had to swap in a set of Corsair’s Dominator DIMMs rated for operation at effective memory speeds up to 1200MHz just to get the system stable with a 360MHz front-side bus.
The XPC SD39P2 is potentially a BIOS update away from true greatness. On the chassis front, Shuttle has achieved near perfection; the SD39P2 has just the right mix of expansion options, effective and quiet cooling, a small footprint, and sleek aesthetics that should appeal to a broad audience. Hardware has always been Shuttle’s strength, and it seems that selling complete systems has only heightened the company’s attention to detail on that front.
Unfortunately, the SD39P2 has a few shortcomings that tarnish its enthusiast-class credentials. A lack of proper support for Intel processors’ C1E halt state, an inability to run four DIMMs at 800MHz, and a largely anemic array of BIOS-level overclocking options limit the SD39P2’s appeal. The C1E halt and DIMM issues are things that Shuttle should be able to fix with a BIOS update, and they should also be able to add a better array of memory bus dividers, more CPU and memory voltage options, and perhaps even the ability to adjust the chipset voltage. Shuttle really needs to address the SD39P2’s BIOS deficiencies to balance that side of the equation with what is an otherwise brilliant chassis.
The SD39P2 may never be a champion overclocker, and it doesn’t need to be, but things like C1E and four-DIMM compatibility at 800MHz ought to work out of the box. Right now, they don’t, and that makes this XPC hard to recommend unreservedly. Yes, you can get around the lack of C1E support by enabling SpeedStep. And yes, if you’re only running two DIMMs, you won’t have to throttle the memory bus back to 667MHz. Even at 667MHz, the SD39P2’s dual memory channels should provide enough bandwidth to saturate the Core 2 Duo’s 1066MHz front-side bus. If you’re willing to live with or work around these shortcomings, there’s much to recommend in the SD39P2. But we’d like to see Shuttle fix these problems.
If Shuttle is able to bring the BIOS up to scratch, the SD39P2 will become quite an attractive platform for those looking to build small form factor systems with Intel’s latest Core 2 processors. The fanfare surrounding SFF barebones boxes may have died down of late, but the SD39P2 has the potential to be a testament to what made them so attractive in the first place.