|
|
| Design, Layout and Installation | ||||
If beauty is in the eye of the beholder, then behold, the SB77G5, up close and personal. Design wise, the SB77G5 is about as efficient as any small form factor PC we've ever had in the lab here. Esthetically, we're also smitten with this XPC's obvious good looks. In our opinion the SB77G5 is sort of like the "Angelina Jolie" of SFF PCs. The first time you see her, she captures you attention and turns your head. As you get closer though, that's when bold-faced sex appeal hits you like a round-house kick upside your head... strictly speaking from a Tech geek's perspective of course.
The SB77G5's two drive bays and front I/O panel, are shrouded by retractable face plates. We installed a black Lite-On DVD-RW drive and a black 3.5" floppy (not that we really needed the floppy, we just wanted to complete the full build-out for you). The optical drive bay has chrome coated pushbutton that lines up inside the case with the eject button on most all standard CD or DVD ROM drives. In fact, we had to install the longer throw actuator pad, on the backside of this button mechanism, in order to properly reach the Lite-On DVD drive's button. Once we swapped out that plastic push pad, the drive tray slid out like a charm, pushing the bay door open and shutting it upon hitting the button for a second time. The backside I/O panel of the SB77G5 is fairly standard issue and it is color coded based on the connection types, as you can see in the shots above. However, drop that second serial port and give us a couple more USB ports on this I/O panel instead, any day. Dropping down inside the chassis, we see the SB77G5 has a larger, redesigned heat pipe cooler that has spring loaded thumb screws that provide excellent retention force down on the CPU socket area. The CPU block itself has a large copper slug in the mid section which definitely helps the thermal transfer coefficient of the entire "ICE" cooling system, as Shuttle likes to call it. The Northbridge has a fairly stout heatsink installed on it, as does the MOSFET power array on the backside of the board, which you can see if you look closely in one of the shots above. The power supply is a 250 Watt model that has a PFC circuit (Power Factor Correction). While its power output rating may seems a bit underwhelming, as you'll see in the pages ahead, it seems it can stand up to the heavy loads put forth by today's high end graphics cards. Finally, at the motherboard I/O chipset level, Broadcom's 5788 chip provides the Gig-E support, VIA's VT6703 provides IEEE 1394 (Firewire) support, and Realtek's ALC658 provides AC'97 5.1 Channel Surround Sound output. Incidentally, we tested the quality of this Realtek audio solution and while it certainly is not going to compete with the likes of an Audigy 2 or Envy24 chipset, it provides respectable quality surround sound for both gaming and digital video applications, with relatively low CPU utilization. A quick check with Right Mark's Audio Analyzer showed dynamic range, THD (total harmonic distortion) and frequency response to be in the "Good" to "Very Good" range, which is competitive with many discrete audio solutions on the market. |
| BIOS Menus and Overclocking Results | |||||||||||
The FB77 motherboard that the SB77G5 is built upon, is equipped with an Award BIOS version that we've all come to know and love over the years. All the features and tweaks you would expect to see in a full sized system implementation, can also be found in the SB77G5's menu options.
There's a good range of ability to tweak bus and interface speeds, as well as manipulate various voltage settings, should you want to try your hand at overclocking. CPU FSB settings of up to 355MHz should allow for "unofficial" support of Intel's latest 1066MHz (266X4) FSB Pentium 4 Extreme Edition processors and we'll show you more on that in the follow overclocking section here. There are also DDR266 (a 3:2 divider ratio) DDR320 (5:4) and DDR400 (1:1) settings, which allow the end user to manipulate FSB speeds to accommodate RAM speeds, keeping them within a reasonable range of their specs in conjunction with an overclock of the CPU. Voltages settings are available in .125 volt increments for the CPU and .1 volt increments for system memory, however we would have liked a bit more high end settings beyond 2.9V for the memory options, even though it is admittedly rare that voltage levels that high are required for stability. Then there's the "secret sauce" that Shuttle employs to keep the SB77G5 and many others in their XPC line-up cool and quiet during operation, "Smart Fan" and "CPU Temp Tag". Smart fan automatically adjusts fans speeds according to the CPU Temp Tag setting in the BIOS. Then, depending on the number of degrees over the CPU Temp Tag setting, the fan will adjust its speed from 950 RPM to over 2100 RPM, in order to compensate for the added heat build up. We found this feature to be a really nice tool in striking that all important balance between acoustics and system or CPU temperature control.
Our efforts in overclocking the SB77G5 were met with impressive results, although frankly we didn't go into this experiment expecting all that much from such a little enclosure housing Intel's 3.6GHz Prescott CPU, which we've come to discover runs just a tad hot under the collar.
However, with a little finesse, a 1.4V CPU core voltage setting and a 2.9V setting for our RAM, we had the SB77G5 running at a stable 3.9GHz CPU speed and our Kingston DDR memory at 436MHz, although we had to set the fan setting in the BIOS to "mid" in order to keep heat in check. This was a 2100 RPM fan speed setting and the resultant noise was completely tolerable but a lot less stealthy than while in "Smart Fan" mode. Our Prescott 3.6GHz processor, running at 3.9GHz, hovered in the 65oC range, which is a bit on the warm side but things were stable regardless. We were more than impressed at the SB77G5's ability to overclock, so we decided to look at one more experiment, from a different angle, that of at 1066MHz front side bus speed. A ghost of things to come perhaps?
We thought it would make sense to look at an optimal setting for a processor, like the 3.46GHz P4 EE or the upcoming 1066MHz FSB Prescott with 2MB of L2 cache, that Intel is rumored to be unleashing in Q1 of '05. So, with that in mind, we took the divider of our 3.6GHz Prescott core, down to 14X from 18X and bumped the system bus speed up to 266MHz, which yields our 1066MHz FSB. With a DDR320 divisor setting in the BIOS, our memory was clocked in at 425MHz and the CPU was set at 3.73GHz with its 14X multiplier on a 266MHz bus. Things were very stable at this speed and it helped lend proof that the SB77G5 should by all rights, unofficially support 1066MHz FSB based Pentium 4 processors moving forward. We contacted Shuttle on this topic and they gave us the standard issue that the 1066 FSB Pentium 4 CPU is not "officially" supported at this time but off the record we got a wink and a nod. That's good enough for us but you'll be running things at your own risk of course. Nothing like a global "cover-your-butt" statement which is always the case in matters of overclocking! |
| Power To Support High End Graphics, Test Setup | |||||||
How we configured our test systems: When configuring the test systems for this review, first, we entered the system BIOS and set each board to their "Optimized" or "High-Performance Defaults". The hard drives were then formatted, and Windows XP Professional (SP2) was installed. When the installation was complete, we hit the Windows Update site and downloaded all of the available updates, with the exception of the ones related to Windows Messenger. Then we installed all of the necessary drivers, and removed Windows Messenger from the system altogether. Auto-Updating and System Restore were also disabled, and we setup a 768MB permanent page file on the same partition as the Windows installation. Lastly, we set Windows XP's Visual Effects to "best performance", installed all of our benchmarking software, defragged the hard drives and ran all of the tests...
High-end, high-powered graphics in an XPC? But what about the SB77G5's 250 Watt Silent X power supply? On the side of the SB77G5 box Shuttle claims it's "designed for use with today's power-hungry, processors, graphics cards and serial ATA drives". So we put that claim to the test. SB77G5 with GeForce 6800 Ultra and X800 XT installed Burn-In Test Specifics: Our results were impressive to say the least. Both configurations, with either the GeForce 6800 Ultra or Radeon X800 XT, were running flat out on our 3.6GHz Prescott CPU, 12 hours later when we checked the system. The demo loop never even had so much as a twitch. We took temp readings with the GeForce 6800U installed and discovered the CPU was at a level 62oC, the NV40 GPU was reading 72oC and the system temp recorded was 41oC. The main system fan was running at 1650 RPM, which is virtually inaudible but the GeForce 6800 Ultra still projected its classical whine from its turbine fan driven HSF. With the Radeon X800 XT in the SB77G5, in this test configuration, you had to lean your head over the system to tell if it was running. So, the SB77G5's power supply is not only up to the task of powering the highest end graphics cards on the market, but the rest of the system also does this with a near stealth-mode acoustical signature. |
| SiSoft Sandra Metrics | |||||||
We began our standard testing with SiSoftware's SANDRA, the System ANalyzer, Diagnostic and Reporting Assistant. SANDRA consists of a set of information and diagnostic utilities that can provide a host of useful information about your hardware and operating system. We ran three of the built-in sub-system tests that partially comprise the SANDRA 2004 suite (CPU, Multimedia, and Memory). All of these tests were run with the Shuttle SB77G5 XPC powered by a Pentium 4 560 processor clocked at its stock speed of 3.6GHz.
In all three of these tests, the Shuttle SB77G5 performed virtually identically to the similarly configured reference system in SANDRA's database. The CPU benchmark that the SB77G5 posted was actually a hair faster than the reference P4 560 system and the same held true for the Multimedia test. Our memory bandwidth scores came in neck and neck with the reference i875 CAS2 system as well. All told, it was a predictable outcome for the SB77G5 in our Sandra tests, which in this case is certainly all good. |
| PCMark 04 Tests | |||||
As another quick reference point, we ran the CPU and Memory test modules of Futuremark's PCMark 04. These test suites run a series of quick scripted routines that comprise many common desktop compute workloads that an average end user would likely invoke at some point in there experience. Here are some additional details on the tests.
"The CPU test suite is a collection of tests that are run to isolate the performance of the CPU. There are nine tests in all. Two pairs of tests are run multithreaded - each test in the pair is run in its own thread. The remaining five tests are run single threaded. These tests include such functions as file encryption, decryption, compression and decompression, grammar check, audio conversion, WMV and DivX video compression."
The SB77G5 steps into the middle of the pack here, falling just slightly short of our i925XE P4 560 reference system but besting the P4 3.46GHz EE with its higher clock speed. There is virtually no advantage afforded by the i925XE chipset versus the SB77G5's i875 implementation, at least in this test case.
"The Memory test suite is a collection of tests that isolate the performance of the memory subsystem. The memory subsystem consists of various devices on the PC. This includes the main memory, the CPU internal cache (known as the L1 cache) and the external cache (known as the L2 cache). As it is difficult to find applications that only stress the memory, we explicitly developed a set of tests geared for this purpose. The tests are written in C++ and assembly. They include: Reading data blocks from memory, Writing data blocks to memory performing copy operations on data blocks, random access to data items and latency testing."
In the memory test module we see the main area where the three platforms are differentiated from each other, system memory bandwidth. The P4 3.46GHz EE and i925XE combination puts up the best score, with its 2MB of integrated L2 cache on board the CPU. The i925XE and P4 560 reference system comes in second with its 533MHz CAS 3,3,3,8 DDR2 memory offering a tangible but smallish advantage over the SB77G5 and it's i875 chipset, along with 400MHz CAS 2,2,2,5 standard DDR system memory. |
| Business and Content Creation Winstone 2004 Tests | ||||
A more "standard" set of tests of more easily recognizable desktop Business and Professional apps, would be Veritest's Winstone test suite, which is up next.
The PC Magazine Business Winstone 2004 test utilizes the following applications in its benchmark.
The PC Magazine Content Creation Winstone 2004 test utilizes the following applications in its benchmark.
If you were looking for any revelations here, sorry there are none to be found. The Shuttle SB77G5 simply puts up a very strong showing versus its i925XE equipped counterparts. The system actually takes a slight lead in the Business Winstone test, which more than likely is simply within the margin of error for this set of benchmarks. |
| 3DMark05 CPU and Windows Media Encoder | ||||||||
We continued our testing with a video encoding benchmark using, Windows Media Encoder 9. In the WME 9 test, we took a 416MB video file and encoded it into WMV9 format. Times were recorded in Minutes : Seconds, with lower times indicating better performance.
The memory bandwidth advantage that the two i925XE systems have, seems to garnered them a 10 second advantage, in round numbers, in this test. The high clock speed of the P4 560, combined with the robust memory bandwidth of DDR2 allowed the i925XE board to take the lead and Shuttle's SB77G5 brought rear. In actuality, the i875 based SB77G5 is about 7% slower in this test, versus a similarly equipped i925XE test system.
Again, we're back with more of a synthetic benchmark analysis. 3DMark05's CPU test module gives us an idea of how these systems will perform in a DirectX gaming environment, with the main emphasis of the test focused on CPU throughput and overall system bandwidth. This test consists of two different 3D scenes that are generated with a software renderer, which is dependant on the host CPU's performance. This means that the calculations normally reserved for your 3D accelerator, are instead sent to the central processor. The number of frames generated per second in each test, are used to determine the final score.
Once again we see our three test systems all right on top of each other, in terms of relative performance in this test. The P4 560 / i925XE combination has a 2% lead over the others, including the Shuttle SB77G5, which is obviously an inconsequential advantage. Perhaps a few different gaming scenarios will show some spread in the field performance picture here. Let's have a look at those next. |
| UT2004 and Wolfenstein ET | ||||||||
Wolfenstein: ET is a free, standalone multiplayer game that is based on the original Return to Castle Wolfenstein, that was released a few years back. It uses a heavily modified version of the Quake 3 engine which makes it a very easy to use benchmarking tool. We ran the test using the "Fastest" setting at a low resolution of 640X480, using 16-bit color and textures. Running this test with a high-end graphics card, at these minimal settings, isolates processor and memory performance, with very little emphasis placed on the graphics subsystem.
As we suspected, the two Prescott core P4 560 systems fell behind the P4 3.46GHz Extreme Edition setup with its 2MB of L2 cache. The P4 EE has a 5% advantage over the fastest P4 560 score and the i925XE based P4 560 test system has a small 3% advantage over the Shuttle SB77G5. However, once again for a SFF PC, these scores are more than respectable.
Epic's Unreal Tournament 2004 give us a slightly different picture on performance oddly enough, versus our Wolf ET test. While the OpenGL based Wolfenstein engine is affected more so by system memory bandwidth, UT2004 seems to be a bit more CPU intensive and the Shuttle SB77G5 put up a strong showing here, just edging out the i925XE reference system with an identical processor installed. |
| Doom 3 CPU and Graphics Benchmarks | ||||||||
Next up we have Doom 3 CPU centric performance numbers for you here and then we'll crank up the resolution just to show you what the SB77G5 can do with a top end graphics card installed in its bread box sized chassis.
Our low resolution, low quality multiplayer Doom 3 time demo test, looked a lot like our Wolfenstein tests, with the P4 EE taking the lead, the P4 560/i925XE system squarely in the #2 slot and the SB77G5 close behind it within a couple of FPS striking distance. If you had concerns over whether or not an XPC was capable of Doom 3 gaming mayhem, then your questions will be answered in our next time demo batch run.
In this Doom 3 test we took our GeForce 6800 GT card right up to its limits, stepping through the resolutions from 1024X768 to 1600X1200, with 4XX AA and 8X Anisotropic Filtering enabled in the game engine. Again, this was the very same setup we used in our stability and burn-in tests, only in that test we ran a 6800 Ultra, just to turn up the heat on the SB77G5 a bit.
1600X1200 with 4X AA and 8X AF, at 40 FPS, in a form factor that about half the size of your average sub woofer, the SB77G5 has the power to handle pretty much whatever you can throw at it. On a side note, our GeForce 6800 Ultra driven scores were 46 FPS at 1600X1200 resolution. You just can't argue with the numbers. |
| Half Life 2 Benchmarks, Performance Analysis and Conclusion | ||||
We thought you would never let us live it down if we didn't fire up the game that everyone on the scene seems to be playing right now, none other than Half Life 2. This time we tossed in our Radeon X800 XT, just to mix things up a bit.
Once again, more than enough power to spare, if you are a Half Life 2 or Counter Strike: Source junkie, is available with the Shuttle SB77G5 as a base system. We actually prefer the more modest power requirements of the Radeon X800 XT for a high end graphics configuration in this XPC. As you can see from our custom demo benchmark scores, it's also a no compromise setup with the ATi's flagship Radeon at the helm.
System Performance Analysis: The Shuttle SB77G5 impressed us on all fronts, from a performance perspective. As you could see in our extensive testing, the system kept pace and occasionally edged out its i925XE counterpart. The i875 chipset that the SB77G5 is built upon, still has plenty of legs left in it and even though 533MHz DDR2 can offer an advantage at CAS 3 settings, the SB77G5 with its DDR400 memory at CAS 2, came within striking distance, even in the memory intensive tests like Windows Media Encoder 9. The other notable that our testing showed is that the SB77G5 should be able to run new 1066MHz FSB P4s, although "unofficially supported", as they become available. You'll just have to adjust the system's memory divider in the BIOS, to keep things running within spec for the DDR memory, which shouldn't be much of an issue for virtually all of the latest modules from the major players.
Reflecting back on our experience with the Shuttle SB77G5, we can't help but be impressed by its stylish design, build quality, efficient layout, easy build-out effort, and extremely quiet operation. Even with a 3.6GHz Prescott core Pentium 4 in its LGA775 socket, the system kept thermals in check and fan speeds well underneath the radar acoustically. The system also demonstrated an ability to run even the highest end GeForce 6800 Ultra and Radeon X800 XT graphics cards, at full stability, on what turned out to be a very robust 250 Watt power solution. Optimally, we think a GeForce 6800 GT or Radeon X800 XT would be your best bet for quiet operation and reasonable thermals inside the system. Regardless, the SB77G5 took our most power-hungry card we had in the lab and ran it full tilt, with the most power-hungry processor we had available, in its socket. If that isn't a ringing endorsement for this XPC's component quality, we don't know what is. The only draw-back we considered for the SB77G5, is something very much out of Shuttle's control. Simply put, Intel's Socket T (LGA775) processors, at many speed bins right now, clock for clock, can't match the price/performance ratios that the Athlon 64 can offer, at this time. Regardless, if you're in the Pentium 4 camp, the SB77G5, with it's i875 chipset and LGA775 socket, does a real nice job of straddling the fence to Intel's latest processors, while enabling you to work with current AGP graphics card solutions and your good ol' DDR PC3200 - PC4000 DDR RAM. That does wonders for bringing total system cost down as well. Bravo Shuttle, bravo! The Shuttle SB77G5 XPC scores a 9 on HotHardware.com's Heat Meter.
|