SiS 672FX Reference Board Review

[nextpage title=”Introduction”]

SiS 672FX is a chipset with integrated graphics targeted to low-end socket LGA775 motherboards. We’ve got from SiS a reference board for this chipset and since motherboards based on the same chipset usually achieve similar performance, you can have a clue on what performance to expect from motherboards based on SiS 672FX.

A reference board is a motherboard manufactured by the chipset maker to make internal tests and also to ship to their partners (i.e., motherboard manufacturers) so they can create their own products based on the given chipset.

It is important to have in mind that not all features provided by the chipset are available on all models based on it. It is up to the motherboard manufacturer to choose which features they want on their models. For example, the reference board had two x1 PCI Express slots, two standard PCI slots and one CNR slot, but products based on SiS 672FX may come with less slots.

Figure 1: SiS 672FX reference board.

SiS 672FX is a low-end chipset with several limitations. Its memory controller allows only single-channel mode, meaning that this chipset accesses the memory at half the speed other chipsets available for the socket LGA775 chipset can. It also only supports DDR2 memory up to DDR2-667, at least officially. On our tests we configured our memory modules at 800 MHz and they worked just fine. This chipset also only supports two memory sockets. Another limitation is that this chipset does not support the new 1,333 MHz FSB, but it accepts all 1,066 MHz-based socket LGA775 CPUs.

As mentioned SiS 672FX has integrated graphics (“on-board video”), based on the Mirage 3+ graphics engine, which is a truly DirectX 9.0 (Shader 2.0) engine. The graphics engine runs at 300 MHz. We asked SiS twice about the number of pixel shader engines and vertex shader engines this chipset has and they ignored these questions. On other reviewing websites we could see that SiS 672FX does not feature a vertex shader engine, this stage being processed by the system CPU, i.e., software-based vertex shader.

Even though SiS 672FX has integrated graphics, it supports one PCI Express x16 slot. Keep in mind that motherboard manufacturers in order to cut costs may offer SiS 672FX-based motherboards without this slot, but this limitation is of these particular motherboard models.

The south bridge chip defines other features that will be found on SiS 672FX-based motherboards. The reference board we got used a SiS 968 south bridge.

SiS 968 main features are:

• One ATA-133 port;
• Two SATA-300 ports, supporting RAID 0, 1 and JBOD;
• Eight USB 2.0 ports;
• Two PCI Express x1 slots;
• Gigabit Ethernet;
• High Definition Audio (eight channels, up to 192 kHz, 32 bits).

Keep in mind that the motherboard manufacturer may choose not to use all features provided by the south bridge chip.

Also, for the LAN port, the manufacturer needs to add a chip on the motherboard to make the interface with the physical layer. On SiS reference board this chip was SiS 196. To cut costs the motherboard manufacturer may use a Fast Ethernet (100 Mbps) chip.

As for the integrated audio, it is very important to keep in mind that the final specs will depend on the codec used, which is a small chip located on the motherboard. On the reference board SiS used a Realtek ALC883, which has a 95 dB signal-to-noise ratio (SNR) and 192 kHz maximum sampling rate for its output and an 85 dB signal-to-noise ratio and 96 kHz maximum sampling rate for its input, both with 24-bit resolution. The output specs of this codec are fair for Average Joe, but the input specs are simple too low for today’s standards.

In summary, the final features of the motherboard depend a lot on the components the motherboard manufacturer chooses.

• More information on SiS 672FX
• More information on SiS 968

Now let’s take a look at its performance.

[nextpage title=”How We Tested”]

During our benchmarking sessions, we used the configuration listed below. Between our benchmarking sessions the only variable was the motherboard being tested.

Hardware Configuration

• BIOS version: 8.00.14 (March 9th, 2007)
• Motherboard revision: N/A
• Processor: Core 2 Duo E6700 (2.66 GHz, 1,066 MHz FSB, 4 MB L2 memory cache).
• Cooler: Intel.
• Memory: 1 GB DDR2-1066/PC2-8500, two Corsair CM2X512-8500C5 modules (512 MB each) with 5-5-5-15 timings.
• Hard Disk Drive: Samsung HD080HJ (SATA-300, 7,200 rpm, 8 MB buffer).
• Video Cards: XFX GeForce 6200 TC 64 MB 64-bit and MSI factory-overclocked  GeForce 8800 GTS 320 MB (NX8800GTS-T2D320E-HD OC).
• Video resolution: [email protected]
• Power Supply: Antec Neo HE 550.

Software Configuration

• Windows XP Professional installed using NTFS
• Service Pack 2
• DirectX 9.0c

Driver Versions

• Intel Inf chipset driver version: 8.3.0.1013
• Intel video driver version: 14.29
• SiS video driver version: 3.81
• Nvidia video driver version: 93.71 (on GeForce 6200)
• Nvidia video driver version: 158.22 (on GeForce 8800 GTS)
• Audio driver version: Realtek R1.62

Software Used

• SYSmark2004 – Patch 2
• PCMark05 Professional 1.1.0
• 3DMark 2001 SE 3.3.0
• 3DMark03 3.6.0
• 3DMark05 1.2.0
• Quake III Arena 1.32
• Quake 4 – Patch 1.3

We adopted a 3% error margin; thus, differences below 3% cannot be considered relevant. In other words, products with a performance difference below 3% should be considered as having similar performance.

[nextpage title=”Overall Performance”]

We measured the overall performance of this motherboard using SYSmark2004, which is a software that simulates the use of real-world applications. Thus, we consider this the best software to measure, in practical terms, the system performance.

The benchmarks are divided into two groups:

• Internet Content Creation: Simulates the authoring of a website containing text, images, videos and animations. The following programs are used: Adobe After Effects 5.5, Adobe Photoshop 7.01, Adobe Premiere 6.5, Discreet 3ds Max 5.1, Macromedia Dreamweaver MX, Macromedia Flash MX, Microsoft Windows Media Encoder 9, McAfee VirusScan 7.0 and Winzip 8.1.
• Office Productivity: Simulates the use of an office suite, i.e., simulates sending e-mails, word processing, spreadsheets, presentations, etc. The following programs are used: Adobe Acrobat 5.05, Microsoft Office XP SP2, Internet Explorer 6.0 SP1, NaturallySpeaking 6, McAfee VirusScan 7.0 and Winzip 8.1.

The software delivers specific results for each batch and also an overall performance result, all in a specific SYSmark2004 unit.

We ran this software in two scenarios. First using the motherboard on-board video, comparing its performance to the only other motherboard with on-board video compatible with Core 2 Duo we had available, which was ECS G33T-M2 (Intel G33 Express).

Then we disabled the board on-board video and installed an overclocked GeForce 8800 GTS from MSI to compare it with a MSI P35 Platinum, which is a high-end socket LGA775 motherboard based on Intel P35 chipset. Our idea was to see if this motherboard would achieve the same performance level of a high-end motherboard when we installed a high-end video card on it. Since on SIS 672FX reference board we could set our memories to run only up to 800 MHz, we also set our memories at this clock rate with MSI P35 Platinum, even though we were using DDR2-1066 memories and that this motherboard allows them to run at 1,066 MHz, otherwise we would have an unfair comparison.

With its on-board video enabled SiS 672FX reference board achieved an overall performance 10.09% lower than ECS G33T-M2 (Intel G33 Express). Its Internet Content Creation performance was 5.37% lower than ECS G33T-M2 and its Office Productivity performance was 14.72% lower.

When we installed our overclocked GeForce 8800 GTS on the reviewed board it kept the same performance level – far lower than a high-end motherboard. This means that by installing a “real” video card on this motherboard it WON’T achieve the same performance level of a high-end or even a mainstream motherboard without on-board video. Under this scenario MSI P35 Platinum achieved an overall score 9.09% higher, an Internet Content Creation score 4.39% higher and an Office Productivity score 13.79% higher.

[nextpage title=”Processing Performance”]

Using the same methodology of the previous test, we measured processing performance using PCMark05 Professional program. This program gives the results in a specific unit and since it includes video performance on its score, the motherboard with the best video will achieve the best results.

With its on-board video enabled ECS G33T-M2 achieved an overall performance 40.97% higher than SiS 672FX, which is simply outstanding.

When we installed our overclocked GeForce 8800 GTS on the reviewed board it was 7.55% slower than MSI P35 Platinum with the same video card installed. This means that by installing a “real” video card on this motherboard it WON’T achieve the same performance level of a high-end or even a mainstream motherboard without on-board video.

[nextpage title=”3D Performance: 3DMark2001 SE”]

To evaluate SiS 672FX’s 3D performance we installed a very low-end video card on SiS 672FX reference board: GeForce 6200 TurboCache with 64 MB and 64-bit interface (from XFX).

We also installed a high-end video card, an overclocked GeForce 8800 GTS from MSI, on it and compared its performance to a high-end motherboard with the same video card installed, MSI P35 Platinum. The goal here was to see if by installing a high-end video card the reviewed board achieved the same performance level of a high-end motherboard.

We ran several programs, as you will see in the next pages. The amount of RAM memory the chipset “steals” from the main RAM memory to be used as video memory was left on the motherboard default value, which was 128 MB for the reviewed motherboard.

The first one, 3DMark2001 SE, measures 3D performance by making DirectX 8.1 simulations, and the results you can see below (we ran it using its default configuration). All the listed video cards were installed on SiS 672FX reference board.

On 3DMark2001 SE the on-board video produced by Intel G33 Express was 346% faster than the one produced by SiS 672FX. Here our GeForce 6200 TurboCache with 64-bit memory interface and 64 MB was 340.93% faster than SiS 672FX.

When we installed our overclocked GeForce 8800 GTS on the reviewed board it didn’t achieve the same performance level of MSI P35 Platinum – this motherboard from MSI was 4.16% faster. This means that by installing a “real” video card on this motherboard it WON’T achieve the same performance level of a high-end or even a mainstream motherboard without on-board video.

[nextpage title=”3D Performance: 3DMark03″]

We followed the same methodology described in the previous page, but now running 3DMark03. 3DMark03 simulates DirectX 9.0 (i.e., Shader 2.0) games, which is fully supported by both Intel G33 and SiS 672FX chipsets.

You can check the results of our benchmarking below. All the listed video cards were installed on SiS 672FX reference board.

On 3DMark03 the on-board video produced by ECS G33T-M2 was 377% faster than the one produced by SiS 672FX. Here the “worst” video card available on the market, GeForce 6200 TurboCache with 64-bit memory interface and 64 MB, was 409.20% faster than the on-board video produced by SiS 672FX.
 
When we installed our overclocked GeForce 8800 GTS on the reviewed board it achieved the same performance level of MSI P35 Platinum. But don’t get excited, as on other programs this didn’t happen.

[nextpage title=”3D Performance: 3DMark05″]

We again followed the same methodology described previously, but now running 3DMark05. This program measures 3D performance by simulating DirectX 9.0c games, i.e., using Shader 3.0. This programming model is used by the latest games but it isn’t supported by Intel G33 nor by Sis 672FX.

It isn’t fair to use this program to evaluate 3D performance of motherboards with on-board video, as they achieve a very low score on this program. We ran it anyway, basically to see the performance achieved by SiS 672FX using a real video card installed and also to compare it to other motherboards with on-board video.

You can check the results of our benchmarking below. All the listed video cards were installed on SiS 672FX reference board.

Here the on-board video produced by ECS G33T-M2 was 33% faster than the one produced by SiS 672FX. The “worst” video card available on the market, GeForce 6200 TurboCache with 64-bit memory interface and 64 MB, was 235.14% faster than the on-board video produced by SiS 672FX.
 
When we installed our overclocked GeForce 8800 GTS on the reviewed board it didn’t achieve the same performance level of MSI P35 Platinum: this board from MSI was 5.77% faster. This means that by installing a “real” video card on this motherboard it WON’T achieve the same performance level of a high-end or even a mainstream motherboard without on-board video.

[nextpage title=”3D Performance: Quake III”]

We used the Demo four available on version 1.32 of Quake III to make our benchmarking with this game. We ran this demo three times at 1024x768x32 resolution and all image quality settings on their default configuration and we picked the middle value for our comparisons, i.e., we discarded the highest and the lowest values.

You can check the results of our benchmarking below. All the listed video cards were installed on SiS 672FX reference board.

Here the on-board video produced by ECS G33T-M2 was 314% faster than the one produced by SiS 672FX. Here the “worst” video card available on the market, GeForce 6200 TurboCache with 64-bit memory interface and 64 MB, was 337.80% faster than the on-board video produced by SiS 672FX.
 
When we installed our overclocked GeForce 8800 GTS on the reviewed board it achieved the same performance level of MSI P35 Platinum. But don’t get excited, as on other programs this didn’t happen.

[nextpage title=”3D Performance: Quake 4″]

Quake 4 is very heavy game that uses the same engine of Doom 3. We used the id_demo001 available on version 1.3 of Quake 4 to make our benchmarking with this game. We ran this demo four times at 1024x768x32 resolution and image quality settings on “low”. The results shown on the chart is an arithmetic average of the collected data. The results are in frames per second. For more information on how to use Quake 4 to benchmark a PC, read our tutorial on this subject.

Here the on-board video produced by ECS G33T-M2 was 365% faster than the one produced by SiS 672FX. Here the “worst” video card available on the market, GeForce 6200 TurboCache with 64-bit memory interface and 64 MB, was 600% faster than the on-board video produced by SiS 672FX.
 
When we installed our overclocked GeForce 8800 GTS on the reviewed board it didn’t achieve the same performance level of MSI P35 Platinum: this board from MSI was 35.66% faster. This means that by installing a “real” video card on this motherboard it WON’T achieve the same performance level of a high-end or even a mainstream motherboard without on-board video.

[nextpage title=”Overclocking”]

The reference board for SiS 672FX allowed us to overclock our CPU, increasing its external clock rate in 1 MHz steps.

On this board we could set the external clock rate at 313 MHz with stability. With this overclocking our Core 2 Duo E6700 which normally runs at 2.66 GHz was running internally at 3.13 GHz, a 18% increase over its standard clock rate. Not bad at all for a motherboard with on-board video.

With this overclocking the performance measured by PCMark05 increased 9.66% but we saw no performance increase on Quake III.

Keep in mind that we could set a higher clock rate but the system wasn’t stable. We only consider an overclocking to be successful after we could run PCMark05 and Quake III at least four times with no crashes.

[nextpage title=”Conclusions”]

SiS 672FX is simply a joke. This is probably the worst chipset for the Intel platform available today.

Not only its graphics engine provides a very low performance: you won’t achieve the same performance level of mainstream motherboards by installing an add-on video card on it. For example, on Quake 4 a MSI P35 Platinum with a GeForce 8800 GTS installed was 36% faster than the reviewed board with the same video card installed. This is simply unheard of.

The major flaw of this chipset is its memory controller, as it does not work under dual channel mode and seems to be badly designed.

SiS chipsets are the cheapest around, making motherboards based on their chipsets also the cheapest. But due to the low performance offered by SiS 672FX we strongly recommend you against buying a motherboard based on SiS 672FX. Pay a little extra and buy any other motherboard.

Just to put things into perspective with SiS 672FX our Core 2 Duo E6700 with 1 GB DDR2 achieved a performance WORSE than a Pentium 4 2.4 GHz with a low-end motherboard with on-board video (VIA P4MA Pro 533 motherboard, VIA P4M266A chipset) with 512 MB DDR on Quake III (53 FPS vs 46 FPS). On 3DMark2001 SiS 672FX was faster than this Pentium 4 configuration, however on this program the same Pentium 4 2.4 GHz using an Intel 865G-based motherboard was 18% faster. This is simply ridiculous: what is the point of buying a Core 2 Duo and install it on such lousy motherboard that will make your computer to run like an old Pentium 4 2.4 GHz? In other words, you will buy a new computer and will get a performance level worse than the performance level provided by computers with on-board video from four years ago.