Page 5

SiSoft Sandra Benchmarks
SiSoftware Sandra (the System ANalyser, Diagnostic and Reporting Assistant) is an information & diagnostic utility. It should provide most of the information (including undocumented) you need to know about your hardware, software and other devices whether hardware or software. Sandra provides similar level of information to Norton SI, Quarterdeck WinProbe/Manifest, etc. The Win32 version is 32-bit and comes in both ANSI (legacy for Windows 98/Me systems) and native Unicode (Windows NT4/200X/.Net) formats. The Win64 version is 64-bit and comes in native Unicode format.

Do note that all the SANDRA benchmarks are synthetic and thus may not tally with real-life performance. The latter stands for whatever your environment is, i.e. which applications you run with what amount of data and so on. It is up to you to decide whether what Sandra measures is what you want to measure.

• Download SANDRA (special Guru3D.com edition)

Here you can find the scores of Sandra.

The test system used is an Albatron PX865PX PROII. It uses a 2.8 GHz Pentium 4 Processor (aircooling). It of course has a quad pumped 4x200 MHz FSB.

Memory is based on non tweaked and suggested (2:2:2:5) timings. Memory timings as you probably know are good for this rule: the lower the better. Memory timings let you know how many cycles it takes for diverse operations internally to the memory with CAS being the most important one. Remeber it like this: 3 is 'okay', 2.5 is good and 2 is by far the best, yet hard to achieve with regular memory.

As you can see from the results performance is really good. We increased the FSB until we could not go any further, unfortunately that limit was roughly 243 MHz making the memory run at 486 (that's still way below the specified maximum). At that speeds tried the memory at 2-2-2-5 yet the system obviously refused to boot with that FSB, aah well... can't have it all eh?. At standard DDR400 we are able to run CAS 2 flawlessly though. You will be able to maintain the fastest timings up-to a 215-220 MHz FSB, then you need to play around with slower timings.

If all this is new to you: overclock your processor to the maximum and with the memory work your way downwards. This means maximum voltage on the memory with the slowest timings. Keep lowering the timings until you get issues and then go one step back.

At one point we where able to overclock the 2.8 GHz processor to 3.5 GHz (FSB 250 MHz). But unfortunately it was unstable, the processor that is.

But the CPU was running so hot that I decided to call that test run off. Anyway, at a 4x243=972 MHz FSB we where able to achieve a memory bandwidth faster than a standard DDR2 533 system. That's pretty amazing stuff for PC3200 memory. We had to slow down the timings to 2.5:3:3:7 though and had increase the memory voltage towards the mainboards maximum 2.85 volts which really is the limiting factor on this mainboard.

Furthermore on this 865 based mainboard we can use a little trick (PEM/PET - Performance Enhanced Memory mode) to enable a little extra memory performance (a shorter optimized data path) bringing performance up-to the Intel 875 chipset standard. We where able to enable that at standard 400 MHz mode, which actually is a mainboard maximum requerment, a lovely extra 300MB/sec is then added to memory bandwidth. The Patriot memory had no problems with this sneaky little tweak.

Of course in the past we've tested quite a few PC3200 modules. Above you can see the standard performance with SPD default settings enabled. All brands of course offer faster memory as time passes.

Let's go have a look at the PC Mark04 results.

PC Mark 2004PCMark®04 is the latest version of the popular PCMark series. PCMark04 is an application-based benchmark and a premium tool for measuring overall PC performance. It uses portions of real applications instead of including very large applications or using specifically created code. This allows PCMark04 to be a smaller installation as well as to report very accurate results. As far as possible, PCMark04 uses public domain applications whose source code can be freely examined by any user. 

Info and download - Download!

Another synthetic benchmark that displays memory performance. We expect this memory to perform at a ~4200 score. And it does precisely that.

I also included the results at slow 3:4:4:8 timings @ 400 MHz (2x200) followed by the standard 2:2:2:5 timings (no PEM, that's not properly stated in the graph). As you can see the score differs only 100. Keep that in mind in our gaming benchmarks as I will show you something. That something being what difference timings actually make in real world applications.