Synthetic Graphics Performance

The 3DMark series of benchmarks developed and provided by Futuremark are among the most widely used tools for benchmark reporting and comparisons. Although the benchmarks are very useful for providing apples to apples comparisons across a broad array of GPU and CPU configurations they are not a substitute for actual application and gaming benchmarks. In this sense we consider the 3DMark benchmarks to be purely synthetic in nature but still valuable for providing consistent measurements of performance.

General Graphics Performance

General Graphics Performance

General System Performance

In our first tests, each P965 platform score is so close that there is no real winner here. In the more memory and CPU sensitive 3DMark03 benchmark we see the same trend with the P965 boards scoring extremely close with the Gigabyte board leading in both benchmarks. We really did not expect to see any real differences but we do know the P965 is very competitive with the 975X in our synthetic benchmarks. Our only real disappointment is the performance of the ASUS P5NSLI board as generally the NVIDIA chipsets shine in the 3DMark scores. We attribute the scores to the fact that the board is based in the budget sector and is tuned for stability and overall system performance rather than games/graphics. As we recognized in our preview of this board, it was very sensitive to memory speed and timings with DDR2-800 being a minimum requirement before the board would perform well. We expect the upcoming 680i to change the results in this test.

General System Performance

Considering our 3DMark results we really did not expect a difference in this benchmark but we found one. The Intel based ASUS boards walked off and left the other boards in a cloud of dust. We typically find some variations in this benchmark due to Hard Disk performance with the NVIDIA chipsets consistently scoring about 9% better than the Intel chipsets. In this case the overall difference was around 4% in the HDD tests where the P965 based boards were performing better than the 975X board. Our 570SLI board did not score well in the multitasking tests where the ASUS P965 boards shined.

After reviewing the individual results we noticed all of the scores between the P965 boards were essentially the same until we reached the final test which consists of multitasking three different applications. The ASUS P965 boards and to some extent the 975X board scored up to 58% better in the File Encryption and HDD Virus section of the test. The File Compression number was around 16% better. We ended up running this particular test on each board several times with different memory and hard drive configurations with the same result. We continued to play detective and once all possible causes were researched we decided to create our own multitasking script to determine if ASUS had tuned their BIOS for this test or if they really had better multitasking performance. The results are on the next page.

Test Systems: Benchmark Setup General System Performance
Comments Locked

62 Comments

View All Comments

  • vailr - Saturday, October 21, 2006 - link

    Re:
    quote:

    Intel in their infinite wisdom decided to pull PATA support from this chipset when over 98% of optical drives are still based on PATA technology.

    So, how about the (yet unreleased) ATI and NVidia Conroe chipset boards?
    Does either chipset include PATA support?
    Thanks.
  • Gary Key - Saturday, October 21, 2006 - link

    quote:

    Does either chipset include PATA support?


    They both have native support for two drives.
  • n7 - Saturday, October 21, 2006 - link

    Gary, always love your reviews!
    I read thru the whole thing, & it was a good read :)

    Meticulous detail, as well great sarcastic humor as well.

    I look forward to the following parts.
  • Sho - Friday, October 20, 2006 - link

    In an earlier AnandTech article, the one about Kentsfield support, it was written that Gigabyte would bring a revision 2.0 of all of their P965 boards to the market in mid-October, including the DS3. The article does not mention whether the board tested was this new rev 2,9 or any other. Could that be clarified?

    And does anybody know what was changed/fixed in 2.0?
  • Gary Key - Friday, October 20, 2006 - link

    Gigabyte has not released any further details on the revision 2 boards except for the fact they were addressing some layout issues and possible BIOS improvements. The only major change we could see them making would be going from a three phase power design on the DS3 to a five phase system as an example. The board we tested is still revision 1.
  • Sho - Saturday, October 21, 2006 - link

    Thanks!
  • dreddly - Friday, October 20, 2006 - link

    'caliper' should be caliber on AB9Pro page

    Great work on this roundup though, impressive job.
  • Puddyglum1 - Friday, October 20, 2006 - link

    quote:

    However, the board is not without its shortcomings. We certainly have an issue with this board...
    Which board? The topic of the previous page was about sound cards vs. onboard audio. Is there a missing page? Why is there a picture of the Asus heatsink and no mention of which board is the preferred of the bunch?

    Just some questions =)

    Great article for Cost/Performance comparison.
  • Puddyglum1 - Friday, October 20, 2006 - link

    Woah, there's a lot more there now. Thanks for the explanation.
    quote:

    Overall, the Gigabyte GA-965P-DS3 performed the best in our benchmarks when not overclocked.

    I just built a workstation for a client using the 965P-DS3, but the board was DOA. I went to a local shop and picked up a 965P-S3 instead (seeing as how the only main feature missing was the solid capacitors of the -DS3), and it performed just as well as the DS3. For $110, a GA-965P-S3 would be the best Cost/Performance of the 965P bunch, in my unresearched opinion.
  • JarredWalton - Friday, October 20, 2006 - link

    Now you're skipping ahead to part 2! :p

Log in

Don't have an account? Sign up now