Gigabyte K8NNXP-940: Stress Testing

We performed stress tests on the Gigabyte K8NNXP-940 in these areas and configurations:

1. Chipset and motherboard stress testing, conducted by running the FSB at 219MHz.
2. Memory stress testing, conducted by running RAM at 400MHz with 2 DIMM slots filled and at 400MHz with all 4 DIMM slots filled at the lowest memory timings possible.

Front Side Bus Stress Test Results:

As is normal in our testing, we ran a full range of stress tests and benchmarks to ensure the Gigabyte K8NNXP-940 was absolutely stable at each overclocked FSB speed. This included Prime95 torture tests, and the addition of other tasks - data compression, various DX8 and DX9 games, and apps like Word and Excel — while Prime95 was running in the background. Finally, we ran our benchmark suite, which includes ZD Winstone suite, Unreal Tournament 2003, SPECviewperf 7.0, and Gun Metal Benchmark 2. At default voltage, 219MHz was the highest overclock that we were able to achieve with the K8NNXP without encountering any reliability issues. We will take another look at the K8NNXP-940 when a new BIOS update corrects the multiplier and FSB adjustment issues.

Memory Stress Test Results:

This memory stress test is very basic, as it simply tests the ability of the KX18D PROII to operate at its officially supported memory frequency (400MHz DDR) at the lowest supported memory timings that our Mushkin High Performance ECC Registered Modules will support:


Stable DDR400 Timings — 2 DIMMs
(1/2 Dual-Channel banks populated)
Clock Speed: 200MHz
Timing Mode: N/A
CAS Latency: 2.0
Bank Interleave: N/A
RAS to CAS Delay: 3T
RAS Precharge: 8T
Precharge Delay: 2T
Command Rate: N/A

We currently have 4 DIMMs each of Registered ECC DDR400 from both Legacy Electronics and Mushkin. The Legacy runs well at 2.5-3-4-5 timings, while the Mushkin has SPD timings of 2-2-3-8. Until we complete our upcoming roundup of Registered (ECC) DDR400 memory, we are using the Mushkin as our test standard. We had no problem at all with absolutely stable performance of the Mushkin High Performance ECC Registered at SPD timings.

Filling all available memory banks is more strenuous on the memory subsystem than testing 1 bank (2 DIMMs) in dual-channel mode, as it tests four DIMMs (2 banks) of Registered ECC memory running 400MHz DDR at the most aggressive memory timings the memory will allow.


Stable DDR400 Timings — 4 DIMMs
(2/2 Dual-Channel banks populated)
Clock Speed: 200MHz
Timing Mode: N/A
CAS Latency: 2.0
Bank Interleave: N/A
RAS to CAS Delay: 3T
RAS Precharge: 8T
Precharge Delay: 2T
Command Rate: N/A

Running both Dual-Channel Banks required the same timings as a single Dual-Channel Bank. It appears you can load the Gigabyte K8NNXP-940 with four DIMMs and still expect the fastest performance that your memory is capable of achieving.

We tested all these memory timings using several stress tests and general applications to guarantee stability. Prime95 torture tests were successfully run at the timings listed in the above charts. We also ran Sciencemark (memory tests only) and Super Pi. None of the three stress tests created any stability problems for the Gigabyte K8NNXP-940 at these memory timings.

FSB Overclocking Results Gigabyte K8NNXP: Tech Support and RMA
Comments Locked

35 Comments

View All Comments

  • juc - Friday, October 10, 2003 - link

    can you try and put in a lower clock opteron and see what type of overclocking you can do w/ it?, is the regular 14x opteron unlocked? it would be nice if it was.
  • Reflex - Friday, October 10, 2003 - link

    #1: A RDRAM version would completely eliminate the advantage of having an on-die memory controller on the CPU as it is very very high latency by design. The A64 thrives on very very low latency/high IPC, and RDRAM does not provide that.

    Honestly, what would be truly ideal is a QDR solution. But everytime I hear about it being close nothing seems to come of it. Too bad...
  • Anonymous User - Friday, October 10, 2003 - link

    Considering the performance gain, money ain't that important :-)
  • Anonymous User - Friday, October 10, 2003 - link

    #1,

    Samsung PC-3200 512 MB DDR SDRAM $125
    Samsung PC-3200 512 MB ECC Reg. DDR SDRAM $174

    +49

    Corsair XMS3200 PC-3200 512MB DDR SDRAM $175
    Corsair XMS3200 PC-3200 ELL 512MB DDR SDRAM $220
    Corsair XMS3200LL-RE PC-3200 ECC Reg. 512MB DDR SDRAM $235

    +15 (+60 compared to slower timings)

    completely unmeaningful to anyone with the money to buy an fx.
  • Anonymous User - Friday, October 10, 2003 - link

    Looks like a cool mobo, and an amazingly fast CPU, but . . .

    Who's going to buy one of these!?!?!?

    The price you'll spend on memory put's this way out of most people's price range! And before you yell at me for saying that, look up pricing for registered modules!

    You could probably buy an awesome Athlon 64 system now, then upgrade your mobo and CPU to FX when the 939 pin version comes out, and still spend less money than paying this ridiculous premium on memory. Plus, it would be upgradable to future FX chips, not an unsupported beast. Anyone remember socket 423?

    Say goodbye to the idea of 'surpassing the 4Gb memory limitation,' unless you have like $10,000 to spend on memory!

    My real question here is why, when the Athlon 64 (non-FX) is such a success, would they make this strange beast?

    What I would LOVE to see (I know you're going to hate this one) is a really tight RDRAM chipset ready when the 939-pin chipset comes out.

    What do you think? Quad Channel 1200Mhz RDRAM on the new FX? Ain't gonna happen, but I can dream.

Log in

Don't have an account? Sign up now