• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

A8N-SLI & Slow SANDRA Memory Benches

RyanLM

Member
I am trying to find out if anyone else with this board is having this issue. I am getting lower than expected Memory performance shown by SANDRA 2005 in the Memory Bandwidth Benchmark. Here is the screenshot:

SANDRA Memory Bandwidth Benchmark (2005)

Seems I am maxing out at 4.9 G/Sec.

I have tried this benchmark on 2 separate A8N-SLI boards, I am using Corsair Pro Ram (the ones with the LEDs rated for 500mhz) I have also tried basic DDR400 TWINX, as well as OCZ Platinum 3200, and basic DDR400 Kingston ram, all give the same results within a few points.

I am using a AMD FX 55, most of the benches that I see out there show it at around 6 GB/Second - Is anyone else having this issue? I have tried each of the bios revisions, all give the same results.

I am currently running in Single Video card mode, and I have tried 2 different Video cards, one Quadro 3400, and one regular GT.

 

I just ran a Sandra Mem test on mine and got 5964 which is what should be expected from a dual-channel A64.
With the A64 (FX too) the only factors for memory are the CPU, the motherboard traces, power delivery, and BIOS settings. Your video card wouldn't have anything to do with it. Are you running at command rate 2T?
 
1) uninstall Sandra, the benchmarking equivalent of AOL :evil:

2) install Doom3 and/or HL2 and play 'til your eyes fall out 😎
 
Nope, mine is fine. I get both bars completely filled when I set it up like your test. I have the bios set for 2-2-2-5 w/ 1t max mem = auto.

Memory is in my signature!
 
Ok, well I have more info now.

It seems that it was reverting to a command rate of 2T instead of 1T. However, I can only get a 1T rate if I only use 2 slots, not all 4 - Is this normal?

I have 4 identical sticks of ram, I really prefer 2 gig of ram for my work, but I can only seem to get full speed with only 2 sticks installed. Is it an ammount of ram thing? or a 4 stick thing?

 
Originally posted by: RyanLM
Ok, well I have more info now.

It seems that it was reverting to a command rate of 2T instead of 1T. However, I can only get a 1T rate if I only use 2 slots, not all 4 - Is this normal?

Unfortunately that is normal. It's the way the A64 memory controller works.

 
No, actually that's how plain generally, electrical signals degenerate when adding capacitive load (read: more trace length and more chips) onto them. Sharp timings can't be run with high bus loads.

DDR400 being fairly edgy as it is, something has to give. Remember that DDR333 standard was the one defined for two-DIMM-per-channel operation; so if you're achieving DDR400 in that configuration, you're already doing better than the standards suggest - even if your at 2T command rate then.
 
Interesting!

So, now would 2 Gig sticks of ram allow me to do 2 Gig at 1T. Or is it a shear ammount of memory that is preventing me from getting 1T.

Thanks for your help!
 
Yeah 2 1 gig sticks should allow you to run at 1t, as long as the sticks can handle it, which im pretty sure they can.
 
Originally posted by: stelleg151
Yeah 2 1 gig sticks should allow you to run at 1t, as long as the sticks can handle it, which im pretty sure they can.

hmm...

um... .that's a good question... i don't know if 2 x 1 gig sticks will allow you to run 1T ...

it might but not as high a HTT as running 2 x 512's or 2 x 256's
 
Yes 1gb modules will run 1T, unfortunely to date nobody has come up with 1gb modules with decent timings. They all have like 3-4-4-8.

Bottom line is to get 2gb you have to give up some performance
 
Originally posted by: ChineseDemocracyGNR
Everyone knows memory bandwidth is much lower at 2T Command Rate, but are there any real-world benchmarks to show that it actually affects performance?

Its easy to test. Run a benchmark, change to 2T, run bench again.
 
I am having the same problem a8n-sli, fx-55, 2x6600gt, 2x512 crucial balistix ddr400 , i have the sticks in the first and third slots getting 4.8-4.9 in the memory test. . 1002 bios and auto memory timings. set the timings to 2-2-2-5 1T and memory bench dropped to 3000 lol.. Put timings back to auto and i get 4890 ..Memory Bandwidth 4890, Cache & Memory benchmark 6800
 
no problem, try using 2-2-210 or 2-2-2-11 and see if it makes a difference. The a64 supposed to work better with these timings.
 
Originally posted by: GuitarDaddy
Yes 1gb modules will run 1T, unfortunely to date nobody has come up with 1gb modules with decent timings. They all have like 3-4-4-8.

Bottom line is to get 2gb you have to give up some performance

Kingston offers a 2GB kit with 2.5-3-3-7-1 timings. It's not cheap but performance loss is minimal. The reason they don't make them is not because they can't but because it would cost to much to manufacture.

-------------- Here --------------
 
Originally posted by: Smoothe
Originally posted by: GuitarDaddy
Yes 1gb modules will run 1T, unfortunely to date nobody has come up with 1gb modules with decent timings. They all have like 3-4-4-8.

Bottom line is to get 2gb you have to give up some performance

Kingston offers a 2GB kit with 2.5-3-3-7-1 timings. It's not cheap but performance loss is minimal. The reason they don't make them is not because they can't but because it would cost to much to manufacture.

-------------- Here --------------

Wow! Thanks for the info. Didn't know anybody made ram like this.
 
Back
Top