My FX-57 hates me, or does it?

[SSFSO]

A few days ago I set about putting together a new computer (A8N-SLI Premium, Dual 7800GTX, FX-57, 2Gigs PC3200, Dual 74GB Raptors) and after a few issues and reformats I managed to convince it to work properly and be stable. I eventually installed Aquamark 3 and ran it a few times scoring around 76,000 to 78,000 with an average or 13,000 on video and 9,000 on CPU. The CPU score seemed low to me so I did a bit of research and found that most people with an FX-57 were scoring about 12,000 to 13,000 on CPU. I downloaded ASUS PC Probe II to see if the temperature might be an issue, but it reported my CPU to be running around 35 to 40 degrees C.

I posted the problem on another forum and asked around. Several people recommended that I download the latest version of my motherboard's BIOS and check to make sure my FX-57 was running at its normal speed. I downloaded CPU-Z and it showed my FX-57 was running around 2,815MHz. My BIOS also showed the FX-57 running at 2,800MHz.

A Few Days Pass...

Evidently I already had the latest BIOS version that ASUS offers. I downloaded ASUS Update and had it check my BIOS version and it showed me as having "ASUS A8N-SLI Premium ACPI BIOS Revision 1011-001" and chipset "NF-CK804." I then headed over the ASUS's website to check if anything more recent was available, but I was unable to find any updates.

I ran Aquamark 3 a few more times and am still getting around 9,000 on my CPU score and 13,000 on my Video score for totals around 76,000 to 78,000. I also ran PCMARK05 and 3DMark03 and got around 4,600 and 28,000 respectively. I'm not exactly sure if those scores match the scores that my system should have, so if anyone could offer any insight I would appreciate it. Also, if anyone else is running a non-overclocked FX-57 could you tell me if you are getting similar CPU scores on Aquamark 3?

I suppose the bottom line is: Is my processor (FX-57) running at what it should be considering my scores on Aquamark 3 and the other Benchmarking programs I mentioned above? The PC seems to be handling well enough on CS: Source and CoD2, but I can't help feeling that I should be getting a higher score on Aquamark 3.

 

[Merc14]

That's an awesome 3DMArk05 score. Are you absolutely positive that the 12000-13000 in Aquamark scores aren't from O/Ced CPU's? Try cranking it up to 3.0 and see if you don't break 10k and then push a little harder, like 3.1 and you should be in the range you were seeing.

 

[SSFSO]

Oops, typed '05 by accident. It was actually 3DMark03. Haven't downloaded and tried '05 yet. I'll go back and edit that in the original post in a moment. I am not 100% sure if those FX-57's were overclocked or not, which is why I would like to know if anyone with a non-overclocked is getting a similar score. Thanks for the suggestion though, I may have to try that soon.

 

[SSFSO]

Well, after a bit more research and a week or so of using this computer without any major issues, I finally found a benchmark of the FX-57 that used Aquamark 3. The processor was not overclocked as far as I can tell and scored 12,605 for CPU while mine is still scoring around the low 9,000's. I am at a loss to figure out what the problem is with my system. Every test I have done shows my CPU to be running at the 2.8GHz it should be and my chipset/BIOS/drivers/etc are all up to date. If anyone has any ideas as to what the problem might be I could appreciate hearing them.

Edit: Benchmark Here

 

[Mirob]

You could get higher scores if you tweeked your memory just right(you didn't say what ram you have.) If things are left at auto it may not even be close, even running slow. Auto may not even be at SPD and alot of memory works tighter than that. Set it to 2-2-2-5, if it boots, that's the good stuff.

It's some work testing deferent timings with super pi and memory bandwith benchmarks. Ram tuned just right can make some difference.

You have one of THE best systems anyone could have there buddy.

edit; The best two gig kits do 2-3-2-5 1T

 

[Merc14]

Are you using 2 sticks of RAM or four? If four then that could be the reason for slightly lower benchmarks. 4 sticks of dual sided RAM with that CPU will default to 2T timing. Also, if two sticks then do you have the RAM in the correct slots for DDR? Should be slot 1 and 3 as you move away from the CPU on the A8N SLI

 

[SSFSO]

It seems I decided it would be cool if I stuck the RAM in the wrong slots when I assembled the computer, so I followed your advice and stuck the two sticks in Slot 1 and Slot 3. Upon running Aquamark I noticed a rather significant increase in FPS and ended up with a score of around 15,000 on GFX and 11,000 on CPU with a total of about 85,000. Thanks for helping me figure out that issue.

As for what type of RAM I am using, it is Corsair XMS 2GB (2 x 1GB) 184-Pin DDR SDRAM Unbuffered DDR 400 (PC 3200). I'll go mess with the RAM and set it to "2-2-2-5" or "2-3-2-5 1T" and report back soon.

Edit: After poking around my BIOS for a minute or two I realized I had no idea what I was doing and had best leave it as it was set for now. What settings do I change to achieve "2-2-2-5" or "2-3-2-5 1T?"

Edit2: As always Google has turned out to be most useful and I think I figured out what I need to change. I checked my SPD in CPU-Z and looked at the SPD Timings Table and saw that both my sticks were set to:
Frequency: 200 MHz
CAS# Latency: 3.0
RAS# to CAS#: 3
RAS# PRecharge: 3
TRas: 8

 

[Merc14]

Glad I could help. Sometimes the problem is the simplest, most obvious thing but we just can't see till someone else points it out. Then you bang your forehead and say duh. LOL.

Start here for 2/3/3/6 1T which is very quick for 2GB. (note: Tcl (CAS) set at 2 works for most everything but made my system crash in BF2 so I set it at 2.5 and BF 2 is stable again). These settings work on my rig but your set up can probably handle tighter timings. Start here and play with it as you move along:

DDR voltage at 2.65
400MHZ
Tcl 2.5 or 2 (see above)
Tras 6
Trcd 3
Trp 3
rest as set in auto
then:
1T/2T to 1T
and over 4GB of RAM stuff set to disable

The above should give you fairly fast benchmarks as well as stability. 2x1GB sticks won't cycle as fast as 2 x512MB sticks. The performance difference is negligible, however, unless you are benchmrking at the highest levels.