A8N32-SLI-Deluxe

[Parad0x420]

Got a Question ..

Will the Asus A8N32-sli Deluxe support a 9800 GTX wih proper power supply ???

I might be able to test something similar.. got a Sapphire HD4870 on the way

Ant follow this LINK

its a very well written guide on OC'ing and BIOS layout of the A8N32-SLI, and guess who the author is? (take Neb's advice about QuiteIce) hehe
i bookmarked it after i picked up this board when my Fatal1ty AN8 SLI died

 

[TheAnt]

thanks for all your input guys, appreciated.

 

[QuietIce]

So how did the OC go?!?!? I'm always curious how chips and boards clock.


I still run several s939's in my SETI farm and my data gathering (when it comes to the A8N32) is legendary around here - I love those boards and grab every little piece of info I can about 'em ...!->

 

[rhaazz]

hey guys,

I hope you can give some pointers about why I can't squeeze more juice out of my system which has the same mobo and proc. However, I have 4GB of Corsair's TWINX2048-3200 installed. And that's where the problem may be.
My FSB is currently running at 230MHz at 12x and HTT at 4. Memory limited to 166 which is a 5:4 ratio I believe. VCore at 1.5, everything else on auto. Once I raise the FSB to 235, my system crashes on me. I tried raising voltage but that didn't help. So I'm thinking it must be the RAM. Any pointers?

 

[QuietIce]

2760 MHz isn't bad for some of the older chips. I run my Opty 180 at 2700 most of the time though it will run 2800, which I use for SETI races.

Have you read through the first few posts of this thread?

What type of chip are you running and do you know what the stepping is?
What's your CPU temp and type of cooling? Do you have the RAM at 2T?
"Need input" ...

 

[rhaazz]

hey qi, thanks for your feedback. I did read the thread you referred to and it helped getting familiar with the BIOS.

I'm not sure what you mean by 'type of chip' but it is the same AMD X2 4800 (Toledo) cpu TheAnt is using. I don't know the stepping right now but I will look it up (CPU-Z does provide the info, right?)

My system is watercooled with a Koolance Exos2 so the cpu temp is around 38.

As far as memory settings are concerned, I left the settings on auto which resulted in 2T when checking CPU-Z.

 

[QuietIce]

4800 is what I was looking for - and the stepping is stamped on the IHS; it is not available electronically. It helps to know what the stepping is so one knows what to expect out of the chip. My neither '05 chip, the Opty 180 or the 4400, will run nearly as fast as the '07 Opty 165's I've got. The older chips stop ~2.8 GHz - the 165's ~3.0 GHz.

Take the settings off auto. Even if you keep the stock settings you should set them manually or the BIOS may change them. I thought I remembered 2T being default but it's been a looong time since I used default for anything. I'm a firm believer in manual settings, otherwise you have no control.

Just for grins take out two RAM sticks and try the OC that way (and at 1T). Then you'll be comparing apples to apples on the OC. Four sticks does change things a lot but there may be other things going on - who knows ...?

 

[rhaazz]

CPUZ gives me the following info about the CPU:
Family F
Model 3
Stepping 2
Ext. Family F
Ext. Model 23
Revision JH-E6

My overall question is this: am I gaining anything in terms of overall system performance, for ex. with playing games, by raising my FSB but lowering my memory throughput? I guess I should run 3dmark and find out....

 

[QuietIce]

No you aren't. You should set it up so the CPU is running as fast as possible and the RAM is running as fast as possible at that CPU speed. As long as you've done that the actual values for the clock, CPU multiplier, and memory ratio don't matter a bit - it's still the same throughput.

The third deciding factor is the HT Link speed. With some newer video you might see a difference in performance with a higher HT Link speed but I still doubt anything above 1000 MHz is going to make a difference.



What are the stock timings of your RAM; 2-3-3-6??? If so that RAM should run 230 MHz @ 3-4-4-8 (at 1:1 = MemClock 200) - though I admit I haven't tried it with four sticks ...

 

[rhaazz]

It is 3-4-4-8. I will set the memclock back to 200 from 166 as you suggested and see what happens.

 

[QuietIce]

It probably won't run like that; my bad, I thought all XMS was lower latency than CAS 3.

But 251x11 at MemClock = 166 will give you ~198 MHz, which shouldn't be a problem ...

 

[rhaazz]

qi, I did what you said and set the timings to 3-4-4-8, leaving the memclock at 166 but lowering the multiplier to 11. This allowed me to raise the FSB to 255. At 260, I'm running into stability problems again. What do you think: more juice? I ran 3dMark'06 benchmarks after every change, and so far the increase in score is marginal. So I hope there is a way to raise the FSB even more. Btw, temperature is at 43 under load, 41 idle.

 

[QuietIce]

As long as your LOAD temps are under 50°C (45°C for water) crank that sucker up!!! The board and the chip will both take it well as long as the heatpipe radiator up by the CPU has plenty of air. I've run my vCore up to 1.35v with CPU Over-volt [Enabled] (= 1.55v total). It takes 1.325v + Over-volt to get my Opty 180 up to 255x11 so go for it as long as your temps are good.

Note at that extreme you may have reboot issues. Some systems don't want to reboot near the top end (a couple of mine won't) but it will shut down and cold boot just fine and is completely stable ...

 

[rhaazz]

cool, thanks.

 

[rhaazz]

ok, I'm back after a bunch of benchmark tests. frankly, i'm quite disappointed by the results and hope to hear that i did something wrong. basically, the increase in performance running 3dMark'06 from stock settings to oc'ing @ 255MHz was a mere 1.4%, from 4574 to 4637! i increase the FSB by 27.5% and only yield 1% overall performance gain? that's nothing. something must be wrong. what are your performance gains and how did you determine them?

 

[QuietIce]

Are you looking at the overall 3DMark score or the CPU part of the score? If your CPU isn't bottle-necking your GPU then OC'ing the CPU has little effect on the total 3DMark score. For a better overall benchmark SiSandra or PCMark are better choices. 3DMark is a video only benchmark while the others test CPU, RAM, HDD, etc., as well.


Personally, I use SETI as a "benchmark". Obviously it can't be run in a few minutes but one day of crunching gives me all the data I need for a comparison. SETI is almost 100% CPU and RAM ...

 

[rhaazz]

i'm actually looking at all the scores 3dMark spits out, including overall and CPU. Since I'm running a 8800GTS 640MB gfx card, I would think that the CPU is the bottleneck, not the GPU. I will run SiSandra and PCMark as well per your suggestion.
what about running built-in benchmarks such as the one in Crysis? I am looking for overall performance improvements in games first and foremost.

Thanks again for your feedback!

 

[rhaazz]

Does Sisandra Lite suffice? Which version should I use?

 

[QuietIce]

Lite is fine and often used since it's free w/no time limit. It doesn't matter which version as long as you stick to that version with the machine you're testing. For these chips the older version is as good as the newest one.

Running in-game tests is fine - anything to test the system as long as you use the same benchmark and settings every time.


BTW - Is the HT Link multi set to 4x ...?

 

[rhaazz]

it was when I tried to boost the FSB to 255. Over the weekend I have moved everything back to stock but manually set the HT multi at 5x and memclock limit to 200 and started to increase the FSB again. With the multi at 5x, all I can get is a FSB of 220. The reason why I did is that I wanted to test the proformance at stock and have a reference point to compare my oc settings to. With SiSandra, PCMark and the in-game tests, I now have more testing to run before increasing the FSB again and lowering the multi to 4x.