3500+ Winchester O/C Results

[PVD]

So I have my rig completely stable with Prime 95 at 2.53 GHZ with still the stock voltages. My RAM is set at 2-2-2-5 right now running fine. My question is, do I now up the multipliers further (right now 11 x 230) until I start to get system instablility, and then up my vcore to +3% ?? My HT is set at 4x.


I am cooling with an xp-90 and a 92mm panaflo


PVD

 

[mickles]

You won't be able to up the multiplier, it is locked, you can move it down if you want, but essentially yes, with HTT.
Up the HTT (FSB equivalent) and up the voltage if necessary. If you find it has trouble booting at higher FSB's (mid 240 to 250 range) try lowering your HT multiplier to 3x.

 

[PVD]

sorry. I meant FSB (HT), not multiplier. Thanks for the tip. Will try. 240 x 11 will be 2640 . Question- Once my sytem acts unstable, should I lower the HT Multiplier to 3x or increase the vcore voltage first??

 

[klah]

It will be a lot easier if you find out how high the cpu, memory, and htt can get individually first before running at them simultaneously: http://www.aoaforums.com/forum...257405&postcount=2

edit: the vcore they list is for 130nm parts. for 90nm cpus reduce that by ~0.15V.

 

[PVD]

when does upping the ram and cpu voltages come into play???

 

[PVD]

ttt

 

[Markfw]

Based on Duvies experience, you will probably need more voltage to get over 2.5 (for cpu and memory) But who knows ? crank up the speed, and when it becomes unstable, up the voltage (both very slowly)

 

[PVD]

I am at 2.53 right now on stock voltage ,markfw900

 

[Duvie]

I would probably imagine that a 4x HTT wont be a factor until after 250 anyways....However at this testing juncture I still would have set it to 3x right now to take it out of the equation...

place ram divider at 166 up to 240 and then drop it after that...The key here is to stay below rated ddr speed as well as keep the timings conservative...isolate the cpu and keep the ram out of it...

Ultimately we should just have the cpu and whether it can do an fsb at a certain vcore...plain and simple...

One thing at the point you are in speed I may try in smaller 5 fsb jumps, casue an 11fsb jump for me took .15v more to run. If you are near the end of the cpu in ocing it could change fast....

 

[PVD]

I am now @ 2585 MHZ with STOCK voltages and am prime95 stable (see below). This RAM and CPU are amazing. Still gonna go higher. Will let you know. BTW- my HTT is at 4x, RAM @ 1T, 2-2-2-5 timings

1st test is at 2520MHZ, second at 2.585 GHZ. Can see the speed difference too. On to 2.64 (11x240)!!


AMD Athlon(tm) 64 Processor 3500+
CPU speed: 2530.42 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE, SSE2
L1 cache size: 64 KB
L2 cache size: 512 KB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
L1 TLBS: 32
L2 TLBS: 512
Prime95 version 23.8, RdtscTiming=1
Best time for 384K FFT length: 17.319 ms.
Best time for 448K FFT length: 20.735 ms.
Best time for 512K FFT length: 23.110 ms.
Best time for 640K FFT length: 29.125 ms.
Best time for 768K FFT length: 35.368 ms.
Best time for 896K FFT length: 42.657 ms.
Best time for 1024K FFT length: 47.817 ms.
Best time for 1280K FFT length: 63.837 ms.
Best time for 1536K FFT length: 76.562 ms.
Best time for 1792K FFT length: 92.430 ms.
Best time for 2048K FFT length: 104.206 ms.
[Fri Dec 17 12:53:33 2004]
Self-test 1024K passed!
[Fri Dec 17 13:09:05 2004]
Self-test 8K passed!
[Fri Dec 17 13:24:06 2004]
Self-test 10K passed!
[Fri Dec 17 13:41:46 2004]
Self-test 896K passed!
Compare your results to other computers at http://www.mersenne.org/bench.htm
That web page also contains instructions on how your results can be included.

AMD Athlon(tm) 64 Processor 3500+
CPU speed: 2585.01 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE, SSE2
L1 cache size: 64 KB
L2 cache size: 512 KB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
L1 TLBS: 32
L2 TLBS: 512
Prime95 version 23.8, RdtscTiming=1
Best time for 384K FFT length: 16.947 ms.
Best time for 448K FFT length: 20.299 ms.
Best time for 512K FFT length: 22.714 ms.
Best time for 640K FFT length: 28.568 ms.
Best time for 768K FFT length: 34.617 ms.
Best time for 896K FFT length: 41.772 ms.
Best time for 1024K FFT length: 46.763 ms.
Best time for 1280K FFT length: 62.390 ms.
Best time for 1536K FFT length: 74.841 ms.
Best time for 1792K FFT length: 92.095 ms.
Best time for 2048K FFT length: 103.539 ms.
AMD Athlon(tm) 64 Processor 3500+AMD Athlon(tm) 64 Processor 3500+

 

[Googer]

IF your cpu has hit the celing, lower the cpu multiplier and raise the fsb (ht) untill the cpu is back up to it's maximum (2585).

 

[PVD]

Well I found my cpu limit, I GUESS. I got to 2607 MHZ on 237 x 11, 2 hour stable on prime95.

Here is the results:


AMD Athlon(tm) 64 Processor 3500+
CPU speed: 2607.10 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE, SSE2
L1 cache size: 64 KB
L2 cache size: 512 KB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
L1 TLBS: 32
L2 TLBS: 512
Prime95 version 23.8, RdtscTiming=1
Best time for 384K FFT length: 16.820 ms.
Best time for 448K FFT length: 20.182 ms.
Best time for 512K FFT length: 22.446 ms.
Best time for 640K FFT length: 28.283 ms.
Best time for 768K FFT length: 34.293 ms.
Best time for 896K FFT length: 41.412 ms.
Best time for 1024K FFT length: 46.358 ms.
Best time for 1280K FFT length: 61.843 ms.
Best time for 1536K FFT length: 74.239 ms.
Best time for 1792K FFT length: 89.701 ms.
Best time for 2048K FFT length: 100.995 ms.


When I went 238x11 it failed after 15 minutes. 240 x11 failed after 1 minute, but it booted up. I guess my next step is to do what Googer said and see if I can get it higher by lowering the multiplier to 10 and raising the fsb as high as it will go? I will need to set HTT at 3x do do this. Will this make a slower system at the same GHZ (2.607) as having my HTT at 4x and my multiplier at 11 x 237?? Thanks guys.


PVD

 

[tterris]

PVD: in your bios under cell menu, what are the exact options you have each thing set to? i'd like to know because i have the same cpu and motherboard as you and i'm having a hard time getting my fsb past 213 stable...

 

[Googer]

Now that you know the upper limit of your cpu. Increase the bus using a 10 cpu multiplier untill you are back up to 2600 mhz or so. Find out how high the mother board will go now that your cpu has hit the celing.
Keep this in mind. When Intel went from a 533mhz bus (3066 CPU) to an 800mhz (3000mhz cpu) the slower cpu on the faster 800mhz bus was on average obout 20-40% faster.despite being 66mhz slower.

PS 66mhz slower reminds me of the old 16mhz days of 386! WOW, back then a 66mhz processor seemed unimaginable. No less 1000mhz!

 

[Googer]

Originally posted by: tterris
PVD: in your bios, what are the exact options you have each thing set to? i'd like to know because i have the same cpu and motherboard as you and i'm having a hard time getting my fsb past 213 stable...

Individual results may vary.

 

[tterris]

i obviously know this goober. thanks for the inciteful help though - i really appreciate it.

 

[Googer]

Your welcome my fellow anandtech buddy!

 

[PVD]

TTerris- I have the RAM at 2-2-2-5 timings. What Ram do you have?? I have the ram at 1t timings. I have the agp lock at 67. ram divider set at 166 mhz. hmm.... can't remember all the other categories. If you need the rest, just shoot me a PM with teh categories and I will be happy to look it up.


PVD



Tomorrow will shoot higher and will post results..................

 

[PVD]

tteris- Also, make sure if you have an SATA drive it is on the 3rd or 4th channel

 

[Googer]

Originally posted by: PVD
tteris- Also, make sure if you have an SATA drive it is on the 3rd or 4th channel

I am assuming that having your Sata Drive one Channel 3or4 must be some thing that is specific to you motherboard model. Am I correct?

 

[PVD]

yeah, I think so. For whatever reason MSI Neo boards OC betterl ike this. Has something to do with overloading the AGP bus or something. I am not entirely sure if I remember it right.



Hey googer- how do memory timings play into my overclock?? Is my RAM setup/timings optimal??

 

[Googer]

In my eyes MSI is a slightly shadey company. I wont go into details. There products have some unique querks to them and they market false technology. Any company that has to change its name might be one to avoid.
Double Post OOOOPS!

 

[Googer]

In my eyes MSI is a slightly shadey company. I wont go into details. There products have some unique querks to them and they market false technology.

 

[tterris]

where did you find the option to adjust your ram divider? are you using the bios version 1.4 phoenix that came with the neo2 platinum board, or are you using something else?

question about oc'ing though.. if you know for a fact that your cpu can overclock to a 240 fsb with standard 11 multiplier, can you go straight to 240 fsb or do you have to work your way up to it at like a few mhz at a time? i went from 200 mhz to 210 and it barely booted, but when i went to open up cpu-z windows restarted on me.

 

[Googer]

Originally posted by: PVD
yeah, I think so. For whatever reason MSI Neo boards OC betterl ike this. Has something to do with overloading the AGP bus or something. I am not entirely sure if I remember it right.



Hey googer- how do memory timings play into my overclock?? Is my RAM setup/timings optimal??

When you start to get some instability increase the timings. Just bear in mind that you can destroy you ram if things are not done right (I have seen it happen first hand). As for optimal setup/timings when overclocking it is unique from system to system, I cannot really say with out seeing you computer first hand.

Remember when this second wave of the overclocking craze begain in 1997-98 (Celeron A), the goal of all overclockers at the time was to spend as little money as they could to get a lot of performance.

I remember OCing 386's that were on their last leg so we could play DOOM in DOS. At that time only old stuff ever got overclocked and no one really considerd OCing a new system like they do today. And now these days they Spend BIG BUCKS to do it. If you ask me; any of the reasons people OC today, just does not make much sence.