• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

AMD superpi score

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
Overclocker550 said:
should I be getting lower than 35s at 262x9 1:1? :confused:

Don't look at the fsb or HTT! Look at the size. 262x9 is 2358 mhz so 35 sec is entirely plausible. In fact, that is good for that mhz. Now if you could run 262 in synch at 2-2-2-5 with BH-5 (and 3.6 VDIMM) :rolleyes: you might get that score down by 1-2 sec. Good luck finding a S939 board that likes BH-5.
 
mantralord said:
No, my RAM is running at DDR333...this Patriot TCCD crap caps out at 240 fsb.

I thought some guys swear by that ram. Maybe they mean't swear at it.

2-3-3-6 I can believe, but at 2.5-3-3-6, it should go to 265-270 or so.
 
I got 32s should be able to get this down a sec or two in a couple days when I get better memory. Look and only 1.3Ghz too! :p
 

Attachments

  • SuperPi2.JPG
    SuperPi2.JPG
    64.5 KB · Views: 357
TimoneX said:
I got 32s should be able to get this down a sec or two in a couple days when I get better memory. Look and only 1.3Ghz too! :p
yep, see it's all about the RAM speed (and controller) and FSB.
 
My ram speed sux. I'm running at DDR436. 262x10 with a 166memclocklock on. Timings are 2.5-3-3-7. Got me some Platinum rev2 on the way though, so we'll see what some real memory speed does. I never see all this alleged damage CnQ does to performance either. With or without Cool 'n Quiet I get the same 32s, so I'm all about lowering the electric bill. :clap: 269Mhz gets me to 31s, but the system isn't 100% stable there.
 
Last edited:
mantralord said:
No, my RAM is running at DDR333...this Patriot TCCD crap caps out at 240 fsb.



lol its your weak ondie controller man, I can hit a little over 250fsb with my ondie stably and can do 3dmark runs in the low 260s with the occasional crashing. I tried 3-4-4-10 timings and it didnt help so this is more proof the rams not maxed.


" I got 32s should be able to get this down a sec or two in a couple days when I get better memory. Look and only 1.3Ghz too!"


yea you used clockgen to downclock your cpu after running superPI at your max oc. either that or photoshop :rolleyes:

"oh, hmm, im getting 34s with DDR514. 3-4-4-8 and single channel tho..."

your 2.6GHz claw is doing bad, I get the same score with my winchester at 2.4 cause your ram timings suck and you probably got winxp too.
 
Aphex_Tom_9 said:
oh, hmm, im getting 34s with DDR514. 3-4-4-8 and single channel tho...

...and what's your memory bandwidth & latency look like? Is that at 1t command? I suspect even at 218Mhz(DDR436) my dual channel bandwidth is higher. I'm right around 6700Mb/s as I recall. This memory(PC4000 EL gold) is capable of much higher speeds but not at 1t command and the bandwidth loss is huge, so this setting is better than much higher speeds and 2t command.
 
TimoneX said:
...and what's your memory bandwidth & latency look like? Is that at 1t command? I suspect even at 218Mhz(DDR436) my dual channel bandwidth is higher. I'm right around 6700Mb/s as I recall. This memory(PC4000 EL gold) is capable of much higher speeds but not at 1t command and the bandwidth loss is huge, so this setting is better than much higher speeds and 2t command.

yeah, im running 1T, single ch. this ram can do DDR580 with 2T, but my chip is a C0 revision, and cant do 2T. my bandwidth is like 3800 IIRC. there lies our difference.
 
Overclocker550 said:
lol its your weak ondie controller man, I can hit a little over 250fsb with my ondie stably and can do 3dmark runs in the low 260s with the occasional crashing. I tried 3-4-4-10 timings and it didnt help so this is more proof the rams not maxed.


" I got 32s should be able to get this down a sec or two in a couple days when I get better memory. Look and only 1.3Ghz too!"


yea you used clockgen to downclock your cpu after running superPI at your max oc. either that or photoshop :rolleyes:

"oh, hmm, im getting 34s with DDR514. 3-4-4-8 and single channel tho..."

your 2.6GHz claw is doing bad, I get the same score with my winchester at 2.4 cause your ram timings suck and you probably got winxp too.

LOL dude you're a hoot. What did you get the generic A64 that came without a little feature called COOL N QUIET?

To accuse another member of cheating on a benchmark that's weak bud.

Why...why must you constantly insult other members and run them & their machines down? Are you really SO competitive you have to act this way? You accuse me of cheating and run down his setup and his OS all within a few minutes. Do you feel better about yourself now?
 
Last edited:
Aphex_Tom_9 said:
yeah, im running 1T, single ch. this ram can do DDR580 with 2T, but my chip is a C0 revision, and cant do 2T. my bandwidth is like 3800 IIRC. there lies our difference.

Yup sounds about right. SuperPi seems to like memory bandwidth and it doesn't seem like larger L2 cache is all that helpful in this test. Have you tried A64 tweaker at all? Doesn't make a huge difference but I picked up a couple hundred MB/s from it.

||Console|| said:
38 sec
I dont think i can get this down much more , board wont let me go any higher

That's a nice CPU ya got there. What BIOS are you running? Have you tried out any of the 2t CPC modded BIOSes?
 
Last edited:
Hmm, my 3000+ winnie at stock with memory 1:1 200 @ 2-2-2-5 dual channel gets 47 seconds. Sad thing is my friend's 2500 barton at stock also gets 47 seconds. I guess I just plain suck.
 
TimoneX said:
LOL dude you're a hoot. What did you get the generic A64 that came without a little feature called COOL N QUIET?

To accuse another member of cheating on a benchmark that's weak bud.

Why...why must you constantly insult other members and run them & their machines down? Are you really SO competitive you have to act this way? You accuse me of cheating and run down his setup and his OS all within a few minutes. Do you feel better about yourself now?



did I say he cheated? no. he benched first then downclocked his cpu
 
Hrm.. so u think it´s cheating to use the prescott patch?? Isn´t i all about to se which computer that calculates a 1mb pi number as fast as posible??? So shouldn´t the app (superpi) take advantage of the whole cpu??? Well anyway without the patch I got 35 secs @ 3.7ghz and at stock speed a got 40 sec!!
 
Overclocker550 said:
did I say he cheated? no. he benched first then downclocked his cpu

Actually you did, you mentioned something about maximum overclocks and reducing speed. Guess that would fall under the catagory of "misleading" then? Guess you forgot about cool n quiet eh? :cool: Does look funny with that 5x multiplyer. 262x10 is my maximum overclock at least 100% stable and I see no reason to run anything that's other than 100% stable.

Scph9002 said:
Hrm.. so u think it´s cheating to use the prescott patch?? Isn´t i all about to se which computer that calculates a 1mb pi number as fast as posible??? So shouldn´t the app (superpi) take advantage of the whole cpu??? Well anyway without the patch I got 35 secs @ 3.7ghz and at stock speed a got 40 sec!!

Dunno if it's "cheating" persay, whether it's fair or unfair. Who cares anyway it's just a benchmark. Besides If it's SSE3 that's giving the majority of the boost that'll be in the AMD camp very soon.
 
Last edited:
Back