SUPER PI scores

[pt]

Just got super pi and decided to run it at the CPU/MEM stock speeds

Edit: can't upload image

the time for the 16K was of 00.279s
and for the 32K was of 00.718s

Is this a good result for a AMD 3000+ at 1800mhz?

 

[Alec§taar]

One thing to note about "Std." version of SuperPI:

It's a SINGLE-THREADED program, & will NOT fully extoll the abilities/virtues present/inherent in DualCore/SMP/HyperThreaded CPU bearing systems!

(Keep this in mind)

IIRC, there IS a way to "rig" the std. model to fully exploit the virtues of SuperPi std. build, but I cannot recall the specifics needed though... sorry on that account/note!

Could I think of a "rig job"? Sure - when you start 1 instance of it?? Using taskmgr.exe to set its affinity to a particular CPU core prior to starting the test will do it... you would run 2 instances of the single thread model of it this way, 1 to each core...



* There IS a version that IS "multi-threaded" designed, & thus, inherently SMP/DualCore ready & capable coded, but, on a single CPU machine?

ALSO, do keep THIS point in mind as well:

Technically, single threaded code RUNS FASTER, & has less overheads in the process scheduler than multithreaded code would, on a SINGLE cpu system (single core that is). The single threaded model will be the "Best/Better" performer on a single core system!

(InfraRed I am fairly certain, has D/L links for that model... name of file is: SuperPi 2004 (download filename = sp2004exe_20060405.cab) iirc! Double-Check w/ InfraRed)

APK

 

[pt]

i used the one on the TPU download section, i have a single core cpu, so i think it works fine, what i want is to compare times on AMD cpu's, but thanks for the advice

 

[cdawall]

50s stock 41s w/ oc -1m
.343s stock .282s w/ oc -16k
.718s stock .594s w/ oc -32k

for a64 3000+ (s754)
2ghz stock ram timings 2.5-2-2-6 1T ddr2100 266mhz
2.44ghz oc ram timings 2.5-2-2-6 1T ddr2100 326.6mhz

 

[dccool879]

my amd 3000+ @ 2.6 ghz is 35 sec for 1M, not sure what it is stock

 

[Deleted member 24505]

heres my pi score-

 

[infrared]

P4 640 @ 4.88ghz = 27.000 superpi

memory 1:1, 650mhz 4-4-4-9

I can do better still when i get some better memory. My target is 26s

 

[pt]

for 1m at stock speeds it took 52.9
Pretty good times Infrared, how much did it do at 5.0?

 

[infrared]

5ghz isn't stable

 

[pt]

with an overclock it does 35,4 s in 1M, a 17 seconds increase

 

[mikelopez]

http://valid.x86-secret.com/show_oc.php?id=109515

 

[pt]

Nice overclock and nice CPU, i want one too
It's yours?

 

[magibeg]

lol, i just find it funny a low end conroe just beat the crap out of everyones scores (yes i realize it is overclocked but still!)

 

[mikelopez]

Nice overclock and nice CPU, i want one too
It's yours?

Thanks, and yes it is mine. Look here:

http://forums.techpowerup.com/showthread.php?t=14404

lol, i just find it funny a low end conroe just beat the crap out of everyones scores (yes i realize it is overclocked but still!)

This isnt even close to what these guys are doing:

http://www.xtremesystems.org/forums/showthread.php?t=106685&page=2

 

[Alec§taar]

Well, single-threaded or not (DualCore user here) I gave it a go, w/ diff. builds too

I took this test, single-threaded code or not (see my first post in regards to my thoughts on that), & used the "stock/oem" model 1.4, & ones also geared to speed up its calculations via the SSE2 &/or SSE3 instruction sets for floating point calculations.

Here are my results:

Super-Pi 1.4 stock/oem model:

Super-Pi 1.4 SSE2 enhanced model:

Super-Pi 1.4 SSE3 enhanced model:

* Oddly, though the SSE2 &/or SSE3 enhanced models in theory OUGHT TO HELP imo, more than they did?

Well - I don't see radical increases, yes some (small) but nothing "mind-blowing" either, about 1.1 seconds diff.!

APK

P.S.=> I kept it @ the 1M calculations mark/19 iterations... & MAN!

You guys have some FAST machines boy... whew! apk

 

[cdawall]

well you all still beat my crummy best of 41s though i am going to try super pi at 2.5ghz

 

[cjoyce1980]

see below

 

[Lekamies]

 

[pt]

What cooling are you using on your 3000?

 

[dccool879]

is it safe to go to 1.7 volts on the 3000+???

 

[pt]

is it safe to go to 1.7 volts on the 3000+???

if you have proper cooling you can go higher than 1.7

 

[dccool879]

i thought it was rule of thumb that over 1.6 starts to reduce ur cpu's life

 

[Lekamies]

What cooling are you using on your 3000?

Watercooling like this(only pump is different), Idle temp 26-28c and 100% cpu usage 36-40c (after over 9 hour divx packing temp was 42c)

over 10 months it has been 1,7v 24/7 without problems

 

[cdawall]

well i still have a stock amd hsf run at a little over 1.7v and the highest temp i have seen was 55C so you should be plenty safe at 1.7v just dont try what i got witha stock hsf you will toast your s939

 

[Boneface]

heres mine at 38s at 1m