Super PI Contest

[imported_michaelpatrick33]

SuperPI8m 5:43

I ran this with the SSE3 patched edition. Like Duvie I watched an extra core sit idly by twiddling its thumbs.

Hey Duvie: just wanted to let you know that I got a final overclock (for right now, LOL) of 2750 at 1.52vcore at 46C under load 2Prime 95's running for 12 hours. I even got my 1gig Patriot dimms to run at 1.5 CAS latency (but they are incredibly touchy with dividers so I can't get them above 200. So I traded bandwidth for latency.

 

[McGeyser]

Duvie, what is your take on Super PI using both cores at 50% max load? Though this test is not multithreaded both cores are crunching calculations, at what appears to be half speed.

Is anyone else getting the same result from X2? Is it a perception error on my part? When I use Prime, I get 100% load on both cores, according to performance tab in task manager.

50% Load SSE3 and SP M 1.4

 

[Duvie]

Originally posted by: McGeyser
Duvie, what is your take on Super PI using both cores at 50% max load? Though this test is not multithreaded both cores are crunching calculations, at what appears to be half speed.

Is anyone else getting the same result from X2? Is it a perception error on my part? When I use Prime, I get 100% load on both cores, according to performance tab in task manager.

With prime I can get 2 instances to run...with superpi it will not run 2 instance...one will always give a non convergent right off the bat...eventhough I can run 1 instance of prime95 and 1 instance of superpi32mb....others hav esaid same thing...tell me how you got it to work....

 

[imported_michaelpatrick33]

Originally posted by: McGeyser
Duvie, what is your take on Super PI using both cores at 50% max load? Though this test is not multithreaded both cores are crunching calculations, at what appears to be half speed.

Is anyone else getting the same result from X2? Is it a perception error on my part? When I use Prime, I get 100% load on both cores, according to performance tab in task manager.

50% Load SSE3 and SP M 1.4

I also get 50% in task manager when running Superpi on my X2

 

[McGeyser]

That is the odd thing, I did nothing special. I made sure all other apps were closed, and simply ran SSE3 8M with task manager open to perfromance tab.

Another interesting thing was if I start Super PI while in the middle of opening Everest or CPUZ, the load seemed to favor core 1 not core 0, and then it would show more utilization on core 1 for the duration of the test.

 

[Duvie]

Yes superpi will only use 50% or waht amounts to 1 core only...2 instance will not use 25 each since I have not been able to get 2 work at same time....


Superpi is sngle threaded...It does not better with Ht and wont do any better with dual core....that is why listing dula and single core in this test is senseless....


let se most of the apps we love to do fast benches for are both single threaded.......superpi & 3dmark.....

 

[Duvie]

Originally posted by: McGeyser
That is the odd thing, I did nothing special. I made sure all other apps were closed, and simply ran SSE3 8M with task manager open to perfromance tab.

Another interesting thing was if I start Super PI while in the middle of opening Everest or CPUZ, the load seemed to favor core 1 not core 0, and then it would show more utilization on core 1 for the duration of the test.

NO....how did you get 2 instances to run simultaneously???

 

[McGeyser]

You saw the pic I posted right? I did nothing special. I made sure all was calm and then started the test. The pic clearly shows each core picking up half of the job, all on it's own.

EDIT: Sorry I should clarify I was only running one instance to get 50% on both cores.

 

[imported_michaelpatrick33]

Hey Duvie here are dual SuperPi running. Let's see if the Intel owners can beat two Superpi8m running simultaneously at these times, LOL. THeir dualcore should be able to match this I would imagine.

Screen of two superpi sse3 patched 8m running simultaneously

Screen showing finished simultaneous superpi sse3 patched 8m

 

[McGeyser]

Michaelpatrick33, you are seeing activity on both cores as well at 50% each?

If we are running at half speed, imagine if they optomized Super PI for Dual core. We should be technically twice as fast. So a score of 5m 52s would really be 3m 16s if we were getting our moneys worth from Super PI. Does anyone argue that assumption?

 

[McGeyser]

michaelpatrick33, how did you get two simultaneous instances to run?? I get the convergence error Duvie mentioned as well, Not Convergent in SQR05

 

[imported_michaelpatrick33]

Originally posted by: McGeyser
Michaelpatrick33, you are seeing activity on both cores as well at 50% each?

If we are running at half speed, imagine if they optomized Super PI for Dual core. We should be technically twice as fast. So a score of 5m 52s would really be 3m 16s if we were getting our moneys worth from Super PI. Does anyone argue that assumption?

Yes, when I run single instance of Superpi I see both cores working for a total of 50% in task manager which equals one full core.

That is why I ran two instances simultaneously and still scored around 6 minutes. My cpu usage in task manager went up to 100%. Notice in my screenshots that I have two superpi's running together and that the task manager in the lower right is all green, LOL.

I would like to see Intel processor users run two superpi instances simultaneously to see what their scores are.

So, the OP shoud have two instances of Superpi running simultaneously scores and single instance superpi scores.

Edit: Notice that my times are still just over 6 minutes for two Superpi8m running simultaneously!

 

[imported_michaelpatrick33]

Originally posted by: McGeyser
michaelpatrick33, how did you get two simultaneous instances to run?? I get the convergence error Duvie mentioned as well, Not Convergent in SQR05

I have the Superpi's running in two different folders. I will have to check to see if I used two different versions at the same time or if simply having them in two different folders allows two simultaneous runs

 

[coomar]

it wouldn't be twice as fast, the best you could probably hope for is 90% efficiency, 2x boost is the ideal, most of the time in multi-threaded apps it will be 50-90% increase

 

[imported_michaelpatrick33]

Originally posted by: coomar
it wouldn't be twice as fast, the best you could probably hope for is 90% efficiency, 2x boost is the ideal, most of the time in multi-threaded apps it will be 50-90% increase

Yes, but I ran two instances simulatneously and got socres of just over 6 minutes on my X2. The Intel owners need to do the same and see what their scores are. The single core AMD64 people as well. That is what the advantage of dualcore can be even in single threaded applications

 

[McGeyser]

Good call MP33! The seperate folders worked. I was able to complete 2 simultaneous SSE3 instances in 6m 04s.

EDIT: I was able to get a solid 100% load on both cores. I would say that a 6m 04s dual run should be the equivalent of a 3m 02s single core run, if Super PI were tuned for dual core. Can anyone disagree, with a valid arguement?

 

[imported_michaelpatrick33]

Originally posted by: McGeyser
Good call MP33! The seperate folders worked. I was able to complete 2 simultaneous SSE3 instances in 6m 04s.

Good times I got 6:10 on my second run for both superpis. I will tinker some more and see what I get. Now let's see single core Dothans run two simultaneous instances and get these times. MUUUHHHHAAAAHHHAHAH, LOL. Just kidding.

 

[imported_michaelpatrick33]

Here is my fastest two complete SSE3 superpi runs

5m 57

Edit: These are simultaneous runs, not separate runs

 

[chinkgai]

i never knew people didnt figure out to put superpi in two different folders and running them simultaneously from each folder

i've been doing that since day one...same thing with prime...two folders

no wonder you didnt want to do superpi tests (2x instances at the same time) duvie :laugh:

 

[imported_michaelpatrick33]

Originally posted by: chinkgai
i never knew people didnt figure out to put superpi in two different folders and running them simultaneously from each folder

i've been doing that since day one...same thing with prime...two folders

no wonder you didnt want to do superpi tests (2x instances at the same time) duvie :laugh:

LOL, Yeah I never bothered to try to put superpi into separate folders like prime95. Though, I never really used it much before. LOL. Oh well there is always a learning process.

 

[McGeyser]

With Prime, I use the same folder but tag the second shortcut as -A1. I never use Super PI, don't the Intel guys like Super PI the most?

 

[Duvie]

Originally posted by: coomar
it wouldn't be twice as fast, the best you could probably hope for is 90% efficiency, 2x boost is the ideal, most of the time in multi-threaded apps it will be 50-90% increase

I see we are speaking from experience knowledge...oops wait a minute...nope!!!

Tested at 2.574ghz

My speed with 1 instance = 5:59

My speed with 2 instances no affinity set =6:16-6:18
My speed with 2 instances w/ affnity set = 6:19-6:21
so it looks like 2 instances was 94.6 to 93.1%...not 2x but above your ideal 90 and more then your 50-90 range!!!

screenies will be linked momentarily...


NOtice how I had one core better then the other....

 

[imported_michaelpatrick33]

Originally posted by: McGeyser
Good call MP33! The seperate folders worked. I was able to complete 2 simultaneous SSE3 instances in 6m 04s.

EDIT: I was able to get a solid 100% load on both cores. I would say that a 6m 04s dual run should be the equivalent of a 3m 02s single core run, if Super PI were tuned for dual core. Can anyone disagree, with a valid arguement?

You usually don't get a 100% boost in performance in multi-threaded aware programs. I thought it ran from 50 to 90% with an occasional 100% boost. I think a previous poster already stated this, lol.

 

[imported_michaelpatrick33]

Originally posted by: Duvie

Originally posted by: coomar
it wouldn't be twice as fast, the best you could probably hope for is 90% efficiency, 2x boost is the ideal, most of the time in multi-threaded apps it will be 50-90% increase

I see we are speaking from experience knowledge...oops wait a minute...nope!!!

Tested at 2.574ghz

My speed with 1 instance = 5:59

My speed with 2 instances no affinity set = 6:16-6:18
My speed with 2 instances w/ affnity set = 6:19-6:21

so it looks like 2 instances was 94.6 to 93.1%...not 2x but above your ideal 90 and more then your 50-90 range!!!

screenies will be linked momentarily...


NOtice how I had one core better then the other....

I went from a 5:43 single instance score to a 5:57 dual instance score (5:57 for both superpi runs). Not too shabby with a 97% range

 

[Duvie]

Originally posted by: McGeyser
With Prime, I use the same folder but tag the second shortcut as -A1. I never use Super PI, don't the Intel guys like Super PI the most?

That is the way I do it as well and it wouldn't work with superpi that way...I think cause the access both files including the log file simultanelously....

Intel guys who run dothans like it and mainly the 1mb variety where they really wax the floor...they become more normal around the higher superpi test....Technonut has a screamin system and it appears to me at least with my testing that SSE3 doesn't do as much for the AMD chips as it does for the INtel chips....