Multicore Pi benchmark

[Vectronic]

Two issues with this multi-core Pi programs - Super-Pi correctly calculates the digits of Pi, multi-core Pi does not.

That was more or less my point (in response to your earlier reply)

The first "calculation" is always the same (practically speaking), regardless of clock settings... leaving it open, and calculating over and over again, it changes.

Even just looking through the posted images, the values are different, pick any 3 of them, they will all be different.

 

[Haserath]

For giggles I tested with all but 1 core off.
[/URL]
i7 920@3.4

 

[Vectronic]

Here is an animation of 5 different posted images...

Values are all different.

P.S. I don't remember Photoshops animation stuff being this annoying before... lol that took WAY longer than it should have.

 

[Idontcare]

Here is an animation of 5 different posted images...

Values are all different.

P.S. I don't remember Photoshops animation stuff being this annoying before... lol that took WAY longer than it should have.

Well that's no good (the lack of repeatability, not your image overlay, your sleuthing work itself is superb :thumbsup

 

[Vectronic]

I think it's creating unique identifiers for later trojan infections, disguised in pseudo PI, that's why the application is so big for what little it does ... I kid... kinda...lol, although it would be an interesting way of encryption/keys.... calculate some sort of CRC based on the PC, find that value in PI, use the offset to identify the computer later... ok I'll shut up.

 

[Idontcare]

I think it's creating unique identifiers for later trojan infections, disguised in pseudo PI, that's why the application is so big for what little it does ... I kid... kinda...lol, although it would be an interesting way of encryption/keys.... calculate some sort of CRC based on the PC, find that value in PI, use the offset to identify the computer later... ok I'll shut up.

:hmm: hmmm, we did just download this thing from a romanian website, didn't we

 

[Edgemeal]

:hmm: hmmm, we did just download this thing from a romanian website, didn't we

YIKES! Just close/delete this thread before we're all infected!

 

[TakeNoPrisoners]

 

[Idontcare]

YIKES! Just close/delete this thread before we're all infected!

We're just kidding

 

[moonbogg]

I could OC higher, but this is what I use my rig at, so heres what it got.
3930k @ 4.3
ram @ 1600
24s 855ms

 

[GTRagnarok]

The 3720QM can turbo up to 3.8Ghz on all cores.

 

[cytg111]

Q9450 @ 2.66

1 min 31 sec

 

[Lonbjerg]

Win XP 64

That is your problem.
That OS is weird and flunky.

Windows XP 64 bit should be banned.

Either go 32 bit XP...or go win7 x64...sorry to say

 

[cytg111]

I think it's creating unique identifiers for later trojan infections, disguised in pseudo PI, that's why the application is so big for what little it does ... I kid... kinda...lol, although it would be an interesting way of encryption/keys.... calculate some sort of CRC based on the PC, find that value in PI, use the offset to identify the computer later... ok I'll shut up.

Nice thinking dude .. ICU.

 

[SPBHM]

i3 2100 with single channel ram

 

[coffeejunkee]

why not run 4/8 instances of single core version at the same time

Or just use Hyper Pi.

 

[Edgemeal]

The 3720QM can turbo up to 3.8Ghz on all cores.

Why Intel spec shows, Max Turbo Frequency=3.6 GHz?

Or you set Turbo all cores to 3.8?

 

[Sable]

2500k @ 4.7

edit:

Ran it again and got 40 seconds.

 

[Sable]

So my initial run was 45 seconds, second was 42 seconds and fastest was 40 seconds. That's quite a range of results.

 

[Gikaseixas]

48.444 on the FX machine
will try the i7 860 and i7 mobile later

 

[Diogenes2]

23.808

System in sig line...

 

[firewolfsm]

86.621 on an i7-2620M at 2.691GHz

 

[Idontcare]

AMD FX-8350, 2.2GHz NB, DDR3-1866 10-10-10-28-CR2, Win7x64 all updates applied. I shutdown and restarted the Multi-Core Pi application between each test.

FX-8350 2.0GHz 52s 908ms
FX-8350 2.5GHz 1m 4s 561ms
FX-8350 3.0GHz 55s 208ms
FX-8350 3.5GHz 1m 0s 885ms
FX-8350 4.0GHz 56s 652ms
FX-8350 4.3GHz 1m 4s 439ms

Intel i7-3770K, DDR3-1866 10-10-10-28-CR2, Win7x64 all updates applied. I shutdown and restarted the Multi-Core Pi application between each test.

i7-3770K 2.0GHz 1m 1s 670ms
i7-3770K 2.5GHz 49s 146ms
i7-3770K 3.0GHz 46s 471ms
i7-3770K 3.5GHz 39s 491ms
i7-3770K 4.0GHz 31s 209ms

All these results are suspect. The AMD results are clearly borked, but if you look at the clockspeed scaling for the Intel processor you will note they are borked as well.

Going from 2.5GHz to 3GHz reduced the process time by a mere 2.7seconds, but going from 3.5GHz to 4GHz reduced process time by nearly 8seconds? :hmm:

I feel confident to declare this benchmark is functioning as a chaotic number generator, both in terms of the computed values of Pi as well as in terms of the computed process time for the bench to run. Can it get more broken than that?

 

[Edgemeal]

I feel confident to declare this benchmark is functioning as a chaotic number generator, both in terms of the computed values of Pi as well as in terms of the computed process time for the bench to run. Can it get more broken than that?

Ya those times don't make any sense, this so-called benchmark is useless!

Sorry for wasting you guys time, I won't even bother updating scores,
THREAD CLOSED!

 

[Makaveli]

AMD FX-8350, 2.2GHz NB, DDR3-1866 10-10-10-28-CR2, Win7x64 all updates applied. I shutdown and restarted the Multi-Core Pi application between each test.

FX-8350 2.0GHz 52s 908ms
FX-8350 2.5GHz 1m 4s 561ms
FX-8350 3.0GHz 55s 208ms
FX-8350 3.5GHz 1m 0s 885ms
FX-8350 4.0GHz 56s 652ms
FX-8350 4.3GHz 1m 4s 439ms

Intel i7-3770K, DDR3-1866 10-10-10-28-CR2, Win7x64 all updates applied. I shutdown and restarted the Multi-Core Pi application between each test.

i7-3770K 2.0GHz 1m 1s 670ms
i7-3770K 2.5GHz 49s 146ms
i7-3770K 3.0GHz 46s 471ms
i7-3770K 3.5GHz 39s 491ms
i7-3770K 4.0GHz 31s 209ms

All these results are suspect. The AMD results are clearly borked, but if you look at the clockspeed scaling for the Intel processor you will note they are borked as well.

Going from 2.5GHz to 3GHz reduced the process time by a mere 2.7seconds, but going from 3.5GHz to 4GHz reduced process time by nearly 8seconds? :hmm:

I feel confident to declare this benchmark is functioning as a chaotic number generator, both in terms of the computed values of Pi as well as in terms of the computed process time for the bench to run. Can it get more broken than that?

Those are some interesting results.

And the AMD ones are totally the uneven core speeds are slower than the even numbers regardless of higher clock speed!!