***** Multicore Pi benchmark - AMD results *****

[FlanK3r]

Here is the link of new benchmark:
http://www.pcgamingxtreme.ro/forum/v...php?f=5&t=1909
benchmark that calculates PI using a multithreaded parallel algorithm written in C++. PI is calculated with 80.000 decimals using Bailey?Borwein?Plouffe formula.

Example of result info

• CPU info, clock, voltage. cooling
• memory info, clock, timings ...
• motherboard info
• printscreen of result + double CPU-Z

AMD FX-8350@4013 MHz 2408 MHz CPUNB, 1.25V cpu voltage, wattercooling Corsair H100
G.Skill TridentX 2400MHz@ 2408MHZ 10-12-12-31-45-1T
Asus Crosshair V Formula-Z

[Zeus]

I will post up some results soon but the outcome of this bench seems pretty inconsistent. Best so far was 45seconds but i'm sure that will be easily improved. Let me concentrate on CB first, once i've hit the limit on that i will try this bench.

[ReKcOlNu]

Here is the link of new benchmark:
http://www.pcgamingxtreme.ro/forum/v...php?f=5&t=1909

Can't extract the exe, using winrar it says archive is unknown format or damaged.

[FlanK3r]

try my alternative link:
http://ulozto.cz/x6ByRgG/multicorepi-zip

[ReKcOlNu]

try my alternative link:
http://ulozto.cz/x6ByRgG/multicorepi-zip

That exe is dated 01/21/2013, v1.0.0.1

I tried the original link again and it is OK now.
Exe is dated 01/22/2013, v1.0.0.2.

Don't have my AMD setup,.. FWIW, i5-3570K @ 4GHz: 42.591 secs.
http://imageshack.us/a/img341/9931/m...i3570k4ghz.png

[FlanK3r]

SO, they changed something... (maybe menu?).

[ReKcOlNu]

SO, they changed something... (maybe menu?).

Not sure you can really call this app a benchmark, times don't make any sense, How can an FX @ 2GHz be faster then a FX @ 4GHz or 4.3GHz?
http://forums.anandtech.com/showpost...2&postcount=48

[FlanK3r]

we can try it personally....

[G.Foyle]

I did something very similar once - a BBP formula multithreaded PI calculator. Mine is without GUI and uses an extremely basic method of time measurement.

Some things to note: BBP formula is extremely slow and bad for calculating PI. It's good if you need to extract one digit of PI at an arbitrary position without calculating anything before it; it's also good for loading all threads at once, since you can just assign 1/n digits to every one of n threads. But if you want a real, useful multithreaded pi calculator, use y-cruncher - it's the fastest, strongest, and a better test of stability than OCCT. Only downside is that it uses a LOT of memory for bigger calculations.

[CrimInalA]

time : 53s 392ms
Relevant system info :
Corsair Carbide 400R 4x120mm Papst 4412GL - 1x120mm Noctua NF-12P -!- Asus Crosshair IV Formula BIOS 2101 with RAID ROM 3.3.1540.17 -!- AMD Phenom II X6 1090T @ 4000mhz 1.45V Cpu-NB 3000Mhz -!- Corsair H70 push Papst 4412 F/2GP -!- 4x4gb Crucial Ballistix Tracers @ 1600mhz 7-7-7 1.55V

[felix_w]

Score : 52sec 301ms

AMD Phenom II X6 1090T @ 4000mhz 1.4875V | Cpu-NB 2800Mhz 1.325V
OCZ Blade LV 1150MHz@ 1066MHZ 5-5-5-18-24-2T 1.84V 4x2GB
Asus M4A79 Bios 4101
All-In-Case Watercooling | D-tek Fuzion v2.0 | Black Ice GTS 1x360 & 2x120 | Scythe Gentle Typhoons 1850rpm

[Zeus]

I don't get this benchmark, i ran it at 5.2G and only still got 48 seconds or something. Best was 45 seconds at 4.7 or 4.8G. WTH?

[FlanK3r]

Hope, it will be fixing. Looks some bug at AMD based CPUs...

[FlanK3r]

new version of program?

http://www.pcgamingxtreme.ro/forum/d...ile.php?id=660

[Daveburt714]

Hey Flank3r,

I found this over on TR, it may be the same one you linked above. It looks similar but the GUI's different...

I didn't do much testing, but at least it seems to respond to clock changes (in the right way) compared to the original version.
http://www.pcgamingxtreme.ro/

It's still a little quirky, especially when it comes to clock speeds. Regardless of what the bench says, actual clocks on this run where ~4.9 Ghz .



Hopefully this bench will turn out to be useful in the long run...
At least the 16.162 sec score looks decent for an FX.

[felix_w]

new version of program?

http://www.pcgamingxtreme.ro/forum/d...ile.php?id=660

Hey Flank3r,

I found this over on TR, it may be the same one you linked above. It looks similar but the GUI's different...

I didn't do much testing, but at least it seems to respond to clock changes (in the right way) compared to the original version.
http://www.pcgamingxtreme.ro/

It's still a little quirky, especially when it comes to clock speeds. Regardless of what the bench says, actual clocks on this run where ~4.9 Ghz .



Hopefully this bench will turn out to be useful in the long run...
At least the 16.162 sec score looks decent for an FX.

Τhe benchmark discussed in the beginning is Pi, the ones you found are prime numbers...

[Daveburt714]

Τhe benchmark discussed in the beginning is Pi, the ones you found are prime numbers...

That's what I get for posting a message right before bed...

[felix_w]

That's what I get for posting a message right before bed...

nah, the Multicore Pi link from anandtech forums is also pointing to Multicore Prime thread in pcgamingxtreme.ro forums....

Flank3r's first post and last post are also linking to Multicore Prime, not Pi.


Quote from hwbot forums :

I just renamed Multi Core PI to Multi-Core Prime back then.

Multi-Core PI is no longer in development due to unsatisfied multi-core scaling.

The new benchmark is Multi-Core Prime, i'm the owner / developer.

I corrected the problem regarding submission using datafile, please sire confirm if everything is allright for now.

And I also change the status to Public (BETA) and create a new thread to discuss benchmark scores.

Thank you all very much.

[FlanK3r]

SO, we can now tetsing multicore-prime?

[felix_w]

There is already a thread for that here, however maybe you can rename this thread's title and make it "Multi-Core Prime - AMD Results".

[felix_w]

Maybe you should not rename the thread's title because i found the reply of the program's author in the other thread, saying exact the opposite from the previous statement, that MC Prime is stopped and that the new benchmark is MC Pi....

OK, thanks all for testing this. The software is optimized for multicore but the algorithm is not that complex and may not reveal 100% accurate the differences between different CPU's.

Having this experience and this goal, I developed a new benchmark, witch use a far more complex multithreaded algorithm. Multi Core PI.

The old benchmark Multi Core Prime is no longer in development. I hope you will find this benchmark more useful and will reveal more accurate performance differences between platforms.

Multi Core PI calculates PI decimals using Bailey?Borwein?Plouffe formula. The benchmark is using a multithreaded algorithm written in C++ and provide excellent parallelism. Multi Core PI is written in Visual C++ using MFC and Win32API.

Anyway...keep benching...

Latest version is x86 only (9 Feb 2013) and is version 2.00 available ---> here