New Benchmark: SuperPrime 0.14beta

[KiD0M4N]

Hey guys,


New release:


Version 0.4beta

28-Sept-2006

- Modified the way results are presented to have a more professional (and intuitive) look.
- Changed startup code to change the default run parameters depending upon the system configuration (processor count.)
- Added a option where user can copy forum quote text directly.
- Included a high resolution timer to provide much more accurate results.
- Changed version number to 0.4beta

Thanks to RiO for convincing me to go for the tabular based view and for the various UI tips. Thanks a ton to TE admins for making www.techenclave.com/forums/superprime/ a reality :D


http://www.techenclave.com/forums/superprime/downloads/SuperPrime-0.4beta.zip

:clap:

Regards,

Karan Misra

ps: slowly but surely i am chugging away towards 1.0

original thread: http://www.techenclave.com/forums/hamara-appna-benchmark-superprime-79345.html

[[timko]]

requirement: .Net 2.0 (http://www.techenclave.com/forums/476290-post44.html)

No thanks :)

[KiD0M4N]

']No thanks :)

o common.... u install a 650 MB os from the beast, afraid of a 20 MB framework? :P waiting for results.

[Durzel]

.NET bloatware :D

Write it in a proper language like C/C++, Borland do a great IDE. :)

[KiD0M4N]

.NET bloatware :D

Write it in a proper language like C/C++, Borland do a great IDE. :)

alright i will keep ur SUGGESTION in mind, but for the while mind to try it?

i will get you guys started:

http://img143.imageshack.us/img143/4834/xsgetstartedjw1.jpg (http://imageshack.us)

:toast: :woot:

ps: and imagine when vista comes, every damn GUY will be running the same bloatware you are dissing now :D

[kiwi]

Conroe @ 3.6Ghz

http://www.xtremesystems.org/forums/attachment.php?attachmentid=51373&stc=1&d=1158450654

[KiD0M4N]

Conroe @ 3.6Ghz

http://www.xtremesystems.org/forums/attachment.php?attachmentid=51373&stc=1&d=1158450654

so i guess with a quad core chip, we should see sub 20secs? if they scale that well i.e. :D

(*dont even wanna imagine clover town pwnage*)

[KaptainBlaZzed]

Here you go.:D

Kentsfield @ 3.255GHz
2 x 512 @ 976MHz (4-4-4-12)

[KiD0M4N]

Here you go.:D

Kentsfield @ 3.255GHz
2 x 512 @ 976MHz (4-4-4-12)

neat result... i will do a 5E8 run to see how it compares with ur mighty kents....

pls do a highest + 5E7 run.... with 4 threads. i wanna see sub 20 secs on it :D (i get around 37 sec average)

edit: can you do a 2 thread run please? cuz i want to see the scaling from 2 threads to 4 threads.

[mat128]

SuperPrime 0.13beta @ www.techenclave.com
--------------------------------------------
Total number of threads: 1
Calculating with resultset: 5E7
Total primes reported: 3001134 (VALID RESULT)
Calculations took: 0 h 3:7:93

Validation url = http://www.techenclave.com/sprime/vldt?p=5E7:1:03793&c=A064B09822

http://valid.x86-secret.com/show_oc?id=122153
Something's wrong I think.. or maybe its just the netburst.. lol.

****************************
edit: re-run it with 2 threads... but I dont have HT and its around the same result so I guess it wasnt an error.

SuperPrime 0.13beta @ www.techenclave.com
--------------------------------------------
Total number of threads: 2
Calculating with resultset: 5E7
Total primes reported: 3001134 (VALID RESULT)
Calculations took: 0 h 3:6:812

Validation url = http://www.techenclave.com/sprime/vldt?p=5E7:2:036812&c=C2657F23D4

[KaptainBlaZzed]

neat result... i will do a 5E8 run to see how it compares with ur mighty kents....

pls do a highest + 5E7 run.... with 4 threads. i wanna see sub 20 secs on it :D (i get around 37 sec average)

edit: can you do a 2 thread run please? cuz i want to see the scaling from 2 threads to 4 threads.

Kentsfield @ 3.255
Ram 976MHZ 4-4-4-12

Total number of threads: 4
Calculating with resultset: 5E7
Total primes reported: 3001134 (VALID RESULT)
Calculations took: 0 h 0:25:484


Total number of threads: 2
Calculating with resultset: 5E7
Total primes reported: 3001134 (VALID RESULT)
Calculations took: 0 h 0:37:359

[Kunaak]

Allendale at 7 x 500
1 gig of ram, cas 4-3-2-8

[Doumz]

Wr for the moment -_-


this was really long ;p

[Lestat]

omg 8hrs hehehe

[KiD0M4N]

Hmm, after seeing the results of quad-core runs via peeps over @ XS (thread here (http://www.xtremesystems.org/forums/showthread.php?t=116184))..... i saw that the algorithm was not scaling as well as I would have liked. Hence I implemented a new dynamic load balancing algorithm, and hence this new beta. Test it for me guys:

Version 0.14beta

17-Sept-2006

- Improved the dynamic load balancing algorithm so that it scales well with > 2 physical cores.
- Characterized the different working set sizes based on their intended purposes.
- Changed version number to 0.14beta

Regards,

Karan

[KiD0M4N]

updated first post with new beta.....

[Ad1tya]

SuperPrime 1E7 2 Threads @ 3915 Ghz.. - 2.625

http://img128.imageshack.us/img128/533/1e7yq9.th.jpg (http://img128.imageshack.us/my.php?image=1e7yq9.jpg)

SuperPrime 5E7 2 Threads @ 3915 Ghz.. - 31.15 s

http://img475.imageshack.us/img475/1859/5e7ty5.th.jpg (http://img475.imageshack.us/my.php?image=5e7ty5.jpg)

Sorry for Sloppy Shots.. Was in a hurry.

[Ehwaz001]

I stopped reading when you mentioned .Net
Seriously, ditch that framework, it's pure junk!

[KiD0M4N]

I stopped reading when you mentioned .Net
Seriously, ditch that framework, it's pure junk!

:slap: :stick:

have you heard of something called "Vista" ?????????? or head burried in sand still? cmon man.... why are you being such a hypocrite :P

neways, major updates:


Version 0.3beta

23-Sept-2006

- Added a concept icon to the benchmark. I think it goes well with the theme :P
- Made number of threads a drop down menu.
- Added tool tips to the controls.
- Added a popup just before launching the benchmark.
- Bunched the various misc functions under Tools.
- Changed the benchmark test labels to use "and" instead of "cum."
- Added a popup message on successful completion of the benchmark.
- Implemented "take screenshot" and save feature in the program (shortcut = Alt + S) and in right click :D
- Added a "Legend" mini section to the main form to make the results more intuitive.
- Working on Publishing and validation.
- Added result save and restore to file.
- Changed version number to 0.3beta


Thanks to Funky for helping me beta test this release. Also many thanks to RiO for giving invalueable suggestions and ideas which were mostly implemented (he knows what i mean ^_^)


I am sure you will like where this bench is heading :P

Regards,

K

[Lane-k]

Question about stress test.. is it check error during it?

[KiD0M4N]

Question about stress test.. is it check error during it?

Yes.... + this is one of the easiest way to load up as many cores you might possibly have (upper limit 16 :D) If you get a "Green" result, that implies your CPU was SuperPrime "what-is-the-working-set-size" stable.

Regards,

K

[[cTx]Philosophy]

Will give it a shot soon..
Rig in sig

[freecableguy]

second system in signature:

Total number of threads: 2
Calculating with resultset: 5E7
Total primes reported: 3001134 (VALID RESULT)
Calculations took: 0 h 0:33:421

Validation url = http://www.techenclave.com/sprime/vldt?p=5E7:2:0033421&c=2BD421BBA5

[KiD0M4N]

second system in signature:

Total number of threads: 2
Calculating with resultset: 5E7
Total primes reported: 3001134 (VALID RESULT)
Calculations took: 0 h 0:33:421

Validation url = http://www.techenclave.com/sprime/vldt?p=5E7:2:0033421&c=2BD421BBA5

hey fcg... thanks for the post but major changes have happened.... do check them out :P


New release:


Version 0.4beta

28-Sept-2006

- Modified the way results are presented to have a more professional (and intuitive) look.
- Changed startup code to change the default run parameters depending upon the system configuration (processor count.)
- Added a option where user can copy forum quote text directly.
- Included a high resolution timer to provide much more accurate results.
- Changed version number to 0.4beta

Thanks to RiO for convincing me to go for the tabular based view and for the various UI tips. Thanks a ton to TE admins for making www.techenclave.com/forums/superprime/ a reality :D


http://www.techenclave.com/forums/superprime/downloads/SuperPrime-0.4beta.zip

:clap:

Regards,

Karan Misra

ps: slowly but surely i am chugging away towards 1.0

original thread: http://www.techenclave.com/forums/hamara-appna-benchmark-superprime-79345.html

[KiD0M4N]

Will give it a shot soon..
Rig in sig

thanks mate.... waiting for your results in the new version (0.4beta)

[BobyTT]

there is my result

http://img241.imageshack.us/img241/3732/3573ui0.jpg

[KiD0M4N]

my benchruns with the new version (somewhat optimized):

http://s2.supload.com/thumbs/default/benchruns_399x8sync.jpg (http://s2.supload.com/free/benchruns_399x8sync.jpg/view/)

(screen shot taken with superprime itself :D)

and new url for SuperPrime: http://www.techenclave.com/forums/superprime/
enjoy :D

Regards,
K

[bachus_anonym]

My best shot at this benchmark, on phase-change cooled Conroe setup...

ASUS P5B Deluxe / Conroe E6600 @ 552x8=4414Mhz, DDR2-1104 4-4-4-4.0 --> details (http://www.clockmehigh.com/cooler/screens/965P/402xtreme/552x8_DDR2-1104_SuperPrime.png)

SuperPrime 0.4beta @ www.techenclave.com
---------------------------------------------------
Working set size: 5E7
Number of threads: 2
Total calculation time: 0 h 0:27.217
Vadidation url: http://www.techenclave.com/forums/superprime/validate.php?145
---------------------------------------------------SuperPrime 0.4beta @ www.techenclave.com
---------------------------------------------------
Working set size: 1E8
Number of threads: 2
Total calculation time: 0 h 1:14.741
Vadidation url: http://www.techenclave.com/forums/superprime/validate.php?385
---------------------------------------------------

... then, this is my 24/7, AM2 system...

ASUS M2N-SLI Deluxe / Orleans 3200+ @ 293x10=2930Mhz, DDR2-976 8-4-4-4.0 --> details (http://www.clockmehigh.com/cooler/screens/AM2/T1000UX2G4/293x10_DDR2-976_SuperPrime.png)

SuperPrime 0.4beta @ www.techenclave.com
---------------------------------------------------
Working set size: 5E7
Number of threads: 1
Total calculation time: 0 h 2:54.127
Vadidation url: http://www.techenclave.com/forums/superprime/validate.php?298
---------------------------------------------------SuperPrime 0.4beta @ www.techenclave.com
---------------------------------------------------
Working set size: 1E8
Number of threads: 1
Total calculation time: 0 h 7:49.5
Vadidation url: http://www.techenclave.com/forums/superprime/validate.php?156
---------------------------------------------------

... and for comparison, Conroe @ 2.93GHz (single core bench), just as above AM2 single core setup...

ASUS P5B Deluxe / Conroe E6600 @ 326x9=2934Mhz, DDR2-1086 8-4-4-4.0 --> details (http://www.clockmehigh.com/cooler/screens/965P/402xtreme/326x9_DDR2-1086_SuperPrime_singlecore.png)

SuperPrime 0.4beta @ www.techenclave.com
---------------------------------------------------
Working set size: 5E7
Number of threads: 1
Total calculation time: 0 h 1:21.433
---------------------------------------------------SuperPrime 0.4beta @ www.techenclave.com
---------------------------------------------------
Working set size: 1E8
Number of threads: 1
Total calculation time: 0 h 3:44.278
---------------------------------------------------

Clearly, Conroe is twice as fast as my AM2 setup, at the same CPU clocks :)

btw, nice benchmark, kidoman :up: Although, validation links don't seem to work properly.

[KiD0M4N]

My best shot at this benchmark, on phase-change cooled Conroe setup...

ASUS P5B Deluxe / Conroe E6600 @ 552x8=4414Mhz, DDR2-1104 4-4-4-4.0 --> details (http://www.clockmehigh.com/cooler/screens/965P/402xtreme/552x8_DDR2-1104_SuperPrime.png)



... then, this is my 24/7, AM2 system...

ASUS M2N-SLI Deluxe / Orleans 3200+ @ 293x10=2930Mhz, DDR2-976 8-4-4-4.0 --> details (http://www.clockmehigh.com/cooler/screens/AM2/T1000UX2G4/293x10_DDR2-976_SuperPrime.png)



... and for comparison, Conroe @ 2.93GHz (single core bench), just as above AM2 single core setup...

ASUS P5B Deluxe / Conroe E6600 @ 326x9=2934Mhz, DDR2-1086 8-4-4-4.0 --> details (http://www.clockmehigh.com/cooler/screens/965P/402xtreme/326x9_DDR2-1086_SuperPrime_singlecore.png)



Clearly, Conroe is twice as fast as my AM2 setup, at the same CPU clocks :)

btw, nice benchmark, kidoman :up: Although, validation links don't seem to work properly.

Hello bachus_anonym,

Thanks a lot for running these tests in this detailed a manner. Truely, these will help me a lot in finetuning the benchmarks.....


- You just took the WR in this Benchmark in 5E7. The next best result is 30.20 (Here (http://www.techenclave.com/forums/hamara-appna-benchmark-superprimeandquot-79345-7.html#post482196))

- It is scaling nicely... I want someone with a nicely OCed quad core rig to show me how this is scaling beyond 2 cores. I dont have access to such L33T hardware hence I cannot really predict. But I have used my imagination to make the algorithm as parallel and scalable as possible. (FYI, yesterday I ran the benchmark in a octa core Xeom MP machine. It was scaling like a beauty there but the results were slower because of Netburst architechture :mad: )

- Yes the conroe ownes in the benchmark, particularly due to the number of ALUs.

- I dont see the cache of the Conroe (vs a Allendate) affecting the score much (atleast not as much as it affects SuperPi) I have seen a 0.3s advantage of Conroe over a Allendale at the same clock. Good for my
benchmark....

- I am working with the admins back at my forum (www.techenclave.com) to get all the validations etc working. After that we will create a hall of frame. Make sure to get to #1 as soon as it comes :)

Regards and Thanks,

Karan
Team Techenclave

[Dumo]

Nice benchmark here Kidoman:toast:

Heres mine,,,

http://img84.imageshack.us/img84/8379/screenshot042gp1.jpg

[KiD0M4N]

Nice benchmark here Kidoman:toast:

Heres mine,,,

http://img84.imageshack.us/img84/8379/screenshot042gp1.jpg

ahha thanks for the appreciation man.....

and btw, u snatched the WR from bachus in a big way (1 sec :P)

:toast: :toast:

ps: just waiting for some quad giants to get cracking on this baby :fact:

[KiD0M4N]

ps: just waiting for some quad giants to get cracking on this baby

:D :woot: :D

[KiD0M4N]

Major release :) This benchi has taken off... thanks to the ocers determination to tackle every new demanding benchmark out there.

No longer just a CPU benchmark.... more of a system benchmark now.

Much has changed, hence the old scores dont apply. I guess you all will appreciate the changes in this version.

Version 0.6

31-Oct-2006

- Adjusted the dynamic load balancing system to give weightage to threads which have been running longer.
- Made parts of the benchmark single threaded to bring in the CPU's performance w/ 1 thread into the score.
- Added Rabin Miller Test into the implementation. Now after a certain range, the new test is being used which clearly gives us a health spee boost.
- Brought in memory into the equation in a big way. If you have fast ram, you shall have fast results :) (hopefully)
- Added Team TechEnclave to the Program title.
- Changed version number to 0.6

Available immd via auto update (download 0.5 version from http://www.techenclave.com/forums/superprime/ and press F5 or Help > Check for updates, 0.6 will be ready to use on next restart) :cool2:

BTW, the current world record is held by coolaler @ 15.6 sec with a 4g Core 2 Quad :) :cool2:

[odin318]

Crashes on me :(

EventType : clr20r3 P1 : superprime.exe P2 : 0.5.0.0 P3 : 45445ad3
P4 : mscorlib P5 : 2.0.0.0 P6 : 4333c8f6 P7 : 1c09 P8 : 78
P9 : system.badimageformatexception

[freecableguy]

more data

[bichi]

SuperPrime v0.6
- QX6700 @ 3.0ghz
- P5WDH (early MCH 1.65 version) BIOS: 1707
- screenshot Photoshoped to show CPU utilization and temperatures during test run, with completed screen showing "finished" time

[bichi]

P5WDH (early MCH 1.65 version) - BIOS 1801 (new, Dec 22 06)
QX6700 @ 3.0ghz

SuperPrime v0.6 test and new 1801 BIOS
- "Above Normal" and 1E8 setting
- test to see new 1801 BIOS effect, if any
- about .3 sec improvement @ 1E8, over BIOS 1707.....
- results panel pasted in after test completion
- main shot taken seconds before test completion, to show CPU load, temps, volts, etc.

[gbomb944]

bump...
Pretty cool bench. Not many people seem to be onboard though. Kind of painful to see how badly the conroes whup my lowly opteron 165 though.

http://img214.imageshack.us/img214/3197/superprimedv0.th.jpg (http://img214.imageshack.us/my.php?image=superprimedv0.jpg)

[TopherTony]

http://img256.imageshack.us/img256/5115/superprime1fn4.jpg

hows that?

[_Slim_Shady_]

I stopped reading when you mentioned .Net
Seriously, ditch that framework, it's pure junk!

Is this based on any actual experience, or just pure opinion? ;)

I write code against this framework everyday, and have used every version since the beta of 1.0. While no framework is perfect .NET makes my job much easier, and more enjoyable.

I would like to understand where you are coming from. Are you a professional developer, and end user, or a hobbyist?

[ShaunO]

Hello, thought I would muck around with my work computer to see what the multi-core scaling was like on an AMD CPU.

The setup:
AMD Athlon X2 3800+ (Windsor 2GHz)
2x 1GB Transcend DDR2-667 5-5-5-15 (128MB Video Shared!)
ASUS M2N8-VMX

The results:
1 Thread (AboveNormal)
1E1: 0.8s
1E6: 1.68s
5E6: 10.169s
1E7: 21.606s
5E7: 1:54.15s (114.15s)
1E8: 3:50.212s (230.212s)

2 Threads (AboveNormal)
1E1: 0.42s - 90% improvement
1E6: 0.623s - 169% imrovement?
5E6: 6.95s - 46% improvement
1E7: 12.326s - 75% improvement
5E7: 1:3.232s (63.232s) - 80% improvement
1E8: 2:6.740s (126.740s) - 81% improvement

Cheers,
Shaun.

[ShaunO]

Home computer. No O/C or anything.

The setup:
Intel Pentium D 2.66GHz (805)
2x 1GB Corsair DDR2-800 C4 (Running DDR2-533 @ 4-4-4-12)
ASUS P5W DH Deluxe

The results:
1 Thread (AboveNormal)
1E1: 0.4s
1E6: 0.841s
5E6: 8.188s
1E7: 21.947s
5E7: 3:37.866s (217.866s)
1E8: 9:51.103s (591.103s)

2 Threads (AboveNormal)
1E1: 0.4s - 0% faster
1E6: 0.447s - 88% faster
5E6: 4.359s - 87% faster
1E7: 11.165s - 96% faster
5E7: 1:49.706s (109.706s) - 98% faster
1E8: 4:54.956s (294.956s) - 100% faster

[Monstru]

Office computer, no o/c;
(Unfortunately CPU-Z gives me some errors here and can't display all the memory info. It's a 1Gb Sycron DDR400 stick.)

Intel Pentium 4 524 Prescott 3066Mhz
1Gb Sycron DDR 400
Asus P5PE-VM

Results:
1 Thread (AboveNormal)
1E1: 00.11s
1E6: 01.41s
5E6: 11.215s
1E7: 23.947s
5E7: 2:7.564s (127.564s)
1E8: 4:21.716s (261.716s)

Will be back with some San Diego an Conroe results from my home computers.

[Truckchase!]

;1724569']No thanks :)

.NET bloatware :D

Write it in a proper language like C/C++, Borland do a great IDE. :)

I stopped reading when you mentioned .Net
Seriously, ditch that framework, it's pure junk!

http://dictionary.reference.com/browse/evolution

You guys just get in from yelling at the kids to get off of your front lawn?

[Monstru]

I also have a skt A Sempron (Thoroughbread) as my quiet web browsing machine, and I wanted to test SuperPrime on this one also. But I get this error and I can't run the benchmark...:(

[ShaunO]

I also have a skt A Sempron (Thoroughbread) as my quiet web browsing machine, and I wanted to test SuperPrime on this one also. But I get this error and I can't run the benchmark...:(



You will need the .NET 2.0 Framework installed.

http://www.microsoft.com/downloads/details.aspx?FamilyID=0856EACB-4362-4B0D-8EDD-AAB15C5E04F5&displaylang=en

[Monstru]

AHA! That solves my problem, I will install it and run the benchmark. Thx :)

[Monstru]

I really like this benchmark, it really uses Dual Cores at their maximum. As you can see in the attached images, I ran the tests with 1 core and then I ran the tests with 2 cores, and the difference is double, as it should be in any aplication. I am really impresed. Any news with the development of the benchmark?

X6800 @ 3412Mhz
2Gb Mushkin XP6400 @ 465Mhz 4-4-4-10
Dfi Infinity P965-S

Results:

1 Thread (AboveNormal)
1E1: 00.3s
1E6: 00.248s
5E6: 2.185s
1E7: 5.646s
5E7: 1:8.982s (68.982s)

2 Thread (AboveNormal)
1E1: 00.1s
1E6: 00.165s
5E6: 1.163s
1E7: 2.889s
5E7: 34.652s



E6600 @ 3479Mhz
2Gb Mushkin XP6400 4-4-4-10
Dfi Infinity P965



1 Thread (Above normal)
1E1: 00.1s
1E6: 00.242s
5E6: 2.130s
1E7: 5.521s
5E7: 1:7.609s (67.609s)

2 Threads (Above normal)
1E1: 00.1s
1E6: 00.160s
5E6: 1.121s
1E7: 2.851s
5E7: 33.948s

[Dualist]

Nice little bench proggie. Just had a quick play..

[Monstru]

As promised, some San Diego action:

San Diego 3700+ @ 2750Mhz
2Gb Kingmax DDR400 2,5-3-3-7
Asus A8N-SLI



1 Thread (Above normal)
1E1: 00.8s
1E6: 00.668s
5E6: 6.425s
1E7: 17.103s
5E7: 2:49.363s (169.363s)

[Toysoldier]

Tried it with 1, 2 and 4 threads.

[Monstru]

CPU: Q6600 @ 3015Mhz
Mobo: Dfi Inifnity P965-S
Ram: 2Gb Mushkin Xp6400 @ 419Mhz 5-5-5-15 2T
Cooling: air - Scythe Ninja
Scor:19.985

[Ace-a-Rue]

bench at the speed in my signature.

[KiD0M4N]

Guys a brand new version (and nearing completion) of the benchmark is going to be released this weekend :P

[KiD0M4N]

As promised guys, http://www.xtremesystems.org/forums/showthread.php?t=167942

[KiD0M4N]

Download link: http://forums.erodov.com/local_links.php?catid=7