New cheat-proof Pi Benchmark: Begin Debug.

[eshbach]

OK, i've got a usable build of my pi calculation benchmark done. It is not susceptable to cheats like SuperPi is. I, however, don't know all the cheats (only a few major ones), and the program probably has a bunch of bugs i haven't found yet.

So, you can all help by downloading the benchmark, trying to cheat, looking for bugs, and posting your results here. I will then fix the program to get it perfected and hopefully we can have a new reliable Pi program.

The program is available for download at www.baughman-eshbach.com/ePiInstall.exe
I only have 40GB/Month of bandwidth, so if XS or someone else could help mirror this, that would be great. This is not neccessary until the final version, of course.

here's a screenshot of the program and the checksum verification program i made to accompany it:



The verifier is available to moderators upon request.

The accuracy and Pi algorithim are NOT finalized, therefore benchmark times are probably inaccurate, but it will still work, times will just not be comparable with times from the final build (1.0).

You MUST have the .NET Framework installed to run this program. If you don't have it, you can get it from microsoft.com for free: http://www.microsoft.com/downloads/d...displaylang=en

You also MUST be connected to the internet in order to run this benchmark.

Features:
Time obtained from remote server.
4 part checksum.
No platform optimizations.

Known Issues (will be posted as they are reported) :
- Should be 4 part checksum - FIXED
- Index out of bounds: 16

Suggested Changes (so far):
- Time in sec:ms format.
- Stop Calculation button.
- A pretty desktop icon.
- No Internet Connection. - Tentatively Changing
- No .NET Framework. - Tentatively Changing

Mirror List:
http://jeff.canyonlake.net/ePiInstall.exe
http://www.drbus.com/files/software/ePiInstall.exe

I will be away for the holidays, but get the list started and i will start fixing things after the new year.

[Karnivore]

Nice effort, I'll give it a shot

Happy Holdays, have a safe one!

[conrad.maranan]

Sounds good. I'll test it out.

[blinky]

make it so we can see hundreths of a second

[charlie]

failed to initialize??

Ahhhh, I see need .NET !!

[Aphex_Tom_9]

what about the windows clock manipulation cheat?

[eshbach]

failed to initialize??

C

the installer or the program? .NET Framework 1.1 installed?

You will need to reinstall the program after installing the framework i think.

[eshbach]

what about the windows clock manipulation cheat?

shouldn't be an issue since it gets time from a server, not the local machine.

[charlie]

hmmmmm... where's the ms linkie?

[eshbach]

hmmmmm... where's the ms linkie?

here: http://www.microsoft.com/downloads/d...displaylang=en

i've added it to the first post too.

[DBlue135]

Even though we're talking on AIM right now and this post is completely unnecessary:

What prevents someone from running the test for say 18 iterations, getting a time, setting it to 20 iterations and then taking a screen shot? Your checker program doesn't take into account the number of iterations.

[conrad.maranan]

What's the exact purpose of Accuracy Level?

EDIT:
Nevermind. I figured it out.

[eshbach]

What prevents someone from running the test for say 18 iterations, getting a time, setting it to 20 iterations and then taking a screen shot? Your checker program doesn't take into account the number of iterations.

OK, this brings up an interesting point. Would you rather have the iterations locked after the run and the program need to be "reset" before running again or just make it a 4-part checksum?

[eshbach]

What's the exact purpose of Accuracy Level?

Well, SuperPi has different levels of accuracy of Pi (16k, 1M, 32M, etc) so i thought i would incorporate a similar feature. The more accurate the number the more likely you are to get errors if your OC isn't stable.

[DBlue135]

Just make it a 4 part checksum. That way you don't have to close the program and open it again after every run.

[eshbach]

Just make it a 4 part checksum. That way you don't have to close the program and open it again after every run.

Well you wouldn't have to close it, just click a "reset" button.

[conrad.maranan]

So far, I recommend the following:

- Time in sec:ms format. For example - 19:03
- Stop Calculation button.
- A pretty desktop icon. I like pretty icons. The graphic you have in the title bar would be nice.

[DBlue135]

It's still abusable with a reset button. Nothing prevents someone from photochopping it to read 20

[eshbach]

It's still abusable with a reset button. Nothing prevents someone from photochopping it to read 20

OK: It is now fixed to include a 4 part checksum.

[Geforce4ti4200]

wow this sounds like a sweet cheat resisant program. You got some programming skills there, I was never really good at programing alot of the programming stuff is 100x harder than that too. one feature I would really like is an "orb" database where you can upload your scores, screenshots can be manupliated by photoshop or whatever. even if he cant cheat, he can still doctor the screenshots anyway. with a database, you upload your score(similar to 3dmark) and it has rankings for your cpu MHz and class. This will also let you compare your time with anyone else out there and greatly encourage competition after this maybe someone can make a program thats both cpu and gpu intensive, perhaps use 3-d graphics that spin or something and make it so it needs cpu power to calculate the movement. itll be like a "3dmark"

[PS2pcGAMER]

Cool program, and an ORB would be cool.

BTW, I mirrored it:

http://jeff.canyonlake.net/ePiInstall.exe

I have 50GB of bandwidth, but they don't really monitor it so it should be plenty.

Dothan @ 1.7 result:

[bh2k]

[Hilppa]

BH2K: Too much overclock

[STEvil]

How about make it so a net connection isnt required?

[CodeRed]

How about make it so a net connection isnt required?

yeah, my benching machine lives in the garage and I have to upload results using sneaker-net (floppies)


Can you elaborate on the technique used to calculate Pi? Also is the program optimised and vectorised to use SSE, SSE2 or SSE3. The microsoft compiler does a crap job of optimising vectorisable code.