CryptoSuite Benchmark Public Beta

I have been working on a new benchmark for CryptoSuite for the upcoming release. It has been improved a lot, it now includes 2 synthetic tests, graphical comparisons and an online component.

There is still some work to be done (especially on the online side), and it still isn't 100% polished, however running this and submitting your results would be very beneficial for us.

This has been tested on Windows 98/2000 and XP, so everyone with Windows should be able to run it just fine.

To test all you need to do is download this ZIP file and extract it to a directory, then run cryptosuite_benchmark.exe and click on Start Test. After the tests have completed you can submit your results. Of course everyone can submit their own suggestions if they have any.

Anyone can download this and test, you don't have to be an existing CryptoSuite user to take advantage of this. Thanks everyone.

http://www.diamondcs.com.au/cryptosuite/downloads/cs_bench.zip

 

Here you go Jason, I have also submitted to your data bank.

Can you allow copy from the text results as it won't do it ATM

Here is a screenie anyway:

 

Thanks Jason. Submitted my results as well.

Nick

 

Here's my slow machine Jason

http://www.buttuglysoftware.com/Jason.jpg

 

My machine is 1203mhz AMD athalon, 128k L1 cache, 256k L2 cache, 256 meg ram, and my score is 8258.

 

Do you mean you want to be able to copy the contents from the listview display? Thanks for the submission.

 

Thanks for your submissions everyone.

 

Yes, much like in CS where you highlight the text and it is automatically copied to the clipboard.

Pilli

 

Mine and submitted...

TAS

 

Here be mine...

 

...and mine.

 

and....

 

Showoff...

 

Sorry, dear. Must keep up with the Joneses.

 

For the CPU experts, is there a relationship between the size of L1 cache and better Integer and FPU performance? From the results, it seems like there is.

Nick

 

I don't believe it is the L1 cache as much as it is the chip architechture. The AMD proccessor is better at integer and floating point because it does more computations per cycle and doesn't have the handicap? of the deep pipeline of the P4.

 

Thanks for the info.

Nick

 

Well admitably I have not added any optimizations for any CPU. However the deep pipeline of the Intel P4 does not like what I am currently doing. I will be optimizing the FPU and INTEGER code for longer pipelines in the next version of the benchmark, however the AMD will also benefit from this optimization, not just the Intel, but it should close the gap. The AMD chips are even struggling with the FPU and INTEGER benchmark currently, but due to its 2/5 shorter pipeline it scores a lot better than the Intel P4.

So in summary, can the current FPU and INTEGER code be better optimized for longer pipelines? Yes. It still shows though that some CPUs have a big advantage over other CPUs when you don't try and write heavily optimized long pipeline code (a lot of compiled code falls into this region).

 

I am very impressed with the Encryption and Hash tests on your CPU. Even though the synthetic benchmark scores are bad (due to the even longer pipeline of the prescott), you can see when you mix memory speed and optimized code, the Prescott has a big advantage. We need some Athlon 64 scores in here now.

 

hmm... turned off a few things and increased mine by around 400..

TAS

 

Now I'm really depressed. I guess that means I shouldn't have been such a penny-pincher when I bought my box.

 

Here is my screenshot:

 

Ok, not too bad, not too exceptional

Wonder what you closed Tassie, me with windows only got only an overall cryptomark score of 13489 and the individual values raised to
9505 -- 5512 -- 15641 -- 18653 in stead of those from shot below.
Guess as the system is still new can use lots of finetuning once i know it a bit better.

 

LOL Here comes mine

P3, 600 mhz, Asus P2B mobo, 512 MB RAM, W 98 SE.

Almost everything was disabled during the test.
I still have to submit.

Cheers, Jan.

 

Thanks, FanJ. I think your just trying to make me feel better.