cudaHashcat+ slower after update

[komrad36]

Hi all,

I'm new here so I apologize if I'm making a rookie mistake. I've done my homework but can't seem to find any mention of this issue anywhere.

So I'm using an nVidia GeForce 780M, stock vBIOS and clocks, to crack WPA2 PSK hashes (-m 2500) on an Alienware 17.

Under The-Distribution-Which-Does-Not-Handle-OpenCL-Well (Kali) Linux with cudaHashCat+ v0.15, default settings, I get 31,000 hashes per second.

Under Windows 7 x64 with cudaHashCatplus-64 v0.15, default settings, I used to get 30,500 hashes using nVidia Forceware 311.48.

However, now that I updated to the latest WHQL 331.65, all other benchmarks perform as well or better, but cudaHashCatplus-64 under Windows produces only 20,000 hashes per second at default settings.

I would just use Linux... however, I'm curious as to what may have happened and thought I would ask you all. In addition, I can OC substantially under Windows to boost performance, but it's not worth it if I'm already losing 33% off my speed.

Things I've tried:
-reinstalling new drivers
-completely uninstalling old drivers first, multiple restarts, then installing new drivers
-reinstalling hashcat

I don't want to downgrade my drivers as the new ones increase performance in games.

Thank you for an awesome tool and forum!

[mastercracker]

Increase your -n and -u values. If you don't know what they are, read the wiki.

[komrad36]

Thanks mastercracker! I do and they help quite a bit. I only left them out for simplicity for this test. My point was that before, the default settings produced about 30k under Windows. Now, it's 20k. Before, using ideal values for -n and -u produced about 39k. Now, it's 26k.

Is there any sort of known issue with newer ForceWare that could cause this? This shouldn't be occurring, right?

[mastercracker]

Is there any sort of known issue with newer ForceWare that could cause this? This shouldn't be occurring, right?

atom or philsmd would probably know better about this but I guess it's possible that with driver changes, some optimization are lost. Sorry, I cannot help further.

[komrad36]

Okay thank you. Out of curiosity, how might I find out more or ask atom or philsmd? Does anyone reading this have similar findings with the 331.65 update?