BRIGHT SIDE OF NEWS About | Advertise | Contact BSN USER Login
| Register
SUBSCRIBE Newsletter | RSS Feeds
Thursday, March 18, 2010
Email this to a friend.
Your friend's e-mail:
Your Name:
Your e-mail:
Message subject:

CUDA-enabled GPGPU app cracks PGP passwords 200x faster than a CPU



The ongoing battle between a CPU and a GPU for computing supremacy isn't showing any signs of relief - in fact, real applications are only started to appear. We've been covering ElcomSoft for some time now, and it is impressive to see that intelligence and counter-intelligence agencies around the world rely on GPU technology.

This is also the reason why Intel is spending three billion dollars on Larrabee. If the company continues to make CPUs only, stream applications such as password cracking will switch to GPU completely, and no more spending dozens of millions on CPU-focused servers.
After Intelligence/Counter-Intelligence agencies move to more advanced code, ElcomSoft moves the technology to the realm of us, consumers.

And what they rolled out today is really impressive. By using a single GeForce GTX285 card, Distributed Password Recovery is 15 times faster when compared to the Intel Core 2 Quad Q6600, a reference CPU of choice for numerous software vendors. GeForce GTX295 will crunch 40,000 passwords But the buck didn't stopped there: by using a computer equipped with four GeForce GTX295 GPUs, PGP Disk Password Recovery returned 160,000 128-bit PGP passwords, some 200 times more than a quad-core CPU. A single GeForce GTX295 will crunch 40,000 PGP passwords in a single second, e.g. 30 times faster than Core 2 Q6600. You can check interesting comparison on the official PGP Disk Password Recovery page.

All in all, it looks like CUDA works like a charm in case of ElcomSoft. And if distributed password recovery servers operate on CUDA software, there is a pretty good idea who is buying all those Tesla SuperComputers and servers.



© 2009 - 2010 Bright Side Of News*, All rights reserved.



Related articles:

Tags:

Share and enjoy :)

  • Digg
  • del.icio.us
  • StumbleUpon
  • TwitThis
  • Reddit
  • Furl
  • Google
  • Technorati
  • Sphinn
  • Mixx
  • Facebook
  • LinkedIn
  • Slashdot
  • Newsvine
  • Ma.gnolia
  • BlinkList
  • connotea
  • Fark
  • MisterWong
  • Netvouz
  • PlugIM
  • Propeller
  • Simpy
  • SphereIt
  • Spurl
  • ThisNext
  • YahooMyWeb
  • co.mments
  • Live
  • MySpace
  • Yahoo! Buzz


Comments:

RE: ATI GPU by: Theo Valich on 4/25/2009
A while ago, I saw a chart with ElcomSoft's code running on Radeon 4870 and 4870X2... both cards were significantly faster than nVidia's card, and almost came close to a 4GPU setup by nVidia - Tesla S1070.

I wrote about that chart in my blog, and you can check it here:

http://bit.ly/gpgpuativsnvidia

Thanks for the feedback... definitely interesting stuff, we'll call some people around and get to the bottom of this story.
Request of edits and links ;) by: Gipsel on 4/24/2009
I guess my comment was quite a bit unclear.
The post on the ATI Stream developer forum was posted by the author of the currently *fastest* avavilable MD5 cracker for CPUs as well as GPUs (both nvidia and ATI) not the creator of the elcomsoft tool.

His CUDA version is already quite a bit faster than the elcomsoft variant. He gets 720 million hashes/s on a GTX280 while elcomsoft only gets 493 Mhashes/s (and 920 Mhashes/s on a dual GPU GTX295). But it is all dwarfed by a single HD4870 obtaining 1286 Mhashes/s.

I guess elcomsoft should try to hire that guy from Belarus ;)
That's cool! by: Gipsel on 4/24/2009
Interesting to see that now in a commercial product. The developer of the GPU code [url=http://forums.amd.com/devforum/messageview.cfm?catid=328&threadid=101209&messid=951670&parentid=922118&FTVAR_FORUMVIEWTMP=Single]posted some comments on the ATI Stream developer forum about that topic[/url]. Interestingly he said he can reach the theoretical performance on both nvidia (without that additional MUL/SFU pipe) and ATI cards. According to the numbers posted there a HD4870 is about 80% faster than a GTX280. That means a single HD4890 would be even slightly faster than the dual GPU GTX295 for this kind of tasks.

I really wonder what happend to the ATI code path in the process.
Leave a comment:

Author:

Title:

Comment:


Enter the code shown above:

(Note: If you cannot read the numbers in the above
image, reload the page to generate a new one.)




Highlight
  • AMD forms Fusion strategy, Offensive for 2011 set
  • PowerDVD 10 creates 3D movies from 2D ones, MKV included
  • Updated GTX 470 / 480 Specs Leak Out
  • Apple loses Director York to an aneurysm
  • Apple loses Director York to an aneurysm
March 20, 2010, 20:00 UTC

Dear Readers,

In order to enable new features for the site, we'll be temporarily offline on Saturday, March 20th 2010 at noon Pacific, 3PM Eastern or 8PM/20:00 GMT/UTC. We should be offline for 15-25min, after which you should be able to see new features.

Thank You for understanding,

The BSN* Team

© 2009 - 2010 Bright Side Of News*, All rights reserved.