Agreed. As a point of comparison, it looks like you can get 650 million/s on a cg1.4xlarge instance [1] (Amazon's GPU computing instance with 2x Tesla Fermi M2

Agreed. As a point of comparison, it looks like you can get 650 million/s on a cg1.4xlarge instance [1] (Amazon's GPU computing instance with 2x Tesla Fermi M2050 GPUs), and it looks like they cost $2.10/hour per instance. So some quick math does show that cracking SRP is only about 572 times slower, if we normalize for cost of the instances on EC2. 1. http://www.nervous.it/lang/en-us/2012/06/cracking-sha1-on-amazon-ec2/ http://www.nervous.it/lang/en-us/2012/06/cracking-sha1-on-am...