Today I spent some time testing the speed of the Cluster GPU instances (cg1.4xlarge) of Amazon EC2.
Install ec2-ami-tools, and the environment variables appropriately. You can find more details at AWSEC2 User Guide
For this test I used the AMI image called: amzn-ami-gpu-hvm-2012.03.1.x86_64-ebs (ami-fd65ba94)
Replace “your-key” with the name assigned to your ssh keys in the AWS Management Console under “Key pairs”.
ec2-run-instances –key “your-key” –instance-type cg1.4xlarge –availability-zone us-east-1a ami-fd65ba94
Now let’s try to crack the password “AAAAAAA”:
echo -n AAAAAAA|sha1sum -|cut -d ‘ ‘ -f 1 | tr [a-f] [A-F] >/tmp/crack-me.txt
yum install nvidia-gpu-sdk
tar jxvf Cryptohaze-Linux_x64_1_31a.tar.bz2
./Cryptohaze-Multiforcer -h SHA1 -c charsets/charsetall -f /tmp/crack-me.txt
Make sure that crack-me.txt contains a list of SHA1 hashes, one per line and uppercase.
I got an average of 650 million/s tries on the cg1.4xlarge instance. YMMV.