I noticed something rather important missing in the realm of Full Disk Encryption, and that was benchmark data. The methodology I followed may be a bit unorthodox but would provide me with a consistant platform to test with.
I installed Windows XP SP2 on VMWare Fusion and applied all the patches available from our WSUS server. After installing the Benchmark application I created a SnapShot so that I could revert to the original install everytime with no deviations in installation, i.e. I was lazy.
I only encountered one major problem with running the Disk benchmark on PC Mark 04 when using TwoFish/RIPEMD160 encryption in TrueCrypt the entire benchmark would come up with the useless Windows Has Encountered a Problem message.
Platform Used
Apple MacBook Pro 15″ (2.4GHz, 4GB RAM, 120GB 5400 HDD, 8600GT 128MB)
Windows XP SP2
VMWare Fusion
CrystalMark 2004r3
Crystal DiskMark 2
PGP Desktop 9.82 Win32
TrueCrypt 5.1
Unfortunately I could not get a trial copy of Utimaco’s software. As soon as I can get one I will post an update.
Results
Click Image for a larger copy
Click Image for a larger copy
Download Raw Benchmark Data & Graphs
I did not expect to see such a large performance hit over unencrypted, even less so such a similarity in performance between TrueCrypt and PGP. Interestingly PGP had more of a hit on overall performance. Points of interest are the ALU and FPU scores, TrueCrypt AES being the least system intensive, and the fact that PGP has the best HDD performance.
2 responses so far ↓
1 KimW // Mar 26, 2008 at 15:44
I did some testing using a HP nc4200 and the results where pretty much the same, but I could not trust them as much thanks to the benchmarks crashing more often than running through on the machine. I do plan on doing it again when I can get a more reliable workstation. I had planned on using PCMark as well, but it would not run the Disk tests on an encrypted drive, even on a VM session.
I use the VM sessions quite a bit and did notice a perceivable difference in performance when the system was encrypted, and the tests showed the difference graphically. xxGERMxx tweeted that he noticed a significant boot time difference which I noticed when the VM was encrypted.
2 Sid // Apr 6, 2008 at 6:49
I don't think that using a VM will make consistent result. Also when using my laptop encrypted with AES (TrueCrypt) I don't see so much difference!
Leave a Comment