ares139 Posted November 16, 2007 Share Posted November 16, 2007 Okay, I just got my new Survivor 16GB (non-GT) in the mail today. Before I ask my question, I just want to say how awesome this thing is. Yeah, it's a little bulky, but it's built like a tank. And for a non-GT model, this thing is fast. As I'm typing this, TrueCrypt is writing a volume at about 11MB/s. My old Lexar JumpDrive Firefly wrote at 1.4MB/s. So yeah, GT or not, this is a major upgrade for me. Very satisfied! Okay, here's the question. First, a little background information. I'm well-versed in TrueCrypt, and have used it on a variety of external drives including a 250GB and 120GB drive. For my Survivor 16GB drive, I want to create a 4GB file container (due to the FAT32 limit) instead of making a partition, so that I can have a copy of TrueCrypt on the drive for portability ("traveler mode" as they call it). Anyway, my question is this: When creating a file container, what cluster size should I pick to optimize data transfer? I noticed the drive itself is preformatted in FAT32 with 8KB clusters. Should I make the container have 8KB clusters? Smaller? Larger?? Does it even matter? Should I reformat the drive with a larger cluster size to optimize access to the large container file? Should the container file also have a larger cluster size?? The TrueCrypt forum was basically bare on the subject. As much as I love TrueCrypt, the developers don't really hang around on the forums, and sometimes when I have asked questions like this in the past, I've gotten - not really rude - but "annoyed" responses, I suppose you could say. Or at least that's the feeling I got. I could be completely wrong, but eh... considering the friendly atmosphere around here, and helpful people that actually work at the company like RAM_GUY, I thought maybe this might be the best place to ask such a question. So has anyone experimented with different cluster sizes inside of a TrueCrypt container file on top of the existing FAT32 filesystem? I thought about formatting the drive itself with NTFS and then making a 15GB file container, instead of just 4GB, but I know that NTFS causes more writes due to the journal and may shorten the life of my drive. Are a few extra writes to the journal REALLY that bad, though? Especially considering the wear-leveling that the drive uses? Thanks for taking the time to read and hopefully respond! Link to comment Share on other sites More sharing options...
Corsair Employees RAM GUY Posted November 16, 2007 Corsair Employees Share Posted November 16, 2007 If you do a chckdesk command from a command prompt it will tell you the cluster size and that would be what I would use. Other wise I would experiment with how you use it for best performance because the file size you use will also affect the performance IE if you use or copy a lot of small files a smaller cluster size will be better but if you copy a lot of large file a lager cluster size might be best. Link to comment Share on other sites More sharing options...
ares139 Posted November 17, 2007 Author Share Posted November 17, 2007 Yeah that's how I figured out that the drive was pre-formatted as FAT32 with 8KB clusters. I created a 4GB FAT32 file container with 8KB clusters as well, and so far the performance seems decent. I used the Twofish algorithm, which is fastest on my machine (I prefer it for other reasons as well). So far, so good. I was just wondering if anyone had done some extensive testing with different cluster sizes on both the drive and a file container, and how they affected each other. Of course, it's possible that I'm the only one geeky enough to care - but I figure there's someone out there as bad as I am. ;): Link to comment Share on other sites More sharing options...
ares139 Posted November 17, 2007 Author Share Posted November 17, 2007 Here's a HDTach of the drive: http://aresfiles.googlepages.com/drive.PNG And here's one of the file container hosted on the drive: http://aresfiles.googlepages.com/container.PNG And here's a benchmark of how fast my machine can encrypt information: http://aresfiles.googlepages.com/truecrypt.PNG As you see, I used 100MB as the test size for both programs. It seems I take a pretty steep performance cut using the encrypted file container. Also, my machine should be able to encrypt fast enough to use the full speed potential of the drive - but it isn't (I'm using Twofish). Anyone know any good speed tricks for running an encrypted container on a flash drive? Thanks! Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.