Truecrypt/veracrypt Vslow write performance <20MB/sec

Associate
Joined
23 Dec 2015
Posts
79
Ever since I have used Truecrypt & Verycrypt I have had severely poor write performance on all my encrypted drives, at the time I assumed it was the old server I was using, which did not have hardware AES. (Q9550)

I have just upgraded it to a Skylake 6700 on a Q170 mb and I still have the same issues, I have tried Windows 2008r2 and Windows2012 r2.
This happens on the onboard controller and the LSI 9211 & the 9207 controller I have just purchased in case the other 9211 was at fault. I have even tried connecting the disks to a VM and it still impact the performance, but I have never seen anyone else report a similar impact!

The disks I have used are WD Green and Red 2, 4 & 6TB disks, which all perform perfectly prior to being encrypted.

I have run some benchmarks to show the difference, and the sequential transfer is equal to copying large files. It takes days to populate a 4TB hard disk.

Unencrypted WD Red 4TB on LSI9207 controller
MInzrVm.png


Encrypted WD Red 4TB in lsi9207
VcbcwiE.png


These is the benchmark to prove AES is enabled.
yqdVu4o.png


And here are the settings for the encrypted volume
qaFqg1Y.png



Has anyone else had a similar problem, I must have ruled out hardware as drives and motherboard / CPU have all been changed. So there either must be something I am doing wrong or this is normal for TC / VC. I have tried formatting the drives with different cluster sizes, and have tried whole disk and container volumes, the speed is always impacted like this.

Various versions of truecrypt and Veracrypt 1 - 1.17
 
Last edited:
Ok, so it turns out that the intel ICH10 and the controller on my z170 MB both have no problems writing to encrypted disk at their full speed.

The problem only seems to happen when I connect the disks to the LSI controllers. So there must be some bug or limit on the LSI HBA's which does not like encrypted traffic being transferred to the disks.

I have tried several firmware and drivers and it is the same, so I have logged a support ticket with LSI.

This would also affect the IBM M1015 controller which is a rebadged LSI9211
 
Back
Top Bottom