I have not seen those posts and cannot speak towards any reported performance issues. I am personally running VeraCrypt on an SSD and I have encountered no performance degredation.
My SSD is not particularly fast as SSDs go - it is a Hynix HFS256G39MND-3520a. It is an m.2 form factor, SATA 3 device. The manufacturer claims 340MB/s sustained write, but I have not seen actual sustained writes faster than 240MB/s.
As far as TrueCrypt is concerned I see no reduction in write speed when using it. I regularly image the drive, sometimes to an identical SSD, for backup purposes. Most of the time I just image the encrypted drive, but when I do an unencrypted image, I see no change in transfer speed, always around the 240MB/s mark. I am using AES to take advantage of the AES.NI hardware acceleration, which on my computer benchmarks around 5GB/s transfer. Doing the math the best theoretical CPU usage I should see during a sustained write is in the 4% range. When copying very large files internally I see about 10% system CPU usage. This is well inside the expected outcome range, considering that number represents VeraCrypt decrypting and reencrypting the file plus the hardware driver.
I have used Veracrypt version 1.19 and am currently running VeraCrypt 1.21 with this drive and I could not be happier with its performance.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I use system encryption. My computer runs Windows 10, UEFI. I have used VeraCrypt with this SSD under Windows 10 releases 1511 and 1709. When installing the initial encryption took about 24 minutes. Also when subsequently decrypting for the upgrade to 1709 and re-encrypting after the process took about 24 minutes each time then as well, so there doesn't appear to be a performance difference between Windows 10 v1511 and v1709.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I've seen multiple older posts citing significant performance issues with VeraCrypt - particularly with regards to SSD drives.
Have these issues been addressed in v1.21?
If not, what's the status on addressing these problems?
I have not seen those posts and cannot speak towards any reported performance issues. I am personally running VeraCrypt on an SSD and I have encountered no performance degredation.
My SSD is not particularly fast as SSDs go - it is a Hynix HFS256G39MND-3520a. It is an m.2 form factor, SATA 3 device. The manufacturer claims 340MB/s sustained write, but I have not seen actual sustained writes faster than 240MB/s.
As far as TrueCrypt is concerned I see no reduction in write speed when using it. I regularly image the drive, sometimes to an identical SSD, for backup purposes. Most of the time I just image the encrypted drive, but when I do an unencrypted image, I see no change in transfer speed, always around the 240MB/s mark. I am using AES to take advantage of the AES.NI hardware acceleration, which on my computer benchmarks around 5GB/s transfer. Doing the math the best theoretical CPU usage I should see during a sustained write is in the 4% range. When copying very large files internally I see about 10% system CPU usage. This is well inside the expected outcome range, considering that number represents VeraCrypt decrypting and reencrypting the file plus the hardware driver.
I have used Veracrypt version 1.19 and am currently running VeraCrypt 1.21 with this drive and I could not be happier with its performance.
I appreciate your telling me about your experience.
If I remember correctly, the issue was specifically regarding the case where an SSD being encrypted was being used as a system drive.
Is your SSD an encrypted system drive, or is it a supplemental data drive?
I use system encryption. My computer runs Windows 10, UEFI. I have used VeraCrypt with this SSD under Windows 10 releases 1511 and 1709. When installing the initial encryption took about 24 minutes. Also when subsequently decrypting for the upgrade to 1709 and re-encrypting after the process took about 24 minutes each time then as well, so there doesn't appear to be a performance difference between Windows 10 v1511 and v1709.