VeryCrypt takes very long to mount after conversion from TrueCrypt volume

[Oliver]

Hi everyone

First, a massive thank you. I'm glad VeraCrypt has been developed and it's working very well.

However, after having converted my old TrueCrypt drives to VeraCrypt drives using the "Change Volume Password" option, VeraCrypt now needs around 4-5 minutes to boot my USB drives.

I have checked various posts on this forum and they confirm that mounting a VeraCrypt device takes a lot longer compared to TrueCrypt due to increased security measures which I agree with, however, is this behaviour normal? 5 minutes? Anything I can do or try to decrease that time?

Many thanks for your support.

Regards

[Mounir IDRASSI]

Hi Olivier,

When using the default settings, the time needed to mount volumes depends on the CPU speed since the security level is static and set to a high value. This level is OK on modern CPU since it takes only few seconds for mount volumes on a typical Core i7 CPU. For example, on a Core-i7 2600K CPU @ 3.40 GHz, it takes between 4 and 7 seconds to mount an SHA-512 volume.

Actually, there are benchmark scripts for this and you can get them at: https://sourceforge.net/p/veracrypt/code/ci/master/tree/Tests/
Here is the output of bench.bat on Core-i7 2600K CPU:

SHA-512 (Normal) = 00:00:03,45
SHA-512 (Hidden) = 00:00:06,84

Whirlpool (Normal) = 00:00:06,78
Whirlpool (Hidden) = 00:00:13,49

SHA-256 (Normal) = 00:00:05,04
SHA-256 (Hidden) = 00:00:10,19

RIPEMD-160 (Normal) = 00:00:10,67
RIPEMD-160 (Hidden) = 00:00:21,36

Wrong Password (PRF Auto-detection)= 00:00:21,61

Based on this, a mount time of 5 minutes is excessive and the only explanation is that you are using a low end CPU.

In cases like this, the PIM feature was introduced to enable users to have a quick mount time when the password is 20 characters or more. You can read more on PIM here: https://veracrypt.codeplex.com/wikipage?title=Personal%20Iterations%20Multiplier%20%28PIM%29

The link above explains how to set a custom small PIM to your existing volumes.

[Testoslav]

Benchmark is great idea, but how does it actually work? I'm getting very different results between sha-512 and whirlpool, but real usage takes the same time with sha-512 and Whirlpool (~10 seconds). Tested with VC 1.17beta on rather old core 2 duo @ 2GHz.

SHA-512 (Normal) = 00:00:04,97
SHA-512 (Hidden) = 00:00:09,94

Whirlpool (Normal) = 00:00:10,80
Whirlpool (Hidden) = 00:00:21,51

SHA-256 (Normal) = 00:00:07,55
SHA-256 (Hidden) = 00:00:14,88

RIPEMD-160 (Normal) = 00:00:00,05
RIPEMD-160 (Hidden) = 00:00:00,04

Wrong Password (PRF Auto-detection)= 00:00:51,51

Btw. there should be some if not exist ripemd skip, or it benchmarks under second, when you forget to download ripemd (or another) container ;)

[Mounir IDRASSI]

The measurements done using the benchmark script are very accurate because they measure the raw time taken by the mount operation without any "visual artifacts". I don't know how you did your measurements because definitely SHA-512 and Whirlpool don't take the same time. You can read the benchmark script to see the correctness of the measurements methodology (I have updated recently to ignore missing test containers).

By the way, the last 1.17-BETA16 introduces a huge performance improvement (~2x speedup) thanks to an optimization discovered by fellow Frenchman Xavier de Carné de Carnavalet from Concordia University.
The table above gives a comparative figure of performances between version 1.16 and 1.17-BETA16 on a Core-i7 2600K CPU @ 3.40 GHz.

PRF	VeraCrypt 1.16	VeraCrypt 1.17-BETA16
SHA-512 (Normal)	3,45 s	1,88 s
SHA-512 (Hidden)	6,84 s	3,55 s
Whirlpool (Normal)	6,78 s	3,32 s
Whirlpool (Hidden)	13,49 s	6,58 s
SHA-256 (Normal)	5,04 s	2,68 s
SHA-256 (Hidden)	10,19 s	5,27 s
RIPEMD-160 (Normal)	10,67 s	6,14 s
RIPEMD-160 (Hidden)	21,36 s	12,23 s
Wrong Password	21,61 s	12,20 s