Speed test with version 1.3 VS encfs4win

ehab

my last test for today was to see the speed difference between BOXCRYPTOR 1.3 against encfsw using Dokan.

BOXCRYPTOR is fast :D can be seen from first results

-----------------------------------------------------------------------
CrystalDiskMark 3.0.1 (C) 2007-2010 hiyohiyo
Crystal Dew World : http://crystalmark.info/
-----------------------------------------------------------------------
* MB/s = 1,000,000 byte/s [SATA/300 = 300,000,000 byte/s]

Sequential Read : 17.401 MB/s
Sequential Write : 16.836 MB/s
Random Read 512KB : 17.438 MB/s
Random Write 512KB : 16.793 MB/s
Random Read 4KB (QD=1) : 12.775 MB/s [ 3119.0 IOPS]
Random Write 4KB (QD=1) : 10.469 MB/s [ 2556.0 IOPS]
Random Read 4KB (QD=32) : 13.753 MB/s [ 3357.8 IOPS]
Random Write 4KB (QD=32) : 11.152 MB/s [ 2722.5 IOPS]

Test : 100 MB [X: 16.9% (346.4/2048.0 MB)] (x1)
Date : 2012/03/26 21:43:13
OS : Windows 7 SP1 [6.1 Build 7601] (x86)

the next table is for encfs4win, looks like when the os uses 4KB the paging swaps hurt the performance.

Sequential Read : 16.129 MB/s
Sequential Write : 10.928 MB/s
Random Read 512KB : 12.179 MB/s
Random Write 512KB : 10.820 MB/s
Random Read 4KB (QD=1) : 0.638 MB/s [ 155.7 IOPS]
Random Write 4KB (QD=1) : 0.450 MB/s [ 110.0 IOPS]
Random Read 4KB (QD=32) : 0.663 MB/s [ 161.9 IOPS]
Random Write 4KB (QD=32) : 0.434 MB/s [ 106.1 IOPS]

Test : 100 MB [Q: 78.9% (69.4/87.9 GB)] (x1)
Date : 2012/03/26 21:46:19
OS : Windows 7 SP1 [6.1 Build 7601] (x86)

so good work boxcryptor.

Robert

Thanks for this information!

--Robert

hpvd

looks good!
a comparison with the same drive and cpu without encryption be very interesting also ;-)

ehab

@hpvd, here you go " same drive - different partition"

Sequential Read : 60.332 MB/s
Sequential Write : 54.344 MB/s
Random Read 512KB : 34.813 MB/s
Random Write 512KB : 30.907 MB/s
Random Read 4KB (QD=1) : 0.577 MB/s [ 140.8 IOPS]
Random Write 4KB (QD=1) : 1.044 MB/s [ 254.9 IOPS]
Random Read 4KB (QD=32) : 0.710 MB/s [ 173.3 IOPS]
Random Write 4KB (QD=32) : 1.080 MB/s [ 263.8 IOPS]

Test : 100 MB [Z: 45.1% (3155.9/6999.0 MB)] (x1)
Date : 2012/03/29 22:30:52
OS : Windows 7 SP1 [6.1 Build 7601] (x86)

neiaberau

I am confused, more than 2500 IOPs, I want to by this disk ....

ehab

@neiaberau, can you suggest another test tool? or if you can also try and post your outputs.

neiaberau

Unfortunately I do not know a proper way of testing, this as this is not in my field of expertise. The problem with the numbers above are, that they are faster than physics allows for hard disks. So they must by partially based on ram/cache.

So two questions arise:
Has boxcrpytor build in a caching algorithm? or has cbfs built in a caching algorithm?
Can we disable these, in order to test only the encrpytion and virtual file sytsem performance?

Caching of reads I considere acceptable, but cached random writes can lead to data loss, I would not be willing to accept.

ehab

@neiaberau, i believe boxcrpytor caches filenames.

definitely cbfs is faster than dokan.

Joe0815

@ehab
would you please do the same with a network share like Strato HiDrive?
Or with an SD-Card. Both are terribly slow over here (writing large files to a BC-Folder on SDXC-Card (Sandisk Mobile Ultra 64GB) takes 5 (in words: FIVE) times it takes to write to that same card unencrypted (in my case: >5MB/s unencrypted, about 1MB/s using BC).

Thank you, Joe

ehab

@Joe0815,

I only read from SD's rarely write, however i do backup my encrypted files to SD and other storage.

I did try HiDrive with BC and found it slow using direct webdav until i met WebDrive from http://www.southrivertech.com/ i have tried it and found it very reliable, it has its cache mechanize that makes working with it delightful.

Joe, if you use HiDrive with BC a lot then try WebDrive trial maybe it works best for you. DO check the cache settings since they are the main area for performance and reliability.

ehab