Why would CrystalDiskMark report a drive as faster (in some tests) with bitlocker on?

Question

I have a cheap Linx 10v64 tablet - Intel Atom x5-z8300, 4GB RAM, 64 GB eMMC module.

I am considering enabling bitlocker on the main (OS) drive but am concerned about the effect on performance given the modest specs of the machine.

I made a 7GB partition for test purposes, and ran CrystalDiskMark :

-----------------------------------------------------------------------
CrystalDiskMark 5.2.0 x64 (C) 2007-2016 hiyohiyo
                           Crystal Dew World : http://crystalmark.info/
-----------------------------------------------------------------------
* MB/s = 1,000,000 bytes/s [SATA/600 = 600,000,000 bytes/s]
* KB = 1000 bytes, KiB = 1024 bytes

   Sequential Read (Q= 32,T= 1) :   140.647 MB/s
  Sequential Write (Q= 32,T= 1) :    89.082 MB/s
  Random Read 4KiB (Q= 32,T= 1) :    33.730 MB/s [  8234.9 IOPS]
 Random Write 4KiB (Q= 32,T= 1) :    15.586 MB/s [  3805.2 IOPS]
         Sequential Read (T= 1) :   156.034 MB/s
        Sequential Write (T= 1) :    96.884 MB/s
   Random Read 4KiB (Q= 1,T= 1) :    10.380 MB/s [  2534.2 IOPS]
  Random Write 4KiB (Q= 1,T= 1) :    15.317 MB/s [  3739.5 IOPS]

  Test : 1024 MiB [E: 0.4% (26.4/7167.0 MiB)] (x5)  [Interval=5 sec]
  Date : 2016/11/15 21:52:26
    OS : Windows 10 Professional [10.0 Build 14393] (x64)

I then enabled bitlocker on the partition, and ran again:

-----------------------------------------------------------------------
CrystalDiskMark 5.2.0 x64 (C) 2007-2016 hiyohiyo
                           Crystal Dew World : http://crystalmark.info/
-----------------------------------------------------------------------
* MB/s = 1,000,000 bytes/s [SATA/600 = 600,000,000 bytes/s]
* KB = 1000 bytes, KiB = 1024 bytes

   Sequential Read (Q= 32,T= 1) :   154.574 MB/s
  Sequential Write (Q= 32,T= 1) :    92.137 MB/s
  Random Read 4KiB (Q= 32,T= 1) :    33.680 MB/s [  8222.7 IOPS]
 Random Write 4KiB (Q= 32,T= 1) :    16.052 MB/s [  3918.9 IOPS]
         Sequential Read (T= 1) :   120.384 MB/s
        Sequential Write (T= 1) :    70.879 MB/s
   Random Read 4KiB (Q= 1,T= 1) :     8.561 MB/s [  2090.1 IOPS]
  Random Write 4KiB (Q= 1,T= 1) :    13.000 MB/s [  3173.8 IOPS]

  Test : 1024 MiB [E: 0.4% (26.6/7167.0 MiB)] (x5)  [Interval=5 sec]
  Date : 2016/11/15 22:07:10
    OS : Windows 10 Professional [10.0 Build 14393] (x64)

I was surprised to see that while some of the measured speeds are slower, the sequential tests with multiple queues are a little faster.

How could this be? I thought bitlocker represented an overhead on the processor - which in this case is only an Atom.

Answer 1

BitLocker may be doing some compression as well as encryption, so it's a trade-off between some CPU overhead vs. smaller data on the HDD. With a faster disk, e.g. SSD, the results would probably be as expected -- BitLocker would be slower.

Consider that magnetic disk drives are far slower than RAM by a factor of hundreds or thousands of times, so a small savings in disk access is worth considerable processing use. BTW, I notice that imaging software such as Macrium Reflect show effective disk write speed of ~10 times faster than the maximum possible disk throughput, due to moderate compression.