What benchmark tests can I do to measure how slower will my MB be after encryption ?

[XPcentric]

I only tried encryption on Linux and I was not satisfied. I want to try disk encryption on Macbook, and I can't do that until I have some tests that can measure the speed before and after encryption so I can compare.

 

[chown33]

I don't know of any tests other than trying it. I don't see a way to simulate or factor out a testable instance of full disk encryption that isn't an actual instance of full disk encryption.

Copy your current system to an empty external disk. Boot from it. Run a disk benchmark. Then turn on full disk encryption, let it finish encrypting, and run the disk benchmarks again. Letting it finish encrypting could be many hours.

If you like the encrypted performance, then you can enable encryption on your main disk, and erase the external. Otherwise keep your main disk as-is and erase the external.

If it's not clear, the reason for running a disk benchmark before encrypting is to establish the baseline. Performance on an external disk won't be the same as an internal disk. The ratio between unencrypted and encrypted, however, should be a good predictor, if everything else is the same.

 

[stefanski]

If your disk is a standard HDD 5400RPM then you don't want to use disk encryption. It will be noticeably slower and lagging when working with large files.

However, if your upgrading to an SSD you will win 2 fold. Your MB will become more responsive and your disk encryption will be performing perfectly.

I am running an SSD in a 20011 MBP 17" and I have disk encryption on. There's no lag, nothing. It is simply perfect and a no brainer in case my laptop should get stolen or lost.

I would highly recommend encryption on laptops but only if the machine is up for the task. A 7200rpm drive should be ok but I never tried it as I went directly to SSD.

 

[JohnnieBBadde]

If your disk is a standard HDD 5400RPM then you don't want to use disk encryption. It will be noticeably slower and lagging when working with large files. [...]

...?

I have some experience with fully encrypted systems (both SSDs and mechanical HDDs) but I have never encountered that circumstances. Why would encryption slow down HDDs in particular? There a two pivotal questions, the first one is if the CPU supports hardware AES acceleration and the second one is how the operating system's driver is handling the en- and decryption process.

If the original poster has a MacBook with a Core iX CPU (>2009) it has hardware AES acceleration and if he or she is using a recent OS X version FileVault full disk encryption doesn't slow down the system in day-to-day work (but yes, there is a slight speed loss that is measurable, e. g. with the free tool Blackmagic Disk speed)

 

[robvas]

Here are some benmchmarks and discussion:

http://osxdaily.com/2011/08/10/filevault-2-benchmarks-disk-encryption-faster-mac-os-x-lion/

Remember if you have a Sandforce SSD which uses data compression, encryption can slow you down a bit.

 

[JohnnieBBadde]

[...]
Remember if you have a Sandforce SSD which uses data compression, encryption can slow you down a bit.

That's a nice euphemism
From "500 MB/s" (only if you're writing 0s with benchmark tools) it will drop to about 250 MB/s (write speed).
But since SF controllers have been pieces of sh** due to their extremely buggy firmwares one should avoid these anyway. Even now there are still critical firmware updates (depending on the manufacturer/distributor the customer may not be able to update it).

Back in 2011 (before that was common knowledge) I bought five 240 GB SSDs with synchronous Intel MLC NAND and SF 2281 controllers, last month the last of them died suddenly, it had only about 20 TB data written to it and the SMART status was without any cause for alarm.

BTW only buy SSDs from manufacturers that also produce the NAND chips in their SSDs and deploy firmware updates for end users (e.g. Crucial (->Micron), Samsung, Intel, Sandisk).

 

[robvas]

That's a nice euphemism
From "500 MB/s" (only if you're writing 0s with benchmark tools) it will drop to about 250 MB/s (write speed).
But since SF controllers have been pieces of sh** due to their extremely buggy firmwares one should avoid these anyway. Even now there are still critical firmware updates (depending on the manufacturer/distributor the customer may not be able to update it).

Back in 2011 (before that was common knowledge) I bought five 240 GB SSDs with synchronous Intel MLC NAND and SF 2281 controllers, last month the last of them died suddenly, it had only about 20 TB data written to it and the SMART status was without any cause for alarm.

BTW only buy SSDs from manufacturers that also produce the NAND chips in their SSDs and deploy firmware updates for end users (e.g. Crucial (->Micron), Samsung, Intel, Sandisk).

Apple uses Sandforce drives in some of the 2012 Airs and possibly other models.

 

[JohnnieBBadde]

Thank you for that information I wasn't aware of that. Let's hope Apple ensures SandForce firmware updates on former models that came with them even in the present.

Then, of course, encryption decreases the drive's speed quite a bit, although real-life data is rarely that well compressible so they were never as fast as advertised (cp. benchmark tools that use random data instead of zeros).

 

[newellj]

I only tried encryption on Linux and I was not satisfied. I want to try disk encryption on Macbook, and I can't do that until I have some tests that can measure the speed before and after encryption so I can compare.

Which Macbook are you using?