Readyboost Benchmark

Posted by Minnaar Pieters 11 Jan 2007















I tried looking around for a Readyboost comparative benchmark test, but I could not find one. I decided to attempt this myself. I tested Readyboost with normal everyday apps to see if they load faster. This is my current system:
Centrino Duo 1.66 GHz, 1 Gig DDR 533 RAM, 100 Gig Fujitsu SATA 8 meg cache HD, and a ATI Mobility Radeon X1400 with 128meg VRAM.

I ran the test with Vista running Aero and no background apps like Skype or so running. It is a fresh install of Vista with only my apps installed. This includes things like Office 2007, Dreamweaver, Picasa, Google Earth, iTunes etc. A pretty bog standard setup. I am using a 1 Gigabyte Apacer card which has its full capacity assigned to Readyboost. I first did all the test without Readyboost, with a full restart between each different test. Thereafter, I used Readyboost. 2 Attempts are made with each app to see the effect that Readyboost might have if the app is already loaded into RAM.

Booting into Vista Enterprise (not using Readyboost):
Attempt 1: 1:38
Attempt 2: 1:34
Attempt 3: 1:32
Attempt 4: 1:30

Booting into Vista Enterprise (using Readyboost):
Attempt 1: 1:32
Attempt 2: 1:29
Attempt 3: 1:31
Attempt 4: 1:30

Small effect, shaved about a second off booting time. I measured this by creating a notepad document that loads up last in the startup procedure.

Dreamweaver 8 (not using Readyboost):
Attempt 1: 13.2 sec
Attempt 2: 3.1 sec

Dreamweaver 8 (using Readyboost):
Attempt 1: 14.5 sec
Attempt 2: 3,1 sec

Weird. I thought this might be wrong, but retested, and indeed it loads slower with Readyboost. Once in memory though, it loads in the same time.

Outlook 2007, using a 1200 meg PST file (not using Readyboost):
Attempt 1: 8,35 sec
Attempt 2: 3,3 sec

Outlook 2007, using a 1200 meg PST file (using Readyboost):
Attempt 1: 7,95 sec
Attempt 2: 3,3 sec

It seems Readyboost does help if there is a lot of data to be loaded. Not much though. Im starting to become dissapointed.

Google Earth (not using Readyboost):
Attempt 1: 9,76 sec
Attempt 2: 5,8 sec

Google Earth (using Readyboost):
Attempt 1: 12,64 sec
Attempt 2: 6,47 sec

Another slowdown. And yes, I checked it again.

iTunes (not using Readyboost):
Attempt 1: 9,40 sec
Attempt 2: 4,93 sec

iTunes (using Readyboost):
Attempt 1: 12,5 sec
Attempt 2: 4,9 sec

Shut down (not using Readyboost):
42 seconds

Shut down (using Readyboost):
41 seconds

Very dissapointing. But there could be many reasons that my results were this inconclusive. I was using a 1Gig SD card, which is not small, but it was the same size as my RAM. Microsoft recommends using 1 and half to 2 times your RAM. So in conclusion, if you want to use Readyboost, you HAVE to make sure you use at least double your RAM amount for RB. Boohoo.

And I was so excited by this little idea.

UPDATE: Thank you for everyone who gave feedback on this post - indeed there seems to be some type of slowdown because I used a SD card. In the meantime I purchased a Readyboost capable flash drive (Kingston Datatraveller R) and there does seem to be a small improvement in general everyday apps. Small disk accesses like loading thumbnails etc do load much faster. While I do like the whole idea behind Readyboost, I would prefer if Microsoft made a better effort with implementing it. If the SD card made no improvement to my speed in Vista, why does it inform me that I can use it to "speed up my system" in the first place? Clearly this aspect needs to be fixed somehow.

Another aspect of Readyboost I do not appreciate is that it takes too long to activate after a boot up. If you boot up Windows with the readyboost drive installed, it takes a few minutes to kick in. Surely, there must be some performance advantage if it was the first background service to be activated by vista? Then other services could benefit from faster load times?
blog comments powered by Disqus
My photo
I am a R&D Analyst in Stellenbosch South Africa who has a immense passion for all things tech related. I embrace technology, open source and web standards, and I participate and contribute to the social web.