Thu 11 Oct 2007
The German print magazine C’t (21/07/106-113) tested the performance of Vista’s ReadyBoost. The results are quite disappointing. In most cases, performance wasn’t improved, and under certain conditions it even slowed down the test computer.
The theory behind ReadyBoost sounds quite promising. The advantage of flash storage is that data can be accessed faster than on a hard drive since no hard disk head positioning is required. A disadvantage of flash memory is that the writing of data is relatively slow. But in combination with SuperFetch this shouldn’t be a problem. SuperFetch monitors which programs you use often and preloads them into the system memory. When you launch an app that is already in the RAM it will be started faster. If there is not enough system memory, SuperFetch can leverage ReadyBoost to use a flash drive to store its data there. Since it copies the data in the background to the USB stick, the poor writing performance of flash memory doesn’t matter.
Well, that’s the theory. C’t used the BAPCo SYSmark to test the performance of ReadyBoost. BABCo SYSmark is a well-known performance metric based on real word applications such as MS Office, Adobe Photoshop CS2 or Windows Media Encoder. They tried 30 hardware configurations with different CPUs, hard disks, and RAM. There was only one configuration where ReadyBoost significantly improved performance by 26%. It was a machine with Core 2 Duo E6420 768MB RAM, and an old hard disk. This is not really a common configuration. If you add more RAM to this system the performance improvement won’t be noticeable.
If the computer has a slow processor or if you work with apps which need much CPU power, ReadyBoost might even decrease the overall system performance. The reason is that ReadyBoost encrypts the data it stores on the flash drive. Encryption and decryption cost CPU power. So if your CPU is already busy, ReadyBoost could slow down your computer.
The conclusion is that ReadyBoost is useless at the moment. However, that might change if flash memory becomes faster in the future. If your Vista machine is too slow, you usually just need more RAM. In some cases it might help if you follow my guide to improve the performance of your Windows Vista installation.
Leave a Comment | 
Subscribe RSS
| 
Newsletter
|
Loading ...
I came to the same conclusion. Tried out RB during different phases of the Vista Beta/RC and then on the final. Didn’t need to run benchmarks (Wasn’t publishing the results ), I could tell it was slower.
As you said, we’ll see about it when(much) faster Flash comes around.
Yeah, I also made this experience before, but I thought that my USB stick was just too slow, or that my system didn’t meet the requirements for ReadyBoost. But the C’t guys tested it under almost every thinkable condition. So it is really seems that ReadyBoost is absolutely useless.
[...] Vista ReadyBoost doesn?t improve performance Apparently in some cases it even hurts performance. (tags: ReadyBoost ) [...]
the encryption / decryption wouldn’t be a problem if MS wrote their apps to make better use of different cores - especially with quad cores - 1 core for the os, 1 core for the services, 1 core for the apps etc etc
I would like to see a performance analysis of SuperFetch in the near future. Since the user has no control over what apps get cached, you end up with a very long boot time as seldom used large apps are loaded into memory.
ReadyBoost performance might be improved if the user were given an option to disable encryption, along with a utility to wipe the RB portion of the flash drive when needed.
[...] | Vía 4SysOps [...]
Bob, I didn’t read yet about performance tests of SuperFetch. But somehow I doubt that it really helps to improve performance. It seems to me that this technology is not yet ready. I am going to post something about ReadyDrive soon. The situation there is not really better.