Does ReadyBoost Really Improve Performance?
By Gabriel Torres on April 6, 2007 - 10:18 AM
This entry is based on preliminary testing we are done here at our lab. As soon as we get more conclusive data we will post a full review about ReadyBoost.
ReadyBoost is a technology brought by Windows Vista that allows any piece of flash memory like pen drives and memory cards to be used as a disk cache improving, in theory, the system performance. However so far we were not able to see any performance improvement on our system. Let’s explore the subject.
First we thought that ReadyBoost would be a conventional disk cache system, so disk performance gain would be noticeable right after installing a pen drive and configuring as a disk cache.
Installing a Patriot 2 GB pen drive on our system and running hard disk drive benchmarking software like HDTach and Sandra, however, didn’t show us any disk performance improvement – on the contrary, performance dropped a little bit.
So we had these considerations on our mind:
So we decided to load some programs and files and measuring the time they took to be loaded, with and without ReadyBoost.
With 2 GB installed on our system (Core 2 Extreme X6800, ASUS P5B, Samsung SP0411N HDD, MSI GeForce 8800 GTS 320 MB) we saw no performance improvement.
We start reading about ReadyBoost on the net and some people are saying that ReadyBoost helps caching the Windows swap file. This file, also known as virtual memory, is used when the CPU needs more RAM memory. When there is no more free RAM, the CPU stores what is in the RAM in this file, freeing up RAM. Since the hard disk drive is slower than RAM, this transition is noticeable by the user. That is why when you install more RAM your system becomes faster: your computer will need to access the swap file less times, as there will be more RAM available.
So we reduced the amount of RAM on our system to 512 MB in order to see what would happen.
Loading Photoshop CS2 for the first time took us around 18 seconds without ReadyBoost and then took us 29 seconds with ReadyBoost. So performance decreased – probably because Windows was caching Photoshop files to our pen drive, we thought.
Then from the second time we loaded Photoshop on our system took 5:30 seconds to load the program without ReadyBoost while this loading time decreased to 4:40 seconds with ReadyBoost. We gained 1 second, but we keep thinking if it was worth it, as our system delayed a lot more to load Photoshop for the first time.
Keep in mind that the performance we gained from loading programs from the second time on – 18 seconds vs. 5:30 seconds – was due to SuperFetch technology, having nothing to do with ReadyBoost.
We tried the same approach with Word 2007 and for our surprise ReadyBoost only warmed our performance. Loading Word 2007 for the first time without ReadyBoost took only 5 seconds, while with ReadyBoost it took around 26 seconds.
Loading Word 2007 from the second time one took us only around 1 second without ReadyBoost, but the loading time increased (i.e. performance decreased) to at least 1:50 second when we enabled ReadyBoost.
How ReadyBoost works and whether it can really improve performance is a total mystery to us. Here it made more bad than good. We will keep making some tests here and we will post a more conclusive review as soon as we have a better conclusion on this technology.