Hi,
In March 2010 I made a post on this forum using the same subject title, but for some reason that thread is now locked so I start here again to give more info for anyone interested, as I have some more info after trying a new card.
First some background info for those new to the subject as this info I have found scattered from various places on the Web.
I have a desktop replacement laptop (HP 8530w) on Windows 7 Ultimate with 8GB RAM, 7200rpm disk, NVIDIA graphics etc, so it is a fast machine without ReadyBoost, but ReadyBoost makes it noticeably even snappier. The Laptop has a built in media card reader which I wasn’t using, so leaving a media card (I choose SDHC) in it dedicated to ReadyBoost was a no hassle option compared to a usb stick that would stick out and get in the way.
Here is a summary of my findings which might help anyone else struggling to find information on this subject as I was.
ReadyBoost uses random read/writes which is very different the sequential/continuous read/writes as used by video/digital photo cameras. The class rating of cards e.g. class 2, 4, 6, 10 etc (the number inside the “C” printed on the cards) relates to their continuous read/write performance, not their random read/write performance. Consequently the class rating (from my experience) isn’t a reliable indicator to ReadyBoost performance.
When you connect a removable storage device to your PC for the first time (or after reformatting it and right clicking “properties” on the drive) it tests its random read/write speeds for ReadyBoost performance. You can view the results of these tests in Event Viewer ( “View Event Logs” - “Applications and services logs” - “Microsoft” - “Windows” - “ReadyBoost” - “Operational”, look for event ID’s of “1000”, you can sort by ID to help find them and the entries are date and time stamped).
The event log enables you to compare different media’s ReadyBoost performance. However, be aware the results have some variation. Also if your machine is busy doing something else it will affect the results. Consequently I test cards after a reboot and test each card 4 times and take an average of the results. Also be aware that ultimately the cards can only perform as well as your card reader and PC hardware, there are many types and some are not so fast so be aware your reader/system might be maxing out, not the card.
Also the format of the card has a smallish effect on the results. I found a Microsoft produced article that recommended exFAT be used for ReadyBoost for, in their words, marginal benefits compared to NTFS. Sadly the article said nothing about allocation unit size and there is some debate over the optimal size (I just stuck to the exFAT default of 32K). If the card is 4GB or less or if it is bigger but you only want to use up to 4GB for ReadyBoost then you can also use FAT32. If your card is bigger than 4GB but you format it as FAT32 ReadyBoost will only be able to use a 4GB cache, to allow ReadyBoost to use more you have to reformat it as either NTFS or exFAT (exFAT is recommended by Microsoft).
Microsoft recommend the ReadyBoost cache is between 1.5x and 3x the physical RAM in your PC, but even 1x makes a difference if the ReadyBoost card is significantly faster than your hard disk. If you have a solid state hard disk you won’t benefit from ReadyBoost.
Here is a post made by Iwod in my previous thread…
There are few things to consider for ReadyBoost SDHC,
-
It must be fast in Random Write
-
SLC Chips would be better.
Silicon Power and Transcend both produce 4GB SDHC card that uses SLC, and SLC are generally superior in Random Write. So if you choose a SLC SDHC card it will solve two problems at once.
The only problem is no manufacture seems to produce SLC anymore because they are much more expensive.
Sandisk Old version of Extreme III ( 20MB/s ) use SLC are according to my research has the fastest Random Write results.
Iwod
---------------------------------------------------------------------------
I have tested the following so far (remember the speeds I quote can only be seen as a guide to comparative performance of the cards, the stated speeds will vary in different machines:
Samsung class 6 8GB MicroSDHC card NTFS random read 7580KB/sec, random write 4196KB/sec.
SanDisk class 10 16GB Exteme III (30MB/s) SDHC card (Fat32) random read 6000KB/s, random write 4000KB/s
SanDisk class 10 16GB Exteme III (30MB/s) SDHC card (NTFS) random read 4132KB/s, random write 3864KB/s.
Samsung class 6 16Gb SDHC card (exFAT) random read 7124KB/sec, random write 4038KB/sec.
Integral Class 10 16Gb Endurance (30MB/s) SLC SDHC card (exFAT), random read 5517KB/sec, random write 8164KB/sec.
So some interesting findings. The Samsung Class 6 is faster at random read/writes (and hence ReadyBoost) than the SanDisk class 10, but very interestingly the Integral’s SLC chip based card is much faster in write than it is in read. Seems strange but I tested it probably 6 times in NTFS, and exFAT and each time the results were virtually identical. I presume this is characteristic of SLC based cards as Iwod stated.
Certainly the Samsung cards made my PC noticeable faster than the SanDisk one, and my new Integral card make the PC noticeably faster (but not by much) than the Samsungs.
So I have to concur with Iwod, you are best buying an SLC based card for ReadyBoost
Hope that helps.
RobinGB