6 April 2007
Six Easy Steps
Currently, there is no official list of devices which are compatible with ReadyBoost. Thousands of users have put their devices to trial by Vista, and compiled unofficial lists. One of the most comprehensive lists has been made by community contributions to Grant Gibson. But just knowing compatibility is worthless - because in this blitzing information age, we know that we all want the best speed. Here’s how to find out Vista’s test results for your device.
- Make sure that Vista is using your flash memory device for ReadyBoost.
- Go to My Computer or File Explorer, clear the address bar at the top and enter this: eventvwr.msc
- Press Continue when Vista asks for permission, and the Event Viewer window will be displayed.
- On the left panel of the Event Viewer, navigate the tree as follows: Applications and Services Logs > Microsoft > Windows > ReadyBoost > Operational
- Double-click the ‘Operational’ section to open it.
- On the right panel, hit the ‘Find’ button and search for the word ‘Speed’. Now close the Find window. In the middle panel are Vista’s speed test results (note: ‘Sequential Write’ appears to be a mistype, and is in fact ‘Random Write’)
How Fast is Your Device?
If you would like to share the Vista speed ratings for your device, we host a database for them in our ReadyBoost - Flash Memory Speed List article.
Speed Ratings
Before a flash memory device is used by Vista for ReadyBoost, it is tested for the minimum requirements, which supposedly are as follows (revised):
- Random read speed of 2.18MB/sec* (2,229KB/sec) for reading a 4K block of data;
- Random* write speed of 1.75MB/sec (1,792KB/sec) for writing a 512K block of data; and
- Capacity of 235MB.
Most devices show lightning fast speed ratings on their packaging. These figures are based on sequential read/write. When ReadyBoost uses a device as additional cache memory, both speed ratings go out the window, and there’s a whole different court for this new ball game. It becomes a test of random access speed, or how fast the device can instantaneously retrieve and write data regardless of what order it is stored in.
*EDIT: It is clear that the minimum Random Read Speed is overstated by Microsoft. Vista has reported read speeds of 2,229KB/s for devices which are still compatible with ReadyBoost.
*EDIT: Thanks to one of our readers, atlr, who has commented below that ‘Sequential Write’ readings reported in the Event Viewer, are in fact, ‘Random Write’ readings. This initiated us to carry out Winsat tests on a Kingston DataTraveler R (Enhanced for Windows ReadyBoost). The Event Viewer reported 8,305KB/s of random write on this basis. The Winsat 512K random write test reported 11.13 MB/s (11,397 KB/s). A Winsat 512K sequential write test reported 16.05 MB/s (16,435 KB/s). Therefore, atlr appears to be correct! It is possible that a discrepancy exists because the former is a quick test, whereas the latter is more comprehensive.
Sources
http://blogs.msdn.com/tomarcher/archive/2006/06/02/615199.aspx
http://blogs.zdnet.com/Bott/?p=186
Greetings. I have been examining ReadyBoost too.
I notice that sequential write values are gathered since that is the value identified in ReadyBoost Windows Event 1000. Documents on Microsoft’s web site are not consistent about whether sequential or random writes are the relevant measurement for judging ReadyBoost performance. Here is a document that talks about measuring random write throughput.
https://microsoft.mrmpslc.com/windowsvistapartner/resources/ReadyBoostKit.zip
The most recent Microsoft recommendation I have found is a “non-volatile flash memory buffer with at least 1GB of storage capacity. The flash memory buffer must also meet the requirements for random reads and random writes specified in the Windows Vista Logo “Storage-0009″ WLP specification:
• 5 MB/sec throughput for random 4k reads across the entire device
• 3 MB/sec throughput for random 512k writes across the entire device”
Users see the most benefits when they have between a 1:1 and 1:3 ratio of RAM to ReadyBoost flash.”
reference: Overview.doc from https://microsoft.mrmpslc.com/windowsvistapartner/resources/ReadyBoostKit.zip
Here are command line utilities to check flash device read/write speeds.
To test read speed:
winsat disk -v –read –ran –ransize 4096 –drive
To test write speed:
winsat disk -v –write –ran –ransize 524288 –drive
The RBTest tool tests a device for “slow regions” on a device.
RBTest.exe –assess
As of 2007-04-05, here are some USB drives that I have read reach random read and write speeds above 8 MB/sec.
US$100 Corsair model CMFUSB2.0-8GBGT “8GB Flash Voyager GT”
US$65-$100 Apacer model AP-HHR4009D2/G “4GB Handy Steno HT203 200X”
US$60 Buffalo model RUF2-S4G-WH “4GB Ultra High Speed”
US$55-$75 Corsair model CMFUSB2.0-4GBGT “4GB Flash Voyager GT”
US$35-$55 Kingston model DTR/2GB “2GB DataTraveler ReadyFlash”
US$35 AVB model UF2-2GJ “2GB AVB NAND SLC 2GB ReadyBoost Flash Drive”
US$35 Buffalo model RUF2-S2G-WH “2GB Ultra High Speed”
US$25-$40 Corsair model CMFUSBTV2.0-1GB “1GB TurboFlash USB 2.0 Drives”
Today’s fastest USB flash dives often consist of Single Level Cell (SLC) flash chips in a dual channel or interleaved design.
Here are other lists of Windows ReadyBoost compatible storage devices:
10 April 2007 @ 2:11 amhttp://www.grantgibson.co.uk/misc/readyboost/
http://www.schrockinnovations.com/enhanced_for_readyboost.php
Is there anyone who tried this test with CF2SATA adaptor like following products?
http://www.dansdata.com/cfide.htm (CF2IDE)
http://www.tomshardware.com/2005/08/18/accelerated_compact_flash/ (CF2SATA)
Sandisk’s Extreme III and IV series of CF card claims there speed exceeds over 20MB/s (III) and 40MB/s (IV)!
But all tests done was using USB2.0/Firewire/PC-Card card readers to connect them.
Since they use PIO4 mode which has maxium speed of 16.7MB/s and may bound to PIO speed of controller chip which scores comparative low real speed than theoratical maxium, all values listed in this site are lower than I expect for CF2IDE/CF2SATA configurations.
Sandisk’s OEM manual, shown following links, may report their product support upto UDMA5 (133MB/s) for ATA configuration and I didn’t test it with suggest config at this time of writing.
http://www.sandisk.com/Assets/File/OEM/Manuals/OEMCFlashPMv12.0.pdf
Other CF cards may variant from this results also, since CF controller interface implementation varies from vendor to vendor.
I want to hear various testing results with CF2IDE/CF2SATA configuration for ReadyBoost performance.
Sincerely,
13 April 2007 @ 5:25 amSeongHan Kim
wmnull@gmail.com
Here is podcast with Alex Kirshenbaum who wrote the ReadyBoost code explaining what it does.
http://www.microsoft.com/technet/community/tnradio/archive/ayers.mspx
Note that the 512kb ***random*** write throughput should be measured.
The word “sequential” in the ReadyBoost Windows Event 1000 text “The sequential write speed is …” is a mistake. I believe the value shown in that event is the correct random write speed measurement. Just the text is wrong.
I encourage anyone to confirm this for oneself. Run the Vista command line utility, winsat, to measure storage device speed. winsat requires Administrator privilege to run. Right-click on Command Prompt icon, then select Run as Administrator
winsat disk –write –ran –ransize 524288 –drive
winsat disk –write –seq –seqsize 524288 –drive
reference: Overview.doc from https://microsoft.mrmpslc.com/windowsvistapartner/resources/ReadyBoostKit.zip
14 April 2007 @ 1:30 amThanks for this excellent article and the comments added by “atlr”. The article mentions discrepancies for test results with the Kingston DataTraveler R, and that it may be due to the “quick test” shown in the Event Viewer vs. the “more comprehensive” winsat. That had me wondering about the rankings since apparently the data entered is based on the quick test shown in the Event Viewer vs. winsat results.
With that in mind, here are relevant results obtained using a LEXAR JD LIGHTNING II 4GB Thumb Drive fresh out of the box with FAT32 format. This further illustrates the discrepancies:
EVENT VIEWER
“The random read speed is 8078 KB/sec. The sequential write speed is 12279 KB/sec.”
WINSAT
Random Read 4k: 6.46 MB/s
Random Write 512k: 16.22 MB/s
Seq Write 512k: 22.66 MB/s
CrystalDiskMark 1.02
Random Read 4k: 8.021 MB/s
Random Write 512k: 8.22 MB/s
Seq Write 512k: 19.049 MB/s
I’ve not posted the results to the chart due to the discrepancies and it’s still not clear to me where this drive really ranks. Curiously, CrystalDiskMark (a seemingly thorough test) reported similar Random read results (8.021 MB/s) as the Event Viewer quick test, however Winsat was lower. The winsat read test was done using the command (run as administrator):
winsat disk -read -ran -ransize 4096 -drive
(with the appropriate drive letter at the end)
Curiously the write speed reported by the Event Viewer is midway between the vastly different figures reported by Winsat (16.22 MB/s) & CrystalDiskMark (8.22 MB/s). I ran this test multiple times and had similar results so I don’t know why there is such a large discrepancy between the 2 “more thorough” test tools. I included the sequential write figures just to be complete and as further evidence that Event Viewer is reporting random not sequential write speed. By the way, the manufacturer (Lexar) claims 21 MB/s (hence the “lightning” name”) which Winsat & CrystalDiskMark essentially confirm, however for Ready Boost purposes that isn’t relevant.
Bottom line is that this leaves me wondering if the Winsat results are the better benchmark and if so perhaps an option/suggestion would be for users to be able to enter that result in an additional box for the chart so that others could view/compare.
22 May 2007 @ 5:14 pm[…] of the drives with the default NTFS filesystem and did a bit of ad-hoc testing in Vista. You can find the raw ReadyBoost benchmark results if you know where to look in the event […]
5 June 2008 @ 6:35 amSome discrepancies found between logged events and winsat but more difference between when using a usb reader for a microSDHC card and the supplied SD-converter in an internal laptop SD slot - read and write speeds went in opposite directions! Formatting to NTFS and FAT32 (as supplied) slowed read speeds but increased write speeds. All very strange behaviour to a pseudo-techie like me. I’ve settled on FAT32 but shall be using the card for other purposes as i don’t expect much performance increase.
4 July 2008 @ 12:27 amPlease note the Winsat syntax given above is incorrect…
winsat disk –write –ran –ransize 524288 –drive
winsat disk –write –seq –seqsize 524288 –drive
the correct format is (for example):
winsat disk -write -ran -ransize 524288 -n 1
(the “1″ indicates the logical disk number of the flash drive, obtained from: administrative tools/computer management/storage/disk management)
4 July 2008 @ 6:16 am