I recently received an email from John Marshall pointing out that everyone was reporting sequential write performance on my ReadyBoost compatibility chart, whilst the official requirements were for random write performance.

I couldn’t get to the bottom of it on the web, so I emailed Matt Ayers at Microsoft, the Program Manager who’s in charge of the ReadyBoost feature.  He explained that early beta versions of Vista measured sequential writes, but that the release version measures random writes.  Everything has been updated in the release version apart from the EventLog entry. 

Matt has filed a bug, so expect that to be fixed in future updates.  Thanks for pointing that out John!

Anyone who’s tried to use the ReadyBoost chart recently will be pleased to know that I’ve given the page a bit of a speed boost.  When I first put up the page it listed the compatibility of about 32 Flash drives.  That list now runs to 880 devices and increases daily, making the HTML page 500Kb — quite a lot to serve over my home cable line!

I’ve now split up the chart so it shows 20 results per page.  There are options to jump to specific makes, and so far no make runs to more than 7 pages, so hopefully that’s an ok compromise for everyone.

Please let me know if you spot any problems!

Just a quick post to clear up any confusion on the ReadyBoost compatibility list.  It turns out that devices which don’t quite meet the Vista storage spec may still show up as ‘compatible’ in Windows Vista (and in this list). 

Dave Pryce emailed to point out that the minimum spec for ReadyBoost is 2.5MB/s read & 1.75 MB/s write, while Microsoft’s ReadyBoost usage guidelines say that for a device to be listed as Enhanced for ReadyBoost it must do around double that - 5MB/s read & 3MB/s write (based on 4k random reads and 512k random writes).

You probably already know that not all Flash devices are created equal, but it’s interesting to also note that devices that don’t quite meet the spec are still deemed compatible in Vista. My recommendation is to look for read/write rates wherever they’re quoted in the ReadyBoost chart - a faster device should result in a more noticeable performance gain.

The ReadyBoost device list is still growing - 364 devices tested so far, 59% of which have worked with Vista ReadyBoost.

Kingston have fared better than average with two thirds of tested devices working successfully.  My own DataTraveler Elite has been running a ReadyBoost cache flawlessly for several months now.

A few days ago I received an email from Ron Bartle who pointed out the inconsistency in reports for the DataTraveler range.  Some users had success with the DataTraveler 1 and U3 devices whilst others had received failure notices for the same drives, even those with identical capacities. 

I contacted Kingston to see if they could shed any light on why that might be.  Their reply doesn’t answer that question directly, but points out that there’s only one Kingston device that is officially ReadyBoost compatible: 

So far there is only one Kingston DataTraveler which is officially compatible with Readyboost which is our DTS.

The reply went on to say that more new ReadyBoost compatible devices are on the way, and that the rest of their existing product line is still in testing for ReadyBoost compatibility. 

So for now it seems the safe option is to go for the DTS.  If you’re feeling brave you could try the DataTraveler Elite - it’s not officially compatible, but it’s the one I’m using as I write this and it’s had no negative reports so far… and it’s around 30% cheaper than the DTS.

The Vista ReadyBoost compatibility list received its 200th submission today.  The list now tracks the ReadyBoost ‘readiness’ of devices from over 45 manufacturers. 

Some fun facts based on the data gathered so far:

• A little over half all tested devices (55%) were confirmed compatible with ReadyBoost.
   
• The highest number of compatible device reports have come from SanDisk users - nearly 20% of all positive reports received.  In total, over 78% of the SanDisk devices were ReadyBoost compatible.
   
• PNY users reported the greatest number of incompatible devices - 12% of all the incompatible devices reported. Most PNY users (55%) found their devices to be incompatible.  Funny then that PNY should choose to mention ReadyBoost as a reason to buy PNY.

If there’s one thing the list does show, it’s that no one manufacturer has ReadyBoost compatibility nailed across their range.  If you’re thinking about buying a new Flash device for ReadyBoost, check the list first.

Finally, many thanks to everyone who have submitted their results so far.  And special thanks to Maverick, Jim Byram and the others who have returned several times with compatibility reports for new Flash devices.  The list wouldn’t exist without all your help.

I recently ordered a Kingston DataTraveler Elite USB Flash drive to try out Vista’s ReadyBoost feature (previously blogged about here).  It arrived in the post from eBuyer today and I decided to run a few semi-scientific tests to see what difference, if any, it made.

Boot Times

I timed Windows startup from cold before and after adding the ReadyBoost device.  I started the timer from the second I pressed the power button, as I wanted to include any time lost through BIOS recognition of the USB device.  The timer was stopped when Vista reached the ’select user’ screen. 

Before: 49.5 secs
After: 51.0 secs (yep, slower)

This slowdown seemed to be 100% down to device recognition time - it spent a bit longer on the initial BIOS screens, and there wasn’t any noticeable improvement in the rest of the boot process.

Adobe Photoshop startup times

Photoshop is always slow to start, especially if you have lots of filters, fonts and extensions installed.  However, even with regular Windows disk caching it’s much faster to start the second time around.  So for this test I’m timing both start and re-start times for Photoshop.  

Before ReadyBoost, first start: 26.5 secs.  Second start: 14.6 secs.
After ReadyBoost, first start: 16.0 secs.  Second start: 11.7 secs.

That’s much more like it!  The promise of ReadyBoost - that it will significantly speed up access for small files - seems to be proven here.  Photoshop requires access to hundreds, perhaps thousands of small files when starting up.   From listening to the HDD it certainly seems to be doing less work, and the application start times have reduced by 40%…. result!

iTunes startup times

I knew Photoshop was going to be an extreme case (like most Adobe apps) so to balance things out I thought I’d try iTunes too.  With a massive music library, iTunes takes several seconds to start up, so it’s a reasonable one to test (unlike, for example, Word which starts so fast that startup times are difficult to judge).

Before ReadyBoost, first start: 8.2 secs.  Second start: 6.8 secs.
After ReadyBoost, first start: 9.0 secs.  Second start: 6.3 secs.

It’s likely that iTunes startup times were influenced by external factors here, rather than ReadyBoost making things slower.  However, I don’t think ReadyBoost will improve startup times for an app like iTunes either.

Conclusion

As Tom Archer blogged previously on MSDN, ReadyBoost is all about speeding access to small, random file requests.  When you’re doing that (e.g. starting an app that requires lots of tiny files) then ReadyBoost makes a significant improvement - at least 40% faster in my tests.  For that alone, it makes it worth the investment in a fast USB Flash drive. 

The only difficulty is finding the right drive!  My Kingston DataTraveler Elite 1GB seems fine, but previous devices I’ve tried (e.g. Maxell 512MB, Kaser, etc) weren’t fast enough to register for ReadyBoost.  If you’re in any doubt, please check out my ReadyBoost compatibility chart.