[PCWorld]

Post your comments for ReadyBoost Flash Drives Lack Significant Boost here

[Liquidwave22]

Well what do you expect this is just another way that Micsoft can market its OS but I was wondering about one thing what if you created a RAID 0 Array made of Fast USB thumbdrives there is no risk of losing Data and wouldnt that speed up the drives performance and help windows??? I have been searching for different projects concerning this but cant find more than one

[BryZad]

"We learned that ReadyBoost does shorten the time it takes to load frequently used programs--but not by much."Inaccurate as hell. It works in conjunction with SuperFetch, which loads frequently used applications into memory and thus allows for those applications to load faster especially if ReadyBoost is enabled. Both work together, yet this article neglected that fact.I buried the digg thread as inaccurate as well.

[sirfergy]

How much RAM is on these computers? The point of ReadyBoost is to aid in low RAM situations. I think it's a glaring omission that was not mentioned.

[davidrools]

Did you try testing with a true High speed flash drive? I imagine that a flash drive which reads/writes in the 30 MBps / 25 MBps, respectively range would increase performance significantly over the drives tested here.

[Scortch]

You need to redo the test, this time try using better flash drives. Try ones that are 19-30MB/sec. Drives such as Kingston Datatraveler II Plus Migo Edition (19MB) or Patriot Extreme performance (32MB) 1GB drives. They make 2G and 4G drives just as fast. You can get these online for really cheap.If you are going to do articles like this, you should really test the whole range of speeds. Not just the slow junk and then say readyboost sucks. Are you one of those Mac freaks?

[AlanStafford]

[quote:c6d744572f]Inaccurate as hell. It works in conjunction with SuperFetch, which loads frequently used applications into memory and thus allows for those applications to load faster especially if ReadyBoost is enabled. Both work together, yet this article neglected that fact.[/quote:c6d744572f]This was a lab-tested project (and it took a heck of a long time to test, because we conducted many, many passes). We certainly understand the relationship with SuperFetch; we read Microsoft's marketing materials too.Based on those claims, we tested specifically for application load times. Our tests showed that it has no perceptible effect.How much RAM: We tested with a PC and a notebook, both of which had 1GB of RAM. That's a reasonable, common amount of RAM, especially with Vista. We also tested the same systems after adding an additional 1GB of RAM (for a total of 2GB each); that showed substantially more benefit than using the flash drives.Speed of flash drives: We tested drives that claimed ReadyBoost compatibility, so yes, they were fast enough to evaluate the technology.I am an ex-Mac freak.

[Junker]

ReadyBoost is designed for PC's with 512MB of ram or less. Your test PC has 1GB that's why it didn't make any difference, you'd think that if you were benchmarking a product you'd actually take the time to read the manual and find out how the product is supposed to work.Edited by Cosmo, no trolling

[Scortch]

You should test readyboost with high speed USB drives as mentioned, so that you can give a FAIR review for Readyboost, before claiming it's junk.Only testing the slow junk, is doing your readers a disservice. You should be informing them of what it takes to be better than using a swapfile from the HD.You should be informing them that in order to really see a benefit, they should buy at least xxx speed flash drives.

[AlanStafford]

The JumpDrive Lightning is Lexar's fastest, at 18MBps, which is pretty close to the spec of one of the drives you cited, so it was a fair review.The Kingston DataTraveler is rated at 5MBps, and although Ridata doesn't say what its transfer rate is, its drive carries the ReadyBoost logo, so it must meet the minimum specs for ReadyBoost (3.5-megabytes-per-second throughput for 4KB random reads, and 2.5-MBps speeds for 512KB random writes). OK, so they're slower than the JumpDrive lightning, but as I said earlier, both of the other drives we tested carried ReadyBoost logos, so they should do the job. Besides, if we're going to test a technology, we're going to test a range of products. Otherwise, if we tested only the fastest drives and as a result declared ReadyBoost a winner, you could walk into a store, you see a ReadyBoost logo, and think, great, this will work, whether it would or not. As for testing with 512MB of RAM: Microsoft says nothing about a minimum amount of RAM. Rather, it recommends that the size of the flash drive be one to three times the amount of RAM you have. ([url]http://windowshelp.microsoft.com/Windows/en-US/Help/12428141-2e4c-43ab-8dd2-a6ed6e3b87761033.mspx_)So, with our 1GB systems, that'd be 1GB to 3GB of flash--and we tested a two 1GB drives and one 4GB drive. The JumpDrive Lightning was both the fastest drive and the largest--so if that drive didn't show any advantage, then I don't know what would.Another thing: Why the heck would you buy a $89 flash drive (the JumpDrive) for a 512MB PC? 512MB is the absolute minimum for running Vista. If you want to run any of the applications that ReadyBoost would launch more quickly, you're going to need more than that, so your money would best be spent on more RAM--in fact, you can get a 512MB module for less than the cost of the JumpDrive. Bumping from 512MB to 1GB of RAM will make much more of a difference than any ReadyBoost drive.I think some of you were really hoping for this technology to work--I certainly was. That's why we devoted quite a bit of space to it in the upcoming issue, and we conducted so many tests.

[AlanStafford]

Correction: The 4GB version of the JumpDrive (the size we tested) has a rated write speed of 21 MBps. The 2GB version has a rated write speed of 18 MBps.

[Scortch]

Why would you pay $89?You can get 4GB flash with 34mB/s transfer for like $43. You can use it for other things too while not at the PC using readyboost.Check out NewEgg. You are paying way too much for things.

[AlanStafford]

We tested flash drives that specifically claimed ReadyBoost compatibility (not just the fastest ones out there), and the JumpDrive was one of them. Granted, the prices we quote are conservative; when we research street prices, we restrict our searches to reputable vendors (not to say NewEgg isn't reputable--it just didn't have this particular drive when we were getting prices).We tested the drives' performance as regular flash drives; those performance tests are at the far right of the test results table (second page of the article, click the table icon at the bottom). Sorry--I know we don't present tables like this very well, so it's easy to skip over them).Also, for those of you who requested specs on the systems we used, how we tested, etc.: We used an HP Compaq dc5750 desktop and an HP Pavilion tx1000 notebook. Both systems had 1GB of RAM for all the ReadyBoost tests, but we also ran tests with 2GB of RAM (just to see how the additional RAM would impact performance). The additional RAM was much more effective than the ReadyBoost drives--at least, with our homegrown benchmark utility, which runs many scripted application processes. (Search for WorldBench for details). We also ran an incredible number of tests to time application launch times.If you want more details, say so and I'll get them from the lab analysts.

[Scortch]

I just think you should have included the 34mB/s results in with the 5mB/s. That's all I am saying.I do totally agree the best thing someone could do, is to make sure they have a minimum of 1GB of memory to start with. You may pay more but, it's worth it.

[dev0lution]

This is why PC World and other pubs are absolutely worthless to a consumer who's looking for answers that will make them an informed purchasing decision. Nowhere in the testing was your methodology described - the specs of the machines tested weren't even listed in the article nor did you manage in your exhaustive testing bother to test the addition of these flash drives to systems with varying memory capacities - such as 512MB, 1GB and 2GB systems. And yes, Microsoft does not mention the installed RAM capacity in regards to Windows ReadyBoost, but they do make recommendations on what the minimum requirement to run Vista is, which is probably what your readers would be looking at a great deal more.The intro mentions that ReadyBoost only works with "the fastest flash drives", when in fact two of the drives tested are nowhere near the fastest drives offered by those manufacturers. Was this mentioned when you threw up a chart showing the percentage increases? No... Plus, you didn't even bother to keep the capacities the same across the board either.What you likely would have found, is that ReadyBoost doesn't offer a noticeable increase in performance as you move into higher memory capacities. Testing at 1GB isn't doing a whole lot to inform the masses who're stuck with 512MB of RAM in their laptops that are either already running Vista or contemplating an upgrade. If the footprint of the apps that were being stored by Superfetch was at or slightly above the 1GB mark, of course you aren't going to be seeing a huge increase in performance - perhaps something that the lack of performance advantage on the fastest and largest capacity drive might have clued you in to. What applications were being used and how many times were they launched while ReadyBoost was enabled?Second, you'd actually publish test results from a benchmark utility that has BETA in the title and one you admit had many tests that interfered or weren't compatible with ReadyBoost? Sounds really trustworthy.. How were application load times tested? Again, with what real-world applications? Details please...Your price check links are inaccurate on the Kingston drive. It brings up a link to a pricing search on a totally different model and capacity. Searching pricegrabber, the capacity you list is more like $15-$20, with the 2GB capacity hovering closer to your $24 list price. The other two are listed with "street price" instead of list. Can't you even standardize what kind of prices you're listing? While you mention in the comments that the Lexar costs nearly 4 times as much, nowhere is it mentioned which drive would be the better value for the performance. But wait, that might actually help a reader make an informed decision....Ironic that you arrived at the same conclusion most of the drive manufacturers and even Microsoft would recommend - ReadyBoost does not boost performance more than adding system memory would. In fact, from Microsoft's own site describing ReadyBoost:"Adding system memory (typically referred to as RAM) is often the best way to improve a PC's performance, since more memory means more applications are ready to run without accessing the hard drive. However, upgrading memory can be difficult and costly, and some machines have limited memory expansion capabilities, making it impossible to add RAM."PC World and at least a couple other "technologyr" publications seek to dumb down complicated technology with the premise of helping their reader make sense of the latest technology. If one was to take this article or even it's headline at face value then you'd conclude that ReadyBoost was a useless smoke and mirrors marketing tool from Microsoft and flash drive manufacturers. Is that necessarily true? Not so much.Will ReadyBoost capable memory drives offer the same performance increase as adding more system memory? No, but even MS acknowledges that. Does ReadyBoost drives work better in systems running the 512MB minimum for Vista as recommended by MS? Who knows since you didn't bother to test that. Do certain applications benefit more from ReadyBoost? Again, didn't test it or if you did, didn't publish the basis for that conclusion. Does ReadyBoost offer some sort of advantage to those who are limited in their memory upgrade options, or offer a free boost in performance to someone who bought one of these drives for portable storage? Yes. Will there possibly be future applications or advantages offered by ReadyBoost down the road? Maybe. Is $15-20 for a 5-6% increase in performance cheaper than adding a $80+ memory module? Yep.You get the point.. article and the test blows. You might want to follow the lead from some of the semi-pro review sites and actually try and follow some sort of scientific testing protocols with full disclosure on the methodology. :rolleyes:

[lilbyrdie]

Alan,I'm glad to see that you're responding to folks.I do find it interesting that your tests only showed a 6 percent increase in launch speed. A while, I had done some wall load time tests myself and I usually saw app load times drop from maybe 9 seconds to 5 seconds with ReadyBoost and from 9 seconds to 7 seconds without ReadyBoost (e.g. just continued relaunching and exiting). It's not a very realistic account because one doesn't often launch applications repeatedly.However, ReadyBoost and SuperFetch are both supposed to be predictive and learn what you're doing. I have noticed that ReadyBoost can take quite a while to prepare even a small drive and during that time, system performance may actually drop some. A 4GB card will take much longer to seed than a smaller card, too. There really isn't any way to know, though, what RB and SF have learned and what they're caching the best. If you pick the wrong app, their caches will likely adjust to some other mix of files and the test then can't be duplicated.I should also note that my tests used a 512MB memory stick in a built-in card reader on a system with 1GB of RAM. The card itself is just barely fast enough to pass the ReadyBoost test and sometimes needs to be retested a few times before Vista will allow it to be used if I remove it.I also wouldn't be surprised to find the Microsoft apps benefit more than other apps and that is mostly what I was testing. Firefox launching was too dependent on loading sessions over the web to be readily testable.In the end, I've felt that ReadyBoost and SuperFetch do work, but testing them is pretty much near impossible so proving it also is. Shaving a few seconds off launch times a day, though, isn't something that most people would notice, anyway. You're better off just leaving your apps running if you have the RAM.

[AlanStafford]

lilbyrdie, I appreciate your considerate response. I understand that many folks' gripes with the article have to do with our not reporting enough information on how we tested the drives and why, so we're working on a how-we-test document. We'll append it to the online article as soon as we're finished (it's lengthy).

[AlanStafford]

We've appended the How We Test information to the story; it's the third page:[url]http://www.pcworld.com/article/id,131742-page,3-c,flashmedia/article.html_

[lilbyrdie]

[quote name='AlanStafford']We've appended the How We Test information to the story; it's the third page:> > It would seem to me that the testing is certainly reasonable enough. It's not a lot of variety in software, but much more and the test wouldn't be all that reasonable, anyway.I did see one article long ago during beta about ReadyBoost that showed almost no improvement in the launch times of CS (or maybe CS2) until the loading of images was involved. When something like 40 images were loaded at launch time, RB helped out quite a bit more. Is it normal to launch Photoshop with the same 40 images over and over? And not editing them so the cache data would be cleared? Unlikely.Clearly, it's best to always just have as much RAM as you can afford. However, on the flip side, if you've got a spare USB key or a memory card for an empty slot that's just laying around, any performance boost RB does give would be free and there would be no point in not using it. Right? That is, are the actually common usage scenarios where you would never want RB to be used and enough of them? I'd also guess it would help more when the difference in performance between the flash memory and the system's hard drive is larger. You might get more out of it on a lightweight laptop with a 4200rpm drive than a desktop with a high speed RAID of multiple 10k RPM disks. Or is reading off of two devices always going to have the potential to improve things, however small?So many questions...Perhaps the conclusion is simply that it's not worth buying a USB key to use ReadyBoost, but if you already have one that will work, and you're not using it for anything else, why not give it a go?Thanks for the additional information.

[AlanStafford]

[quote:d1f379c51b]Perhaps the conclusion is simply that it's not worth buying a USB key to use ReadyBoost, but if you already have one that will work, and you're not using it for anything else, why not give it a go? [/quote:d1f379c51b]Excellent, excellent questions and analysis. Your last post prompts two things: On the Photoshop launching: You've hit on one of our big testing rationales, which says that we don't test things that normal people wouldn't do--so no, we're not going to test launching 40 images. That's also why we didn't test with 256MB of RAM--it's below Vista's system requirements, and few people are going to run Vista that lean unless they're experimenting.Second, the why not give it a go thing: Although we can't test every single configuration and scenario, you should be able to reach conclusions based on what we test. That is, even though we didn't test a laptop with a 4200-rpm drive, you should be able to conclude that, even if you were to use ReadyBoost with that particular configuration, it probably wouldn't have a huge benefit. There's no tipping point at which you go from having little or no benefit to having a huge benefit. Thanks for your post.