Does Windows ReadyBoost have a meaningful impact on performance?

39

11

Does Windows ReadyBoost technology, available in Windows since Vista, have a meaningful impact on performance?

Are there well-known benchmarks that compare systems with, and without?

Does the speed of the USB flash device have a measurable impact on the performance gain?

What usage scenarios may yield a gain, and what circumstances may yield no gain?

Chris W. Rea

Posted 2009-07-15T18:35:39.507

Reputation: 10 282

It always seemed to me to be sort of a kludge. If you want your applications to start up fast, put them on a SSD. – Brad Gilbert – 2009-07-16T02:57:50.350

1

www.anandtech.com did a vista performance review a couple of years ago and looked at this. Apparently readyboost is only of any use on PCs with 512MB of RAM. If you have any more than this, (and you really should), you won't notice any difference. ReadyBoost Performance

– Shane Kearney – 2009-07-15T20:59:12.250

I found it sped up some things (shaved 10 seconds off the boot time) on a 1GB machine, but slowed down others (iTunes playback etc). In the end I didn't feel it was worth having a USB key sticking out of the laptop to warrant it. My theory is that since Readyboost encrypts the data it caches, and older iTunes DRM'd songs also require decryption there was too much CPU load to play the tunes without glitches. – Matthew Lock – 2009-09-03T04:18:38.370

As far as recommendations, Microsoft recommends using 1 to 3 times the amount of RAM you have in your system. See here.

– None – 2009-07-15T18:59:28.623

Answers

26

I was very skeptical, having 6GB of ram on my 8730w laptop running 64bit Windows 7 RC. But, since SD cards is so cheap now, I went out and bought a Panasonic Class 10 (22MB/s) 8GB SD card and put it into my laptop and enabled ReadyBoost. To my surprise, it was quite a noticeable performance gain. One must understand though, that it's a cache kind of performance gain, meaning that you will notice a quite substantial improvement the 2nd, 3rd time you start an application. Nevertheless, I'm most certainly keeping the SD card in my laptop, it's "hidden" away and doesn't stick out like an USB stick.

Magnus

Posted 2009-07-15T18:35:39.507

Reputation: 978

2Actually, yes, you DO need to measure everything. Confirmation bias and placebo effect are amazingly effective. – Jamie Hanrahan – 2016-05-31T21:08:01.157

6"it was quite a noticeable performance gain" - how much gain are we talking about? Did you do measurements? How did you rule out placebo effect, and the normal effect of Windows' built-in system cache, from affecting the results you saw? – thomasrutter – 2013-05-22T05:43:21.780

@thomasrutter, it's much faster, trust me, you don't have to always measure everything to know that it is much faster. Especially when things are 10 times faster and you have been using a PC everyday for 10 years straight, it becomes obvious. – David d C e Freitas – 2014-03-29T04:55:24.000

4I've changed the accepted answer to this one. With Windows 7 I seem to benefit more noticeably than with Vista. – Chris W. Rea – 2010-01-13T21:36:13.293

10

Yes, it does. You'll see a significant increase in performance whenever the system is I/O bound. Disk-heavy applications like Visual Studio and Apache OpenOffice start up noticeably faster compared to without ReadyBoost—and this is on a system with 8 GB of physical memory!

The most obvious gain in performance I've noticed is when the system resumes from hibernation. My computer has always been unresponsive due to heavy disk activity for the first five or so minutes after resuming, and ReadyBoost produced a vast improvement in system performance and responsiveness during this time.

bwDraco

Posted 2009-07-15T18:35:39.507

Reputation: 41 701

7

We just made a detailed comparison on two computers: one with Windows 8.1 and one with Windows 7.

The improvements we measured when enabling ReadyBoost on a system with low amounts of RAM, are the following:

  • Opening media files like photos, music or video is slightly faster (approximately by 2%).
  • The loading of web pages and the use of Office applications is slightly faster (approximately by 2%).
  • Your system's boot timings are improved (up to 7%).
  • Your most used applications start faster (by 10 to 15%).

ReadyBoost had no positive impact when playing games or running applications that are CPU or GPU intensive.

You can find the detailed testing procedure plus all the results, here: Does ReadyBoost Work? Does It Improve Performance for Slower PCs?.

Corporate Geek

Posted 2009-07-15T18:35:39.507

Reputation: 1 882

1

ReadyBoost was designed to be faster access memory than most laptop hard drives.

Sort of between the system RAM and the hard drive.

Most laptop hard drives spin at 5400 rpm and have access times of about 10 to 15 ms for random reads and writes. By comparison flash memory has access times of 4 to 6 ms for equivalent random reads and writes.

What ReadyBoost was designed to do was place the most often or recently used 2 to 8 GB of data on the flash memory where accessing it would be at least twice as fast as reading it from the hard drive.

Joe

Posted 2009-07-15T18:35:39.507

Reputation: 1

1

Like so many things relating to performance, it has a lot to do with what you're doing and the other components in your system.

If you have a low-end PC, which I doubt anyone on a site called "Super User" would :-), it's probably beneficial. For those with capable systems with at least 2G memory (and 4-8G is becoming more common), ReadyBoost probably doesn't add enough performance increase to warrant dedication of an entire flash drive and more important USB port.

Memory is cheap, I just doubled my HTPC upgrading to 4G, using the same exact memory I built the system with originally (2x Corsair twinxsomethingfast) for less than half the price I paid a year ago ($35 vs ~$75). I have a 4G memory flash drive that I'd used with ReadyBoost prior, and it "feels" faster with the actual memory. Boot times are irrelevant, my system is a media/gaming PC and runs pretty much 24x7.

jtimberman

Posted 2009-07-15T18:35:39.507

Reputation: 20 109

2True, I don't have a low-end PC myself, though us "super users" often end up building lower-end PCs for others on a budget, even if our own boxes are impressive :-) – Chris W. Rea – 2009-07-16T11:27:05.097

1Good point! Though most often my 'old' system gets repurposed for someone else, and sometimes I put Linux on the old one before giving it away :). – jtimberman – 2009-07-16T14:01:13.913

1

It does improve performance by storing program launch data which RAM does not, or which is cleared when not needed, or via system powerdown.

The simplified results of those few tests that were performed have been used by every website since, and its rubbish (Google it and 99% will tell you it offers no performance increase, few have actually tried it), totally misleading.

I've tested myself using Windows 7 (x64), 9 GB DDR3 (only ever reached about 6 GB in use) and a 16 GB Sandisk micro SD card.

The speed programs launched improved drastically, from 3-5 seconds to under 1.

That's where the performance gains are, and that's why it was created. Tt will not give read or write performance increases for other files a program might use, it is dedicated to program launch.

It always seemed to me to be sort of a kludge. If you want your applications to start up fast, put them on a SSD.

I did, a small cost effective one :) we all know a 128 GB SSD is not enough for windows over a couple of years use, what with program updates and installations, anything over that has scandalous prices. Not to mention the serious bugs that have sprung up with a whole host of SSDs. 16 GB card cost around $15 at the time, much more affordable option.

OKI

Posted 2009-07-15T18:35:39.507

Reputation: 1

0

ReadyBoost provides an enormous advantage when you are writing alot to the hard disk. I have a fairly decent machine. 5GB RAM with a quad-core 3GHz processor. With ReadyBoost I often see a tenfold increase in write speed when performing I/O intensive tasks like copying large amounts of data of compressing/decompressing archives. When not writing alot of data I do not notice any performance benefit, but I understand that if your system is short on memory, ReadyBoost can speed up evn the most mundane tasks.

Guru

Posted 2009-07-15T18:35:39.507

Reputation: 1

0

I have a class 10 50mb/s 16gb SD card the card reader in my laptop is usb3 driven. Even with 4gb of ram, the performance is outstanding... That being said, a USB 2 or class 4 SD card won't do a thing. I tested using a random class 4 SD card I had lying around and the performance was negligible. I will upgrade to ssd later. Gotta remember it's cache not ram and not storage. I have windows 8.1.

Couple of things to make the best of readyboost.

Format to exFAT and not NTFS, use the whole device, the more the better. exFAT does less sequential checking therefore faster. You don't need a journaled file system for 1 big file. There's nothing to journal lmao. Oooo 1 file... Performance exFAT wins.

Make sure it's a class 10 SD if you plan on using SD. Class 4 will offer nothing.

Make sure you're using USB 3. SD or otherwise. I chose SD because it's a laptop and I don't use the SD port. USB 3 USB stick in my desktop.

The bigger the size the better. I saw the biggest performance when copying files while using applications. Basically it allows me to install and copy things while working on other stuff without hitting a wall. Windows will also page to it if need be like when using the hard drive, it will cache paging parses.

The longer you use it the better readyboost and superfetch work together to be the best it can be.

Chris

Posted 2009-07-15T18:35:39.507

Reputation: 1

You mention going for class 10. But that is just about throughput to/from the card. Did you test which performance has the most impact? E.g. IO latency, #IOPS or throughput? – Hennes – 2014-12-18T06:38:56.320