Using Readyboost on a RAMDISK to improve disk access times..even with an SSD!

[LaptopTweaker]

I have enabled readyboost on a RAMDISK for two Lenovo thinkpads using Samsung SSD's that were bound by SATA 2 speeds: T500 and T61.

This is what I did:

1. Install Dataram Ramdisk ( I was using V4_4_0_RC34). I also use Softperfects Ramdisk but it does not work with the Readyboost feature.

2. Set up a ramdisk size based on your installed ram. I only used a small 260MB ramdisk size and stayed with the default FAT format. Even a small readyboost cache of 260MB gives noticeably snappier performance and does not take up much system RAM. I also recommended ticking the option the label the device as it will be easier to find in the next step.

3. Open Regedit and go to: HKLM (Local Machine)->SOFTWARE->Microsoft->Windows NT->CurrentVersion->EMDgmt. Find your device and left click on it( it should be labelled RAMDISK). In the right hand pane of regedit, double click "device status" and set it to 2.

4. Also in Regedit, go to: HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\services\rdyboost\AttachState. In the right hand panel, I set all the variable values to "0". I got this trick from http://hatsoffsecurity.com/2015/05/31/force-enabling-readyboost-windows-78/

5. Open services.msc and set Superfetch to automatic and start it.

6. Open "my Computer" and right click the Ramdisk drive you created and select Properties.

7. Select the Readyboost drive tab and set it to "Use this disk" or "dedicate this device". I dedicated the entire 260MB to Readyboost with good results.

I also enabled readyboost on a T61 with 6GB of ram with windows 7 and got the exact same performance boost as the T500 machine. The t61 is running an 840 EVO SSD as the system disk and it has been modified with the Middleton BIOS so it is also running at SATA II speed. However, rapidmode chews up too many cpu cycles and RAM so I keep it off for this laptop.

I may even turn off rapidmode on my T500 thinkpad as the T61 with readyboost and the 500GB 840EVO is every bit as fast as the T500 with a 256GB 850 EVO with rapidmode and readyboost enabled. It would appear that any SSD throttled by SATA 2 could benefit from this readyboost trick.

I am curious if anyone else wants to try this out and see how well it improves system responsiveness for only a small RAM usage.

Mouse clicks on menus and general OS responsiveness is improved over the normally speedy SSD performance by having readyboost enabled on a RAMDISK.

 

[Solandri]

Pretty much all modern OSes use unallocated RAM as disk cache. So all you've done is create a second disk cache which reduces the amount of RAM available to the OS disk cache by the exact same amount. Benchmarks are programmed to bypass this OS disk cache because they're meant to measure drive performance, not cache performance.

The only speedup that's possible by doing what you're doing is:

- Disk read cache that can't be unallocated. e.g. If you've got 8 GB of RAM and your programs need 8 GB, now they only have access to 7.74 GB. The remaining 260 MB is in your RAM cache. Meaning the disk reads are faster, but now your programs have to swap to the pagefile where they wouldn't have had to if your ramdisk weren't there. However, reads that occur after your programs use 8GB of RAM will be faster since they'll still be cached (whereas the OS disk cache would've given that RAM to the programs).

- Enabling write-back caching. Yes that'll speed up writes to the drive. But if there's a power failure, any data that's in the cache but not yet written to the SSD/HDD will be lost. Write-back caches are normally turned off by default for this reason. You're only supposed to turn it on if you absolutely know what you're doing and aren't concerned about data loss/disk corruption in the event of a power failure.

Samsung RAPID is basically a disk cache. So now you've got three disk caches - Readyboost, RAPID, and the OS disk cache - all doing the same thing using the same RAM. It'll actually be worse than just the OS disk cache because now you might have the same disk file cached 3x in RAM, wasting space.

 

[LaptopTweaker]

Thanks for the tech talk Solandri. I agree that I am creating additional caches in RAM. However, I am also experiencing improved system response on both a T500 and a T61 laptop despite the small amount of "wasted" ram. Both are limited by SATA 2 speed and one laptop has rapidmode enabled and one does not. I wanted to ensure there were no negative interactions with the rapidmode caching.

I am am not very interested in the reasons why this should not work as well as it does.
However, you have a point about the potential risk for losing cached data. However, both of these old laptops have decent batteries but I would be cautious about a desktop without a UPS.

I am more interested in talking with people who actually try adding a RAMdrive readyboost feature to their system and report the results. I propose that a small 260GB RAMdrive should work on a system with only 4GB of RAM as well but I will wait and see if anyone responds with actual usage experience.

As you are likely aware, SATA 2 speeds are old tech so the modern SSD's installed in these older laptops cannot hit the advertised performance levels. This method gets you a bit closer on older hardware with sufficient ram to spare. New life for older hardware?

Let me know if you actually try this caching method. I am sure you can "risk" 260MB of precious RAM to find out.

 

[k1114]

Can you record and time it? That will get a more accurate reading of the improvement. Sata 2 only limits sequential which has nothing to do with system responsiveness. The benefit of ssds themselves is that access times above hdd. People often confuse bandwidth with speed but access times are unaffected by sata version. Ram is faster access but normal activity should not be affected.

 

[LaptopTweaker]

I am not an expert in benchmarking systems so I was hoping that another expert would chime in.

I have been using the Dataram Ramdisk as a readyboost cache for several years as even a small cache really speeds up an old HDD system. Not all ramdisks can be used with readyboost. Softperfect Ramdisk cannot recognize the readyboost feature. However, I also use the Softperfect ramdisk as a temp file cache because the free Dataram product does not allow more than one disk to be created at once.

My curiousity of using readyboost with an SSD as the system disk was intrigued because an HDD system with readyboost on a ramdrive could appear almost as responsive as an SSD system if you exclude boot time. Only recently have I figured out how to enable readyboost on a system with the OS on an SSD and I am pleased with the results. I am currently typing this on the T500.

To be perfectly clear, the CPU and graphics card performance will not be improved. I have noticed faster application loads (after caching), better tab responsiveness in Firefox, faster mouse click responsiveness and a smoother overall experience.

Is there a benchmark method that measures hard drive caching improvements? How about faster mouse clicks? I do not know how to time these events.

I would like to see the data from someone who can to validate the real benefits. In the meantime, I am enjoying using my old laptops just a little more.

I also have a more modern W530 laptop (from work) that has an HDD, 12GB RAM and a 24GB mSata. I have created two different partitions on the mSata: one for pagefile and one for readyboost. I also added readyboost on a ramdisk that was a noticeable improvement over when I only had the Msata configured for readyboost.

The key is that ready boost needs a fast device to provide benefit. You do not get any faster than a ramdisk (on old hardware anyway).

The W530 works awesome (2.5 years) except for the boot time could be better with an SSD.

 

[k1114]

There is no benchmark because everything could be cached since windows already caches any extra ram not being used. So there is no clear and easy way to know what is cached to benchmark it. Hdds also have their own cache which can be seen when benchmarking hdds when you get burst speeds much higher than sustained. You claim it's helping all the time so record the screen when clicking something. Count the frames on how long it takes the screen to change. Compare with what you have without readyboost. Ram caching improves all systems regardless of sata version since that's not what's limiting it.

 

[sccman]

I can definitely say it works great. But I don't know if I agree that the hard drive caches some of its memory in RAM. Every time I run Task Manager, I see that my computer only uses a measly 2.1 GB of RAM. My computer currently has 16 GB of RAM and I've used 6.85 GB of that as Readyboost. I've noticed on the Performance Monitor on Windows that it even utilizes about 80% of Readyboost's cache when it runs.

Overall I think a Readyboost RAMDisk works great if your computer has excess RAM that you aren't even using. You utilize the RAM's fast speed to take the work off the slow Hard Drive.

 

[Laptop_Tweaker]

Thanks for the confirmation of performance improvement with your system sccman. It is nice to see that you see improvements even with 16GB of RAM. It just shows that by focusing on system bottle necks, small improvements can be made that positively impact user experience on newer gear.

You never mentioned if you enabled readyboost on your ramdisk with an ssd or with an hdd? I have done both methods and got better system responsiveness as a result. Despite my screen name I have also enabled readyboost on a ramdisk with older desktops too.

My laptops with older motherboards can not make a 6GB ramdisk as they are limited to 8GB of system RAM so you are fortunate to have that much excess RAM. However, I am only using a fraction of the size of ram disk to get results. You could use a smaller ramdisk if your software usage starts to demand more system ram. It is a balance between setting aside system ram to a ramdisk or letting the system use that RAM.

 

[sccman]

I currently use a HDD. I use an 8GB RAMDisk on my RAM for my Temp and Temporary Internet Files and the rest for Readyboost. The other 8 GB is used for games like Fallout 4 which use 8 GB RAM minimum.

I don't exactly have a numeric data from a benchmark, but my game loads just seem intuitively faster than with the Hard Drive alone. What could I use to test the actual speed of the HDD and RAMDisk Readyboost?