Created a RAM Disk, want to hear your re-directory ideas and input

BusyBeaverHP

Limp Gawd
Joined
Jun 22, 2009
Messages
432
I created a 4GB RAM-Disk using the free version of DataRAM RamDisk.

The purpose of my RAM-Disk is to offload temporary/miscellanous writes from my SSD onto memory. I do not need to back up the image file of this RAM-Disk upon shut down.

So far: I've redirected Photoshop's scratch disks preference to the RAM-Disk, redirected the TEMP and TMP folders to my RAM-Disk as well.

Are there any other temporary folders/cache that could be put onto the RAM-Disk without affecting the program's stability?

I'm using Google Chrome, and is looking to put their temporary files onto the RAM-Disk, but want to store the stuff like auto-fills and bookmarks on the SSD... which folders should I transfer or not transfer?

I'm looking to hear some ideas from those who have been using RAM-Disks and what directories have you put on there? Thanks ahead of time!
 
Last edited:
Is this to gain speed or to save writes to your SSD? In either case, you're basically wasting your time. I cannot say how much performance you might gain in Photoshop, but I sure hope it's capable of using your RAM without the user forcing it.
 
I think this is more of an effot to save writes. Basically things they may not benefit from SSD speeds, seems like a good practice to move as much off the SSD that you comfortable with. May not do a whole lot but a bunch of little things can add up a bit. Also something to tinker around with which is why some of us are here
 
The purpose of creating the RAM-Disk is to offload temporary/miscellanous writes from my SSD onto memory, as mentioned in OP. Speed isn't an issue here.

Photoshop allows you to designate the scratch disk(s) for its temporary files. The default scratch disk is "C:\", in which case you'll actually have to create a RAM-Disk to make Photoshop fully take advantage of the extra memory.

In my Photoshop usage, I can use anywhere from 200-1000+MB of scratch disk in a session, so that does contribute significantly to write/erase cycles of SSD's, which is why it's definitely not a waste of time to create a RAM-Disk for this purpose.

Is this to gain speed or to save writes to your SSD? In either case, you're basically wasting your time. I cannot say how much performance you might gain in Photoshop, but I sure hope it's capable of using your RAM without the user forcing it.
 
I'm using Google Chrome, and is looking to put their temporary files onto the RAM-Disk, but want to store the stuff like auto-fills and bookmarks on the SSD... which folders should I transfer or not transfer?

Most people just move the cache located at C:\users\...\appdata\local\Google\Chrome\User Data\Default\Cache by default. To move the chrome cache follow this.

I personally use the lazy method though: I just copy the entire Google folder to my ramdrive, and put a junction in the original location. This only adds ~200MB or so extra to the ramdrive (not counting however big the 'cache' folder is).
 
Last edited:
The purpose of creating the RAM-Disk is to offload temporary/miscellanous writes from my SSD onto memory, as mentioned in OP. Speed isn't an issue here.

Photoshop allows you to designate the scratch disk(s) for its temporary files. The default scratch disk is "C:\", in which case you'll actually have to create a RAM-Disk to make Photoshop fully take advantage of the extra memory.

In my Photoshop usage, I can use anywhere from 200-1000+MB of scratch disk in a session, so that does contribute significantly to write/erase cycles of SSD's, which is why it's definitely not a waste of time to create a RAM-Disk for this purpose.

I do this also and agree it is not a waste of time.

My browser caches sit on the RAM disc. I had originally moved thse tmp user files but I found some programs gave me issues installing when it couldn't find those folders in the default location so I moved them back.

Because of these moves my SSD currently only has 2.5Tb's of writes and its been in service since dec 2009. While most other people that i've seen that kept the browser and everything writing to the SSD is sitting at 3.5-4TB's writes. So it does make a difference.
 
I put my temp and cache on my ssd also. I have a first gen Torqx that needs all the help it can get...

241sd8h.png
 
The question is not if it makes a measurable difference or not, the question is if it makes sense or not.

Sure, you can save a few writes moving your browser cache to RAM, but in reality, where we live after all, your drive will most likely (the only reason I don't say never, is because it's possible if you really try) not die from wearing out, and if it does, your browser cache will probably only have contributed to a very minor portion of the RE cycles. And it was probably about time getting a new SSD anyways, since we're probably well into the 2020's by that time.

Regarding the Photoshop cache, as per the official adobe documentation, scratch disks are only used when no memory is available. Hence, RAM drive = useless. There's even an option you can set that tells Photoshop how much memory you'll allow it to use.
 
The question is not if it makes a measurable difference or not, the question is if it makes sense or not.

Sure, you can save a few writes moving your browser cache to RAM, but in reality, where we live after all, your drive will most likely (the only reason I don't say never, is because it's possible if you really try) not die from wearing out, and if it does, your browser cache will probably only have contributed to a very minor portion of the RE cycles. And it was probably about time getting a new SSD anyways, since we're probably well into the 2020's by that time.

Regarding the Photoshop cache, as per the official adobe documentation, scratch disks are only used when no memory is available. Hence, RAM drive = useless. There's even an option you can set that tells Photoshop how much memory you'll allow it to use.

Actually I believe this is the question the OP asked.

"I'm looking to hear some ideas from those who have been using RAM-Disks and what directories have you put on there? Thanks ahead of time!"




This is the amount of writes to my drive after moving the browser cache off it.
 
Man, even 10TB, or 20TB of host writes is pretty much nothing. Assuming that you're not experiencing super high WA, it's all good.

Even on a 40GB or 60GB drive... it's nothing.

Here is my 64GB Samsung 830. At this point (December), it had 88,000+ GB of host writes.
ipnykyspg2p7fzbkt7q1.jpg
 
I put my minecraft server on a ram drive using that software. Supposedly it helps with all the chunk loading/unloading. I don't know if it helps or not but I thought hey why not.
 
Wow how long have you owned this drive?

I've used it for about 2 years now, I didn't discover ssdlife until about a year ago and almost had a heart attack as it showed that it would die in a few months from then. Once I moved my dev sql to my velociraptor I "gained" over a year of estimated life.
 
ssd life isn't accurate.

http://www.xtremesystems.org/forums/showthread.php?271063-SSD-Write-Endurance-25nm-Vs-34nm

basically that number is until you reach MWI. MWI does not mean the drive will die though. people go hundreds of tb past MWI and are fine.

That's where my Samsung was running. If you look at the last page, I just started testing a 256GB 830 -- and it wrote 1TB an hour for the first 24 hours.

Once a drive gets to MWI = 0 (or it's equivalent) you basically still have forever to go. SSDlife estimates the daily MWI/host write usage to tell you when the drive is going to die -- except MWI is not death. It's not even close.

From an endurance perspective, MWI is practically useless. But a drive could die from a panic lock or some other malady. The more you space out the wear, the less it "hurts" the drive.

Another thing to consider: host writes aren't a great indicator of much. If you're getting really high WA, you could have much more wear than the number of host writes would indicate. Usually, MWI is based on PE cycles consumed. In a drive with 3000PE cycle rated NAND, MWI will decrease every 30PE cycles. So if you have a drive with 90 average PE cycles used you'll get a MWI of 97.

In some drives, reallocated blocks count against MWI too -- Indilinx drives come to mind.

But the PE cycle ratings are silly conservative. Really, I'd expect most drives to get about 3 or 4 times depending on how good the NAND is.

But as wear accumulates past a certain point, what you're really doing is taking those PE cycles out of the retention time. If you get enough wear on a drive, it will not survive being powered off for more than perhaps one month or even a few days.
 
a31205d158a07927fad2.png


This is a smart shot of the Vertex Turbo I'm also running over at the XS endurance thread.

This has about 160,000GB on it, no reallocations, and is averaging 84MB/s every second of every day (in this case, just over the past 20 days).
 
a31205d158a07927fad2.png


This is a smart shot of the Vertex Turbo I'm also running over at the XS endurance thread.

This has about 160,000GB on it, no reallocations, and is averaging 84MB/s every second of every day (in this case, just over the past 20 days).

Impressive!!
 
Impressive!!

The first one I tried testing like this actually died from a bad NAND device. It only made it to ~120TB. But the one from the picture above has only used a little over 2800 PE cycles. But the Vertex Turbos use 51nm Samsung NAND rated at 10000PE cycles. But I shouldn't think it would die then either.

The idea that you need to use a RAM drive to lessen wear on a SSD is somewhat dubious. If you need it for one reason or another, then that's fine. But most SSDs are capable of handling way more wear than is suggested.
 
The idea that you need to use a RAM drive to lessen wear on a SSD is somewhat dubious. If you need it for one reason or another, then that's fine. But most SSDs are capable of handling way more wear than is suggested.

I'm sure I don't need it, but I have ram to spare. And on the upside I don't have to clean out my temp directories any more.
 
If you could link me the specific document that you read, it would help me get to the bottom of this.

From my user point of view, Photoshop CS5 does indeed use the scratch disk assigned to it, and the write is significant. Any Photoshop user can tell you this is true. I just saw my RAM-Disk take 1.2 GB of write from Photoshop scratch files. That is a 1.2 GB write/erase session that I have excluded from my physical drive.

Regarding the Photoshop cache, as per the official adobe documentation, scratch disks are only used when no memory is available. Hence, RAM drive = useless. There's even an option you can set that tells Photoshop how much memory you'll allow it to use.
 
I have done this on a few computers. For linux I have my chrome cache in tmpfs; it gets deleted at every reboot. I have small SSDs in RAID1 and 16GB of ram so it makes sense. I also use tmpfs for /tmp and compile in /tmp. On windows I also have chrome cache in a ram drive.
 
Have TMP & TEMP variables set to a Ram Drive & I just changed my Firefox cache to a Ram Drive and went to a few websites, closed browser, went back -- definitely a noticeable difference.

The biggest apparent speed gain -- Try going to a website where you are going to download something to install, like CNET. Save file to Ram Drive instead of your MyDocs-->Downloads, then run the install routine. Lightening fast.

The nice thing is you don't have to drill down through the users directory to clean this stuff up when it grows.
 
If you could link me the specific document that you read, it would help me get to the bottom of this.

From my user point of view, Photoshop CS5 does indeed use the scratch disk assigned to it, and the write is significant. Any Photoshop user can tell you this is true. I just saw my RAM-Disk take 1.2 GB of write from Photoshop scratch files. That is a 1.2 GB write/erase session that I have excluded from my physical drive.

I just googled, and it popped up as one of the first hits.
 
Back
Top