SSD hard drive question with a twist..

[nynor]

Originally Posted by BlueSkyJaunte

You guys must be looking at some seriously hi-res porn.

and you aren't?!

[GH85Carrera]

The little sideways arrow that the big arrow points to is where you can change the setting. I like to see the scratch sizes. There are several options.

[nynor]

glen, that photo rocks!

[BlueSkyJaunte]

Originally Posted by nynor

and you aren't?!

Not at "work" .

[GH85Carrera]

Originally Posted by nynor

glen, that photo rocks!

That is a little tiny piece of just one image of many hundred. It is a 1.3 gig image. I can "zoom in" a lot more.

[nynor]

Originally Posted by BlueSkyJaunte

Not at "work" .

well, that sucks.

[slodave]

Looks like we are going to get a SSD and dedicate it as a scratch disk.

[slodave]

More tests. Opening PS creates a 410 meg file on the scratch disk. Open a 500 meg image it stays at 410. Apply a filter that seemed intensive and you can seethe scratch file grow - to about 2 gigs. We closed the image, but the scratch file stayed at 2 gigs. Next, we did a photo merge with 8, 35(roughly) meg images. After the merge was complete, the scratch file ended up at 15.5 gigs and remained even after the large image was closed. Upon shutting PS down, the file was deleted.

One thing I just did, was open a 35 meg RAW image. Under image resizing, it shows the image as being 103.4 meg pixel size. In the lower left corner, the scratch window in the lower left of the image showed 619 megs. The PS temp file on the current scratch disk then showed 590 megs. That was without any manipulation, just opening and image.

We did repeated a couple of tests with the efficiency meter up. It never we t under 100%, but that scratch file sure did grow while applying certain filters or merging multiple files.

[stomachmonkey]

Originally Posted by slodave

More tests. Opening PS creates a 410 meg file on the scratch disk. Open a 500 meg image it stays at 410. Apply a filter that seemed intensive and you can seethe scratch file grow - to about 2 gigs. We closed the image, but the scratch file stayed at 2 gigs. Next, we did a photo merge with 8, 35(roughly) meg images. After the merge was complete, the scratch file ended up at 15.5 gigs and remained even after the large image was closed. Upon shutting PS down, the file was deleted.

One thing I just did, was open a 35 meg RAW image. Under image resizing, it shows the image as being 103.4 meg pixel size. In the lower left corner, the scratch window in the lower left of the image showed 619 megs. The PS temp file on the current scratch disk then showed 590 megs. That was without any manipulation, just opening and image.

We did repeated a couple of tests with the efficiency meter up. It never we t under 100%, but that scratch file sure did grow while applying certain filters or merging multiple files.

PS is set by default to use up to 70-75% of installed/available RAM. You can set it higher but that can result in not enough RAM available for system or other app resources creating a potentially unstable environment.

PS is supposed to gracefully give RAM back if the system needs it but what's supposed to happen and what does happen are not always the same thing.

What PS holds in RAM is really just cache for the scratch file. Meaning its always going to create a scratch file but there is no drive thrashing as long as the scratch fits in allocated RAM. Efficiency just means PS is able to use all the RAM it's allowed to and that nothing else is dipping into its bucket.

You can clear the scratch manually via Edit->Purge. Useful when you are not paying attention and run out of space and can't even save the file. You purge scratch with the doc open.

The reason a dedicated scratch drive makes a big difference is the read/write heads are not hunting/thrashing around the disc dealing with PS + OS + whatever else you have running read/writes. Even a lot of idle apps will do periodic auto saves.

[slodave]

I'm done with all the research. That is exhausting. Tom's Hardware has some info on the Caviar Black drive that is currently the scratch drive. His tests show an average read/write at around 85 for average speeds. The SSD drive we are looking at shows average read/write times at around 250 mb's a second. That's a huge difference and should really speed PS up.

When I build his next PC, it'll have a SSD C: drive, a SSD scratch drive, plus two larger normal drives for storage.

[RWebb]

Apple has this interesting hybrid mini out now - at a guess it is continually writing in and out between the SSD and the platter based HDD...

[Joeaksa]

Originally Posted by slodave

When I build his next PC, it'll have a SSD C: drive, a SSD scratch drive, plus two larger normal drives for storage.

So when do you think that this will be the standard for all higher end computers?

The data transfer sounds great...

[jyl]

They now make hybrid drives, a small amount of NAND in a rotating disk HDD. Tests show a decent, not world-beating, speed up for far less cost than a large SSD or small SSD + HDD. This would not work for the OP's Photoshop purpose though.

[azasadny]

My desktop is not a "powerhouse" by any means, but it's fast enough, reliable and does everything I need it to do. You're correct to build the system based upon your specific requirements... I can't affordably add more RAM (4GB is installed, would cost $300 to upgrade to 16GB), but this PC will do for a few more years...

[VFR750]

PC Development in Windsor steered me towards the ssd. They now recommend it as the base.

I added a 1TB SATA drive for my data (J-drive) to the 256GB SSD wich is my C-drive. Any videos I edit are copied to C, and converted and saved to C. Much faster.

I think this the beginning, soon it will be the standard. It is just too fast for high end users to ignore.

[slodave]

Originally Posted by Joeaksa

So when do you think that this will be the standard for all higher end computers?

The data transfer sounds great...

I think the cost per gig needs to go down a bit more. For one thing, the prices really need to be comparable to one another. You've got one manuaturer offering a 120 gig drive for $100, another for $150, and another over $200. When you look at the specs for each SSD drive, they are almost the same.

I don't spec high-end machines from Dell or HP, so I'm not sure if they even offer this type of scenario using normal drives.

Realistically, to have the best machine using my PS example, would be to have 3 SSD drives and then one or two normal 1,2,3 TB drives for storage. One SSD would contain the OS, one would be used for the OS cache, one for PS, and one for all programs, PS, Word, Excel....

[slodave]

Little bit of disappointment at the moment. After a couple hours researching SSD drives, we settled on an Intel 120 Gig SSD. The max read value under SATA III is supposed to be 550 MB/s, the max write, 500 MB/s. The price was also right. We went to Fry's, picked up the drive and a PCIe SATA III controller - which claims 6 Gb/s. Got it all installed and redid some of the PS tests. Virtually no difference in speed when applying a certain filter. But, when I redid the 8 image panorama, it did speed up, though not Earth shattering. Since the pano test requires PS to open 8 images that are on a platter drive, I redid that test again, copying the 8 images to the SSD. I realize that the scratch file is on the SSD, but I tried anyway. Only a second faster.

I finally broke down and grabbed a free HD speed test utility and ran it on the SSD. It recorded a max read value of around 175 MB/s. Write was about 117MB/s. I decided to run the test against one of the TB platter drives. Read was 100 MB/s, write was about 90 MB/s.

Tomorrow, I want to plug the SSD into a SATA II controller that's builtin on the MB and see if that makes any difference. So far, I am at a loss to explain why the SSD is performing slowly.

[svandamme]

an SSD should never be used as a scratch disk, it's way to slow for writes , and a scratch disk is all about read writes..

SSD should be your OS disk, your program files disk.. your write once read often disk.


You need more ram, not rom.


The Sata controller on board, most MObo's only have 1 or 2 fast ones... yes it makes a difference in speed, but it won't make a scratch disk on SSD work well.
You will constantly be hitting the limits of the onboard cache buffers.

[slodave]

That may be, but the benchmark test was slow too. That should not be.

[svandamme]

primary controller will be faster but not fast enough to scratch