Solid State Hard Drive-v-Sata Drive-test & thoughts

Recently purchased a Solid State Drive (Intel 160GB X25-M) to function as a primary (boot) drive in my main desktop.

I'd read a number of reports highlighting the gains to be had against a more conventional sata hard drive, but none particularly related to the real world of processing that togs employ regularly.

Therefore, I thought I would run a number of tests, comparing like for like, SSD against a conventional sata drive, looking at a number of applications that are generally important to post processing images.

For the test set-up, I installed windows 7 (64bit) on two hard drives within the same desktop, first drive a conventional 1TB sata drive (no RAID configuration), second hard drive, 160 GB SSD, the drives in a dual boot setup. With virtually bare bone drives with just the operating system , I loaded CS4 on each, configured similarly, with memory usage at 80 % on each (5737mb), with a separate partition on each individual drive as a scratch disk.

I removed all peripherals except for monitor, keyboard and mouse.

The base desktop configuration- Quad processor (i7 2.8 GHz), 8GB Memory, twin Nvidia 9500GT Graphics cards.

I ran each test 3 times (except for 1-5 and test 10) and took the average of the 3 tests for each drive.

Test breakdown:

Tests 1 to 5 relate to the Windows Experience score:
1-Processor: calculations per second
2-Memory (RAM): memory operations per second
3-Graphics: Desktop performance for Windows Aero
4-3D Business and gaming graphics
5-Primary hard disk-Disk data transfer rate
6-Windows start up, from pressing 'enter' at the dual boot screen to actual desktop appearing on screen.
7-CS4 start-up, from clicking the CS4 icon to CS4 opening-no plug-ins installed.
8-Dragging 7 tiff (125mb total) files into CS4 to open simultaneously, identical tiff files each on respective source drive (sata or SSD)
9-Copy 30 raw files (214mb) from SD card (via card reader) to source drive (either sata or SSD)
10-Loading 1 tiff file of 105mb( 4992 x 3328 pixels) and applying surface blur filter-radius 100, levels 255
11-Same file-box blur filter applied- 999 pixels
12-Copy 4 video files (6.59 GB) from source drive to source drive i.e. from 1 folder in sata drive to another folder on same drive and partition, similarly with SSD.
13-Copy same video files from a separate internal sata drive to source drive (sata or SSD)
14-Copy same files from source drive to separate sata drive.

Test Results (tests 1-5 higher is better, 6-14 lower is better)

Am I pleased I took the plunge and splashed out -you bet

If you really want to take advantage of SSD, the OCZ vertex is the way to go. Way faster and consistent.

jeppoy wrote in post #8889851
If you really want to take advantage of SSD, the OCZ vertex is the way to go. Way faster and consistent.

Depends upon which reviews you read, and this post isn't about the merits of particular SSD , it's about comparing SSD against a conventional sata drive.

Well then the traditional HDD is more reliable than your SSD except the OCZ vertex...and hopefully you get a good one and if it's not then i hope you don't lose your data....good for you if it works fine for a long period of time.

Where I really notice the difference is in opening Lightroom when it wanted to do a backup of the catalog. On the old HDD machine this would take ages (didn't time it, but around a minute) so I had it setup to do it once a week. With the SSD an 18,000 image catalog does the full test, optimise and backup thing, then opens LR in around 6 seconds! It's now setup to do the backup once a day.

A lot of the tests you've done (1, 2, 3, 4, 10 and 11) are hardly testing the SSD at all - but the other hardware. Still, the other tests do show how much difference an SSD makes when it comes to opening programs and data.

Thanks! These sort of well-done tests are helpful; much more so than anecdotal reports.

>>with a separate partition on each individual drive as a scratch disk

If it is on the same physical drive, what difference does putting the scratch on a separate partition make?

Do you have Lightroom? Can you run some tests importing RAWs to DNG and rendering 1:1 previews? If not, can you run import and render/export tests with whatever your normal workflow is? I'm curious if an SSD would speed up the day to day operations with RAWs like zooming in 1:1 and then going to the next frame at 1:1 to compare, etc.

hollis_f wrote in post #8890069
A lot of the tests you've done (1, 2, 3, 4, 10 and 11) are hardly testing the SSD at all - but the other hardware. Still, the other tests do show how much difference an SSD makes when it comes to opening programs and data.

I agreethat particularly 1,2, 3 and 4 do not test the SSD, as they were part of the WI set, I included them, more to highlight the SSD has no impact on the result.

Tests 10 and 11 are elevant (IMO) if the process brings the scratch disk into play.

TheHoff wrote in post #8890083
Do you have Lightroom? Can you run some tests importing RAWs to DNG and rendering 1:1 previews?

The DNG conversion and preview creation are both, I would imagine pretty heavily dependant on processor and memory speed. However, it sounded interesting, so I thought I'd give it a go.

I took 100 50D raw files (about 2GB) and copied them onto my HDD (WD 500Gb Scorpio Blue - used as my data drive). I also created a new LR (v2.5) catalog on that drive and imported all 100 images using the following settings:

File Handling: Copy as DNG and add to catalog
Copy to: Original folder
Organise: By original folder
No backup
Original filename
No develop settings, metadata or keywords.
Initial Previews: 1:1
Show preview unchecked.

I timed how long it took to convert and import the images and how long to generate the previews.

I then copied the raw files to the SSD (after deleting the DNGs) and created another new catalog on the SSD and imported the images from the SSD using the same settings.

Here's the results:

• For the HDD it took 3' 06" for the convert/import step and 7' 54" to generate previews.
  
• For the SSD it took 2' 52" for the convert/import step and 4' 33" to generate previews.

Thanks! I thought the preview creation might be diskbound but I wasn't sure. After it imports and demosaics the RAW file it then has to write out a much larger TIFF-formatted preview. I assume for files not in the preview cache, the same sort of speed benefit would be obtained when you're clicking through images and then zoom in for a 1:1 look (nearly a 45% speed increase -- nice!)

One disadvantage of the MBPs... on my old Windows laptops I could see when something was accessing the HDD and knew when it was slowing things down. No lights and no vibration on the Mac so I can't tell unless I open up an iStat monitor (which doesn't work well on SL).

My question: what sata drive? pretty sure raid o of two velociraptors would give you more interesting results than an old 1tb WD green 5400rpm drive.

basroil wrote in post #8892571
My question: what sata drive? pretty sure raid o of two velociraptors would give you more interesting results than an old 1tb WD green 5400rpm drive.

1TB SERIAL ATA 3-Gb/s HARD DRIVE WITH 16MB CACHE (7,200rpm), which is almost new.

When I was researching which SSD to purchase, most of the SDD's in the tests seem to be a lot faster than the velociraptors they were pitched against, although as mentioned in a previous post, it depends which test results you read , and were generally not pitched at image processing tests.

Here's a typical benchmark test , one of a number I looked at before chosing the Intel SSD

http://www.behardware.​com …l-x25-m-v2-postville.html

Lesmac, brand of HD would be more helpful. It's almost a crime to put something like a seagate 7200.10 against a velociraptor, even more so with an ssd.
http://hothardware.com …formance-Analyzed/?page=7
from there (though i wish they had better tests), you can see that a single velociraptor will be pretty close to a x25-m, but much more space and much cheaper, and with raid 0, it can probably be pretty good against a x25-m (since I/O numbers aren't that high for photo processing, for certain server apps, then an ssd will just destroy even 15k SAS drives)

My PS 64 bit opens in under 2 secs from a Sata drive. I do run the Scratch drive on a ram disk. I wonder if that helps it with load?

I know that until I fill that scratch file up all the way and then photoshop starts using my secondary disk as additional scratch.... Photoshop operations seem almost instant. Filters and edits happen so fast you can really notice the lag.

lens pirate wrote in post #8915030
My PS 64 bit opens in under 2 secs from a Sata drive. I do run the Scratch drive on a ram disk. I wonder if that helps it with load?

I know that until I fill that scratch file up all the way and then photoshop starts using my secondary disk as additional scratch.... Photoshop operations seem almost instant. Filters and edits happen so fast you can really notice the lag.

Probably using vista 64 bit or windows 7 64bit then, have to love superfetch. Slightly different than the numbers posted above though, since that's actually already in memory rather than reading off a disk.

Funny: I was going to post that Intel released a new firmware for the SSD and started looking for a link. Found this instead http://blogs.zdnet.com​/hardware/?p=6020

Looks like it is best in general to wait a few weeks before applying a new firmware to these SSD. Plagued by bad luck, programming and PR these are.

Having said that, I am waiting for my X25M 80GB SSD (arriving tomorrow) for my laptop and hope to watch it fly. I am also upgrading my desktop to a velociraptor (arriving in two days). I am used to two raptors in Raid0 at work and the setup is really snappy. I keep thinking that an SSD is a guilt inducing luxury on the desktop. So far Win 7 64bit has done a great job in speeding up my desktop, hopefully the raptor will do the trick without me wanting an SSD for a while.