Ad
Ad
Ad
Pages: « 1 2 [3] 4 »   Bottom of Page
Print
Author Topic: PC specs for processing large photos  (Read 16784 times)
kaelaria
Sr. Member
****
Offline Offline

Posts: 2227



WWW
« Reply #40 on: December 14, 2005, 08:04:51 PM »
ReplyReply

The Raid drives are regular western digital caviars, 120GB ea.  

Like I said I don't remember the specifics of the script, but the scratch was hundreds of MB at peak, and constantly churning for several minutes.

I do want to do it again, as I have now gone from 1GB to 2GB ram and would like to gauge the difference as well.  I'll have to search for it soon.
Logged

BernardLanguillier
Sr. Member
****
Offline Offline

Posts: 7941



WWW
« Reply #41 on: December 14, 2005, 08:46:08 PM »
ReplyReply

That'd be great, many of us deal with images that are larger than 1 GB (scanned 4*5 with adjustement layers or larged stitched images with masks), and the scratch can typically get up to 3 or 4 GB.

Regards,
Bernard
Logged

A few images online here!
kaelaria
Sr. Member
****
Offline Offline

Posts: 2227



WWW
« Reply #42 on: December 14, 2005, 09:08:18 PM »
ReplyReply

I know the test wasn't that large - but if you can tell me how to construct such a test I'll be glad to ru it.
Logged

dmerger
Sr. Member
****
Offline Offline

Posts: 686


« Reply #43 on: December 14, 2005, 10:55:56 PM »
ReplyReply

Kaelaria, you are a bald face liar.  Where did I clearly say I agree with the articles I referenced?   Put up or shut up!

Tim thought that the most relevant quote form one article was a benchmark test result, when the whole point of the article was that benchmarks are not representative of actual real world performance.  Furthermore, I thought the conclusion was a more relevant synopsis of the article than one line taken out of context. I did not, however, express any agreement or disagreement with the conclusions of the articles.  

Kaeleria, please show me where I've " admitted [I haven't} EVEN HAVE OR HAVE TRIED A RAID0 ARRAY."  This claim of yours is another bald face lie.  Again, put up or shut up!

I have never pretended to be a computer guy, whatever that means.  If you don't like the results of the tests done by AanandTech and Storage Review, and you think that your over four years of experience makes you more qualified than them, that's fine with me.  Maybe you are more qualified.  But I know that you are a bald face liar, and I have no reason to suspect the same from the people at AanandTech and Storage Review.

DiaAzul, you may be correct.   I'd like to see an actual, in-depth test of RAID 0 for Photoshop scratch disk.  I've tried, but have been unable, to find such a test report.  For me, and me only, I'm not ready to try RAID 0 on my home computer for my Photoshop scratch disk until I see some evidence that it would provide a significant benefit.   Your theories may be correct, but as demonstrated in the articles, real world performance can be different than what is commonly expected.
Logged

Dean Erger
kaelaria
Sr. Member
****
Offline Offline

Posts: 2227



WWW
« Reply #44 on: December 14, 2005, 11:12:54 PM »
ReplyReply

Quote
Tim, you should do some research on actual RAID 0 performance for a desktop PC.  Every test report that I've read found no significant performance increase.  This topic has been discussed on this forum in the past.
[a href=\"index.php?act=findpost&pid=53326\"][{POST_SNAPBACK}][/a]

Here's you clearly agreeing...plus what I already quoted above...
Logged

kaelaria
Sr. Member
****
Offline Offline

Posts: 2227



WWW
« Reply #45 on: December 14, 2005, 11:16:57 PM »
ReplyReply

Quote
Tim, I haven't looked into RAID 0 for quite some time.    I'd add a RAID 0 if I thought it would speed up Photoshop for me. [a href=\"index.php?act=findpost&pid=53390\"][{POST_SNAPBACK}][/a]

And here's you admiting you have zero actual experience of which to comment on.
Logged

BernardLanguillier
Sr. Member
****
Offline Offline

Posts: 7941



WWW
« Reply #46 on: December 14, 2005, 11:17:01 PM »
ReplyReply

Calm dowm gentlemen, every exchange like this is a brick that contributes to turning this forum in a replica of DPreview...

Cheers,
Bernard
Logged

A few images online here!
kaelaria
Sr. Member
****
Offline Offline

Posts: 2227



WWW
« Reply #47 on: December 14, 2005, 11:18:27 PM »
ReplyReply

Quote
Kaelaria, you are a bald face liar.  Where did I clearly say I agree with the articles I referenced?   Put up or shut up!

Kaeleria, please show me where I've " admitted [I haven't} EVEN HAVE OR HAVE TRIED A RAID0 ARRAY."  This claim of yours is another bald face lie.  Again, put up or shut up!

I have never pretended to be a computer guy
[a href=\"index.php?act=findpost&pid=53584\"][{POST_SNAPBACK}][/a]


Proven, proven, and - yeah, that's quite obvious - now stop giving computer 'advice'.
Logged

kaelaria
Sr. Member
****
Offline Offline

Posts: 2227



WWW
« Reply #48 on: December 14, 2005, 11:19:50 PM »
ReplyReply

Quote
Calm dowm gentlemen, every exchange like this is a brick that contributes to turning this forum in a replica of DPreview...

Cheers,
Bernard
[a href=\"index.php?act=findpost&pid=53587\"][{POST_SNAPBACK}][/a]


Perfectly calm.  I just like correct information on a given subject, not armchair quarterback BS.
Logged

dmerger
Sr. Member
****
Offline Offline

Posts: 686


« Reply #49 on: December 15, 2005, 12:17:02 AM »
ReplyReply

Kaeleria, it's amazing that you can say such nonsense publicly.  My statements that you quote in no way support your allegations.  It seems obvious that you don't understand the meaning of simple english words, you're irrational, or you simply have no hesitation to make obviously false statements.
Logged

Dean Erger
kaelaria
Sr. Member
****
Offline Offline

Posts: 2227



WWW
« Reply #50 on: December 15, 2005, 07:00:12 AM »
ReplyReply

To anyone reading this for raid0 information...it definitely has it's uses.
Logged

SJAFSA
Newbie
*
Offline Offline

Posts: 2


« Reply #51 on: December 15, 2005, 08:42:51 AM »
ReplyReply

Quote
Video Card:
Adobe systems have stated several times that Photoshop does not utilize the video card. (search for the documents on tweaking PS for performance) The video card only matters in its ability to run the monitor resolution you need to run at a decent refresh rate (75Hz or more). Refresh rates do not apply to LCDs.
The Matrox cards are beneficial over others only if you are using CRT monitors and the usefulness only really applies to the legibility of text onscreen. Matrox cards are know for producing the sharpest image possible for analog CRTs. If you are using LCDs this does not matter as the signal is digital.

One video card will not help you photoshop better over another video card.

Windows Vista will be mostly Vector based (mathematically rendered and not with bitmap images) and hardware accelerated if your system can handle it. Much like the Macs are today but more-so.

Buying a fast video card that is "Vista Ready" is only important if you a) later upgrade to Vista (which may or may not turnout to be a valuable thing to do) and b ) want to have the pretty interface special effects that Vista will be capable of.

If neither of these items matter to you than don't worry about the video card too much. Any of the old 128MB Nvidia or ATI cards will do you well and only cost you about $30. If you only use one display, many of the motherboards out there with built-in video will do the trick.

A fast video card is not a requirement for upgrading to Vista.
Ram:
Max out the ram. Buy Windows XP Professional as that allows for 3GB of Ram and Windows XP Home limits you to 2GB. XP Pro 64-bit will allow up to 16GB of ram but driver and software support for it is minimal meaning your experience could be rather problematic.

CPU:
Defiantly get dual-core system at a minimum. Honestly, PS is optimized to take advantage of multiple cores and it shows (YMMV of course). I have several clients with single-core 3.2Ghz P4s that are easily only 1/5th the overall performance of my Dual 2.3Ghz Mac (and that was before I upgraded it from 1.5GB of RAM to 3.5GB). It drives me nuts because I have to use the confounded things regularly.

It should be noted that a dual-core system won't always be faster than a single core system per-se. It all depends on the actions you're doing. It will make things run more smoothly overall and that is just as important. Sheer speed doesn't mean squat if the system comes to a halt if you try to browse a folder in Bridge while PS is batch processing in the background (just one example).

Dealing with large files, a dual-core system with as much ram as you can squeeze into it will make the difference between tapping your fingers waiting for crap to happen and getting things done. I would not recommend skimping on your computer, especially if you deal with large files. You'll only end up with a frustrating experience. A more powerful system may cost a bit more but it'll last you longer and make you happier.
[a href=\"index.php?act=findpost&pid=52899\"][{POST_SNAPBACK}][/a]
Logged
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #52 on: December 15, 2005, 08:51:55 AM »
ReplyReply

How about easing up on the verbal pissing contest, and devising a RAID 0 performance test with real-world relevance to Photoshop users (like doing a batch conversion of 300 RAW images with some sharpening and other common image processing operations) and posting the results of RAID 0 vs single drive? He said / she said arguments about how many drive platters can dance on the head of a RAID controller in a non-Photoshop context aren't particularly useful.

Logged

dandill
Jr. Member
**
Offline Offline

Posts: 84


WWW
« Reply #53 on: December 15, 2005, 08:52:14 AM »
ReplyReply

One caution about video cards: While PS may not need a sophisiticated card, the video display may.

For example, I have an Eizo CG210 (1600x1200) and it specifically takes advantage of an ATI 5100 card (on XP Pro), for hardware control of the display that is used when calibrating with ColorEyes Display and for the live rotation feature.
Logged

Dan Dill
kaelaria
Sr. Member
****
Offline Offline

Posts: 2227



WWW
« Reply #54 on: December 15, 2005, 08:54:14 AM »
ReplyReply

Quote
How about easing up on the verbal pissing contest, and devising a RAID 0 performance test with real-world relevance to Photoshop users (like doing a batch conversion of 300 RAW images with some sharpening and other common image processing operations) and posting the results of RAID 0 vs single drive? He said / she said arguments about how many drive platters can dance on the head of a RAID controller in a non-Photoshop context aren't particularly useful.


[a href=\"index.php?act=findpost&pid=53610\"][{POST_SNAPBACK}][/a]


Write me a script and tell me exactly what hardware configs you'd like tested and I'll be happy to do that tonight.
Logged

SJAFSA
Newbie
*
Offline Offline

Posts: 2


« Reply #55 on: December 15, 2005, 08:57:47 AM »
ReplyReply

Hi.

I went through the same frustrations with my computer and I finally bought the perfect photography computer for me. Here are the specs.

Dell Precision Workstation 650. Dual 3.1G processors 4 Gig memory (two for Photoshop and two for everything else), two 15K SCSI hard drives in Raid 0 73Gigs each. One IDE? (I forget the latest name they use) 400Gig hard drive. Two Dual layer DVD burners 16X. I've attached a Canon 4990 scanner, i9900 printer, Colorvision for color management of monitor and printer, Eizo L997 20.4" monitor for pictures, Dell Ultrasharp 19" monitor for Photoshop tools. 6 x 8 Wacom Intuos tablet.

Incidentally, my camera equipment is all Nikon - D2x camera, 17-55mm f2.8, 12-24mm f4, 70-200mm f2.8 vr, 105mm micro, 50mm f1.8, tc2e II.

I've been using it for about eight months and am still thrilled with the speed. The real key is the 15K drives in the Raid 0 array. It's expensive, but I have stopped shopping for equipment and have removed all frustration from my workflow which employs RAW processing at 20megs per picture.

Steve Abramson
Logged
DarkPenguin
Guest
« Reply #56 on: December 15, 2005, 09:42:42 AM »
ReplyReply

Pom mentioned batch converting wedding photos (in this or another thread).  Run on some 300 photos that would seem like a pretty valid test.

The problem with this is that most people only have one (decent) machine to run this on.  So if you'd like multiple data points we'd need the same RAW files winging around.  (Or maybe just one duped 300 times.)

15k SCSI drives are just hella fast.  In a RAID or not.  But for the price and noise of two of them I could run a second machine.
Logged
jimhuber
Full Member
***
Offline Offline

Posts: 147


WWW
« Reply #57 on: December 15, 2005, 10:55:34 AM »
ReplyReply

I should know better than to wade into this mess, but I am a "computer guy", professionally for almost 20 years. So here goes...

RAID 0 writes exactly the same data to each drive in a pair of drives, so write speed is exactly the same as a single drive. But reading from the drives is independent, so read speed is theoretically doubled. "Theoretically" meaning that if the transfer bus can handle it and the controller has sufficient buffer RAM, you get doubled read performance.

In Photoshop terms, if you're batch processing a set of RAW captures into some other format, how much of the total time is spent reading from and writing to the drive(s)? As the resulting file size increases in relation to the source file size the benefit of RAID 0 fades into obscurity... remember: reading is faster, writing is unchanged. So if you're reading 20 MB files and generating 20MB files the RAID 0 setup will take AT LEAST 75% of the time the single drive setup will take (read time is 50% of total time, write time is 50% of total time, RAID 0 read performance exactly double). If you're reading 20 MB files and generating 80 MB files, the RAID 0 setup will take AT LEAST 90% of the time the single drive setup will take (read time is 20% of total time, write time is 80% of total time, RAID 0 read performance exactly double).

But then we get into how long does it take to write X bytes versus read them? On most drives, writing takes more than twice as long because the drive itself writes the data and then reads it back to verify it's correct. So then the above examples become even worse.

Out here in the real world, processing time ain't zero, either... far from it. The first example (20MB in, 20 MB out) may really be 25% read time, 50% processing time, then 25% write time. So then the total time of RAID 0 versus single drive becomes almost 90%. The second example (20 MB in, 80 MB out) may really be 10% read time, 60% processing time, and 30% write time. The total time of RAID 0 versus single drive becomes 95%.

So RAID 0 is really only applicable to applications that read a lot and write very little. Databases are the classic example - most are searched endlessly but updated only rarely. Photoshop, at least in my use, writes more than it reads. Bridge building thumbnails will be much faster, but not most of the workhorse functions, with a RAID 0 setup.

Conclusion: RAID 0 offers little benefit for Photoshop. Buy a faster single drive. Use separate drives for operating system, data, and scratch. Use the fastest transfer bus available: SCSI320 or FireWire.

I personally use 15,000 rpm SCSI internal and FireWire external drives in a Dell Precision Workstation 340 with a 2.4 GHz Pentium 4 processor with 1 GB of RAM (2002 vintage). For my raw captures from a Canon Rebel XT, 1 GB of RAM running Windows 2000 has proved to be plenty, and I work in ProPhoto RGB.

And now back to your regularly scheduled photographic content...
Logged
dandill
Jr. Member
**
Offline Offline

Posts: 84


WWW
« Reply #58 on: December 15, 2005, 11:01:23 AM »
ReplyReply

Quote
I should know better than to wade into this mess, but I am a "computer guy", professionally for almost 20 years. So here goes...
[a href=\"index.php?act=findpost&pid=53622\"][{POST_SNAPBACK}][/a]
Clear as a bell, and very helpful. Thanks!
Logged

Dan Dill
jimhuber
Full Member
***
Offline Offline

Posts: 147


WWW
« Reply #59 on: December 15, 2005, 11:01:43 AM »
ReplyReply

Oh yeah, if you want a RAID setup that writes faster than a single drive use RAID 5 with SCSI drives. That's what professionals use in servers. The more drives you add, the better your performance. The data is "striped" across multiple drives, plus "check data" so if any one drive fails the array still functions.
Logged
Pages: « 1 2 [3] 4 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad