Dual Athlon MP, 2 GB of RAM, Windows XP pro.
I was opening few large (~250MB) PSD files, 8-bit, CMYK, lots of layers (40+) with styles. Some PS7, some PS CS/ Mac. They take a wile to open on Mac (~15 sec), took a while to open in PS7/ Win (~15 sec), I expected ~15 sec to open in CS/ Win
Instead I had to wait 3.5-4 minutes for them to open. This is AFTER updating layers (where applicable) and RE-SAVING. I turned off network and all the usual suspects, no joy.
This is clean XP install, no AV crap and no other utilities!
THE ONLY THING that makes (huge) difference is lowering PS ram allocation below magic 1GB:
997 MB= ~15 sec.
1.1GB= 3.5-4 min.
Im fully aware of the new memory management scheme in CS; increased Tile size with more than 1000MB of RAM allocated to PS. This was supposed to enhance performance when working with large files.
I see the opposite here. Program becomes practically unusable with large files when memory allocation is set above 1GB. No setting above 57% (1013MB) makes any appreciable difference.
Here is bizarre twist to this: I have three 10K rpm U-160 SCSI HDs and one 7200 rpm IDE. I was trying to move PS scratch from one SCSI drive to another nothing would make any difference. Hell even putting PS scratch and Windows Page File on the same drive didnt make any difference one way or the other.
Then I put PS scratch on my slowest, fragmented and 60% full IDE drive and that same large file opened in 1 min 15 sec. Still much slower than expected but way faster than 3.5 minutes when PS scratch was on faster SCSI drive. I tried this with few other large files and the same pattern emerges.
What the hell is going on here? Does PS CS have problem with SCSI drives (in addition to multi-processor and 2GB RAM)?
What the hell is going on here? Does PS CS have problem with SCSI drives (in addition to multi-processor and 2GB RAM)?
Maybe you were on the right track when switching cables, etc?
What’s your physical drive/controller configuration and what’s where? (ps scratch vs windows scratch) where’s the cd/dvd? where’s the ide hd? there’s a lot of combinations, but that’s the only thing that jumps out at me. is the slow ide primary master or secondary master on the bus w/o a cd hooked to it?
Dave,
IDE drive is primary master (nothing else on that channel), CD is on another IDE channel. CD burner is SCSI but I already disconnected that, I have external FW CD/DVD combo.
All SCSI drives are on the same controller but this shouldnt and never has been a problem. This card can connect 15 devices.
I did run SiSandra drive benchmark tests, Im not sure how reliable these are but SCSI drives came ahead of IDE by pretty healthy margin as they should.
Oh and PS scratch is always on different physical drive than Win page file. One time I put them on the same SCSI drive just for the hell of it and it didnt make any difference either way.
Len,
Thanks for the link.
Yes I have looked at it long and hard, many times.
I still don’t understand why assigning slower drive as PS primary scratch disk improves performance. I still don’t know why increasing PS allocated RAM beyond 1GB brings program to its knees with large files when it’s supposed to do the opposite.
I hope Adobe folks do.
Could that be the case because the PSscratch and pagefile are on disks chained to the SAME controller? (I am perfectly aware that SCSI should avoid that with Command Queuing…)
What if you put the pagefile on the IDE? and the scratch on either (and/or both) SCSI?
I still don’t understand why assigning slower drive as PS primary scratch disk improves performance.
I have a gut feeling it’s related to it’s location and isolation on the primary ide channel. As a test, maybe you can pull the secondary slave off line, put an ide drive as secondary master, and assign that as the ps scratch disk (assuming your windows swapfile is on primary master.) Also make sure you’re using the jumpers to set the position (Master/Slave) in the channel, not just the Cable Select (CS) jumper.
Could that be the case because the PSscratch and pagefile are on disks chained to the SAME controller?
That shouldn’t matter Pierre, especially with SCSI drives but you never know, something is funky here.
Dave,
As a test, maybe you can pull the secondary slave off line, put an ide drive as secondary master
I’ll try that later when I get home. This activation makes me paranoid though I already had to activate 4 or 5 times on this box (I lost track already).
I’m not sure what you’re getting at with this though. Are you trying to slow down my IDE drive? Or do you think my IDE configuration is affecting SCSI controller?
Activation aside, certainly there has to be an issue. Assume someone who has only taken up a computer AND Photoshop, certainly ALL this CRAP about changing drives, memory allocation, sorting out a "File Browser" problem with having to have "large Image processing" box unchecked etc etc. How can they work at all.
Len, Dave, Chris etc etc. Get the bloody program to work.
It REALLY IS BROKE. Fix it and enough of the very poor workarounds for a program that’s delayed to the point , half the world’s crying.
Photoshop IS Broken, Fix IT
No Regards
Not Mark
Mark,
Why not talk to Adobe direct if you have a problem. For many of us it works fine.
Len and Dave both do an excellent job helping us all on this forum and are not employed by Adobe and receive no payment.
Lets see if you have enough good manners to apologize for your inappropriate rant.
Best regards,
Ken
Mark,
Chris is the only one on that list that works for Adobe, and being here is not part of the job requirement.
Bob
I’m not sure what you’re getting at with this though. Are you trying to slow down my IDE drive? Or do you think my IDE configuration is affecting SCSI controller?
Hmm… rereading all your specs. I missed this:
Here is bizarre twist to this: I have three 10K rpm U-160 SCSI HDs and one 7200 rpm IDE.
I was thinking you had a single scsi and multiple ide’s. Nevermind my last suggestion. Have you tried removing all but one device from the scsi chain and seeing if that works better? There’s obviously a bottleneck somewhere.
Sorry if it sounds like I’m reaching, but I am. I’m just trying to think of some ideas that might help you get to the bottom of this or might help the adobe guys to recognize something.
Len, Dave, Chris etc etc. Get the bloody program to work.
LOL! What Ken and Bob said! Mark, I know I post here a lot. That’s because I hang here and pick up tips from the ar-teests that frequent these demesnes. I’m a PS hobbyist who’s trying to learn the program. I’m a programmer by trade (not for Adobe) and a former PC/Network tech (looooong time in the corporate trenches), so I happen to have some insights to share on hardware configs, operating system configs and general operation (and malfuntions) with windows programs. IOW, I do what I can. I get paid in knowlege. I know you’re frustrated but if you can’t help you’re doing worse than nothing and exacerbating the problems by writing rants in the middle of technical discussions.
If you want "Expert" help, here’s the link for you:
<
http://www.adobe.com/support/expert_support/main.html>
Dave
Have you tried removing all but one device from the scsi chain and seeing if that works better?
Thats exactly where my fishing expedition was headed when I was forced to activate PS (again) hence my frustration expressed in the opening post. Makes trouble shooting PITA doesnt it?
I started disconnecting one SCSI drive at the time. Given number of activations I already had to do on this box I fear Ill get activation denied eventually and get screwed at the wrong time.
There’s obviously a bottleneck somewhere.
Frankly I think bottleneck is PS new memory management (remember problem disappears below 1GB of PS allocated RAM) but I was more than willing to go the extra mile just to make sure I dont have something screwy here.
Sorry if it sounds like I’m reaching, but I am. I’m just trying to think of some ideas that might help you get to the bottom of this or might help the adobe guys to recognize something.
Thats OK Dave, nothing to be sorry for. Im trying to get to the bottom of this too just hope Adobe folks get there first. 😉
Thats OK Dave, nothing to be sorry for. Im trying to get to the bottom of this too just hope Adobe folks get there first.
<nodding>
Another possible troubleshooting scenario, Andrew , is to try the similar tests in Photoshop 7 . I expect that the SCSI chain works just fine under Ver 7.01 ( I have a similar config to yours, Ultra320 host adapter with 15k drives, plus an IDE RAID 0 array ).
I do suppose, though I think it’s not likely, that a single flawed SCSI device could cause these problems . Certainly , the activation requirements do nothing to hasten the troubleshooting process.
🙁
Brent
Sorry, Guys. It WAS a drunken rant. I DO appreciate all you do and have benefited in the past. Again I do apologise.
Regards
Mark
That’s cool mark. Drunken rants are what make helping strangers on the internet fun… um, wait, no… that’s what makes BBQ’s fun… nevermind.
🙂
Dave,
I was out of kilter but i re-read my post and although I did have strong aggression, (No more Whisky) I did think about the newbie/ novice and with all intents and purposes, if it was me, I wouldn’t have a clue what people were talking about. it’s just with previous releases no-one ever had these issues crop up until now.
Anyway, i’m staying out of the argument til .01 comes out.
Regards
Mark
Hope May’s a good ‘un for BBQ’s
Brent,
Another possible troubleshooting scenario, Andrew , is to try the similar tests in Photoshop 7 . I expect that the SCSI chain works just fine under Ver 7.01
I dont have PS 7 installed since I switched OS to XP pro few weeks ago. When I did use PS 7 (under Win 2000) it was practically on the same hardware (I added one SCSI drive). Ive never had to wait 3.5 minutes to open 250 MB file in PS 7.01. The same files were opening in about 15 sec as they do in CS with lower RAM allocation. Then again 7 used the same Tile size regardless of allocated RAM.
As I mentioned earlier I did run SiSoftware read/ write benchmarks for all drives and SCSI drives outperformed IDE by a margin I would expect. Is this test reliable enough to assume nothing is wrong? I dont know but I dont see any sign of these HDs misbehaving outside of PS CS.
Andrew
i’m staying out of the argument til .01 comes out.
Dont be a stranger here Mark,
You might need to hold your breath for a while waiting for .01
Stephanie Schaefer "Best guess on CS2 upgrade date?" 4/7/04 2:41pm </cgi-bin/webx?14/17>
Andrew,
Re-reading your posts, it seems to be a memory management issue. Perhaps using over 1GB of ram causes the scsi I/O area to be "virtually relocated" ( I just synthesized this concept <G> ) in the memory map. It would be nice if you’d be a guinea pig and install Pshop 7 and see if you still have problems at the 1 GB boundary of RAM .
I’m only running 1GB RAM total , so I can’t be any useful help on that score ( nor have I been tempted to upgrade to CS ) .
🙂
Brent
if it was me, I wouldn’t have a clue what people were talking about.
I know mark, but I was speaking directly w/andrew who does seem to have some idea of what he’s about. 😉 I try to temper my advice for who i’m addressing, usually.
brent’s advice on the memory thing seems valid too. (not sure about the virtual relocation concept though! <g>) but there’s definately something flaky going on w/cs’s memory mgmt over 1 gig. unfortunately i too only have a gig of ram so can’t mess around any to check it out. also it can’t be ALL the memory scheme’s fault because of the speed increase when going to a drive on primary master controller. there’s some interaction going on with the memory and the scratch disk that we can’t see from the outside (of the code). chris has said the memory mgmt has remained virtually unchanged from 7 to cs, but "virtually" can cover a lot.
Dave,
Granted regarding the memory aspect but surely coding (Which I know Skank) can be streamlined. It’s an obvious fact that as machines have got faster the coding has become more slack with a come what may attitude.
We ALL know PS has got slower it’s the excuses that annoy me and when we have to delve into "Virtual Memory" to fix a problem there has to be something wrong. It’s the I’ll stick with PS no matter what that and Oh! NO! Ps isn’t at fault attitude that’s hard to bear.
We’ve got probably the best programmers here bar none so this should be easy to fix. It’s also plain to see the collusion between MS and Adobe and when they start to blame OS problems with software issues, someone is going to loose "PALM"
I think that may also have sparked my original "RANT". It’s like out of the hands of the minors, yet the minors put them where they are.
So! Believe it or not, I’m not getting involved ‘cos look what Andy said,
Dont be a stranger here Mark,
You might need to hold your breath for a while waiting for .01
Stephanie Schaefer "Best guess on CS2 upgrade date?" 4/7/04 2:41pm<
So I’ll have a beard getting the "File Browser, Opening Files, Waiting on filters, No PSD views in windows, and all the other problems being fixed.
BTW, I’m on Czech beer tonight LOL!
Regards
Sometimes Mark
It’s an obvious fact that as machines have got faster the coding has become more slack with a come what may attitude.
no it’s not obvious at all. coding has gotten HARDER with faster machines and new technoligies.
It’s the I’ll stick with PS no matter what that and Oh! NO! Ps isn’t at fault attitude that’s hard to bear.
In this case (and this is probably related to the 2 gig memeory thing they’re working on) it probably is PS’s fault. Nothing wrong with trying to troubleshoot it though.
We’ve got probably the best programmers here bar none so this should be easy to fix.
The 2 parts of that statement are not necessarily in agreement. And "Easy" is always relative.
I and most of the users here realize there are problems with CS. There were problems with 7 when it was released, and 6, etc. We’re just kinda spinning our wheels here trying to come up with work arounds for the problems until the fixes are released. (Which’ll hopefully be soon!) But if some "easy" problems turn out to be not so easy, well then, pop a czech beer for me too cuz it’s gonna be a bit of a wait.
Good Dave, this will get cracked, and I don’t mean that in the negative. I’ve been using PS since V5 and I regards it as the best but it really IS V.V. Slow compared to other V’s and i’m using a new PC. I really didn’t expect this at all.
Nothing else left but the "Masters" on TV and the coming "Good friday" soccer program on UK Sky.
Can’t wait, bet I get drunk before you, unless you get up 6 hours earlier.
Regards
Becoming more like Mark
I’m now inclined to be slightly – that’s "slightly" – forgiving towards Adobe. There is an obvious problem with 2GB DC DDR SDRAM, PS and Windows XP addressing. I experienced the problem with 7.0.1 before CS, so it’s not just a CS problem, being hugely disappointed at having built a (then) kick-arse cutting edge machine only to find that the advertised handling of 2GB was not what it was cracked up to be. Even more disappointed when the initial response from Adobe – "Huh?" – kind of indicated that the pre-release testing of CS maybe wasn’t quite what it should have been. This should have been picked up as 2GB RAM setups were on the cusp of being mainstream by then.
I’m also anxiously awaiting the release of the final of SP2 for XP – though having absolutely no delusions that MS will have done anything to help with its memory management as I suspect SP2 will be entirely focused on patching all the security holes in XP that make it such a hacker’s delight. Given the general quality of MS’ software, I don’t reckon that’s likely to be too successful, either. Monopolies are not condusive to quality, value or customer service, after all.
With regard to the 2GB RAM thing in PS, I’m tempted draw a parallel with Ford’s recent trouble with one of their SUV’s and its tyres – though Ford didn’t manufacture the tyres themselves, I believe that they eventually admitted liability to the point of replacement. It took them a mighty long time to come to the party, though. Pity that no-one is likely to get killed running out of RAM, as Adobe might then be in a similar position. So I’ve just got used to the idea that the dot patch won’t fix the problem and paying for the upgrade to 9 in a year or so will be the only real chance at a proper fix.
Let’s also not forget that the IT industry is still the most arrogant on the planet, but it’s nice to see a glimmer of normality returning when the richest man in the world is no longer a meglomaniac software seller but a miserly old alcoholic furniture salesman from Sweden.
Fred.
Brent,
It would be nice if you’d be a guinea pig and install Pshop 7 and see if you still have problems at the 1 GB boundary of RAM .
I dont need to re-install PS 7 to answer your question. PS 7 ran flawlessly on 2GB RAM system with memory allocation as high as 90%. There was no penalty for going above 1 GB.
There is no mystery here. In PS CS memory allocation determines Tile size.
PS CS creates 144 MB temp file (before you even open an image), that temp file becomes 571 MB once you exceed 1000 MB of PS allocated memory. Opening 235 MB multi-layered PSD: 1.2 GB temp file with less than 1000 MB of allocated RAM, 2.5 GB with more than 1000 MB. This is way more than PS 7 regardless of allocated RAM.
We have been told not to worry, not to read too much into these numbers, its all good and meant to improve performance.
Obviously this new memory management isnt always working as expected. Is SCSI another piece of a puzzle here? Sure looks that way but I also have two other known suspects here (MP and 2GB of Ram).
FWIW: I disabled two of my three SCSI drives (it improved remaining drives R/W index by about 10%). Makes no difference CS still likes slower IDE drive for scratch better, even with Windows page file on the same IDE drive. This is insane.
Andrew,
Only reason I can think of to re-install ver 7.01 ( besides the fact that it works < cheap shot ! > ) is for side-by-side comparisons to guard against the remote possibility that this is all caused by a subtle failure in scsi hardware . I increased my system RAM to 1.5 GB to test large images with 1.2GB devoted to Photoshop , to see if there were any obvious slowdown symptoms on my Win2k box with Pshop 7.01, but I have no issues with the system speed.
The ball is in the court of the coming update, I think.
🙁
Brent
The ball is in the court of the coming update, I think.
Indeed.
