Lightroom 3.3 Performance Feedback

[Tom Hogarty]

Please use this discussion topic for your feedback on Lightroom 3.3 RC and the final Lightroom 3.3 release when it becomes available.  The Lightroom team has tried very hard to extract useful feedback from the following discussion topic but due to the length and amount of chatter we need to start a new, more focused thread.  Please post specifics about your experience and be sure to include information about your hardware configuration.

Regards,

Tom Hogarty

Lightroom Product Manager

[Jeff Schewe]

imajez wrote:

Jeff has said on several occasions he is not an Adobe employee, though he certainly does advise and give them useful feedback.

Correct, I do not work for Adobe (otherwise there would be the employee tag by my name). But I do sometimes work with Adobe–primarily ACR, LR and Photoshop. Most recently working on image sharpening in ACR & LR (including the output sharpening in Print and Export).

Look, there is a very clear line in the sand here for me. If you need to work on a hundred or a thousand images and optimize the tone, color and detail of the images, that's what Lightroom was designed to do. If you want to take your hero selects and spend time "perfecting" them which would include substantial retouching, masking and composites, you don't really want to do that in Lightroom–that's what Photoshop is for. Two separate tools with two different working philosophies.

The fact that Lightroom can do as much as it does is a testament to the ingenuity of the engineers who came up with the parametric editing approach-largely Thomas Knoll and the ACR team and Mark Hamburg and the Lightroom team. And yes, the very existence of the spot healing tool (originally designed to eliminate sensor dust spots) tends to encourage users to use it for purposes it was really designed for such as retouching. The Adjustment Brush was similarly designed to do softer, local adjustments of tone and color that you couldn't do with a global tool. It is not designed to take the place of accurate masking and adjustments that you can do in Photoshop with adjustment layers. The fact that Lightroom users tend to try to finish off and "perfect" an image in Lightroom is where the parametric paradyne tends to bog down.

Also, don't read what I write as an indication that neither I nor the Lightroom team fails to care about the future of Lightroom. We do...but general ranting and Pi$$ing & Moaning™ and attacking Adobe and presuming motives that don't really exist is simply unproductive...this thread is a prime example of a thread that was started to get specific information which then spun completely out of control and became a waste of bandwidth...oh, some good things did come out but it was precious little.

If you want to have a positive impact, when you find a bug, report it. The bugs that get fixed are the bugs that can be duplicated by the QE team. If they can't duplicate, they can't fix it. If a particular bug is cropping up for a number of people on various platforms and is deemed mission critical, those are the bugs that get the most attention and are more likely to get fixed. Dot releases are primarily designed for new camera additions and maintenance updates (which include bug fixes where possible), not for changes in the functionality of the app.

It's STILL not clear why some people have very poor performance while the vast majority don't.

Personally, for me, LR 3.x is just about as fast as 2.x except when you factor in the greater processing requirements of the Process 2010 version and the addition of Lens Corrections. It should also be noted that new cameras tend to produce yet larger files. I'm using ACR/LR for processing 60MP P65+ MF files and I can tell you that adding a ton of spots and Adjustment Brushes will really slow down processing...but I don't really try to "finish" those images in ACR or LR. I use LR for selection editing and global adjustments (with the odd locale adjustment when needed) and do the heavy duty imaging on my hero selects in Photoshop and take them back to LR for organization.

If you are having issues, a dispassionate description of the problem with enough information (platform, OS, camera, etc.) is the most useful thing you can contribute...all the ranting and raving is really rather counterproductive because it just makes noise without any useful signal...

And I would caution against trying to read the tea leaves about Adobe's alleged motives about anything. Everybody I know who works on Camera Raw and Lightroom are highly dedicated and very hardworking.

[imajez]

Jeff Schewe wrote:
The fact that Lightroom can do as much as it does is a testament to the ingenuity of the engineers who came up with the parametric editing approach-largely Thomas Knoll and the ACR team and Mark Hamburg and the Lightroom team. And yes, the very existence of the spot healing tool (originally designed to eliminate sensor dust spots) tends to encourage users to use it for purposes it was really designed for such as retouching. The Adjustment Brush was similarly designed to do softer, local adjustments of tone and color that you couldn't do with a global tool. It is not designed to take the place of accurate masking and adjustments that you can do in Photoshop with adjustment layers. The fact that Lightroom users tend to try to finish off and "perfect" an image in Lightroom is where the parametric paradyne tends to bog down.

The fact is that it can now be be quicker and easier to finish off an image in LR compared to using PS. Insisting one uses PS as that is what you are meant to do is a retrograde step and is the argument used  by traditionalists trying to defend something that is no longer applicable as things have changed. Not to mention that LR users may not even have PS to work with.

Using grad, filter, some dabs of the adjustment brush and a bit titivating with the targeted adjustment tool is a good use of LR's toolset, not an abuse of the programme as you seem to be saying.

Blaming the end user for problems of the tools not doing the job properly is not a good PR move. Besides if it were indeed the issue of parametric editing as a paradigm failing under heavy usage, ACR would have the same problems as LR. It doesn't from my experience. Images that cause LR to grind to a halt and crash work just fine in ACR.

Also images that now cause issues did not do so when I edited them originally, yet in 3.3 and 3.4 there is an serious issue.

I'll try a new LR catalogue to see if it's a catalogue or dot release problem when I get back home.

[Jeff Schewe]

imajez wrote:

Besides if it were indeed the issue of parametric editing as a paradigm failing under heavy usage, ACR would have the same problems as LR. It doesn't from my experience. Images that cause LR to grind to a halt and crash work just fine in ACR.

Well, Lightroom doesn't crash for me. But my experience indicates that the same images tend to behave about the same way in both ACR 6.x and LR 3.x's Develop module. I often go back and forth working on an image in LR (usually first) and then often in ACR (sometimes as a Smart Object) and the pain points of a lot of healing spots, Process 2010, noise reduction, and local adjustments seem about the same regardless of the application. That's the way parametric edits tend to add up to performance hits. If you are seeing a much slower LR performance than ACR edits, then that is a clue that something else is going on. If you wish to find that out and have the LR team do something, then bug reports; with all required system info, is what can help find the issues.

And, until you provide that to the LR team, continued harping isn't gonna get you anywhere...look at the most message from Dan Tull (one of the brave LR engineers that does make an effort here in the forums) for the type of info required...

The bug reports, when properly documented do indeed end up in the bug database and those bug reports whose info is sufficient to document and duplicate by QE get worked on...

[GaryRowe-gcvt07]

Having followed this thread since its inception (and the all of the old one too!), I can say that most of the griping I've read here is about the fact that issues that *have* had bug reports submitted have not been addressed, and some across many major releases (not just point releases). Therefore, it is disingenuous to say that people should submit bug reports instead of griping here - stop trying to put a 'spin' on things! It is the lack of response from Adobe over these already-reported issues that is the problem, not that there are bugs. There are always bugs. And they are the responsibility of the company who developed the software. If the company seems to ignore reported bugs, then how can you expect the customers to keep quiet?

The bare facts are that LR3, touted as being amazingly fast ("warp speed" is how Adobe described it), is NOT so for some people, and the best I've seen reported by people here is that LR3 is only as fast as LR2.7 - yet Adobe advertise a speed improvement for LR3? Couldn't this be taken as a bit misleading?

It also has many niggling bugs that shouldn't be there if there is decent testing in place in the development cycle, especially those that should never have got through regression testing.

Face facts - if a version of a program has bugs and is unusable for some people, then it doesn't matter how brilliant or hardworking the development team is, the program *must* be judged on its merits and flaws. And it is plain wrong to try to whitewash the obvious and apparent flaws in LR3.

We all paid money for LR3, and therefore expect 'fitness for purpose'. If LR3 can't supply that for some people, then they have every right to be annoyed. And gripe. And complain. And hold Adobre responsible for their actions. *And* inactions.

Gary

[JW Stephenson]

Jeff Schewe wrote:

I often go back and forth working on an image in LR (usually first) and then often in ACR (sometimes as a Smart Object) and the pain points of a lot of healing spots, Process 2010, noise reduction, and local adjustments seem about the same regardless of the application. That's the way parametric edits tend to add up to performance hits. If you are seeing a much slower LR performance than ACR edits, then that is a clue that something else is going on. If you wish to find that out and have the LR team do something, then bug reports; with all required system info, is what can help find the issues.

I have a different experience on my W7 machine (8/16 cores, 12GB memory, etc). and have been reporting this lag as a bug since v3.0.  In my case, the local adjustments, particularly the clone tool, never lag in ACR but have significant lags in LR3 after a many are in place.  As you state, something else is going on if ACR accessed directly compared to being accessed via LR simply do not behave the same.  I have submitted data, on request, on numerous occasions to Adobe engineers since my first bug report and assume they are still working on this issue.  There are a lot of places performance can be tuned in v3 that still are sub-par to v2 but this one causes me headaches as it can lead to the dreaded "(Not Responding)" window and a 15-30 second wait to recover.  On the bright side, since upgrading to v3.3 and turning hyper-threading off I am always able to recover if I wait long enough.

Jeff

[TK2142]

Jeff Schewe wrote:

If you want to take your hero selects and spend time "perfecting" them which would include substantial retouching, masking and composites, you don't really want to do that in Lightroom–that's what Photoshop is for.

Oh, I do want to do substantial image editing in LR. Here's why:

1. I don't want to pay for PS. It may be worth it for graphic designers and professionals but to an amateur enthusiast the price tag is way too high for what I need from it.
   
2. I don't want the destructive editing paradigm. I know I can use (adjustment) layers, etc. but LR's solutions for making later tweaks to editing operations works better for me so why should I adopt the PS way when I don't need it for my editing requirements?
3. I don't want the workflow that creates two versions of an image; one pre-edited in LR to be imported by PS and another exported by PS with potential post-edits in LR. There is no seamless editing history in one place anymore.
4. I don't want to run LR & PS in parallel on my machine. One of them on their own is sufficiently resource hungry.
   

I don't accept the premise that LR is just a DAM application with some image editing capabilities thrown in which one is supposed to use in homoeopathic doses only. You accept that premise because you believe that it is technically impossible to support extensive image editing within the parametric editing paradigm. But that's were you are wrong. There is no inevitable divide between LR and PS.

Your idea that two applications are needed plays well into the hands of those who sell both applications. However, there is no technical argument to support the "two application" assumption.

Jeff Schewe wrote:

The fact that Lightroom can do as much as it does is a testament to the ingenuity of the engineers who came up with the parametric editing approach-largely Thomas Knoll and the ACR team and Mark Hamburg and the Lightroom team.

The IQ from ACR is testament to the ingenuity of the engineers involved. LR's performance regarding image adjustments as I perceive it on my hardware, however, is not the result of someone's genius. As I said before, re-rendering times will inevitably be long for extensively edited images within a non-destructive paradigm (which need not be a problem if one has good previews) but there is no need for editing operations to become slower. When do you experience performance breakdown? During image re-rendering or when editing the image? Can you explain to me why editing has to become slower? I have the technical background to know it does not need to so I'm happy for your explanations to arrive. Of course optimisation techniques are required to cope with a large number of edits in a parametric editing approach but it is not impossible as you make it appear.

Jeff Schewe wrote:

The fact that Lightroom users tend to try to finish off and "perfect" an image in Lightroom is where the parametric paradyne tends to bog down.

It only "bogs down" because of LR's implementation. Not because it wouldn't be possible in principle.

Jeff Schewe wrote:

We do...but general ranting and Pi$$ing & Moaning™ and attacking Adobe and presuming motives that don't really exist is simply unproductive...

You can keep that phrase without tagging a "TM" on it. You are quite a show when it comes to dealing with customer frustration.

Jeff Schewe wrote:

this thread is a prime example of a thread that was started to get specific information which then spun completely out of control and became a waste of bandwidth...oh, some good things did come out but it was precious little.

People have been trying to express multiple times that they were eager to help, e.g., by contributing to threads like this, but there is only so much disappointment about bugs being kept and new ones being introduced, people are willing to take. If the LR team really is the heroic squad fighting against all odds as you describe them to be, maybe they/Adobe/you should be asking the users for more understanding and try to keep their faith one way or the other. Being told what a moaning, ignorant and useless bunch of whiners we are is only adding insult to injury.

Jeff Schewe wrote:

If you want to have a positive impact, when you find a bug, report it. The bugs that get fixed are the bugs that can be duplicated by the QE team. If they can't duplicate, they can't fix it.

Reporting bugs is what many here did before they became frustrated. You present the situation as if it were not possible to reproduce performance problems or other problems by the QE team. I find that odd. Clearly bugs have been acknowledged and are not hard to reproduce but are still not fixed. Also, there should be a way for the QE team to obtain hardware which shows particular problems, e.g., performance problems. Many users here would volunteer to drive their hardware to the engineers. Your "they can't fix it" really doesn't have much of a leg to stand on when that leg is a "impossible to reproduce" argument for problems that are widely observed among many users.

Jeff Schewe wrote:

If you are having issues, a dispassionate description of the problem with enough information (platform, OS, camera, etc.) is the most useful thing you can contribute...all the ranting and raving is really rather counterproductive because it just makes noise without any useful signal...

Again: Others and I provided many dispassionate problem descriptions, even free suggestions for how to improve the product ("feature requests"). The "ranting and raving" is a reaction to a perceived non-action. It is only "noise" to those who do not understand that the way Adobe treats a large number of LR customers is not acceptable in the eyes of those customers. Those who accept that not all is rosy in LR-land, must see that the "ranting and raving" is valid feedback. You don't simply accept "realities" yourself in other contexts either, do you?

[Shocker2000]

Here's the additional info to provide as well:
- exactly how are you importing (add from folder, copy, move, convert to DNG?)
- in the case of copy/convert/move what are the source and destination volumes?
- what are all the other options for the import (secondary backup, rename, etc)
- exactly what kind of photos are they and are there any existing adjustments?
- have you done exactly this import with 3.3 and has anything else changed besides moving to 3.4?
- can you post the contents of the system info dialog in the help menu (be sure to remove your serial number from the copied output before posting)
- it sounds like you're seeing lots of CPU usage (in task manager)? What other data does task manager show?
 
Also, if anybody else has seen anything similar, the same battery of questions applies and it'd be interesting to get more info. On a new thread, preferably so we can keep import similar symptoms together instead of in (yet another) dogpile thread.
 
Thanks -- DT

• I am performing an import from an external CF card reader, copying to c:\users\[username]\pictures\2011\2011-03-12
• No secondary backup or rename is used
• Pictures are RAW from 5D Mk2, no existing adjustments
• I have not tried this import with 3.3

Lightroom version: 3.4 RC [733717]
Operating system: Windows Vista Service Pack 1 (Build 6001)
Version: 6.0 [6001]
Application architecture: x64
System architecture: x64
Physical processor count: 2
Processor speed: 1.8 GHz
Built-in memory: 2984.3 MB
Real memory available to Lightroom: 2984.3 MB
Real memory used by Lightroom: 655.7 MB (21.9%)
Virtual memory used by Lightroom: 695.0 MB
Memory cache size: 105.5 MB
System DPI setting: 96 DPI
Desktop composition enabled: Yes
Displays: 1) 1280x800
Serial Number:

Application folder: C:\Program Files\Adobe\Adobe Photoshop Lightroom 3.4 RC
Library Path: C:\Users\mmalleck\Pictures\Lightroom\Lightroom 3 Catalog-2.lrcat
Settings Folder: C:\Users\mmalleck\AppData\Roaming\Adobe\Lightroom

• I could not even bring up Task Manager until the import was finished. I could see the CPU meter in the system tray, but Lightroom was blocking making Task Manager visible. The right click menu on Task Manager would eventually bring up the menu, but I couldn't choose Restore to view processes. CPU meter was flashing and not solid. During the import, no other applications would respond.
  

[Shocker2000]

Dan,

I continued to use 3.4 RC to sort through and process approximately 870 photos. Once imported, I found the performance of 3.4 RC to be on par with my favorite release 2.7 (Windows 64-bit).

Pictures were of a Taekwondo tournament - all RAW, no RAW+JPG or JPG. Camera is Canon 5D MkII.

My workflow was: Import > Set White Blance > Flag Photos > Filter Flagged > Apply Noise Reduction > Exposure Compensation > Color > Sync Settings to All Images > Crop > Final Adjustments for Individual Photos > Export

The only performance issues I found were on the import, particularly in creating the previews, and backing up the catalog. During these operations, Lightroom blocked other applications. I couldn't even start Perfmon to view performance. Even tried Ctrl+Shift+Esc to bring up Task Manager - nothing. While Lightroom continued and succeeded in its operations, this blocking problem is new behavior on my system.

Thanks

Mark

[Shocker2000]

Dan,

This is perfect timing. I'm going to be shooting a taekwondo tournament tomorrow for 12 hours. I'll have plenty of pictures to process.

I'll be sure to report back here any improvements or issues.

Thanks for the update!

Mark

[teho59]

Samoreen wrote:

Hi,

JW Stephenson wrote:

After some internet searching I found out about a program called the Windows Event Viewer (type Event Viewer in the start menu).  From this lead, I found the Windows Event Viewer was recording massive"information" entries...

.....

...

CloseKey    HKLM\Software\Microsoft\Windows NT\currentVersion\Time Zones\Romance Standard Time    SUCCESS       

OpenKey    HKLM\Software\Microsoft\Windows NT\CurrentVersion\Time Zones\Romance Standard Time\Dynamic DST    NOT FOUND

When I say continuously, this really does mean it doesn't stop accessing the same keys. As soon as it is done with the last line of the above block, it starts again. This madness stops as soon as I stop using the local adjustment brush. Now tell me what this tool has to do with the Time Zone settings of my system. Here again, this is static information, so there's no need to read it again and again. It won't change during a Windows session.

Performance problems you said? No wonder that LR becomes slow if it spends time doing useless things with the registry.

Hi,

It seems I do have the same symptoms for registry reading. What I  wonder about is the last line above here says "NOT FOUND". Thats must be why it tries so often to get it? If it was found it would stop checking?

Does anyone have that entry in their registry? 

It seems to be a couple of other entries to that LR is looping to get and is responded with NOT FOUND.

I am running LR 3.3, I did not uninstall before each upgrade. Win7 64 bit. Q6600 cpu, 4GB memory and ASUS GTX 275 graphical card.

Update: I was in develop when this started to loop, I went back to grid view and it is still looping and my cpu is running at 10-25% constantly and I am not doing anything in LR. The folder I selected has 50 images from a Canon 7D at ISO 1600. I have notices the high iso images are the worst images, they are more slow than 100 iso.

- Terje

[Samoreen]

Hi,

teho59 wrote:

It seems I do have the same symptoms for registry reading. What I  wonder about is the last line above here says "NOT FOUND". Thats must be why it tries so often to get it? If it was found it would stop checking?

No. When a program looks for information in the registry, it's not unusual that the requested registry key or value doesn't exist. I would even say that this happens all the time in many cases. In most situations, when a key or value doesn't exist, this means for the calling program or module that a defaut value should be used. Not all possible registry keys and values are actually present in the registry. I would say fortunately because this would make the registry even bigger. Using default value for keys and values that are not changed that frequently is a good way to save registry space.

In our case, the request sent to the registry is not made by Lightroom itself but by a system routine that is itself called by LR. In no case a system routine would request again and again the same registry value because it didn't find it the first time.

So you shouldn't worry about this NOT FOUND warning in the registry monitor.

[teho59]

Samoreen wrote:

...

So you shouldn't worry about this NOT FOUND warning in the registry monitor.

Thanks, I wont.

An update to the looping. I have about 50000 images in my database. Most of them was with Camera Calibration profile "2003". The slow one was "2010". Now I accidentely changed all images to "2010" when I had all images selected. After that LR was working at 10-25% cpu and this "Time"-access to the registry looped all the time. I stopped LR, restarted my computer and started LR. Immediately LR was up to 10-25% cpu and it looped to the registry. After some time (1-2 hours) it was finished and cpu was at zero and no more looping either. Now when I work with my "slow" images LR seems faster (maybe I only think so) and there is no looping against the registry either.

I don't know if there could be some connection with the old "2003" profile and the looping. But as long as LR was working changing profiles it looped to the registry. I think LR seems faster but I have not done any measures to confirm that.

-Terje

[thewhitedog]

One caution with applying the 2010 option in the Camera Calibration Presets panel: If you have made any previous adjustments to images using Noise Reduction (in the Detail panel) the 2010 version of your 2003 settings may be drastically different, as this option was expanded significantly in Lightroom 3. To deal with this issue, I created a custom preset in the 2010 version of Noise Reduction that produces what I consider satisfactory results in most cases (though they are still different from and superior to the Lightroom 2.x version). I can then click this item in the User Presets panel in the Develop module to insure standardized results. For what it's worth, I also included some modest sharpening in this preset to compensate slightly for the softening effect of the noise reduction. I know this can be confusing since the 2010 preset is three panels removed from the Detail panel in the Camera Calibration Panel and it might not be obvious at first that they are connected.

[Samoreen]

Hi,

IMHO, the following question should be added to the Lightroom bug report form (or should be asked when users are reporting problems with LR):

If you have upgraded from a previous version, did you uninstall the previous version before installing the new one?

It would be interesting to see if there's a relationship between the way LR has been installed (upgraded) and the fact that a given user is hit by these performance problems. If there's an obvious correlation, we might have discovered something interesting.

I recently got in touch with someone who has access to the test versions of LR and he told me that Adobe always recommend to uninstall the previous version before installing the new one. If I remember well, the installer doesn't make such a recommendation. If this is necessary the installer should proceed by itself with the full uninstallation of the previous version before installing the new one (at least, if it has the same major version number).

My two cents...

--

Patrick

[Lee_Jay-ZyZk56]

Samoreen wrote:

I recently got in touch with someone who has access to the test versions of LR and he told me that Adobe always recommend to uninstall the previous version before installing the new one.

That's for an entirely different reason and has nothing to do with the current situation.

[Samoreen]

Hi,

Lee Jay wrote:

That's for an entirely different reason and has nothing to do with the current situation.

Maybe but the reason is irrelevant. The fact is important: this tester never had performance problems and has always uninstalled the previous version before testing a new one.

--

Patrick

[Lee_Jay-ZyZk56]

Samoreen wrote:

Hi,

Lee Jay wrote:

That's for an entirely different reason and has nothing to do with the current situation.

Maybe but the reason is irrelevant. The fact is important: this tester never had performance problems and has always uninstalled the previous version before testing a new one.

Like I said, it has nothing to do with the current situation.

I have had performance problems (some pretty dramatic), and I have always uninstalled the previous version.

Next idea.

[teho59]

Lee Jay wrote:

Samoreen wrote:

Hi,

Lee Jay wrote:

That's for an entirely different reason and has nothing to do with the current situation.

Maybe but the reason is irrelevant. The fact is important: this tester never had performance problems and has always uninstalled the previous version before testing a new one.

Like I said, it has nothing to do with the current situation.

I have had performance problems (some pretty dramatic), and I have always uninstalled the previous version.

Next idea.

Lee, I think you are a bit wrong. The thing is that Adobe and other software vendors  wants us to uninstall old versions to avoid conflicts as we may see here. And as I see it the uninstaller does a poor job cleaning up the registry as most vendors do, not only Adobe.

The registry is Microsofts worst idea so far in windows history in my opinion.

- Terje

[Samoreen]

Lee Jay wrote:

I have had performance problems (some pretty dramatic), and I have always uninstalled the previous version.

That's why I suggested to get some statistics about this. Also, it's interesting to know how the previous version was uninstalled. Just using the uninstaller is probably not enough in some cases. It's because I don't want to draw definitive conclusions from my case (or from your case) that it's interesting to collect more information. After all, if I had waited for help from Adobe, I would still be struggling with a non usable software.

--

Patrick

[PhilBurness]

Samoreen wrote:

Hi,

Lee Jay wrote:

That's for an entirely different reason and has nothing to do with the current situation.

Maybe but the reason is irrelevant. The fact is important: this tester never had performance problems and has always uninstalled the previous version before testing a new one.

--

Patrick

Indeed, and if the testers are asked to uninstall previous versions prior to testing, then who tests the upgrades, and how much testing do upgraded systems get? Other than the beta testing community?

It's not in the user notes that previous releases should be uninstalled, and indeed the installation process does not uninstall previous versions.

The reason testers are asked to uninstall previous versions may be due to a completely different issue, but, if they do uninstall prior to updating then they may well not experience the performance problems that others are seeing, and may well be convinced that these problems are due to hardware issues etc.

Phil

[bob frost]

PhilBurness wrote:

Indeed, and if the testers are asked to uninstall previous versions prior to testing, then who tests the upgrades, and how much testing do upgraded systems get? Other than the beta testing community?

It's not in the user notes that previous releases should be uninstalled, and indeed the installation process does not uninstall previous versions.

The reason testers are asked to uninstall previous versions may be due to a completely different issue, but, if they do uninstall prior to updating then they may well not experience the performance problems that others are seeing, and may well be convinced that these problems are due to hardware issues etc.

Phil

I've always considered it good practice to uninstall previous versions and reboot before installing a new version. Some programs are now doing this for you, but not all. Deliberately running multiple versions of the same program seems masochistic to me! I remember when Nikon's Capture and View programs had all sorts of problems because of leftovers, and Nikon produced specific Cleaners to solve some of the problems.

I use CCleaner frequently to clean my Registry. I used to use JV16, but that got too complicated and hairy. I also go around and get rid of old folders that old versions often leave - with or without stuff in them.

No performance problems with LR 3.3 and Win7x64 except on one machine with an old version of an Nvidia driver - then the brush and spotting went berserk. An updated driver cured that.

Bob Frost

[PhilBurness]

bob frost wrote:

PhilBurness wrote:

Indeed, and if the testers are asked to uninstall previous versions prior to testing, then who tests the upgrades, and how much testing do upgraded systems get? Other than the beta testing community?

It's not in the user notes that previous releases should be uninstalled, and indeed the installation process does not uninstall previous versions.

The reason testers are asked to uninstall previous versions may be due to a completely different issue, but, if they do uninstall prior to updating then they may well not experience the performance problems that others are seeing, and may well be convinced that these problems are due to hardware issues etc.

Phil

I've always considered it good practice to uninstall previous versions and reboot before installing a new version. Some programs are now doing this for you, but not all. Deliberately running multiple versions of the same program seems masochistic to me! I remember when Nikon's Capture and View programs had all sorts of problems because of leftovers, and Nikon produced specific Cleaners to solve some of the problems.

I use CCleaner frequently to clean my Registry. I used to use JV16, but that got too complicated and hairy. I also go around and get rid of old folders that old versions often leave - with or without stuff in them.

No performance problems with LR 3.3 and Win7x64 except on one machine with an old version of an Nvidia driver - then the brush and spotting went berserk. An updated driver cured that.

Bob Frost

Bob,

Generally I would too, however, there has been advice on this forum that there is no need to, just upgrade. The fact that Adobe appear to support different versions on the same machine (admittedly, not the same catalog), if they didn't it would tell you so in the upgrade section of the Read-Me file (wouldn't it?) - or at least warn you during the installation process.

The above may sound naive on my part, but it is equally naive on the part of Adobe to assume everybody is going to do uninstall, clean the registry, flush the cache(s), reset parameters, etc.etc.

And, I don't remember these kinds of problems from previous upgrades, until I went from 3.0 beta to 3.0 final

Phil

[Ian Lyons]

The above may sound naive on my part, but it is equally naive on the part of Adobe to assume everybody is going to do uninstall, clean the registry, flush the cache(s), reset parameters, etc.etc.

Remember, the the general mass of users had access to a public beta prior to the official version of Lr3.3. So we could be looking at the left overs from the public beta, a fluke or a rogue installer. Point is - it is not clear what has happened here. That being said, there should be no need to clear the registry between versions, beta or official. That goes for any software released to the public.

[bob frost]

Ian Lyons wrote:

That being said, there should be no need to clear the registry between versions, beta or official. That goes for any software released to the public.

I'm glad that you said "should be no need". It would be nice if there was 'never' a need, but software is written by human beings, who are fallible despite some thinking they are perfect! So software will rarely ever be perfect. That's why some planes flown by computer have three computers running in parallel with three sets of software written by different software teams, so that the same bug is very unlikely to occur in two or more computers, so they run on majority voting!

Programs HAVE been sold that needed Registry cleaning between some upgrades - Nikon Capture and NikonView are two that I personally know about. After lots of complaints Nikon produced a Registry Cleaner specifically for the purpose. It cured the problems. So I wouldn't assume that it is never needed. So I just take avoiding action and clean my Registry regularly. Even that won't be perfect, and manual cleaning is occasionally necessary to solve problems.

Bob Frost

[Lee_Jay-ZyZk56]

PhilBurness wrote:

Indeed, and if the testers are asked to uninstall previous versions prior to testing, then who tests the upgrades, and how much testing do upgraded systems get?

Let's say you, Phil, get an early version of 3.4.  That version is designed to upgrade your 3.0 to 3.3 installation.  So you just install it and it does its job upgrading you to 3.4 (early beta).  Next, you get a slightly newer version of 3.4.  Like the last one, it's designed to upgrade your 3.0 to 3.3 installation, but you don't have 3.0 to 3.3 anymore, you have 3.4.  3.4 isn't designed to upgrade 3.4.  So what do you do?  You uninstall 3.4 (early beta) and install the slightly less early beta you now have.

If you're smart, you're doing this on a machine that isn't in your critical workflow, just a machine you use for testing, otherwise, early releases can do all sorts of ugly things and your workflow could be seriously impacted.  So, let's assume you, Phil, are smart.  You receive several more early betas which cause you to uninstall and then re-install on your test machine.  Finally, you get what you expect to be the release version.  You install it on your test machine and everything seems fine so you go ahead and install it on your work machine, allowing it to upgrade that system as would any general user.  Now you've tested the upgrade process both ways (direct upgrade and uninstall/reinstall), and from both an early version and from the release version, and you've done it on two different machine.

The process is roughly the same for any software from any company, and I've done this myself from companies other than Adobe that release early betas to the general user community (engineering software is often done this way).