| T O P I C R E V I E W |
| rogojin |
Posted - Jul 23 2007 : 1:01:50 PM
Hi,
I posted some time ago, complaining about performance issues with the tomato. I mentioned then that I believed it had something to do with my carting my sources around between machines, but I've subsequently come to believe that there is very little correlation there. My problems do seem to be compounded by running multiple IDEs with dependent sources at the same time (even just 2 at a time can choke the tomato-- it begins to parse about 1 source file per second, and these are small files).
In general, the problem I'm having with the tomato is that it makes my IDE completely unusuable for large periods of time. If I watch the devenv.exe process handles in Process Explorer, I notice that files are getting created and destroyed frequently in etc\\vs8\\CPP\\VA_x.tmp.
While this is happening, my IDE is pretty much unusuable. I can't type in it at all. It can go on for minutes, and all the while my disks thrash madly. I have quite decent hardware, as I've detailed earlier. I defrag my disks often, and I keep my sources and my OS on separate disks, with the partitions located at the start of the disks. I keep my OS clean also, I run no AV software or anything of that nature. The problem might be compounded by the internal VS intellisense, and if I get frustrated enough with the tomato that I end up turning it off for a couple days/weeks, then I'll let you know about that. Anycase-- I don't understand why you can't make the tomato parsing stuff happen at a lower priority, and not have to open/flush/close so many OS files. I even get audio stuttering from time to time, while the tomato is doing its thing. This really shouldn't happen. It's making my life really difficult when working on larger projects. Smaller projects are fine. They seem to parse up and get done with quickly enough. But particularly if I modify a source file that is referenced by a lot of others (around 200 others), my system comes crashing down, and I'm better off in notepad. I don't really know what I can do to help you diagnose the issue.. other than continually moaning about it, so that it remains at the top of your collective consciousnesses.
I like the product, but these issues are making it unusable.
Ben |
| 5 L A T E S T R E P L I E S (Newest First) |
| feline |
Posted - Aug 01 2007 : 08:52:36 AM
In my experience the major bottle neck is often the CPU, I say this since with a complex project and a CPU meter running you often see the CPU hit 100% and stay there while the IDE's intellisense parser is running when you first load the IDE.
The "best" solution to this is often to load the IDE and the solution, and then go make a coffee while the IDE hammers the machine into the ground.
To delay VA you may find that using the menu item:
VAssistX -> Enable/Disable Visual Assist X
is enough to help.
The problem with directly delaying VA is that VA takes longer to start working. I know this sounds obvious, but you will not get any enhanced syntax highlighting, etc, until VA has done the parsing it needs to do.
I wish I had a better answer for you, but this is why so many of our users disable the IDE's intellisense and never look back. |
| rogojin |
Posted - Jul 31 2007 : 4:02:26 PM
Hi,
I'm getting those same issues again- this time it is the startup thing. I've just tried disabling the IDE's intellisense, as you describe, and that does alleviate the problem a lot. I'm guessing it's the two fighting for the hard disk that really kills performance. So in conclusion.. all I can suggest is:
- Put some kind of control on the menu or something that allows the user to temporarily disables the Tomato's parser. That way one can explicitly let the IDE get it's business out of the way.
- Investigate the reasons why the disk access patterns are causing so much thrashing/flushing. If feasible, maybe you could reduce the number of files touched, or warm up the OS file cache to all the files in a directory that you are going to parse..I dunno..
But thanks for the comprehensive reply. You are right about the Intellisense- maybe I should just get myself a raptor. |
| feline |
Posted - Jul 23 2007 : 1:49:41 PM
quote:
Originally posted by rogojin
Please ignore my ranting and raving. The problem seems to disappear a couple minutes after booting up, syncing, and leaving the IDE alone.
The joys of cross posting  I replied before seeing this.
If you run a CPU meter or watch the IDE status bar is there a lot of parsing / scanning going on and listed on the IDE status bar, and a corresponding CPU spike?
Both the IDE and VA will want to parse things when you first load the IDE and open a solution, so either of them could be causing a lot of CPU activity.
In this specific situation it is best to let the parsing finish before getting on with your work. However you should not run into the same situation once the IDE has been running for half an hour or longer. |
| feline |
Posted - Jul 23 2007 : 1:41:02 PM
Here is the previous thread:
http://forum.wholetomato.com/forum/topic.asp?TOPIC_ID=6441
One thing really stands out for me, which is your comment "But particularly if I modify a source file that is referenced by a lot of others (around 200 others), my system comes crashing down"
I would not expect this behaviour from VA, but this is exactly the behaviour I have seen, and been frustrated by, with Visual Studio 2005 and its C++ intellisense parser. For me when this happened I could not even move the caret in the IDE, often for a couple of minutes at a time, and this would happen several times a day. Editing header files that were included by many source files was always a good way to trigger it for me.
Can you try disabling the IDE's intellisense scanner as described here:
http://docs.wholetomato.com?W133
and see if this makes any difference? You should know fairly quickly if this helps. |
| rogojin |
Posted - Jul 23 2007 : 1:38:17 PM
Please ignore my ranting and raving. The problem seems to disappear a couple minutes after booting up, syncing, and leaving the IDE alone. |
|
|
