3.0 Overall Performance

saki2048

Bit poster
Hi,

is it just me or has the performance degraded severely compared against the 2.5 version, especially when it comes to disk performance? :confused:

Comparing against the same copy of a virtual machine (prior updating to 3.0) on my MBP I'd say that 3.0 has lost approx 30% compared between the versions.

TO test this I've used the same external FireWire Disk on both machines and duplicated the local subdirectory containing nearly 2GB data.

brgds
Sascha
 
3.0 is a huge disappointment across the board. It seems less stable and slower than 2.5, not to mention all the broken "features"
 
I stopped playing with Parallels when I started using vista with bootcamp natively....

I installed ver 3.0 and didnt enable direct x and was pretty disappointed... then I found out about direct x and enabled it and my experience seems better than 2.5...

No I havent tried any games.... and I'm a bit miffed that aero doesnt work yet but overall its not too bad....

but OTOH, I'm typing this on another system having enabled beryl/compiz in ubuntu and I'm thinking that in a short time I wont be using vista anyways....

so maybe 3d acceleration in parallels for me will be a moot point in the near future.
Beryl/compiz support... now yummy!
 
Hm, Bootcamp never was an option as I do most of my stuff under Mac OS X and want to keep it that way. If I wanted to have another windows Laptop there had been easier and cheaper options :/

For Parallels I do not really care about Games or EyeCandy for Vista (don't even bought it yet). Actually I don't even like Aero as it looks ugly to me - even if it works I would switch it off like I did with "Candy-Windows" for XP, too.

For my Job I need MS Visual Studio as our GUI is written for Win32 in C++ - and that's very disk intensive so the degration there hits me more than everything else :(

Unless the next updates I think that I'll stick with 2.5 though and hope for the best ;(

The performance of the version reminds me of the VMWare Beta when the Debug code wasn't switchable or still switched on. Maybe someone just forgot to get rid of it...

brgds
Sascha
 
Just to chime in - Idle CPU usage and memory usage in 3.0 is up over 2.5.

Same VM, same processes.

3.0 - 20% - 30% usage CPU idle.
2.5 - 7% - 15% usage CPU idle.

168M VM used to take 225-240M in non coherence mode and 250M in coherence, now takes 276M in both modes.
 
Last edited:
My experience in terms of performance has been nothing but positive since upgrading to 3.0. Startup, for one, is much faster. Also, the general responsiveness of applications (both Windows and Mac OS X apps running alongside one another) seems much better to me in this version.

Maybe I am just lucky...
 
I've had no problems with the upgrade or performance, but 3.0 is a little porkier in the resource requirements department vs 2.5.

My problem is that I am limited to 2GB of RAM in my MBP and i normally run at least a dozen apps before parallels is started. OS X does not like to swap at all.

Maybe I should just go get one of the new 4GB MBPs.
 
Shidayu said:
My experience in terms of performance has been nothing but positive since upgrading to 3.0. Startup, for one, is much faster. Also, the general responsiveness of applications (both Windows and Mac OS X apps running alongside one another) seems much better to me in this version.

Maybe I am just lucky...


You are not the only one lucky.

I have been using version 3 and its betas for the last 3 weeks and it has been going very well for me.

My only problem left is an issue when using multi monitors of different resolution in coherence mode. Then parallel has a tendency to cut off the window on my higher resolution screen to match my laptop resolution.
 
saki2048 said:
For my Job I need MS Visual Studio as our GUI is written for Win32 in C++ - and that's very disk intensive so the degration there hits me more than everything else :(

brgds
Sascha


I use Visual Studio every day under Parallels. But I use a fixed size VM (not expanding) and I used iDefrag to make sure it was contiguous. It's still not native performance but it's very acceptable (at least for me.)

Plus, stop compiling soo much. :D j/k
 
Shidayu said:
My experience in terms of performance has been nothing but positive since upgrading to 3.0. Startup, for one, is much faster. Also, the general responsiveness of applications (both Windows and Mac OS X apps running alongside one another) seems much better to me in this version.

Maybe I am just lucky...
We're two, then.
I share your experience since installation of 3.0 12 hours ago. It is especially pleasant to see how quick double clicking my .XLS files in Mac open up in Excel 2007 in the preloaded VM. I also use application which create PDF files on the fly and pass them up to the default viewer to show them. I just changed file association of PDF files in XP VM to Mac Aperçu (how have they named 'Aperçu' in english??) and voilà! PDF output comes up on the screen quicker than through the bulky Adobe Reader. Only Foxit Reader comes close.
 
simplicity said:
I use Visual Studio every day under Parallels. But I use a fixed size VM (not expanding) and I used iDefrag to make sure it was contiguous. It's still not native performance but it's very acceptable (at least for me.)

Plus, stop compiling soo much. :D j/k

I'm doing pretty much the same - but by just installing and upgrading to 3.0 I've lost a lot of performance. That's all I said. Obviosly the results differ as some people above claim that it got faster than 2.5.

I'll give it tomorrow one more shot on my iMac, but up to now the update did not fare so well for me.

brgds
Sascha

PS: I'm also quite happy with the 2.5er performance for compilation, even if I compile a lot ;) - but not with 3.0 yet...
 
I haven't noticed any degradation in performance with 3.0 It seems pretty much the same to me. I'm using a MacBook Pro with 2GB memory.
 
saki2048 said:
I'm doing pretty much the same - but by just installing and upgrading to 3.0 I've lost a lot of performance. That's all I said. Obviosly the results differ as some people above claim that it got faster than 2.5.

I'll give it tomorrow one more shot on my iMac, but up to now the update did not fare so well for me.

brgds
Sascha

PS: I'm also quite happy with the 2.5er performance for compilation, even if I compile a lot ;) - but not with 3.0 yet...

Yes, after doing some quick checks myself it does appear 3.0 eats alot more CPU. When my VM is 99% idle, Parallels is chewing up 50%. It used to chew up maybe 8-9%....
 
simplicity said:
Yes, after doing some quick checks myself it does appear 3.0 eats alot more CPU. When my VM is 99% idle, Parallels is chewing up 50%. It used to chew up maybe 8-9%....

Glad it's not only me. You'll also notice memory usage is up by 20M-30M over 2.5
 
simplicity said:
Yes, after doing some quick checks myself it does appear 3.0 eats alot more CPU. When my VM is 99% idle, Parallels is chewing up 50%. It used to chew up maybe 8-9%....

What kind of numbers do you see if you run top in a terminal?
 
simplicity said:
Yes, after doing some quick checks myself it does appear 3.0 eats alot more CPU. When my VM is 99% idle, Parallels is chewing up 50%. It used to chew up maybe 8-9%....

Can't confirm that - the load has remained the same for me, so still my main concern is the disk performance.

Those divergent results are, um, interesting :eek:

I think I'll give it again a try when the first update is out...

brgds
Sascha
 
Count me as being happy with this version however I'm also suffering for inflated CPU in windows. This is being caused by ParallelsToolsCenter.exe, it seems OK when booting and logging on but once I started task manager CPU would go to 100% in the VM and as a results 50% in OS X. My temp solution was to remove ParallelsTools from startup and hope the next updated tools will work better. Once done performance seems better than 2.5
 
I pre-ordered 3.0 based on my excellent experience with 2.5 however 3 is definately a bit wonky, no really serious issues but my XP vm does take much longer to start, cpu usage in mac os x while 99% idle in xp is 30% on my macbook pro 2.33 with 3 gigs of ram which is much higher than 2.5 (it used to sit around the 5 to 10% mark)

Its perfectly usable tho and im sure, just like the previous releases, parallels will get a handle on these issues and fix them.
 
Last edited:
Back
Top