OSX Users - Performance advantages of Lion/Mountain Lion over Snow Leopard ?

[Dysanfel]

Performance is the same for me.

[TAFKAT]

Bump.

[ksandvik]

Might even be that any newer drivers might not work well in SnowLeopard -- there's just so much time QA people could test across multiple OS versions... Plus you most likely won't get reported bugs fixed in case you report these as SnowLeopard ones.

[T_R_S]

Since Lion has already been out for a year I have bought new software that does not even work on S.L.
Buy a new system drive and try the new OS out first before committing.
I always do that first as there are always so many variables.

[TAFKAT]

The install will be on a new HD , I do not plan on touching my working SL install, just needing to get some feedback whether there has been any noticeable performance improvements on the Lion variants over Snow Leopard

Thanks

[greyskull]

My production machine (mac pro 2008) is still on 10.6.8. Running PT 9.05

Lion made my c2duo laptop miles slower in day to day operation but I've yet to try taxing tools on it.

[ksandvik]

Originally Posted by TAFKAT

The install will be on a new HD , I do not plan on touching my working SL install, just needing to get some feedback whether there has been any noticeable performance improvements on the Lion variants over Snow Leopard

Thanks

Yes there will be as OS and drivers are work-in-progress.

[emeline-rec]

I stupidly installed lion and subsequently MT Lion (with no improvement) on my 2011 MBP and Logic has become very buggy and slow compared to snow leopard. I've noticed pops and clicks recorded in my audio even with high buffer settings. Latency appears higher also.

Could either be Lion/MT Lion, The new lion drivers for my interface, logic 9.1.7 update.

Suffice to say tonight I've gone back to Snow Leopard and the older drivers on a dedicated recording partition, hopefully this will sort me out.

As has been said, create a new partition on your system drive and throw MT lion on it and see how you get on. It's full of great features for general computing and icloud integration, but for recording snow leopard feels more stripped down and lean to me.

[zephonic]

I didn't notice much performance difference between 10.6.8 and 10.7.4

[dr_Jezz]

Originally Posted by emeline-rec

I stupidly installed lion and subsequently MT Lion (with no improvement) on my 2011 MBP and Logic has become very buggy and slow compared to snow leopard. I've noticed pops and clicks recorded in my audio even with high buffer settings. Latency appears higher also.

Could either be Lion/MT Lion, The new lion drivers for my interface, logic 9.1.7 update.

Suffice to say tonight I've gone back to Snow Leopard and the older drivers on a dedicated recording partition, hopefully this will sort me out.

As has been said, create a new partition on your system drive and throw MT lion on it and see how you get on. It's full of great features for general computing and icloud integration, but for recording snow leopard feels more stripped down and lean to me.

In a very similar way, I upgraded to SL and went back to 10.5.8 - nothing I can't live without compared to my current (yet relatively old) reliable set up, with SL on a separate drive just to use the Apple Store if and when I want to purchase anything. There are so many of us (silently) happy with our systems in the way that everyone agrees you should if you're happy and want a "hardware equivalent" tool. I've even just upgraded my old Titanium Powerbook with max RAM and a new 7200 drive to bring it up to date for running 10.5.8 for surfing and very basic stuff, and you know what? It works great for that. I've got the feeling that OS upgrades nowadays are more to 'outdate' old systems and software than to offer much advantage - it's a point that's been aired many times, but is no less relevant for it. IMHO etc.

[basmartin]

ML is a little bit snappier on my system than Lion was, boot time is about 20 sec faster, no major issues yet.

[TAFKAT]

Thanks Guys,

Keep em coming.. :-)

[Tui]

Originally Posted by dr_Jezz

I've got the feeling that OS upgrades nowadays are more to 'outdate' old systems and software than to offer much advantage - it's a point that's been aired many times, but is no less relevant for it. IMHO etc.

Yup, it seems that way. For the life of me, I cannot understand why Apple deemed it necessary to break Rosetta support, for example. There are some lovely older programmes that don't have a modern equivalent.

[inthere]

Boot time Lion-17.5 seconds
Boot time Mountain Lion-14.2 seconds

This is with an OWC Mercury Extreme 480 GB SSD in a 2011 MBP 2.2 ghz w/16GB RAM.

Logic multicore benchmark test Lion-66 tracks
Logic multicore benchmark test Mountain Lion-72 tracks

Programs and plugins seem to be launching noticeably faster but I have no way of measuring because
they were launching almost instantaneous anyway both before and after Mountain Lion because of the SSD. Safari is a LOT faster and almost useable now. Still only half as fast as Chrome at best but at least it's not the anchor it was in Snow Leopard and Lion. Tabs are weird and take some getting used to, and there are some flat-out dumb workflow changes (scrolling tabs???)

ALL programs and plugins are working. I had Preview freeze up on me the 1st day on a PDF file, but I can't reproduce the freeze. Metric Halo ULN-8 soundcard working perfectly and without a hitch, as well as the Focusrite VRM Box. All Thunderbolt peripherals working fine(Hard drives, monitor)

[jrhager84]

It's a shame my old ass Mac pro can't handle mountain lion...

Sent from my SAMSUNG-SGH-I777 using Gearslutz App

[inthere]

*Also*

My install was not a clean install, I just put it on over Lion. Didn't repair or verify permissions, nothing. Just installed and went on with my business. Total installation time was about 5 minutes after download.

[TAFKAT]

Originally Posted by inthere

This is with an OWC Mercury Extreme 480 GB SSD in a 2011 MBP 2.2 ghz w/16GB RAM.

Logic multicore benchmark test Lion-66 tracks
Logic multicore benchmark test Mountain Lion-72 tracks

Any idea what the result would be in SL ?

[manowar84]

same performances between snow leopard and moutain lion with logic for me

[RyanC]

Seems about the same here...

I was going to say it would be good to refresh these a bit, one thing I would be interested in would be your results on a cross platform d-verb test, or even opening up some different plugs that the standard dawbench stuff and including some aax in the mix.

Mainly because on my machine on a d-verb test OSX absolutely crushes w7. At 1024 I got 760 dverbs in osx and 450 on w7 and the results continued to scale like that at low latencies. I'm out of the loop on w7 optimizations etc, but it left me with a lot of questions. Not doubting your methods but I was wondering about the plugins used in dawbench, IE is it possible they are just coded better for windows more so than windows being more efficient across the board?

[inthere]

Originally Posted by TAFKAT

Any idea what the result would be in SL ?

No idea, sorry

[Tui]

Originally Posted by manowar84

same performances between snow leopard and moutain lion with logic for me

If I may ask, on what kind of machine?

[TAFKAT]

Originally Posted by RyanC

Seems about the same here...

I was going to say it would be good to refresh these a bit, one thing I would be interested in would be your results on a cross platform d-verb test, or even opening up some different plugs that the standard dawbench stuff and including some aax in the mix.

Mainly because on my machine on a d-verb test OSX absolutely crushes w7. At 1024 I got 760 dverbs in osx and 450 on w7 and the results continued to scale like that at low latencies. I'm out of the loop on w7 optimizations etc, but it left me with a lot of questions. Not doubting your methods but I was wondering about the plugins used in dawbench, IE is it possible they are just coded better for windows more so than windows being more efficient across the board?

Hey Ryan,

I have a whole range of 3rd party plugins being used in the sessions as that is a fairer comparative when using other DAW hosts, DVerb for the most part was near useless as I could tap the 320 instances too easily and of course the question of code optimisation cross platform raises its head, but its no different for DVerb IMO.

I never tested it on OSX , so I can't really comment on your results.

I have added some new higher resource plugins recently from Softube and will be adding more again from Slate and Waves.

That gives us a very wide pallet to draw from.

If the results are consistent across all the 3rd party plugins, either all the developers are equally incompetent at OSX/Win coding, or there are other variables at play.. :-)

You can download the sessions and have a play yourself, I'll be interested in your results.

Sessions available Here

If you check out Part IV / V of the reports you will see that there wasn't a huge variable with PT9 on OSX/Win anyway, which raised my suspicions about hidden buffering but I had no evidence to support that theory at the time.

BTW: What version of Protools are you referring to, what audio hardware, also what system are you testing on, are you using BootCamp or OSx86 ?

In regards to the results scaling at the lower latencies in Protools, well that raises some interesting points. PT9 had some very odd behaviour under ASIO with some 3rd party plugins where performance took a U turn at 128 samples , performance was not even matched at 256 to what I achieved at 032.

There was obviously some added buffering at the lower latencies being employed but the process did not remain consistent at the moderate to higher latencies. PT10 brings even more variables into play as AVID quietly introduced a hybrid playback engine which has a fixed playback buffer of 1024 on Windows, I have no idea what it is on OSX as I haven't explored that as yet, there is no reason that the fixed playback buffer isn't higher on OSX to compensate for the inefficiencies imposed by Core Audio. What that means is that in PT10 on Win7 , I achieve identical results at 032 all the way to 256 across all the 3rd party plugins I tested ( I haven't released those results as yet as I need to do a stack more testing ). The question then is what happens to the hybrid buffering when input monitoring /track arming ala Logic , as you can't play VI's in real time or input monitor via a guitar amp sim at 1024 , for example.

Lots to explore in the coming months/years.. :-)

My main motivation for the thread is to be clear that what ever OSX version I will be testing on next is the best performing, as I don't want the peanut gallery accusing me of try to skew the results. If I stick to 10.6.8 then I'll have the complaints about it being an older version of the O.S , if I use Lion/Mountain Lion and its performs worse than Snow leopard, I will get the complaints about not using the best performing OSX version, so I need to be clear... :-)

At the moment I am leaning towards Mountain Lion as I have consulted a few trusted colleagues experienced with OSX DAW performance, and they have all said to steer clear of Lion.

Originally Posted by inthere

No idea, sorry

No problem

Cheers

Keep the experience coming guys, all greatly appreciated.

[RyanC]

Intersting...I'm running HD|N with a lynx aurora, osx86 (tonymac) on a 3930k @4.6 PT10.2. I also have a 2nd 3930k as a VE slave.

I know you thought you saw evidence of the hybrid buffer via VEP but I'm not seeing that here unless I'm not looking in the right place. I've been sorta on the fence about going back to win at some point but I really can't until PT goes 64bit mostly for omnisphere mem server. Also on HDN the amount of plugs certainly does go down as the buffer goes down on both w7 and osx. I actually wish PT did have a hybrid buffer or there was a way to turn that on, etc...

I'm very interested in the new round of tests and would be glad to run them on my machines when I get a minute.

[mattsearles]

one thing I have heard is moving from SL to Lion... is it's break your soft raid.. which has me a little nervous about it

[TAFKAT]

Originally Posted by RyanC

Also on HDN the amount of plugs certainly does go down as the buffer goes down on both w7 and osx. I actually wish PT did have a hybrid buffer or there was a way to turn that on, etc...

O.K , I wonder if its only an ASIO/Core Audio thing with the hybrid buffer.

Its definitely a fixed buffer using the RME HDSPe card on Win7, the 1024 value was reported by VEP no matter what buffer setting, whereas it changes in VEP when using Cubase according to the set buffer. It made sense to me when I saw the results across all the plugins.

This is a preliminary worksheet with my X58/i7 920 Test/Report rig , I am planning on expanding the testing to my X79/Xeon 1650 dev system in future as well, but you will get the idea.



Apart from a few plugin instances @ 032 for the CSP and EMP , it pretty obvious we are dealing with a fixed buffer.

Can you confirm that VEP reports a different buffer value that corresponds to the buffer set on the HD Native Card ?

O.T - So you have OSx86 running on X79 which Apple does not natively support..., hmmmm, interesting, but thats a whole other discussion and not one I will have online... ;-)

I digress..

[RyanC]

Originally Posted by TAFKAT

O.K , I wonder if its only an ASIO/Core Audio thing with the hybrid buffer.

Its definitely a fixed buffer using the RME HDSPe card on Win7, the 1024 value was reported by VEP no matter what buffer setting, whereas it changes in VEP when using Cubase according to the set buffer. It made sense to me when I saw the results across all the plugins.

This is a preliminary worksheet with my X58/i7 920 Test/Report rig , I am planning on expanding the testing to my X79/Xeon 1650 dev system in future as well, but you will get the idea.



Apart from a few plugin instances @ 032 for the CSP and EMP , it pretty obvious we are dealing with a fixed buffer.

Can you confirm that VEP reports a different buffer value that corresponds to the buffer set on the HD Native Card ?

O.T - So you have OSx86 running on X79 which Apple does not natively support..., hmmmm, interesting, but thats a whole other discussion and not one I will have online... ;-)

I digress..

Strange...I'm new to VEP so where do I check for that? I see delay comp change in PT. The number in the bottom corner of the VE server app changes. Even routing through VEP with the send changes all the delay comp in PT.

OT, fwiw osx screams on x79. Full vanilla installs on both giga ud5 and asus p9x79 (no dsdt's). No sleep or power management but those things always seem to cause problems for audio IME. Either way they are much more efficient than my tape machine...or a toaster. I do have an ac controlled machine room. I have been very happy with this setup.

Some video guys have dual e5 hacks running...doesn't seem worth the cost to me for now, especially without eist/turbo, but interesting nonetheless.

[manowar84]

Originally Posted by Tui

If I may ask, on what kind of machine?

macbook pro 15" ,i7 2.0ghz (4+4 core), 8gb ram, 7200rpm hd

[TAFKAT]

Originally Posted by RyanC

Strange...I'm new to VEP so where do I check for that? I see delay comp change in PT. The number in the bottom corner of the VE server app changes. Even routing through VEP with the send changes all the delay comp in PT.

I need to get in front of VEP/Protools when I get some clear air to confirm that for you, I only use it for testing/benching with PT , but I do remember the value in the VEP maintaining 1024 no matter what buffer was set via the RME. The results were also identical on the DAWbench VI - Kontakt 4 bench no matter what buffer was set , so its clearly a fixed buffer.

I also remember some major problems when adjusting the added buffering in VEP, with Cubase I could set it to zero , with PT anything below default (2) would send it into spasms. PT still has a long way to go with VI implementation, no matter what work arounds are employed IMO.

I'll drop back with some more info and some screenies when I can get to it.

BTW: Unless I missed it, you didn't mention what OSX version you are running.

I Digress,

Back to OSX performance variances ( if any ) from Snow Leopard to Mountain Lion

[Tui]

Originally Posted by manowar84

macbook pro 15" ,i7 2.0ghz (4+4 core), 8gb ram, 7200rpm hd

Cheers. Which OS was pre-installed on your MBP? 10.6? It seems machines that were built before Lion run better under Snow Leopard.