0.2a performance
Moderator: Bob L
-
DominicPerry
- Posts: 3049
- Joined: Mon Dec 19, 2005 12:00 pm
- Has thanked: 1 time
- Been thanked: 1 time
Re: 0.2a performance
I am curious... if you are dropping buffers... do you get latency drift after an hour or so... or are the buffers being corrected properly?
And... try using the RealTime Priority option... does that improve things... otherwise... try 2 or 3 buffers and see if you can stop the loss.
At least now... you can read them... and they should be correctede automatically... but I would still suggest running the system in a mode where there are zero dropped buffers after an hour or so.
Bob L
And... try using the RealTime Priority option... does that improve things... otherwise... try 2 or 3 buffers and see if you can stop the loss.
At least now... you can read them... and they should be correctede automatically... but I would still suggest running the system in a mode where there are zero dropped buffers after an hour or so.
Bob L
Re: 0.2a performance
Just ran a quick two hour test tonight with my (now standard) test setup.
SAC with 2x64, 32 loaded channels, while recording 32 channels in SAW.
0 slipped buffers.
SAC MT load at 62-65%
CPU running rather high between 80-90% (with Task Manager open), but I've got 22 processes running as well.
The only time I got dropped buffers was when I opened Task Manager, but running SAC in RealTime priority solved that problem.
Putting Saw into record or record ready mode causes SAC to jump about 10% higher. That's with Saw using mme and Sac using asio.
DanT
SAC with 2x64, 32 loaded channels, while recording 32 channels in SAW.
0 slipped buffers.
SAC MT load at 62-65%
CPU running rather high between 80-90% (with Task Manager open), but I've got 22 processes running as well.
The only time I got dropped buffers was when I opened Task Manager, but running SAC in RealTime priority solved that problem.
Putting Saw into record or record ready mode causes SAC to jump about 10% higher. That's with Saw using mme and Sac using asio.
DanT
Re: 0.2a performance
That's because MME loads the cpu heavily in the background... outside of SAC and SAW... therefore with the loads that high... you are cutting into the SAC processing time.
For every open MME device, the system starts another thread, or more... and they are all then competing for cpu time.
Bob L
For every open MME device, the system starts another thread, or more... and they are all then competing for cpu time.
Bob L
-
Iain Westland
- Posts: 746
- Joined: Sat Mar 29, 2008 6:36 am
Re: 0.2a performance
still, that will be fixed when the hooks are in place, wont it Bob, eh? hint hint, whats that over there? oh, its a nudge 
thanks for all the good work sir
thanks for all the good work sir
life will bring what life brings, without rain, sun is a curse
Iain Westland info@paddedcellrecords dot co dot uk
Uk 796 then 200 plus 5712
Iain Westland info@paddedcellrecords dot co dot uk
Uk 796 then 200 plus 5712
Re: 0.2a performance
Can't fix the way MME works... no... but you won't be using a driver model in SAW once the hooks are in... hopefully, you will be directly connecting to SAC.
Bob L
Bob L
Re: 0.2a performance
So, does that mean that if you were to have to shut SAC down momentarily that you would interrupt your Saw recording as well?Bob L wrote:Can't fix the way MME works... no... but you won't be using a driver model in SAW once the hooks are in... hopefully, you will be directly connecting to SAC.
Bob L
DanT
Re: 0.2a performance
Well, knowing that, even though the hooks sound like the best idea for most applications, I hope you'll also allow them to stay independent.
If I'm recording and decide that I need to shut down SAC to reconfigure something, or for whatever reason, I'd hate to lose part of my recording as well.
DanT
If I'm recording and decide that I need to shut down SAC to reconfigure something, or for whatever reason, I'd hate to lose part of my recording as well.
DanT
Re: 0.2a performance
Of course the idea is that SAC is the FOH mixer as well as the Monitor Mixers... so shutdown is not an option. 
Of course you can record direct to SAWStudio and monitor right inside SS if that's all you are doing... then there is no reason to run SAC... why strain the machine.
The idea of recording to SAWStudio using the MME and use SAC on ASIO at the same time on the same machine... is not the best concept in the world... while it does work for a demonstration... I do not think that is how I would attempt to capture an important live performance.
Bob L
Of course you can record direct to SAWStudio and monitor right inside SS if that's all you are doing... then there is no reason to run SAC... why strain the machine.
The idea of recording to SAWStudio using the MME and use SAC on ASIO at the same time on the same machine... is not the best concept in the world... while it does work for a demonstration... I do not think that is how I would attempt to capture an important live performance.
Bob L
Re: 0.2a performance
I am currently planning on using SAC as a simple live-to-two-track mixer (as in streaming audio feed, video sound feed, safety to dat, etc.), in my case with no need for ellaborate monitor sends, at least until I have the need to run sound (doesn't happen that often, but who knows with SAC on board?).Bob L wrote:Of course the idea is that SAC is the FOH mixer as well as the Monitor Mixers... so shutdown is not an option.
Of course you can record direct to SAWStudio and monitor right inside SS if that's all you are doing... then there is no reason to run SAC... why strain the machine.
The idea of recording to SAWStudio using the MME and use SAC on ASIO at the same time on the same machine... is not the best concept in the world... while it does work for a demonstration... I do not think that is how I would attempt to capture an important live performance.
Bob L
The advantage I see to using SAC for mixing while recording in SAW is that one does not interfere with the other. If I stop recording, SAC still sends live audio... If I stop SAC, SAW continues recording.
Using SAW by itself for recording and mixing has been troublesome in the past, what with the live audio signal being interrupted for this or that chore that requires stopping SAW, or while setting up a fresh session. The new monitoring options have helped a lot, but not completely.
I'm really enjoing the fact that both of these programs are residing happily side-by-side, doing their own thing and one not relying on the other. I had hoped that you would see the benefits of this even with the hooks you are planning, which is very cool in and of itself! I think it has its place.
Considering what I have described for my needs, what would you suggest then?
DanT
Re: 0.2a performance
Well... you can continue doing what you are doing... or use two separate machines... or... once you start SAC... I see no reason to stop it... the idea is it runs just like a live console... you can access everything you need in SAC without stopping the engine.
Bob L
Bob L
Re: 0.2a performance
I suppose you're right... I imagine if the console goes down there's isn't much to record anyway, is there.
Maybe it's just because right now I'm fiddling with settings a lot while doing both tasks and like the independence.
DanT
Maybe it's just because right now I'm fiddling with settings a lot while doing both tasks and like the independence.
DanT
-
Bruce Callaway
- Posts: 1078
- Joined: Wed Jun 09, 2004 12:00 pm
Re: 0.2a performance
I noticed that I started to get glitches in the sound when SAC had worked fine previously. This could be caused by simply loading task manager and letting it run. Very strange. I thought I must have changed my system in some way that caused this problem. I was testing this on a dual core laptop that had previously run for hours with no problems.
So after a lot of tests, I accidentally discovered this was happening when I set SAC and sawstudio to different CPUs
This had not caused any issues before. So if I now leave SAC/sawstudio set for both CPUs, no problems at all (so far). Is this a result of 0.2a? Is anyone else finding this?
So after a lot of tests, I accidentally discovered this was happening when I set SAC and sawstudio to different CPUs
Re: 0.2a performance
Task Manager will interupt SAC's threads unless SAC is set for RealTime Priority on the Options menu... Task Manager is High Priority which will step on SAC at Normal priority as it polls.
Bob L
Bob L
-
Bruce Callaway
- Posts: 1078
- Joined: Wed Jun 09, 2004 12:00 pm
Re: 0.2a performance
The unexpected item for me was that if I left the CPU affinity alone that is, the XP default setting for SAC and SAW to use both CPUs, SAC worked flawlessly and nothing including task manager caused any problem.Bob L wrote:Task Manager will interupt SAC's threads unless SAC is set for RealTime Priority on the Options menu... Task Manager is High Priority which will step on SAC at Normal priority as it polls.
Bob L
Re: 0.2a performance
Lots of unexpected things can happen in a dual cpu rig... as I said before... its voodoo. 
Try as you may... you will most likely never be able to understand how the cpu's are being distributed and what threads and pieces of threads go where... and when and in what order.... don't even waste your time.... Windows is in charge... not you... or me.
Bob L
Try as you may... you will most likely never be able to understand how the cpu's are being distributed and what threads and pieces of threads go where... and when and in what order.... don't even waste your time.... Windows is in charge... not you... or me.
Bob L
-
Bruce Callaway
- Posts: 1078
- Joined: Wed Jun 09, 2004 12:00 pm
Re: 0.2a performance
Thanks Bob, windows voodooo
Well I hope it stays that way because SAC was rock solid at 3 x 64. I can go lower to 2 x 64 but I am trying to be a little conservative 
So after a month or more of testing SAC/Saw with my new mobile rig (RME Digiface/Presonus Digimax 96K/ADA8000), I am about to put it to good use by providing FOH/monitors (SAC) and recording (Sawstudio) for a conference in Sydney this weekend. This will entail five sessions of approx 2-3 hours per session over Friday night and Saturday.
The main recording task is the conference speakers. I will be providing MP3s of the sessions shortly after they finish to the organisers who will sell them to customers who bring along memory sticks to the conference. On top of that, I will be doing a live band recording on Sunday night.
Should be an exciting weekend for all the right reasons
So after a month or more of testing SAC/Saw with my new mobile rig (RME Digiface/Presonus Digimax 96K/ADA8000), I am about to put it to good use by providing FOH/monitors (SAC) and recording (Sawstudio) for a conference in Sydney this weekend. This will entail five sessions of approx 2-3 hours per session over Friday night and Saturday.
The main recording task is the conference speakers. I will be providing MP3s of the sessions shortly after they finish to the organisers who will sell them to customers who bring along memory sticks to the conference. On top of that, I will be doing a live band recording on Sunday night.
Should be an exciting weekend for all the right reasons
Re: 0.2a performance
I am a little leary of doing the SS recording from the same input devices... although it is working if you have enough cpu power.
Hopefully, the direct hooks will ease the load for that scenario.
Bob L
Hopefully, the direct hooks will ease the load for that scenario.
Bob L
-
Bruce Callaway
- Posts: 1078
- Joined: Wed Jun 09, 2004 12:00 pm
Re: 0.2a performance
I had the same concern/fear however the setup has tested very well so far mixing/recording for up to 3 hours at a time. My laptop uses an Intel T7500 2.2Ghz dual core CPU with 4 Gig RAM. I use very little EQ and comp with no vst plugins and the load on SAC is under 10%. I had SAC running at a "dummy" 20% load for testing purposes and it was all OK.Bob L wrote:I am a little leary of doing the SS recording from the same input devices... although it is working if you have enough cpu power.
Hopefully, the direct hooks will ease the load for that scenario.
Bob L
In case of disaster, I have a backup. There is an existing sound system at the conference venue. The existing mixing desk stays in place and the output from my rig will patch into it. If my setup fails to work as expected, we patch the snake connections back into the existing desk and the existing system is back up and running.