What's up with multi core really poor performance under 128 buffer?

[TNM]

I am really shocked about this.. I tested S1/Cubase/Logic in the same scenario, and all pass. S1 can do 16 or 32 samples no problem, Cubase 32, Logic 32.

In pro tools, if i put my buffer under 128 and arm just a SINGLE instrument track, ALL cores get used like crazy.

You have to see this to believe it.. Is this why some people say to use VE pro for virtual instrument?

The thing is, at 128 buffer, Pro tools is performing better than the other DAWs.. it can play through stuff the others literally can't (same midi file and synth preset).. and with pro tools, I am even using blue cat patchwork to host the vst in this test (padshop pro), so, extra overhead, whereas Cubase is running it's VST3 version natively.

I have tested the buffer issue with 20 virtual instruments today and all the same.. whether xpand, fab filter twin, even boom.. as soon as a 64 or 32 buffer is chosen via core audio and an instrument track is armed so it can be monitored and played via midi keyboard, all cores go mental.

Here is a pic, showing a project with only ONE instrument.. just a brand new project with *nothing* else whatsoever.

Look at the 64 buffer pic vs the 128.

Now, i expect higher usage at 64 buffer, but surely, for ALL cores to become engaged for ONE instrument like this, it's got to be a nasty BUG?

Can anyone on windows or mac confirm their findings, and if so, what interface? This test is confirmed with Apollo Twin, and Mac laptop on board audio. confirmed on both my imac and macbook. Under 128 and pro tools goes crazy.

anyway, pics are attached.. If anyone wants me to make a vid, i can do that too.

Cheers and i hope we can sort this out, so Avid can perhaps fix it. Pro tools is very efficient at 128 buffer and above.. no issues at all, including with VI's.

The reason I need 32 or 64, is because i use my virus control plugin, and the way the virus works, it is minimum double your buffer plus whatever the plug in latency is, to play live via the USB connection and the plugin. I suppose i could switch to 64 only when the virus is armed, then back to 128 for everything else.
I LOVE PRO TOOLS like you have no idea. I just want this app to be the best. I just can't get used to any other DAW after how my brain is so wired to PT now.

PS if you look at my menubar in the pics, you will see istat showing Pro tools engaging all cores heavily when a low buffer is chosen, yet only one core higher than others when 128 is chosen.. This really doesn't make any sense.. Even if 32 is chosen, pro tools should only be using the one live core when an instrument is armed, whichever core it has put it on. All of the cores turning on just has to be a bug from a sensical point of view.

http://duc.avid.com/attachment.php?a...1&d=1514858180
http://duc.avid.com/attachment.php?a...1&d=1514858180

[Darryl Ramm]

What actual problem are you trying to solve? Pro Tools CPU meters are often an aberration. What actual “poor performance” are you getting? AAE 6101 errors?

I doubt almost anybody here could actually describe what those meters measure to start with.


Sent from my iPhone using Tapatalk

[gonsdeejay]

Your iMac Right Now has almost 7 Years old!!! with a 3rd Generation Intel processor, right now if you get a brand new iMac has a 6th generation, and for other side, Why the aberration to Use your DAW under 256?

Today Intel sale i7 8ht Generation!!!! or i9...So if you get that kind of Processor maybe you will work at 128 or lower, I work with my MacBook Pro late 2011, at 512 without Issues but on my Custom PC with 7th Generation i7 windows 10. I work great at 256, and Mix my 80- 90 tracks at 512 with high performance!!!!

Cheers

[TNM]

Originally Posted by gonsdeejay

Your iMac Right Now has almost 7 Years old!!! with a 3rd Generation Intel processor, right now if you get a brand new iMac has a 6th generation, and for other side, Why the aberration to Use your DAW under 256?

Today Intel sale i7 8ht Generation!!!! or i9...So if you get that kind of Processor maybe you will work at 128 or lower, I work with my MacBook Pro late 2011, at 512 without Issues but on my Custom PC with 7th Generation i7 windows 10. I work great at 256, and Mix my 80- 90 tracks at 512 with high performance!!!!

Cheers

the macbook pro is only 2 years old. The chip is actually been only superceded by one model, and until last year the very top 2,8ghz that was used, was exactly the one i have. The pics are the macbook pro.

are you saying you have a mac, use a core audio interface (not hd), and at 64 and 32 samples, you can use a VI and all cores are not engaged when you put the track into low latency monitor mode? That's great, please show me, so i can present it to Avid as a bug on my machine, to see if they can help me.

[TNM]

Originally Posted by Darryl Ramm

What actual problem are you trying to solve? Pro Tools CPU meters are often an aberration. What actual “poor performance” are you getting? AAE 6101 errors?

I doubt almost anybody here could actually describe what those meters measure to start with.


Sent from my iPhone using Tapatalk

do you do anything other than troll?

I clearly said istat meters in my menu bar is showing pro tools is actually engaging all the cores, for one single plugin. I mean i couldn't be any clearer.. Look at the 64 buffer pic, in my menu bar, and see all the 40%.. istat is showing real cpu use like activity monitor.

Yes, if i were to add just one more synth, there'd be dropouts.

If the project were a little busier, i wouldn't be able to monitor any synth at 64 buffer, let alone 32.

I am trying to show that pro tools engages all cores for one plugin, when the buffer is at a certain level..

So i am trying to show what i believe is a bug.

if your machine is not doing that, please show me, that would be great.

Then i know there is something wrong with both of my machines..

or it could even be a bug in the way pro tools is handling core audio.

Why not just stop answering in my topics, it's like you don't have a life and just have to jump in immediately to attack and belittle people. You ain't all that smart Darryl, get over yourself. You are going back on mute. I mistakenly took you off, what a fool I was.

[Darryl Ramm]

Boom on one mono instrument track, playing back prerecorded MIDI track. 48 kHz sample rate, built-in output playback engine. Runs at 64 sample IO buffer with one core running at ~2%. MacBook Pro 15" 2016. Pro Tools 12.8.3. Yawn.

Not properly optimized, with WiFi left on (like on your computer), I get a AAE 6101 error at 64 samples, but runs stable at 128 samples.

But of course I'd not want a lot of VIs at such a small IO buffer size...

The first thing you should be suspecting are all the other (many?) plugins installed on your system. Done troubleshooting there?

You fill up DUC with all these meandering confused posts. The common issue to all the different problems you have is guess what... *you*. You need to work out how to quietly go troubleshoot you systems before running around yet again claiming Pro Tools is broken...

[TNM]

why the "old cpu" argument is not valid... Same test with same midi file with same synth on SAME computer, with different DAWs

Cpu usage out of a possible 800% (8 threads) - using istat and activity monitor, NOT the daw meters:

Pro tools Playing one VI at 64 samples - 285%
Cubase playing same VI at 64 samples - 75%
Logic Playing same VI at 64 samples - 65%
S1 Playing same VI at 64 samples- 46% (!)

Note, in all cases, as all these DAW's have hybrid buffers, the instrument track was armed and therefore playing at the chosen 64 samples.

At 128 samples:

pro Tools - 55%
Logic - 55%
S1 - 38.8%
Cubase - 80% !!

Note, S1 uses lowest real CPU of all DAWs, Cubase uses highest.. S1 is the absolute first to tip over and crackle and pop, in all scenarios..Therefore, this is proving that S1's audio engine is very inefficient - it uses the least real CPU, but it's audio engine overhead is way worse than the others. Cubase falls over next, pro tools next, Logic last. S1's meter is always maxed out when it starts to spit and pop..so it's a useful meter to follow when using S1.. however, it also once again shows the inefficiency, as it has no correlation to real world CPU usage.

Now this is where it gets weird.. If i engage 4 of these identical instrument tracks in real time in pro tools, the usage stays roughly the same, as in the real cpu usage (as well as pro tools' meter).. This is why i think it's a simple bug in the coding, that is for some reason engaging multiple cores when simply monitoring a single track, when the buffer is lower than 128. Also note, pro tools' performance meter bears the closest real world usage to the mac's real cpu meter. This in itself is very interesting. Whatever it's doing, whether good or bad, the real world cpu meter is mimicking it.

As far as a DAW cpu meter not being relevant, EVERY Daw's internal meter is VERY relevant, because one learns over time, at what level their DAW's meter has to be at for the DAW to tip over. It's all relative. For example, at 128 buffer, i *know* pro tools falls over for me when the main meter is over 90%. At 64, over 60%
At 32, over 50%.
With Logic, if a live track is not armed, i know that all threads have to show completely full before any issues. In fact, they often show completely full and logic will keep going pop and click free. Therefore Logic's meter is inaccurate and always has been.

Overall, DAW meters are very useful indeed - what DAW meters measure is available audio buffer time, not cpu time.. and the meters fill up the more effects and instruments used.

I have discovered 2 serious bugs in pro tools the past couple weeks, and I feel it's important to bring these things out in the open. Perhaps it's just that others using core audio are not using a 64 sample buffer. I would presume, that HD native uses a proprietary integration that doesn't have this issue.

Because i use apollos, i have no need to monitor audio tracks at a low buffer in pro tools itself. However, i do like to be able to play my VI's at a low buffer which is how i discovered this issue.

The way pro tools is managing cores when the latency is below 128, is very inefficient and puts much unnecessary strain on the cpu.. As well as causes fans to go nuts. Why does it need to engage 7 threads out of a possible 8, to play an instrument that is actually only designated one thread on playback?

By design, the way DAWs work, they can not split single effect/instrument instances over multiple cores. Usually, it's one track per core till the 8 or 16 or whatever your computer has are used, then it cycles over and so on and so on. This is why, Pro tools activating all the cores to use just one instrument track doesn't make any coding sense at all.

I'd just love to hear from someone else using core audio or asio, at 64 and 32 buffer, whether pro tools engages all the cores for one live VI, like it does here.
Not smart ass responses about my computer (again it's all relative.. if it's a bug, it will do the same on a 16 core mega beast, it will just use less overall cpu but all cores should still engage).. not snarky who cares and condescending attacks.. i'm trying to get to the bottom of something here, so i'm only interested in people who are also actually interested in the performance of Pro Tools.

[bartosz idzi]

Hi, i attached my results. I'm on Dual 2,26 5520 Xeon 2009 Mac Pro with RME Hammerfall card under OS X 10.9.5. PT 12.7.1 @32 samples buffer.

[Drew Mazurek]

PT has never been anywhere near as CPU efficient as other DAWs. There are many competing theories as to why, but if this is news to you, you haven't done your research.

[musicman691]

Originally Posted by Drew Mazurek

PT has never been anywhere near as CPU efficient as other DAWs. There are many competing theories as to why, but if this is news to you, you haven't done your research.

And this is one of the reasons a number of people host things inside VEPro.