KVR Audio

Dewdman42 · Post by **Dewdman42** » Wed Feb 03, 2016 9:41 pm

Machinesworking, I haven't missed the mark on anything. Calm down. Take a deep breath. Communication appears to still be broken...

There are two completely different buffers under discussion now. Which one are you referring to?

The audio card buffer can have the size set by the DAW, but it does not fluctuate from track to track or moment to moment as you seem to be indicating. There is a low level hardware hand off of the data in that buffer from the DAW to the sound card and the size does not vary while playback is happening.

The internal process buffer, is a different buffer, can have a different value, and since the DAW has exclusive access to it, it can be changed however the DAW sees fit, from track to track, moment to moment and any other hair brained optimization scheme the DAW engineers think of.

machinesworking · Post by **machinesworking** » Wed Feb 03, 2016 10:15 pm

Dewdman42 wrote:Machinesworking, I haven't missed the mark on anything. Calm down. Take a deep breath. Communication appears to still be broken...

There are two completely different buffers under discussion now. Which one are you referring to?

The audio card buffer can have the size set by the DAW, but it does not fluctuate from track to track or moment to moment as you seem to be indicating. There is a low level hardware hand off of the data in that buffer from the DAW to the sound card and the size does not vary while playback is happening.

The internal process buffer, is a different buffer, can have a different value, and since the DAW has exclusive access to it, it can be changed however the DAW sees fit, from track to track, moment to moment and any other hair brained optimization scheme the DAW engineers think of.

Again, you just simply cannot admit you weren't following what others were talking about. Pretending that other people are confused isn't helping your case at all. Nobody here ever claimed that a DAW could make an audio card transmit multiple buffer settings. Everyone but you was talking about the DAWs ability to add buffers (latency) to tracks that were unarmed etc.

It's not the end of the world, don't be that archetypical nerd that can't be slightly off, have a slight misunderstanding in a conversation, without flipping out and attempting to deny what everyone else can plainly see.

Dewdman42 · Post by **Dewdman42** » Wed Feb 03, 2016 10:20 pm

oh for sure I have had problems following what you have had to say, but not so much the other people they have all been clear. Do you agree that we agree now?

ObsoleteAcc99 · Post by **ObsoleteAcc99** » Wed Feb 03, 2016 11:09 pm

[DELETED]

Dewdman42 · Post by **Dewdman42** » Wed Feb 03, 2016 11:40 pm

well that is exactly what I was saying a few days ago ThoeM, and made exactly the same argument you're making now; to aMUSEd, who didn't like it. And I hope you are right that MOTU's nextgen pre-rendering is more like freezing. And I also hope if that's what it is that Apple and everyone else will follow the lead and do the same thing in theirs! Unfortunately I am not so sure that is what it actually is...

And I agree, the best of both worlds would be to use automatic and complete freezing....whether they are frozen to memory or frozen to disk...such that The CPU only needs to crunch DSP for any given section of a track once...at the moment or shortly thereafter the input midi or audio has been set in stone, combined with more optimal real time performance when a track is enabled for record. That would be best of both worlds.

Unfortunately in the DP 9.02 tests I did in the past few days, the non real time performance was not even remotely close to freezing the tracks. Its possible the FirstGen PreGen they've had for the past 7-8 years simply didn't work that well, I don't know, but what I can say is that its not even close to frozen track performance. and now they are coming out with NextGen version of theirs...which they say is more efficient. We'll see!

The open question is whether pre-rendered tracks are saved somehow, even if temporarily; whether in memory or on disk...similar to freezing...so that the redundant DSP is eliminated. I originally thought that is what MOTU was coming out with now based on the way it was described, but as we have explored it over the past few days..I'm not so sure that is what it actually is..more like their own take how to process the plugin DSP a little further ahead of time on non-record-enabled tracks...which is pretty much what everyone has been doing in some form or another. And now MOTU is coming out with their NextGen version of that. They say its really efficient. We'll find out soon. I hope its freezing as you say and I said earlier...but until I see that kind of performance or hear them actually state it that way...I'm not sure we can assume that.

machinesworking · Post by **machinesworking** » Thu Feb 04, 2016 12:13 am

Dewdman42 wrote:oh for sure I have had problems following what you have had to say, but not so much the other people they have all been clear. Do you agree that we agree now?

Rhetorical questions are cheap ploys, power grabs that mean nothing.

I'm the thread starter, invested in the subject for that reason. I have no interest in arguing with people about buffer settings etc. especially when they aren't following, but for some godawful reason you do. If something someone says isn't clear to you, ask. Don't go on a five page explanation about how hardware buffer settings work because you think other people don't know etc. It's patronizing.

The original conversation we struck up, I simply said that maybe developers haven't come up with NextGen PreGen until now because it's harder to make that seamless than you think it is. The fact that became a topic of debate is mind blowing if you objectively look at how straightforward that reply is.

Dewdman42 · Post by **Dewdman42** » Thu Feb 04, 2016 12:18 am

I have no interest in arguing with people about buffer settings etc.

Then by all means please stop.

Orbit-50 · Post by **Orbit-50** » Thu Feb 04, 2016 12:30 am

So, Payton Manning is definitely going to need hip replacement surgery. I think that this might be it for his NFL career. Discuss.

machinesworking · Post by **machinesworking** » Thu Feb 04, 2016 12:32 am

Dewdman42 wrote: Unfortunately in the DP 9.02 tests I did in the past few days, the non real time performance was not even remotely close to freezing the tracks. Its possible the FirstGen PreGen they've had for the past 7-8 years simply didn't work that well, I don't know, but what I can say is that its not even close to frozen track performance. and now they are coming out with NextGen version of theirs...which they say is more efficient. We'll see!

{Note, as long as you're in this thread and comment, I will comment back, I have no reason to be petty, I hope you feel the same, please keep it on the subject at hand.}

Tests not to failure are useless, different DAWs report CPU use differently. Every test I've ever done for Logic, DP, and Live has had significantly different results if I did not tax the CPU 100% then back a plug in or two off and note the amount of plug ins etc.

This directly in my experience has been shown to be true of daily use. Live gets 60% of the performance that DP can get in to failure tests, but DP does not look that great with a just a few plug ins running, it has seemingly equal performance to Live's CPU meter.

Live is a great example, at lower track counts running a test that gets to 55% CPU on my Mac Pro it will read 60% on someone's Mac Mini, but if the test is done to failure (then one back for audio quality etc.), the Mac Pro gets 6 times as many plug ins. Live reports CPU in an arbitrary way, (or a curve instead of a straight percentage) so does DP.

Run a test with and without Prerendering to failure, and compare those results to Logic if you're curious. Although I believe that the Logic VS DP with pre rendering on result will be the same, I would bet that without pre rendering on Logic beats DP easily.

Dewdman42 · Post by **Dewdman42** » Thu Feb 04, 2016 12:43 am

I agree about a max out test being useful. I started to do one today but it was too time consuming..and I don't have time really to do it. And that would in fact be very interesting to compare against Logic.

But the statement you quoted above by me is still valid and yes I think you're being petty, but whatever.

Even with only a few tracks...when you freeze them, the CPU use drops 350%. Pre-gen does not have the same effect. Therefore we can easily see very clearly that FirstGenPreGen does NOT accomplish the same performance gain as freezing tracks..not even close by orders of magnitude. My system could handle about 35 tracks non frozen instruments today before coughing and sputtering... maybe a little less actually. On the other hand I know I can do many many more frozen tracks then that.

machinesworking · Post by **machinesworking** » Thu Feb 04, 2016 1:59 am

Dewdman42 wrote:I agree about a max out test being useful. I started to do one today but it was too time consuming..and I don't have time really to do it. And that would in fact be very interesting to compare against Logic.

I bellieve max out tests are the only truly useful tests for the reasons I mentioned. A straightforward reason to think this is dead easy to test yourself. With DP's transport running on your computer compare CPU use between DP and OSX's built in CPU meters. You would think that DPs built in CPU read out would be lower right? considering OSX is going to add in the rest of the processes running on your computer etc., but it's much much higher. Right now DP is running at 80% CPU with 25 heavy plug ins loaded (Maschine, Altiverbs etc.), but Activity Monitor is showing only 15% CPU use across only three out of 12 cores and a possible 24 threads. 305% of a possible 1200% <Activity Monitor really needs to stop giving only one readout...

Even with only a few tracks...when you freeze them, the CPU use drops 350%. Pre-gen does not have the same effect. Therefore we can easily see very clearly that FirstGenPreGen does NOT accomplish the same performance gain as freezing tracks..not even close by orders of magnitude. My system could handle about 35 tracks non frozen instruments today before coughing and sputtering... maybe a little less actually. On the other hand I know I can do many many more frozen tracks then that.

Sure, pre-rendering as it stands isn't giving great results, but my guess that DP without pre-rendering would be significantly worse, (than your finding of 5% less than Logic) is based on tests I did years ago with DP5, Logic 7, and Live 5. Giving Logic 7 the score of 100%, DP5 (before pre-rendering) came in at that time at 80%, Live at 60%. Again, nothing like what freeze tracks does, and nowhere near what NextGen is promising.

DP7 on up comes in close to Logic these days, when I tested Logic 8 VS DP7 it was pretty much a tie. I think that initially pre-rendering was MOTU's answer to Apple/Emagic's buffer(latency) tricks, but they now seem to have perfected it in the coming update. Hopefully they have, I'm pretty excited to find out. The 9.02 update is cool, but there are a few bugs, I have one with VSTi MIDI out, and there are video bugs. Still, good to see them push it out there, hopefully that means the NextGen PreGen update might happen before June.

Dewdman42 · Post by **Dewdman42** » Thu Feb 04, 2016 4:21 am

The midi out bug is concerning. You must be the dude from Motunation that was complaining about that with maschine? Anyway, that seems like Motu will need a 9.03 rather soon. I plan to test a couple VSTi's with midi out soon to see if I have any of those kinds of problems, just no time this week.

machinesworking · Post by **machinesworking** » Thu Feb 04, 2016 4:59 am

Dewdman42 wrote:The midi out bug is concerning. You must be the dude from Motunation that was complaining about that with maschine? Anyway, that seems like Motu will need a 9.03 rather soon. I plan to test a couple VSTi's with midi out soon to see if I have any of those kinds of problems, just no time this week.

Yeah Maschine and any VSTi that can send MIDI to the host seems to have ghost inputs. The problem is there with Reaktor and Geist ads well. They updated the 9.02 download, Mr. Miller from MOTU said it fixed the video bugs, no mention of VSTi in bugs. Downloading right now.

IMO you should always rename the damn download, at least 9.021 or something?

Dewdman42 · Post by **Dewdman42** » Thu Feb 04, 2016 5:04 am

I tested and have same problem with Jamstix. Submitted a support ticket just now. Hopefully they will fix it quick like they did the video bug.

incubus · Post by **incubus** » Thu Feb 04, 2016 7:05 am

Too much tit for tat with this.

Let's try to make the most important part for anyone and everyone. They are working on trying to make the host better suited for TODAY'S musician(s).

Can we at least agree on that?

DP9 set to get an update, claiming 4X CPU performance...