Funny enough, I have 6.07 on my rig (also 32bit), and it behaves pretty much the same - though the load is about 10-20% lower(!). The video is currently being encoded, and short before uploading.
Take a closer look here:
Cubase 6.07 behavior with 3rd party plugins
Please right-click/download
ca 44MB, mp4, no audio, 1680x1020px
I think I need to contact TwoNotes about that as well.
Though, I encountered that after installing C7 on my rig, and using the most current RME drivers on it as well (utilizing ASIO 2.3!, the same as C7), every installed Cubase version is messed up.
So four scenarios (again):
1) Reinstall
2) contact 3rd party dev to improve things
3) get a new rig
4) Hope that Steinberg get's their act straight, because there is a bug in the ASIO system (Theo and I are not imagining things! even a tech confirmed my findings, even if it's not made public)
I am not imagining things. There is something massively wrong and I can't constantly switch to 1024 samples in order to work somewhat stable and on low load.
Cubase 7.06 is out
-
- KVRAF
- 14739 posts since 19 Oct, 2003 from Berlin, Germany
-
- KVRAF
- 14462 posts since 16 Feb, 2005 from Planet Earth, Somewhere
thanks Compyfox...
Appreciate the videos.
I hope you believe that I am really trying to help.
However I looked at both videos and looked at cpu usage at 256 at certain similar points in each video.
Firstly I think you know that cpu usage although related to asio load is not synonymous with asio load, so maybe I am missing why you showed cpu usage from the task manager in the cubase 6 video when you didn't in the cubase 7 video. Please tell me what I am missing with that if I am missing something.
I of course eyeballed the asio load at various points.
Cubase 6
Idle ~ 20%
Just Before Gtr1 comes in ~35%
Just after Gtr1 comes in ~52%
when Gtr2 starts to play ~70%
At Bar 13 ~70%
At Bar 31 ~55%
Cubase 7 (which the asio load meter is much larger and easier to see)
Idle ~20%
Just before Gtr1 comes in ~35%
Just after Gtr1 comes in ~55%
when Gtr2 starts to play ~65%
At Bar 13 ~68%
At Bar 31 ~55%
so they are roughly the same... where do you get the 10-20% lower figure from? You can't compare the cpu task manager to asio manager directly.
I really don't see the bug you speak about.
I think it was you who have said to me before, that same type of projects in Reaper vs Cubase uses less cpu. I don't use Reaper but I accept that, so maybe Cubase needs to be better optimised if it is at all possible. But I don't' see any bugs comparing the two videos.
sorry
rsp
Appreciate the videos.
I hope you believe that I am really trying to help.
However I looked at both videos and looked at cpu usage at 256 at certain similar points in each video.
Firstly I think you know that cpu usage although related to asio load is not synonymous with asio load, so maybe I am missing why you showed cpu usage from the task manager in the cubase 6 video when you didn't in the cubase 7 video. Please tell me what I am missing with that if I am missing something.
I of course eyeballed the asio load at various points.
Cubase 6
Idle ~ 20%
Just Before Gtr1 comes in ~35%
Just after Gtr1 comes in ~52%
when Gtr2 starts to play ~70%
At Bar 13 ~70%
At Bar 31 ~55%
Cubase 7 (which the asio load meter is much larger and easier to see)
Idle ~20%
Just before Gtr1 comes in ~35%
Just after Gtr1 comes in ~55%
when Gtr2 starts to play ~65%
At Bar 13 ~68%
At Bar 31 ~55%
so they are roughly the same... where do you get the 10-20% lower figure from? You can't compare the cpu task manager to asio manager directly.
I really don't see the bug you speak about.
I think it was you who have said to me before, that same type of projects in Reaper vs Cubase uses less cpu. I don't use Reaper but I accept that, so maybe Cubase needs to be better optimised if it is at all possible. But I don't' see any bugs comparing the two videos.
sorry
rsp
-
- Banned
- 1373 posts since 5 May, 2007 from Finland
Compyfox wrote:This is certainly a bug, because if you look closer, I turn the ASIO guard off, and the behavior is the same.
I raise the buffer to 1024 and the load is unholy low. I raise it even more, and the load is back up again.
Sorry - I don't take it.
I refuse to take it.
It is some weird behavior that should not(!!!) happen. Period!
My rig isn't a damn single core from like 10 years back! It's a friggin iMac 11.3 (i7 870) equivalent!
I'll see what I can do about C6.x. Currently only have 6.0.4 on my rig.
My guess: You have an i7. That could be the problem.
To elaborate more: power saving features of the processor are messing up your stability.
Why does it take more CPU when run at higher buffer values? Because your processor is throttling down as it get's the job done before the time-slice is over. If the time-slices are lower(lower buffer) it won't run into the "idle"-mode power saving feature. Test this by running some cpu-stressing program in the background(with lower priority.. 7zip could be great) and see how Cubase reacts.
Also ASIO Guard makes a huge difference no matter what ASIO buffer value it's run on. The biggest advantage is that the plugins get to run on other processor cores(and ahead of time) so they won't touch the real-time process. This way you can get a lot more out of your processor. Check out my Kontakt test for example:
(and mind you, this test was on a low-end sub-200$ CPU)
-
- KVRAF
- 14462 posts since 16 Feb, 2005 from Planet Earth, Somewhere
I assume the video wasn't to show that asio guard makes a difference as it doesn't compare the asio guard on and off.mkdr wrote:.........
Also ASIO Guard makes a huge difference no matter what ASIO buffer value it's run on. The biggest advantage is that the plugins get to run on other processor cores(and ahead of time) so they won't touch the real-time process. This way you can get a lot more out of your processor. .........
Have you personally seen asio guard make a difference irregardless of asio buffer values?
My credible information is as above.. it only makes a differences at lower values.
rsp
-
- KVRAF
- 14462 posts since 16 Feb, 2005 from Planet Earth, Somewhere
However Compyfox, mkdr hit on something that I just realised on checking your video you aren't using Steinberg's power energy modes.. in your settings..it may be as simple as that.
Please Enable:Modus for ......... below the asio guard.
that on my system makes a huge difference.
rsp
Please Enable:Modus for ......... below the asio guard.
that on my system makes a huge difference.
rsp
sound sculptist
-
- KVRAF
- 14739 posts since 19 Oct, 2003 from Berlin, Germany
I changed in my OS (Power Management) that the CPU is always on full load, so I didn't need the switch - though BIOS wise HT is on, which throttles the Watt usage of my i7 (CoreCPU confirms that). But it's worth a try - can't hurt much more anymore.
Regarding WoS III, I filed a ticked either way. Maybe there is something wrong.
And in terms of the ASIO meter:
Well, C7's one is major funked up and hard to read. I interpret the spikes as 90% in C7, and about 80 to 90% in C6.
CPU wise I do not(!) have any problem. The project with WoS barely scratches 30% and 1,6GB of RAM. I had a test project by Steinberg that ate 4,5GB of RAM, and 39% of CPU power, 24 channels, and that thing crapped out after I copied channels to reach above 32. Even then, my CPU stayed below 40%, but the ASIO system barfed on me.
Theo and I aren't insane - there is something borked with ASIO 2.3. Something that was not there with 2.0 (Cubase 6, ASIO engine/drivers are overwritten with a more recent engine/drivers - no rollback!). And the infamous MMCSS switch doesn't help me, because that's for "Multimedia Applications through ASIO modules".
Since I have an onboard audio module, I run everything multimedia related through there.
Regarding WoS III, I filed a ticked either way. Maybe there is something wrong.
And in terms of the ASIO meter:
Well, C7's one is major funked up and hard to read. I interpret the spikes as 90% in C7, and about 80 to 90% in C6.
CPU wise I do not(!) have any problem. The project with WoS barely scratches 30% and 1,6GB of RAM. I had a test project by Steinberg that ate 4,5GB of RAM, and 39% of CPU power, 24 channels, and that thing crapped out after I copied channels to reach above 32. Even then, my CPU stayed below 40%, but the ASIO system barfed on me.
Theo and I aren't insane - there is something borked with ASIO 2.3. Something that was not there with 2.0 (Cubase 6, ASIO engine/drivers are overwritten with a more recent engine/drivers - no rollback!). And the infamous MMCSS switch doesn't help me, because that's for "Multimedia Applications through ASIO modules".
Since I have an onboard audio module, I run everything multimedia related through there.
-
- Banned
- 1373 posts since 5 May, 2007 from Finland
Well if it's the same as running with a higher latency... Have you noticed how raising the latency affects the performance?zvenx wrote:I assume the video wasn't to show that asio guard makes a difference as it doesn't compare the asio guard on and off.mkdr wrote:.........
Also ASIO Guard makes a huge difference no matter what ASIO buffer value it's run on. The biggest advantage is that the plugins get to run on other processor cores(and ahead of time) so they won't touch the real-time process. This way you can get a lot more out of your processor. .........
Have you personally seen asio guard make a difference irregardless of asio buffer values?
My credible information is as above.. it only makes a differences at lower values.
rsp
And yes. I have seen a difference with and without ASIO Guard. Just haven't made a video of that. Maybe i should.
-
- Banned
- 1373 posts since 5 May, 2007 from Finland
And here we go:mkdr wrote: And yes. I have seen a difference with and without ASIO Guard. Just haven't made a video of that. Maybe i should.
Running on 1024 buffer on RME.
-
- KVRAF
- 14739 posts since 19 Oct, 2003 from Berlin, Germany
zvenx wrote:Please remind me of your soundcard?
rsp
RME HDSPe, Firmware rev20, drivers 3.36 (already ASIO 2.3). According to a Steinberg tech, v3.0x and v3.14 max should NOT introduce such loads. But I couldn't confirm that without a reinstall, since rolling back didn't work.
I also still have the random "ASIO peak" issue on load.
Just got home... will see if I can take a closer look at the Steinberg's power energy modes, though I remember it not working at all for me. Same with the MMCSS switch (as mentioned earlier).
Regarding ASIO Guard having influence on the performance... all good and fine. But what good does that bring me, if the "realtime" readout is low, but the Guard (AVG) shows a very high value. Introduce another channel, and your system comes to a grinding halt.
Been there...
-
- KVRAF
- 14739 posts since 19 Oct, 2003 from Berlin, Germany
Not if Steinberg points to RME in terms of issues, and RME towards Steinberg.
And now, Steinberg confirms an issue on Mac with the Core2ASIO drivers (see here). Dare I even say that I'm not that far off the pond with my issue.
And now, Steinberg confirms an issue on Mac with the Core2ASIO drivers (see here). Dare I even say that I'm not that far off the pond with my issue.
-
- KVRAF
- 14739 posts since 19 Oct, 2003 from Berlin, Germany
Currently, yes. But as I was saying earlier, I think these issues have way deeper roots.