Cubase 10, Buffer Size and Real Time Performance Issue Ryzen Threadripper 1950X
- KVRAF
- 5913 posts since 17 Aug, 2004 from Berlin, Germany
There have been 2 known problems recently. One problem came from Windows, but I think it was fixed in the last update and had something to do with USB.
Then there was a problem with Cubase and very high CPU peaks with VST3 plugins... that was also fixed in the last update.
So: is Windows and Cubase the latest version?
Then there was a problem with Cubase and very high CPU peaks with VST3 plugins... that was also fixed in the last update.
So: is Windows and Cubase the latest version?
| Links-
- KVRAF
- 35685 posts since 11 Apr, 2010 from Germany
Apart from that you are talking about another CPU than the OP has: "Der Ryzen 3700x liegt in Anwendungen ca 15-20% vor dem Intel" - Which Intel CPU exactly? There are Intel CPU's which are much more performant than the 3700X.enroe wrote: Thu Feb 13, 2020 9:22 amMmmhh ...chk071 wrote: Thu Feb 13, 2020 8:43 amNo, it isn't. It's a comment based on long time experience with AMD hardware. If it works for you, fine, it never really worked for me.
It's also well known that AMD aren't known for their excellent single core performance (especially the CPU in question sucks at it, and that's obviously why the OP overclocked it). Most AMD CPU's provide mutli core performance, but lack on single core performance. If you want to deny that, then I can't help you, because, then you deny facts. And, the authority, and urgency of your reactions tells me that you're the typical AMD apologist. The latter is my personal opinion, also formed by experience.
Back to topic.seems your personal opinion is very special: [url=<span class="skimlinks-unlinked">https://forum.cockos.com/showpost.php?p ... t=17</span>]click me![/url]
And, no, my opinion is not special. Check benchmarks. Check hard facts. No AMD has the single core performance of high end Intel CPU's, and there are very few which can compete with the top Intel CPU's in terms of multi core performance.
And, again a personal opinion: The fact that you can read about "AMD-Gläubiger" everywhere should also tell you something.
-
AdvancedFollower AdvancedFollower https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=418780
- KVRian
- 1342 posts since 8 May, 2018 from Sweden
Intel is slightly faster per single-core, but it's not an enormous difference, like Intel is 4x faster or something. They are nearly the same and the main difference is that Intel cores run at a higher clock speed generally.
I have a 1st gen Ryzen 1800X running at the stock 3.6 GHz (3.7 - 4.1 GHz with boost) and I can use way more than 24 VST's and I use a buffer size of 256. If the OP is experiencing problems with twice the amount of cores, 4x the buffer size and a higher clock speed, something else is obviously wrong and bashing a brand because you have an agenda isn't exactly helpful.
I have a 1st gen Ryzen 1800X running at the stock 3.6 GHz (3.7 - 4.1 GHz with boost) and I can use way more than 24 VST's and I use a buffer size of 256. If the OP is experiencing problems with twice the amount of cores, 4x the buffer size and a higher clock speed, something else is obviously wrong and bashing a brand because you have an agenda isn't exactly helpful.
Take a single oscillator, producing a drone. Send it to the wave shaper, altering the tone.
This can be a triangle, Sawtooth or a square. Modulate the pulse width, nobody will care
This can be a triangle, Sawtooth or a square. Modulate the pulse width, nobody will care
-
- KVRAF
- 35685 posts since 11 Apr, 2010 from Germany
Well, that would be insane, wouldn't it?AdvancedFollower wrote: Thu Feb 13, 2020 9:44 am Intel is slightly faster per single-core, but it's not an enormous difference, like Intel is 4x faster or something.
Well, no, they're not nearly the same. Most Intel CPU are signinifcantly faster than AMD CPU's in single core performance.AdvancedFollower wrote: Thu Feb 13, 2020 9:44 am They are nearly the same and the main difference is that Intel cores run at a higher clock speed generally.
For multi core, there are a couple of AMD CPU's which are similarly fast, most notably the Threadipers 3970X and 3960X.
Ok, we're talking about real high end here, the question is how relevant that is for end users, who buy in the 2.000 € range. Nonetheless, I just can't stand spreading untruths, and that's what zealots often do.
-
- KVRAF
- 1524 posts since 6 Nov, 2012
But either way, you patch spectre&meltdown and half the performance of intel CPU. No refund by the way.chk071 wrote: Thu Feb 13, 2020 9:34 amApart from that you are talking about another CPU than the OP has: "Der Ryzen 3700x liegt in Anwendungen ca 15-20% vor dem Intel" - Which Intel CPU exactly? There are Intel CPU's which are much more performant than the 3700X.enroe wrote: Thu Feb 13, 2020 9:22 amMmmhh ...chk071 wrote: Thu Feb 13, 2020 8:43 amNo, it isn't. It's a comment based on long time experience with AMD hardware. If it works for you, fine, it never really worked for me.
It's also well known that AMD aren't known for their excellent single core performance (especially the CPU in question sucks at it, and that's obviously why the OP overclocked it). Most AMD CPU's provide mutli core performance, but lack on single core performance. If you want to deny that, then I can't help you, because, then you deny facts. And, the authority, and urgency of your reactions tells me that you're the typical AMD apologist. The latter is my personal opinion, also formed by experience.
Back to topic.seems your personal opinion is very special: [url=<span class="skimlinks-unlinked">https://forum.cockos.com/showpost.php?p ... t=17</span>]click me![/url]
And, no, my opinion is not special. Check benchmarks. Check hard facts. No AMD has the single core performance of high end Intel CPU's, and there are very few which can compete with the top Intel CPU's in terms of multi core performance.
And, again a personal opinion: The fact that you can read about "AMD-Gläubiger" everywhere should also tell you something.
Code: Select all
https://youtu.be/stM2CPF9YAY?t=263
https://youtu.be/MwXY2RXmmUQ?t=229- KVRAF
- 3017 posts since 8 Jun, 2018
the AMD - Intel discussion is void, here. wow, i speak..
my intel i7-6700k performs quite well still..
and AMD, has great stuff. i like intel, but i am not blind to want AMD has achieved now.
the problem lies elsewhere. the OP seeks help, not a discussion that based on opinions.
and every system can have it's problems. it is combination of factors.
some people have given good advice. also bios up to date, all latest drivers for the motherboard (not from windows, but from the manufacturer, even when the windows drivers are newer!).
the high performance mode, very important.
try latencymon, for troubleshooting.
my intel i7-6700k performs quite well still..
and AMD, has great stuff. i like intel, but i am not blind to want AMD has achieved now.
the problem lies elsewhere. the OP seeks help, not a discussion that based on opinions.
and every system can have it's problems. it is combination of factors.
some people have given good advice. also bios up to date, all latest drivers for the motherboard (not from windows, but from the manufacturer, even when the windows drivers are newer!).
the high performance mode, very important.
try latencymon, for troubleshooting.
Primoridal Music: sadà\exposadà - Indusrial & Expanding Your Mind Hurts: Sound Brut
-
- KVRer
- Topic Starter
- 16 posts since 24 Feb, 2014
Both are at latest versions the only thing I have not done is upgrade my Cubase to 10.5 but from what I know there was an update to 10.0. which also fixed similar issues? Should I upgrade the edition too?4damind wrote: Thu Feb 13, 2020 9:32 am There have been 2 known problems recently. One problem came from Windows, but I think it was fixed in the last update and had something to do with USB.
Then there was a problem with Cubase and very high CPU peaks with VST3 plugins... that was also fixed in the last update.
So: is Windows and Cubase the latest version?
Windows is always updated.
I agree here that I'd rather keep the discussion on which is better AMD or Intel to another topic. I have made my choice and I will not change it and rather I want to work with it and solve the problems around it.WasteLand wrote: Thu Feb 13, 2020 10:21 am the AMD - Intel discussion is void, here. wow, i speak..
my intel i7-6700k performs quite well still..
and AMD, has great stuff. i like intel, but i am not blind to want AMD has achieved now.
the problem lies elsewhere. the OP seeks help, not a discussion that based on opinions.
and every system can have it's problems. it is combination of factors.
some people have given good advice. also bios up to date, all latest drivers for the motherboard (not from windows, but from the manufacturer, even when the windows drivers are newer!).
the high performance mode, very important.
try latencymon, for troubleshooting.
Regarding drivers they are all up to date but I need to check if the bios has a newer version than October last year that is when I last updated it.
Yes already double checked high performance mode is active!
- KVRAF
- 3017 posts since 8 Jun, 2018
cubase 10.5 had vst3 problems, not with all plugins. cubase 10.5.10 fixes this, this is confirmed on the steinberg forum, it wasn't stated in the release notes, not all things were there strangely enough.Sh4d0wStrider wrote: Thu Feb 13, 2020 10:30 amBoth are at latest versions the only thing I have not done is upgrade my Cubase to 10.5 but from what I know there was an update to 10.0. which also fixed similar issues? Should I upgrade the edition too?4damind wrote: Thu Feb 13, 2020 9:32 am There have been 2 known problems recently. One problem came from Windows, but I think it was fixed in the last update and had something to do with USB.
Then there was a problem with Cubase and very high CPU peaks with VST3 plugins... that was also fixed in the last update.
So: is Windows and Cubase the latest version?
Windows is always updated.
I agree here that I'd rather keep the discussion on which is better AMD or Intel to another topic. I have made my choice and I will not change it and rather I want to work with it and solve the problems around it.WasteLand wrote: Thu Feb 13, 2020 10:21 am the AMD - Intel discussion is void, here. wow, i speak..
my intel i7-6700k performs quite well still..
and AMD, has great stuff. i like intel, but i am not blind to want AMD has achieved now.
the problem lies elsewhere. the OP seeks help, not a discussion that based on opinions.
and every system can have it's problems. it is combination of factors.
some people have given good advice. also bios up to date, all latest drivers for the motherboard (not from windows, but from the manufacturer, even when the windows drivers are newer!).
the high performance mode, very important.
try latencymon, for troubleshooting.
Regarding drivers they are all up to date but I need to check if the bios has a newer version than October last year that is when I last updated it.
Yes already double checked high performance mode is active!
but cubase 10.5.10 performs better than 10.5. (i use cubase 10.5.10 pro by the way, on windows 1909).
o you have cubase 10, there seemed to be the same problem in 10.0.5, if i am correct, with vst3. i didn't notice it, because it happens not with all vst3 plugins. in 10.5 i had no problems with vst3, other problems i had..., but just before 10.5.10, i used a vst3 plugin that used a ridiculous load. when playing the project it disappeared.
so i have not experienced myself, better check the steinberg forum, about 10.0.5, the maintenance update.
Primoridal Music: sadà\exposadà - Indusrial & Expanding Your Mind Hurts: Sound Brut
-
- KVRian
- 751 posts since 9 Mar, 2001
Singlecore speed is not important in the long run. Intel won't come up with 6 GHz CPU in the near future you know. All OS's and software developed today is built for multiple cores.chk071 wrote: Thu Feb 13, 2020 9:55 amWell, that would be insane, wouldn't it?AdvancedFollower wrote: Thu Feb 13, 2020 9:44 am Intel is slightly faster per single-core, but it's not an enormous difference, like Intel is 4x faster or something.Noone would buy AMD then.
Well, no, they're not nearly the same. Most Intel CPU are signinifcantly faster than AMD CPU's in single core performance.AdvancedFollower wrote: Thu Feb 13, 2020 9:44 am They are nearly the same and the main difference is that Intel cores run at a higher clock speed generally.
For multi core, there are a couple of AMD CPU's which are similarly fast, most notably the Threadipers 3970X and 3960X.
Ok, we're talking about real high end here, the question is how relevant that is for end users, who buy in the 2.000 € range. Nonetheless, I just can't stand spreading untruths, and that's what zealots often do.
It's just dumb not to check the latest AMD Ryzen's right now. How hard it is to build systems and do your own benchmarks? I did it and Ryzen 3950 won the price vs performance by far.
How can you be so sure that your previous impressions of previous MODELS applies to these new models?
My 3950x is definitly not significantly slower in single CPU, seriously, I have tested it. The 9900 was less than 1-2% faster single core compared 3950x system I have built (using CPUZ benchmark for measurement). That is NOT significantly faster. But it IS signifacantly FASTER than Intels 9900 in total. Just install fast memory in your Ryzen system. Very important.
What *I* care about is the number of VST plugins and instruments I can run and still have the system stable, cool, quiet and rapid in response. I have that. My Ryzen 9 3950X system is very cool and quiet. No issues so far.
Windows 10 {1903} + Cubase 10.5 + Ryzen 3950x + RME = incredibly nice combo..
Multiple cores are the future and has been for many years. There is not much to be done with singlecore speed. That is why Intel is WAY behind now. I don't think Intel will have 6 GHz CPU's next year or the year after that. Software that can only use single core is going away. Even with gaming..
Last edited by Freaqpeak on Thu Feb 13, 2020 12:37 pm, edited 1 time in total.
-
- KVRAF
- 35685 posts since 11 Apr, 2010 from Germany
In a ideal world, they would be. The reality is that most applications are not built for optimal use of multiple cores.cnt wrote: Thu Feb 13, 2020 12:32 pm Singlecore speed is not important in the long run. Intel won't come up with 6 GHz CPU in the near future you know. All OS's and software developed today is built for multiple cores.
You can see that in some games, where the physics calculation is done on a single core, e.g. some racing simulations. Most people who want to run those choose Intel CPU's, for a reason, which is the single core performance.
I think most soft synths and effects also run on a single core, unless they have a multicore option. So, yeah, it's also very important for audio software.
- KVRAF
- 5913 posts since 17 Aug, 2004 from Berlin, Germany
Only the latest Cubase update (10.5.10) fixed this VST3 bug. It's not in the release notes but confirmed by Steinberg employees that its fixed. So its highly recommend to update to the latest Cubase version and test this again. The other things have already been mentioned, Steinberg has this own energy saving mode in the settings (this mode optimized for audio...) I would enable this and of course Asio Guard at "standard" (I would not use the highest settings because there are some known problems with some midi plugins).Sh4d0wStrider wrote: Thu Feb 13, 2020 10:30 amBoth are at latest versions the only thing I have not done is upgrade my Cubase to 10.5 but from what I know there was an update to 10.0. which also fixed similar issues? Should I upgrade the edition too?4damind wrote: Thu Feb 13, 2020 9:32 am There have been 2 known problems recently. One problem came from Windows, but I think it was fixed in the last update and had something to do with USB.
Then there was a problem with Cubase and very high CPU peaks with VST3 plugins... that was also fixed in the last update.
So: is Windows and Cubase the latest version?
Yes, Intel was often the preferred CPU for audio stuff but I don't know if this recommendation is still valid. There is this guy making the DAW Benchmark stuff, maybe he has more updated information how the DAWs perform with the newest AMD CPUs.
| Links-
- KVRian
- 751 posts since 9 Mar, 2001
Of course I agree that it is where we are today in gaming and that there might be scenarios where a more little single core speed makes a difference also in DAW usage, but would argue that it is so rare that 98% (?!) of DAW users should not even have to take it into account.chk071 wrote: Thu Feb 13, 2020 12:35 pmIn a ideal world, they would be. The reality is that most applications are not built for optimal use of multiple cores.cnt wrote: Thu Feb 13, 2020 12:32 pm Singlecore speed is not important in the long run. Intel won't come up with 6 GHz CPU in the near future you know. All OS's and software developed today is built for multiple cores.
You can see that in some games, where the physics calculation is done on a single core, e.g. some racing simulations. Most people who want to run those choose Intel CPU's, for a reason, which is the single core performance.
I think most soft synths and effects also run on a single core, unless they have a multicore option. So, yeah, it's also very important for audio software.
One might ask why single core important is for a DAW in a real use case?
Let me ask you this, have you EVER taken a whole CPU CORE to the maximum using a VSTi or VST with a Intel 9700 or a Ryzen 3900+? Or even one channel?
I don't know of a single VST plugin that takes more than 1 core on a modern CPU.
Even the most CPU intensive plugins I have doesn't do it. Like Diva (which actually has multicore option). The only way to make this unlikely scenario happen is to load more than 8 Insert FX of the most CPU intesnive software I have, the SGA1566 set to 16x oversampling and high CPU setting. It took away 1 total core. Yes, I actually did test this..
-
- KVRian
- 751 posts since 9 Mar, 2001
Here's my Cubase DAWbench benchmarks for my 3950x system from december. Tested Reaper one time as well.
viewtopic.php?f=7&t=536839
I was going to do more benchmarks and tests but I haven't gotten the time. Anyways.
In one of the tests I have SGA1566 set up with DAWbench settings at 128 samples. The 9900K @ 5 GHz can run 194 instances. The Ryzen 3950 system I built can run 404 instances. It's twice as fast. For real... Single-core speed is just not important for most cases.
viewtopic.php?f=7&t=536839
I was going to do more benchmarks and tests but I haven't gotten the time. Anyways.
In one of the tests I have SGA1566 set up with DAWbench settings at 128 samples. The 9900K @ 5 GHz can run 194 instances. The Ryzen 3950 system I built can run 404 instances. It's twice as fast. For real... Single-core speed is just not important for most cases.
-
- KVRer
- Topic Starter
- 16 posts since 24 Feb, 2014
I just reinstalled and upgraded to the 10.5.10 version and it definitely gave an improvement although I am not sure whether it could be better. I have attached a screenshot of the real time load and average load on the current project which has 24 VST all are Kontakt Instances each one is a seperate one with a single MIDI Track. All of them have a Fab Filter EQ and 8 of them have Fab Filter R on them. This is with a Buffer Size of 512 which is an improvement. Also this is with Asio Guard on Low.4damind wrote: Thu Feb 13, 2020 9:32 am
Only the latest Cubase update (10.5.10) fixed this VST3 bug. It's not in the release notes but confirmed by Steinberg employees that its fixed. So its highly recommend to update to the latest Cubase version and test this again. The other things have already been mentioned, Steinberg has this own energy saving mode in the settings (this mode optimized for audio...) I would enable this and of course Asio Guard at "standard" (I would not use the highest settings because there are some known problems with some midi plugins).
I think I need to download DAWBench and test with it but the clean install with the update definitely did something.
You do not have the required permissions to view the files attached to this post.
- KVRAF
- 5913 posts since 17 Aug, 2004 from Berlin, Germany
Yes, to have some comparable values for your system compared to others, this DAWBench is a good choice. If I read this correctly the user with the 3950x could run >400 instances? The 1950X will not perform as well as the 3950x but I expect results near 70%, so we speak about ~300 instances.
| Links