I own an Phase 26 USB soundcard and on my laptop concerning the performance it makes no difference whether I use the built-in soundcard via asio4all or the external with it's own asio drivers.druid wrote:... USB performance will suffer, whereas not so much with the PCI soundcard in my PC, but it didn't take as long to fill the CPU up of course!...
FR theme: Performance/Graphics issues
-
- KVRist
- 233 posts since 14 Mar, 2005 from Cologne
-
- KVRAF
- 1601 posts since 24 Jun, 2004 from Australia
My understanding of USB was that it took a fair hit to CPU usage, so that the more USB devices one has, the more CPU use could be expected if they were all operating at peak capacity. I thought that's one reason why some prefer FireWire. So I figured if I max out my CPU, USB will suffer. It's alright, if I didn't stubbornly operate at 96khz all the time, I probably wouldn't run into much problem at all. :P
-
- KVRian
- 507 posts since 8 Dec, 2004 from Belgrade
Yes, there is CPU hit when using USB devices, and that's not so good for audio work.
But the problem is that XT2 isn't so good at lower latencies, even with PCI cards. I'm working with 6ms latency, and all is fine, but can't go lower under moderate processor load. It's rather old PC, AMD643000+, but in other hosts, and xt1.41, I can work at 1,5 latency without problems. Just moving objects in comp makes audio skip...
But the problem is that XT2 isn't so good at lower latencies, even with PCI cards. I'm working with 6ms latency, and all is fine, but can't go lower under moderate processor load. It's rather old PC, AMD643000+, but in other hosts, and xt1.41, I can work at 1,5 latency without problems. Just moving objects in comp makes audio skip...
-
- KVRist
- 233 posts since 14 Mar, 2005 from Cologne
I am not an expert concerning these matters and don't want to enter in a discussion about the subject. Nevertheless it seems to me that this isn't really a problem of the main cpu workload. With a 1,5 Ghz Laptop and 752 MB ram I don't really own the newest or most qualified machine for producing music. Anyhow everything is working fine here. So my guess is, that it might be a problem of how the interaction of graphics is done on a special computer. But as I said already: I am a laity concerning computer hardware and so this is just an assumption.druid wrote:My understanding of USB was that it took a fair hit to CPU usage, so that the more USB devices one has, the more CPU use could be expected if they were all operating at peak capacity. I thought that's one reason why some prefer FireWire. So I figured if I max out my CPU, USB will suffer. It's alright, if I didn't stubbornly operate at 96khz all the time, I probably wouldn't run into much problem at all.
-
- KVRAF
- Topic Starter
- 1958 posts since 16 Jan, 2005 from France's Dirty South
Jorgen, please give us a heads up on this !
I've checked out a couple of apps that generate system specs reports (like the freeware sandra lite).
Users with problems could post some of these reports, the software can output xml and txt files ; problem is, the default report is absolutely huge, maybe you could check it out and tell us which parts would be most helpful to you ?
cheers
I've checked out a couple of apps that generate system specs reports (like the freeware sandra lite).
Users with problems could post some of these reports, the software can output xml and txt files ; problem is, the default report is absolutely huge, maybe you could check it out and tell us which parts would be most helpful to you ?
cheers
-
- KVRAF
- 1601 posts since 24 Jun, 2004 from Australia
I can't believe people feel the need to work at such low latencies! I am more than happy using 512 sample delay even 768 sometimes, which results in .. um .. I think something like 11ms. I admit I don't play much, most is played direct from the computer, but 3ms seems as close to perfect as anyone would need.
horn: It may depend on what delay and samplerate you're using. My USB thing works great if I'm down at 48 or 44.1 kilohertz, and I typically use 512 delay (my USB interface switches to half for lower samplerates like that though, it's really annoying...).
Anyway. I'd be willing to run those system spec reports programs and quote parts of it if it would help.
horn: It may depend on what delay and samplerate you're using. My USB thing works great if I'm down at 48 or 44.1 kilohertz, and I typically use 512 delay (my USB interface switches to half for lower samplerates like that though, it's really annoying...).
Anyway. I'd be willing to run those system spec reports programs and quote parts of it if it would help.
-
- KVRAF
- 10815 posts since 26 Nov, 2004 from UK
at low latency i can reproduce the crackling,
i use 512 so never noticed before,
but if i knock the the buffer down to 128 i can make it crackle with a track that only uses 30 cpu
Subz
i use 512 so never noticed before,
but if i knock the the buffer down to 128 i can make it crackle with a track that only uses 30 cpu
Subz
-
- KVRian
- 889 posts since 10 Jul, 2004
i dont have any graphic issues.
but ext2 standalone uses up to 20% of cpu IDLE>(1 empty audio+1 empty midi)
if i load ext2 vst inside xt 1.4.....the whole thing dont use more than 8%.
conclusion.....the standalone is flawed with some seroius performance issue.no other sequencer i ever tried uses more than 5% idle.
but ext2 standalone uses up to 20% of cpu IDLE>(1 empty audio+1 empty midi)
if i load ext2 vst inside xt 1.4.....the whole thing dont use more than 8%.
conclusion.....the standalone is flawed with some seroius performance issue.no other sequencer i ever tried uses more than 5% idle.
-
- KVRist
- 233 posts since 14 Mar, 2005 from Cologne
Ok you're right. Because of my personal needs, I normally never use 96kHz. Tried it and... it crackles.druid wrote:horn: It may depend on what delay and samplerate you're using. My USB thing works great if I'm down at 48 or 44.1 kilohertz, and I typically use 512 delay (my USB interface switches to half for lower samplerates like that though, it's really annoying...)
On 48kHz with 256 latency everything's fine. No glitches no crackles even though I added one instance of ERA and an internal synthezizer to my last breadboard construction.
-
- KVRist
- 344 posts since 14 Oct, 2005
hi, sorry to drag this issue up again but reading through jens' poll thread i noticed that abi stated that in xt2 he gets crackles with 30 percent cpu usage according to task manager and with sonar le and samplitude le he gets crackles with 50 percent cpu usage but can use a sample rate of 128 or 64 (i presume the sample rate would be higher in xt2)so is 50 percent cpu acceptable before getting crackles or does samplitude le and sonar le have a similar problem? i would have expected at least 80 percent plus before crackles and glitches appeared. of course it will probably depend on other factors to like hard drive speed and ram etc.
any views?
john
any views?
john
-
- KVRAF
- 25007 posts since 12 Jul, 2003 from West Caprazumia
Samplitude, just as Logic, are highly optimized - they perform buch better than Sonar.
In Samplitude you can go to 95% without having crackles and the gui stays as fast as with 5% - and with Logic it's very similar.
Those programmes are the proof that it's not just theroretically but also practically possible to get such a performance. But at least Samplitude's engine has been written in Assembler which is said to allow code-optimization which is hard to achieve with a high level language.
B.t.w.: with eXT 1.4 I can go up very high as well until crackles appear - but the gui becomes totally sluggish.
In Samplitude you can go to 95% without having crackles and the gui stays as fast as with 5% - and with Logic it's very similar.
Those programmes are the proof that it's not just theroretically but also practically possible to get such a performance. But at least Samplitude's engine has been written in Assembler which is said to allow code-optimization which is hard to achieve with a high level language.
B.t.w.: with eXT 1.4 I can go up very high as well until crackles appear - but the gui becomes totally sluggish.
-
- KVRAF
- 25007 posts since 12 Jul, 2003 from West Caprazumia
No, not at all - though it's carefully assembled (I don't buy complete machines - they usually don't work too good). 
-
- KVRAF
- 5350 posts since 8 Aug, 2003 from Berlin Germany
Ok,
I just started testing my ssytem...
I am not getting the 20% cpu hit for a blank project that you all are.
I have 18 audio tracks playing and have only reached 16% for the energyXT2.exe thread in my task manager. No clicks, pops, or stutters or graphics issues.
Dell Latitude D620 laptop: T2400 @ 1.83GHz x2, 1GB RAM, integrated simtel audio w/ ASIO4All, NVIDIA Quatro NVS 110M supporting a second monitor.
I just started testing my ssytem...
I am not getting the 20% cpu hit for a blank project that you all are.
I have 18 audio tracks playing and have only reached 16% for the energyXT2.exe thread in my task manager. No clicks, pops, or stutters or graphics issues.
Dell Latitude D620 laptop: T2400 @ 1.83GHz x2, 1GB RAM, integrated simtel audio w/ ASIO4All, NVIDIA Quatro NVS 110M supporting a second monitor.