Convolution Reverb for NVidia and ATI GPUs - saving CPU time
-
Nils Schneider Nils Schneider https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=45370
- KVRist
- Topic Starter
- 279 posts since 22 Oct, 2004 from Neuss, Germany
This is a link that doesn't need a nameserver:
http://81.169.130.123/files/nsGPUImpuls ... TSetup.exe
http://81.169.130.123/files/nsGPUImpuls ... TSetup.exe
https://k1v.nilsschneider.de - Kawai K1 emulated as VSTi/AU
https://heatvst.com - Android Synthesizer with full VST integration
https://gpuimpulsereverb.de - Use your GPU as reverberation DSP
https://heatvst.com - Android Synthesizer with full VST integration
https://gpuimpulsereverb.de - Use your GPU as reverberation DSP
-
- KVRist
- 253 posts since 14 May, 2005 from Sweden
Thanks. No worries.
- KVRAF
- 16385 posts since 22 Nov, 2000 from Southern California
What are the correct cards? I'm looking at the following:LeVzi wrote:The latest Nvidia drivers for the correct cards come with CUDA stuff already in it.
http://www.newegg.com/Product/Product.a ... 6814150320
- KVRAF
- 9453 posts since 17 Sep, 2002 from Gothenburg Sweden
All NVIDIA GPUs released since Nov 2006 are CUDA-capable.
- KVRAF
- 2138 posts since 8 Feb, 2007
ONLY if they have 256 MB (and up) of internal memoryjupiter8 wrote:All NVIDIA GPUs released since Nov 2006 are CUDA-capable.
CUDA-Enabled GPU Products
- KVRAF
- 9453 posts since 17 Sep, 2002 from Gothenburg Sweden
Tp3 wrote:ONLY if they have 256 MB (and up) of internal memoryjupiter8 wrote:All NVIDIA GPUs released since Nov 2006 are CUDA-capable.
CUDA-Enabled GPU Products
Wikipedia wrote:note that many applications require at least 256 MB of dedicated VRAM
So it's not ONLY if they have 256 MB (and up).nVidia wrote:Many CUDA programs require at least 256 MB of memory attached to the GPU.
- KVRAF
- 2138 posts since 8 Feb, 2007
Ummmjupiter8 wrote:So it's not ONLY if they have 256 MB (and up).nVidia wrote:Many CUDA programs require at least 256 MB of memory attached to the GPU.
Check out CUDA-Enabled GPU Products page and Then this (for example)
Do you see card with less then 256 MB
Enlighten me pliz...
I have an Quadro FX 560 card.
In theory, a fairly strong card (for my type of uses)
In reality... not enough RAM --> not CUDA for me
-
- KVRAF
- 2945 posts since 23 Dec, 2002
Here is a technical question regarding Cuda enabled cards and the 4 gig switch. FOr those of us who are locked into the world of XP 32 bit, the 3 gig switch has been a godsend. I have three uad-1 cards which are PCI cards. Each card needs to have some of the memory allocated , I believe it is about 128 megs per card which is then subtracted from the available memory. If I add a 256 meg cuda enabled card do I have t subtract a further block of 256 megs of memory from the 3 gig switch line?
Incidentally , Nebula also supports Cuda enabled cards for some of the presets.
Incidentally , Nebula also supports Cuda enabled cards for some of the presets.
-
- Banned
- 1374 posts since 5 May, 2007 from Finland
I don't understand these statements at all. If you play anything, you'll notice the latency. If you change any parameter, you'll notice the latency. It's always there, the compensation only means that it makes the output sound like there's no unsync between the tracks.Nils Schneider wrote:Yes, it does indeed or course
If the plugin is used in a sequencer with latency compensation you won't notice the latency at all.
So yes, you WILL notice it.
But it's not that bad
-
Nils Schneider Nils Schneider https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=45370
- KVRist
- Topic Starter
- 279 posts since 22 Oct, 2004 from Neuss, Germany
Of course, if you play live, you will notice it for sure.mkdr wrote:I don't understand these statements at all.Nils Schneider wrote:Yes, it does indeed or course
If the plugin is used in a sequencer with latency compensation you won't notice the latency at all.
But I create my tracks in Cubase and the compensation means, that the plugin has its correct timing.
https://k1v.nilsschneider.de - Kawai K1 emulated as VSTi/AU
https://heatvst.com - Android Synthesizer with full VST integration
https://gpuimpulsereverb.de - Use your GPU as reverberation DSP
https://heatvst.com - Android Synthesizer with full VST integration
https://gpuimpulsereverb.de - Use your GPU as reverberation DSP
-
- Banned
- 1374 posts since 5 May, 2007 from Finland
Yes, i do the same. But don't you ever "play" the notes in to the sequencer? If not then any kind of latency has no difference at all, except to the reaction time of you pressing the "play" button and the playback starting.Nils Schneider wrote:Of course, if you play live, you will notice it for sure.mkdr wrote:I don't understand these statements at all.Nils Schneider wrote:Yes, it does indeed or course
If the plugin is used in a sequencer with latency compensation you won't notice the latency at all.
But I create my tracks in Cubase and the compensation means, that the plugin has its correct timing.
So i don't understand why to care about latency if not for the things that the latency actually has some effect on.
Ah i, think i took the sentence a bit out of it's context. The question was basicly about the syncing of tracks. And that is what indeed happens. Sorry.
I've just seen that thing said too many times...
Sorry for going offtopic
..and being stupid
- KVRAF
- 16385 posts since 22 Nov, 2000 from Southern California
Giancarlo from Acustica advises to use a card with as much memory as possible. It's probably more important with Nebula than with this plug-in since Nebula's impulse collections are much larger than standard impulses.Tp3 wrote:ONLY if they have 256 MB (and up) of internal memory
btw, I'm now looking at getting the following card since users are reporting it to have a quieter fan:
http://www.newegg.com/Product/Product.a ... 6814143154
-
Nils Schneider Nils Schneider https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=45370
- KVRist
- Topic Starter
- 279 posts since 22 Oct, 2004 from Neuss, Germany
For my IR plugin, even 256MB are enough, as the memory consumption is not the problem. The processing power will be maxed out before the memory.
https://k1v.nilsschneider.de - Kawai K1 emulated as VSTi/AU
https://heatvst.com - Android Synthesizer with full VST integration
https://gpuimpulsereverb.de - Use your GPU as reverberation DSP
https://heatvst.com - Android Synthesizer with full VST integration
https://gpuimpulsereverb.de - Use your GPU as reverberation DSP
-
- KVRAF
- 10815 posts since 26 Nov, 2004 from UK
got a GeForce 9500 GT
thanks Nils!
enjoying all 3 of your gpu plugins
Subz
thanks Nils!
enjoying all 3 of your gpu plugins
Subz