RTX Voice on AMD Fails | Log

Recep Baltaş

Recep Baltaş

Technopat
Tam Bir Technopat'sınız!
Katılım
14 Ağustos 2010
Mesajlar
79.057
Makaleler
289
Çözümler
2.268
Yer
İstanbul
Daha fazla  
Cinsiyet
Erkek
Profil Kapağı
1522743131
Tried to get NVIDIA's RTX Voice on an AMD GPU and it turns out to be using CUDA, thus failing with AMD GPUs.

Here's the log:

Kod: 
2020-04-30,20:16:45: Logging session started
2020-04-30,20:16:45: App version: v0.5.12.6
2020-04-30,20:16:45: Initializing microphone pipeline.
2020-04-30,20:16:45: Initializing effect
2020-04-30,20:16:48: Mic denoiser initialized successfully.
2020-04-30,20:16:48: Initializing playback pipeline.
2020-04-30,20:16:48: Initializing effect
2020-04-30,20:16:48: Playback denoiser initialized successfully.
2020-04-30,20:16:49: Starting MicCapture Threads
2020-04-30,20:16:49: Stopping MicCapture threads
2020-04-30,20:16:49: MicCapture stopped
2020-04-30,20:16:49: Default endpoint is: Mikrofon (HD Pro Webcam C920)
2020-04-30,20:16:49: Coercing WAVEFORMATEX with wFormatTag = 0x0000fffe to int-16
2020-04-30,20:16:49: Selected mic endpoint props: Channels: 2, Sample rate: 16000, Bits/Sample: 16, Buffer size: 352, period: 10 ms, latency: 10 ms
2020-04-30,20:16:49: Device Period: 100000 Allocated buffer size(ms): 40 size(bytes):2560
2020-04-30,20:16:49: webrtc resampler enabled (1), input (320), output (960)
2020-04-30,20:16:49: Starting capture client
2020-04-30,20:16:49: Initialized IAudioClient to 32b59020
2020-04-30,20:16:49: MicCapture thread start
2020-04-30,20:16:49: Framesize10ms 640
2020-04-30,20:16:49: MicCapture ready
2020-04-30,20:16:49: Starting playback threads
2020-04-30,20:16:49: 5 playback endpoints found
2020-04-30,20:16:49: 0: Playback device: Hoparlör (Realtek High Definition Audio)
2020-04-30,20:16:49: render client started
2020-04-30,20:16:49: 1: Playback device: Realtek Digital Output (Realtek High Definition Audio)
2020-04-30,20:16:49: render client started
2020-04-30,20:16:49: 2: Playback device: Realtek Digital Output(Optical) (Realtek High Definition Audio)
2020-04-30,20:16:49: render client started
2020-04-30,20:16:49: 3: Playback device: 2 - ASUS VP247 (AMD High Definition Audio Device)
2020-04-30,20:16:49: render client started
2020-04-30,20:16:49: 4: Playback device: Speakers (NVIDIA RTX Voice)
2020-04-30,20:16:49: RenderCapture buffer allocated (ms) 40 (bytes) 7680
2020-04-30,20:16:49: render client started
2020-04-30,20:16:49: Selected playback device (default): Hoparlör (Realtek High Definition Audio)
2020-04-30,20:16:49: Selected playback endpoint props: Channels: 2, Sample rate: 48000, Bits/Sample: 16, Buffer size: 2136, period: 10 ms, latency: 10 ms
2020-04-30,20:16:49: RenderDenoiser thread started
2020-04-30,20:16:49: Framesize10ms(render) 1920
2020-04-30,20:16:49: Started playback threads
2020-04-30,20:16:49: Stopping MicCapture threads
2020-04-30,20:16:49: MicCaptureThreadMain Duration(ms) 92, captured(bytes) 6400
2020-04-30,20:16:49: MicPushThreadFunction Duration(ms) 92, captured(bytes) 24960
2020-04-30,20:16:49: MicPushThreadFunction stopped
2020-04-30,20:16:49: DenoiserThreadMain Duration(ms) 92, denoised(bytes) 23040
2020-04-30,20:16:49: MicCapture thread Stop
2020-04-30,20:16:49: MicCapture stopped
2020-04-30,20:16:49: Set endpoint name for capture
2020-04-30,20:16:49: Starting MicCapture Threads
2020-04-30,20:16:49: Stopping MicCapture threads
2020-04-30,20:16:49: MicCapture stopped
2020-04-30,20:16:49: 3 capture endpoints found
2020-04-30,20:16:49: 0: Capture device: Mikrofon (Realtek High Definition Audio)
2020-04-30,20:16:49: 1: Capture device: Mikrofon (NVIDIA RTX Voice)
2020-04-30,20:16:49: 2: Capture device: Mikrofon (HD Pro Webcam C920)
2020-04-30,20:16:49: Could not find device with specified name (Default Device), will use default device
2020-04-30,20:16:49: Default endpoint is: Mikrofon (HD Pro Webcam C920)
2020-04-30,20:16:49: Coercing WAVEFORMATEX with wFormatTag = 0x0000fffe to int-16
2020-04-30,20:16:49: Selected mic endpoint props: Channels: 2, Sample rate: 16000, Bits/Sample: 16, Buffer size: 352, period: 10 ms, latency: 10 ms
2020-04-30,20:16:49: Device Period: 100000 Allocated buffer size(ms): 40 size(bytes):2560
2020-04-30,20:16:49: webrtc resampler enabled (1), input (320), output (960)
2020-04-30,20:16:49: Starting capture client
2020-04-30,20:16:49: Initialized IAudioClient to 32b772c0
2020-04-30,20:16:49: MicCapture thread start
2020-04-30,20:16:49: Framesize10ms 640
2020-04-30,20:16:49: MicCapture ready
2020-04-30,20:16:49: Capture restarted successfully
2020-04-30,20:16:49: Stopping playback threads
2020-04-30,20:16:49: PlaybackCaptureThread Duration(ms) 17, captured(bytes) 0
2020-04-30,20:16:49: PlaybackCaptureThread Duration(ms) 17, denoised(bytes) 1920
2020-04-30,20:16:49: RenderDenoiser thread stopped
2020-04-30,20:16:49: Stopped playback threads
2020-04-30,20:16:49: Set endpoint name for capture
2020-04-30,20:16:49: Starting playback threads
2020-04-30,20:16:49: 5 playback endpoints found
2020-04-30,20:16:49: 0: Playback device: Hoparlör (Realtek High Definition Audio)
2020-04-30,20:16:49: render client started
2020-04-30,20:16:49: 1: Playback device: Realtek Digital Output (Realtek High Definition Audio)
2020-04-30,20:16:49: render client started
2020-04-30,20:16:49: 2: Playback device: Realtek Digital Output(Optical) (Realtek High Definition Audio)
2020-04-30,20:16:49: render client started
2020-04-30,20:16:49: 3: Playback device: 2 - ASUS VP247 (AMD High Definition Audio Device)
2020-04-30,20:16:49: render client started
2020-04-30,20:16:49: 4: Playback device: Speakers (NVIDIA RTX Voice)
2020-04-30,20:16:49: RenderCapture buffer allocated (ms) 40 (bytes) 7680
2020-04-30,20:16:49: render client started
2020-04-30,20:16:49: Device not found, starting playback on default device
2020-04-30,20:16:49: Selected playback device (default): Hoparlör (Realtek High Definition Audio)
2020-04-30,20:16:49: Selected playback endpoint props: Channels: 2, Sample rate: 48000, Bits/Sample: 16, Buffer size: 2136, period: 10 ms, latency: 10 ms
2020-04-30,20:16:49: RenderDenoiser thread started
2020-04-30,20:16:49: Framesize10ms(render) 1920
2020-04-30,20:16:49: Started playback threads
2020-04-30,20:16:49: Render restarted successfully
2020-04-30,20:17:57: Stopping MicCapture threads
2020-04-30,20:17:57: MicPushThreadFunction Duration(ms) 68043, captured(bytes) 13071360
2020-04-30,20:17:57: MicCaptureThreadMain Duration(ms) 68043, captured(bytes) 4355200
2020-04-30,20:17:57: MicPushThreadFunction stopped
2020-04-30,20:17:57: DenoiserThreadMain Duration(ms) 68043, denoised(bytes) 13069440
2020-04-30,20:17:57: MicCapture thread Stop
2020-04-30,20:17:57: MicCapture stopped
2020-04-30,20:17:57: Set endpoint name for capture
2020-04-30,20:17:57: Starting MicCapture Threads
2020-04-30,20:17:57: Stopping MicCapture threads
2020-04-30,20:17:57: MicCapture stopped
2020-04-30,20:17:57: 3 capture endpoints found
2020-04-30,20:17:57: 0: Capture device: Mikrofon (Realtek High Definition Audio)
2020-04-30,20:17:57: Selected capture device: Mikrofon (Realtek High Definition Audio)
2020-04-30,20:17:57: Default endpoint is: Mikrofon (Realtek High Definition Audio)
2020-04-30,20:17:57: Coercing WAVEFORMATEX with wFormatTag = 0x0000fffe to int-16
2020-04-30,20:17:57: Selected mic endpoint props: Channels: 2, Sample rate: 48000, Bits/Sample: 16, Buffer size: 1056, period: 10 ms, latency: 10 ms
2020-04-30,20:17:57: Device Period: 100000 Allocated buffer size(ms): 40 size(bytes):7680
2020-04-30,20:17:57: webrtc resampler enabled (1), input (960), output (960)
2020-04-30,20:17:57: Starting capture client
2020-04-30,20:17:58: Initialized IAudioClient to 2e8e3a10
2020-04-30,20:17:58: MicCapture thread start
2020-04-30,20:17:58: Framesize10ms 1920
2020-04-30,20:17:58: MicCapture ready
2020-04-30,20:17:58: Capture restarted successfully
2020-04-30,20:18:26: Stopping playback threads
2020-04-30,20:18:26: PlaybackCaptureThread Duration(ms) 96395, captured(bytes) 0
2020-04-30,20:18:26: PlaybackCaptureThread Duration(ms) 96395, denoised(bytes) 1920
2020-04-30,20:18:26: RenderDenoiser thread stopped
2020-04-30,20:18:26: Stopped playback threads
2020-04-30,20:18:26: Set endpoint name for capture
2020-04-30,20:18:26: Starting playback threads
2020-04-30,20:18:26: 5 playback endpoints found
2020-04-30,20:18:26: 0: Playback device: Hoparlör (Realtek High Definition Audio)
2020-04-30,20:18:26: render client started
2020-04-30,20:18:26: 1: Playback device: Realtek Digital Output (Realtek High Definition Audio)
2020-04-30,20:18:26: render client started
2020-04-30,20:18:26: 2: Playback device: Realtek Digital Output(Optical) (Realtek High Definition Audio)
2020-04-30,20:18:26: render client started
2020-04-30,20:18:26: 3: Playback device: 2 - ASUS VP247 (AMD High Definition Audio Device)
2020-04-30,20:18:26: render client started
2020-04-30,20:18:26: 4: Playback device: Speakers (NVIDIA RTX Voice)
2020-04-30,20:18:26: RenderCapture buffer allocated (ms) 40 (bytes) 7680
2020-04-30,20:18:26: render client started
2020-04-30,20:18:26: Selected playback device: Hoparlör (Realtek High Definition Audio)
2020-04-30,20:18:26: Selected playback endpoint props: Channels: 2, Sample rate: 48000, Bits/Sample: 16, Buffer size: 2136, period: 10 ms, latency: 10 ms
2020-04-30,20:18:26: RenderDenoiser thread started
2020-04-30,20:18:26: Framesize10ms(render) 1920
2020-04-30,20:18:26: Started playback threads
2020-04-30,20:18:26: Render restarted successfully
2020-04-30,20:21:12: creating effect
2020-04-30,20:21:13: LoadLibrary(AFX) done
2020-04-30,20:21:13: cudaGetDeviceCount() failed
2020-04-30,20:21:13: Failed to create Effect
2020-04-30,20:21:13: NvAFX_CreateEffect() failed.
2020-04-30,20:21:13: initDenoiserOnly() failed
2020-04-30,20:25:53: creating effect
2020-04-30,20:25:53: cudaGetDeviceCount() failed
2020-04-30,20:25:53: Failed to create Effect
2020-04-30,20:25:53: NvAFX_CreateEffect() failed.
2020-04-30,20:25:53: initDenoiserOnly() failed
2020-04-30,20:26:08: creating effect
2020-04-30,20:26:08: cudaGetDeviceCount() failed
2020-04-30,20:26:08: Failed to create Effect
2020-04-30,20:26:08: NvAFX_CreateEffect() failed.
2020-04-30,20:26:08: initDenoiserOnly() failed
 
This is classical marketing tactic. When DirectX 11 was released, a lot of advertisements were made that it contained many innovations. However, with a very small setting, DirectX 11 games were easily playable on DirectX 10 supported cards.
 
Cruslan dedi:
This is classical marketing tactic. When DirectX 11 was released, a lot of advertisements were made that it contained many innovations. However, with a very small setting, DirectX 11 games were easily playable on DirectX 10 supported cards.
Genişletmek için tıkla...
It's not the same thing at all.

When it comes to AI - accelerated applications, such as RTX Voice, CUDA is playing important role on both creating and running AI system. CUDA based GPU's computing capacities are very strong and efficient. I'm also using CUDA for ML applications on TF.

So, It's not about marketing. It's all about technology.
 
SideWinder dedi:
It's not the same thing at all.

When it comes to AI - accelerated applications, such as RTX Voice, CUDA is playing important role on both creating and running AI System. CUDA based GPU's computing capacities are Very strong and efficient. I'm also using CUDA for ML applications on TF.

So, It's not about marketing. It's all about Technology.
Genişletmek için tıkla...

AI is not powerful at background works (RTX Voice etc.). For example, when RTX Voice is open, GPU usage of 2080 Ti become %10 about.
So, still, GPUs are not enough at AI working or render etc.

Maybe, Next GPUs will be more strong than current GPUs for this works.
(Sorry for MY English mistake(s).)
 
Son düzenleyen: Moderatör:
Abu! dedi:
It's very interesting. My GTX 1060 GPU usage becomes max %5 when i open RTX Voice.
Genişletmek için tıkla...
I did not said; Mesut Çevik said :)
Bu içeriği görüntülemek için üçüncü taraf çerezlerini yerleştirmek için izninize ihtiyacımız olacak.
Daha detaylı bilgi için, çerezler sayfamıza bakınız.
04:12
 
Salih METİN dedi:
AI is not powerful at background works (RTX voice etc.). For example, when RTX voice is open, GPU usage of 2080 Ti become %10 about.
So, still, GPU's are not enough at AI working or render etc.

Maybe, Next gpus will be more strong than current gpus for this works.
Genişletmek için tıkla...

Sorry, but thats not true. You can't measure the power or capabilities of rendering like that.

This article is a great example about how power GPU's are when it comes to ML:

medium.com

Introduction to TensorFlow — CPU vs GPU

Dear reader,
medium.com medium.com

Final comparison graph:

TF.PNG


So, you can't say "Oh, GPU usage is low, so GPU's are not good at rendering", only looking to usage statics.


Salih METİN dedi:
I did not said; Mesut Çevik said :)
Bu içeriği görüntülemek için üçüncü taraf çerezlerini yerleştirmek için izninize ihtiyacımız olacak.
Daha detaylı bilgi için, çerezler sayfamıza bakınız.
04:12
Genişletmek için tıkla...

I watched and couldn't understand anything.

You want to use real-time voice clerification system which powered by AI, than expecting not to create any load to GPU or CPU.

Maybe in future updates, usege or load can be optimized but you can't eliminate that.
 
Son düzenleme:
SideWinder dedi:
Sorry, but thats not true. You can't measure the power or capabilities of rendering like that.

This article is a great example about how power GPU's are when it comes to ML:

medium.com

Introduction to TensorFlow — CPU vs GPU

Dear reader,
medium.com medium.com

Final comparison graph:

Eki Görüntüle 551827

So, you can't say "Oh, GPU usage is low, so GPU's are not good at rendering", only looking to usage statics.
Genişletmek için tıkla...
I got it wrong.
Both render, game and RTX voice is not for RTX lower. Even, Mesut Çevik said "I'm using RTX 2080TI (Smiles) but GPU usage become %10 about.
The usage is too much for this card. (Both game, RTX Voice and streaming...)

For example, Nvidia says "Ray Tracing uses the AI cores (or Tensor cores but I don't mean CUDA.) While (or when) we open Ray Tracing, starting FPS drops (reduction).
(I don't know, why Nvidia put Tensor cores in the RTX 2060.)
Sorry for "rendering". I did not want, mean that.
 
I think it's now becoming a race of features provided with Machine Learning. More than raw FPS power, drivers and software technologies supported by hardware are becoming more and more relevant for consumer market. RT Cores and similar CPU or GPU features will become more important. It seems RTX Voice is a good and valid point to consider when buying a new GPU. We will, of course test it.
 

Benzer konular

Recep Baltaş
WoT enCore Fails with Ray Tracing ON | Core i9-10900
Mesaj
2
Görüntüleme
734
Recep Baltaş
Recep Baltaş
Ali Güngör
Neil Spicer Interview on AMD Console and PC Solutions
Mesaj
1
Görüntüleme
828
ProfTech
ProfTech
Recep Baltaş
[Success] AMD Radeon Driver On NUC's RX Vega M GH
Mesaj
2
Görüntüleme
1B
Recep Baltaş
Recep Baltaş
RTFM
Concerns on TurkNet port forwarding
Mesaj
2
Görüntüleme
550
Ali Güngör
Ali Güngör
Recep Baltaş
On The Road - Truck Simulator
Mesaj
4
Görüntüleme
841
eyre
eyre

Yeni konular

Yeni mesajlar

Geri
Yukarı