rigaya
9.04 Version--mt-cuda is very unstable
StaxRip-v2.50.3 Under the same coding instructions, the 9.04 crash probability occurs more frequently than 9.00.
--vbr 1750 --codec h265 --preset p7 --profile main --vbr-quality 26 --multipass 2pass-full --qp-max 28:28:30 --qp-min 22:24:24 --aq --aq-temporal --bref-mode each --ref 8 --lookahead 32 --lookahead-level 3 --gop-len 60 --slices 1 --multiref-l0 7 --multiref-l1 3 --cu-min 8 --cu-max 32 --weightp --colormatrix bt709 --colorprim bt709 --transfer bt709 --output-thread 1 --mv-precision q-pel
Well, it may not be a problem with --mt-cuda. The error will still occur after removing it.
Error Video encoding (v2.50.3 [v2.48.0])
NVEncC 9.04
D:\PlutoUser\Downloads\StaxRip-v2.50.3-x64\Apps\Encoders\NVEncC\NVEncC64.exe --avsdll D:\PlutoUser\Downloads\StaxRip-v2.50.3-x64\Apps\FrameServer\AviSynth\AviSynth.dll --vbr 1750 --codec h265 --preset p7 --profile main --vbr-quality 26 --multipass 2pass-full --qp-max 28:28:30 --qp-min 22:24:24 --aq --aq-temporal --bref-mode each --ref 8 --lookahead 32 --lookahead-level 3 --gop-len 60 --slices 1 --multiref-l0 7 --multiref-l1 3 --cu-min 8 --cu-max 32 --weightp --colormatrix bt709 --colorprim bt709 --transfer bt709 --output-thread 1 --mv-precision q-pel -i E:\x.vpy -o E:\x.h265
--------------------------------------------------------------------------------
E:\x.h265
--------------------------------------------------------------------------------
weighted prediction with B frames unsupported.
NVEncC (x64) 9.04 (r3432) by rigaya, Sep 29 2025 12:41:39 (VC 1944/Win)
OS Version Windows 11 x64 (26100) [UTF-8]
CPU AMD Ryzen 9 5900X 12-Core Processor [4.65GHz] (12C/24T)
GPU #0: NVIDIA GeForce RTX 3090 (10496 cores, 1695 MHz)[PCIe3x16][581.57]
NVENC / CUDA NVENC API 13.0, CUDA 13.0, schedule mode: auto
Input Buffers CUDA, 44 frames
Input Info VapourSynthMT r72(yv12)->nv12 [AVX2], 1600x900, 2997/100 fps
Vpp Filters copyDtoD
Output Info H.265/HEVC main @ Level auto
1600x900p 0:0 29.970fps (2997/100fps)
Encoder Preset quality
Rate Control VBR
Multipass 2pass-full
Bitrate 1750 kbps (Max: 11520 kbps)
Target Quality 26.00
QP range I:22-28 P:24-28 B:24-30
QP Offset cb:0 cr:0
VBV buf size auto
Split Enc Mode auto
Lookahead on, 32 frames, Level 3, Adaptive I, B Insert
GOP length 60 frames
B frames 3 frames [ref mode: each]
Ref frames 8 frames, MultiRef L0:7 L1:3
AQ on (spatial, temporal, strength auto)
CU max / min 32 / 8
VUI matrix:bt709,colorprim:bt709,transfer:bt709
Others mv:Q-pel
Video encoding returned exit code: -1073741819 (0xC0000005)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
StaxRip.ErrorAbortException: NVEncC 9.04
D:\PlutoUser\Downloads\StaxRip-v2.50.3-x64\Apps\Encoders\NVEncC\NVEncC64.exe --avsdll D:\PlutoUser\Downloads\StaxRip-v2.50.3-x64\Apps\FrameServer\AviSynth\AviSynth.dll --vbr 1750 --codec h265 --preset p7 --profile main --vbr-quality 26 --multipass 2pass-full --qp-max 28:28:30 --qp-min 22:24:24 --aq --aq-temporal --bref-mode each --ref 8 --lookahead 32 --lookahead-level 3 --gop-len 60 --slices 1 --multiref-l0 7 --multiref-l1 3 --cu-min 8 --cu-max 32 --weightp --colormatrix bt709 --colorprim bt709 --transfer bt709 --output-thread 1 --mv-precision q-pel -i E:\x.vpy -o E:\x.h265
--------------------------------------------------------------------------------
E:\x.h265
--------------------------------------------------------------------------------
weighted prediction with B frames unsupported.
NVEncC (x64) 9.04 (r3432) by rigaya, Sep 29 2025 12:41:39 (VC 1944/Win)
OS Version Windows 11 x64 (26100) [UTF-8]
CPU AMD Ryzen 9 5900X 12-Core Processor [4.65GHz] (12C/24T)
GPU #0: NVIDIA GeForce RTX 3090 (10496 cores, 1695 MHz)[PCIe3x16][581.57]
NVENC / CUDA NVENC API 13.0, CUDA 13.0, schedule mode: auto
Input Buffers CUDA, 44 frames
Input Info VapourSynthMT r72(yv12)->nv12 [AVX2], 1600x900, 2997/100 fps
Vpp Filters copyDtoD
Output Info H.265/HEVC main @ Level auto
1600x900p 0:0 29.970fps (2997/100fps)
Encoder Preset quality
Rate Control VBR
Multipass 2pass-full
Bitrate 1750 kbps (Max: 11520 kbps)
Target Quality 26.00
QP range I:22-28 P:24-28 B:24-30
QP Offset cb:0 cr:0
VBV buf size auto
Split Enc Mode auto
Lookahead on, 32 frames, Level 3, Adaptive I, B Insert
GOP length 60 frames
B frames 3 frames [ref mode: each]
Ref frames 8 frames, MultiRef L0:7 L1:3
AQ on (spatial, temporal, strength auto)
CU max / min 32 / 8
VUI matrix:bt709,colorprim:bt709,transfer:bt709
Others mv:Q-pel
Video encoding returned exit code: -1073741819 (0xC0000005)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
在 StaxRip.Proc.Start() 位置 D:\Code\Github\staxrip\staxrip\Source\General\Proc.vb:行号 367
在 StaxRip.NVEnc.Encode() 位置 D:\Code\Github\staxrip\staxrip\Source\Encoding\NVEnc.vb:行号 191
在 StaxRip.GlobalClass._Closure$__._Lambda$__162-7() 位置 D:\Code\Github\staxrip\staxrip\Source\General\GlobalClass.vb:行号 581
在 System.Threading.Tasks.Parallel.<>c__DisplayClass4_0.<Invoke>b__0()
--- 引发异常的上一位置中堆栈跟踪的末尾 ---
在 System.Runtime.ExceptionServices.ExceptionDispatchInfo.Throw()
在 StaxRip.GlobalClass.ProcessJob(String jobPath) 位置 D:\Code\Github\staxrip\staxrip\Source\General\GlobalClass.vb:行号 590
There seems to be a problem with version 9.0.4. I manually reverted the Nvencc version in StaxRip-v2.50.3 from the default 9.0.4 to 9.0.0, and I was able to transcode a file without crashing. Previously, it couldn't complete a full file transcoding.
The title has some errors, but testing suggests the issue isn't related to --mt-cuda, as the crash also occurs when removing that line in the test above.
This problem appeared again, and I began to wonder if this problem might have nothing to do with the NVencc version.
I rolled back StaxRip to version 2.48.3, and after four hours of encoding, the problem didn't reoccur. I then upgraded 2.48.3's default NVencc 8.1 to 9.0, and after another two hours of encoding, the problem didn't reoccur either.
At this point, I began to suspect that the issue was caused by other changes in the StaxRip 2.50.3 update.
The option --cuda-mt 1 can be unstable, so it's strongly discouraged for anything other than testing purposes. However, if instability was already present in version 8.11, it's unlikely to be the cause this time.
Regarding the update to StaxRip, I'm unfamiliar with the specific details. However, looking at the release log, it seems there are updates related to VapourSynth, which might potentially be the cause of the issue.
The option --cuda-mt 1 can be unstable, so it's strongly discouraged for anything other than testing purposes. However, if instability was already present in version 8.11, it's unlikely to be the cause this time.
Regarding the update to StaxRip, I'm unfamiliar with the specific details. However, looking at the release log, it seems there are updates related to VapourSynth, which might potentially be the cause of the issue.
I'm now absolutely certain that the issue lies with StaxRip 2.50.3. I've been using NvencC 0.9.0 on 2.48.3 for over 24 hours, and the aforementioned issue has not recurred. As for the --cuda-mt 1 issue you mentioned, I previously used this command with 2.48.3 and 0.9.0 without any issues, using NVENC API 13.0 and CUDA 13. I initially suspected NVENC was because during the 2.50.3 update, I noticed that it had been crashing frequently after being updated to 0.9.4. I subsequently rolled back to 0.9.0, which worked without crashes for an hour, thinking the issue was resolved. However, I discovered I was wrong; the issue had not been resolved.