Beta CUDA backend fails to decode MPEG-4 Part 2 videos

[ladhiet]

🐛 Describe the bug

Description

VideoDecoder with set_cuda_backend("beta") fails immediately on any MPEG-4 Part 2 (mpeg4) video with "Requested next frame while there are no more frames left to decode" on the very first get_frame_at() call. H.264 videos work fine with the same setup. The "ffmpeg" CUDA backend and CPU decoding both handle MPEG-4 videos correctly.

Minimal reproducible example

import subprocess
import tempfile
from pathlib import Path

import torch
import torchcodec.decoders

# Generate a 1-second mpeg4 test video
tmpdir = tempfile.mkdtemp()
path = Path(tmpdir) / "test.mp4"
subprocess.run([
    "ffmpeg", "-y", "-f", "lavfi",
    "-i", "testsrc=duration=1:size=320x240:rate=30",
    "-c:v", "mpeg4", "-pix_fmt", "yuv420p", str(path),
], capture_output=True, check=True)

# CPU: works
dec = torchcodec.decoders.VideoDecoder(path.as_posix(), device="cpu")
dec.get_frame_at(0)  # OK

# Beta CUDA: fails
with torchcodec.decoders.set_cuda_backend("beta"):
    dec = torchcodec.decoders.VideoDecoder(path.as_posix(), device="cuda")
dec.get_frame_at(0)  # RuntimeError: no more frames left to decode

# FFmpeg CUDA: works
with torchcodec.decoders.set_cuda_backend("ffmpeg"):
    dec = torchcodec.decoders.VideoDecoder(path.as_posix(), device="cuda")
dec.get_frame_at(0)  # OK

Environment

Tested on two configurations, both reproduce the issue:

• torchcodec 0.11.1+cu128 / torch 2.11.0+cu128
• torchcodec 0.11.1+cu126 / torch 2.11.0+cu126
• GPU: NVIDIA GeForce RTX 4090
• OS: Linux

I was unable to test on CUDA 13 so this is probably the first thing to do with the above example.

From what I can see, the issue seems related to the flush that occurs on the first frame access. We noticed that BetaCudaDeviceInterface::flush() sends CUVID_PKT_ENDOFSTREAM to the NVCUVID parser before any data packets have been sent. For H.264, the h264_mp4toannexb bitstream filter injects SPS/PPS headers inline so the parser can recover after the flush. MPEG-4 Part 2 doesn't have an equivalent mechanism — its codec configuration (VOS/VOL headers) lives only in extradata and is never embedded in the bitstream, so the parser can't re-initialize.

Versions

PyTorch version: 2.11.0+cu128
Is debug build: False
CUDA used to build PyTorch: 12.8
ROCM used to build PyTorch: N/A

OS: Ubuntu 24.04.1 LTS (x86_64)
GCC version: (Ubuntu 13.2.0-23ubuntu4) 13.2.0
Clang version: Could not collect
CMake version: version 3.28.3
Libc version: glibc-2.39

Python version: 3.12.11 (main, Apr 8 2026, 13:55:11) [GCC 13.2.0] (64-bit runtime)
Python platform: Linux-6.8.0-59-generic-x86_64-with-glibc2.39
Is CUDA available: True
CUDA runtime version: 12.6.85
CUDA_MODULE_LOADING set to:
GPU models and configuration: GPU 0: NVIDIA GeForce RTX 4090
Nvidia driver version: 570.133.20
cuDNN version: Probably one of the following:
/usr/lib/x86_64-linux-gnu/libcudnn.so.9.5.1
/usr/lib/x86_64-linux-gnu/libcudnn_adv.so.9.5.1
/usr/lib/x86_64-linux-gnu/libcudnn_cnn.so.9.5.1
/usr/lib/x86_64-linux-gnu/libcudnn_engines_precompiled.so.9.5.1
/usr/lib/x86_64-linux-gnu/libcudnn_engines_runtime_compiled.so.9.5.1
/usr/lib/x86_64-linux-gnu/libcudnn_graph.so.9.5.1
/usr/lib/x86_64-linux-gnu/libcudnn_heuristic.so.9.5.1
/usr/lib/x86_64-linux-gnu/libcudnn_ops.so.9.5.1
Is XPU available: False
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True
Caching allocator config: N/A

CPU:
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 48 bits physical, 48 bits virtual
Byte Order: Little Endian
CPU(s): 32
On-line CPU(s) list: 0-31
Vendor ID: AuthenticAMD
Model name: AMD Ryzen 9 7950X 16-Core Processor
CPU family: 25
Model: 97
Thread(s) per core: 2
Core(s) per socket: 16
Socket(s): 1
Stepping: 2
Microcode version: 0xa601206
CPU(s) scaling MHz: 63%
CPU max MHz: 5881.0000
CPU min MHz: 400.0000
BogoMIPS: 8983.98
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good amd_lbr_v2 nopl nonstop_tsc cpuid extd_apicid aperfmperf rapl pni pclmulqdq monitor ssse3 fma cx16 sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb cat_l3 cdp_l3 hw_pstate ssbd mba perfmon_v2 ibrs ibpb stibp ibrs_enhanced vmmcall fsgsbase bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local user_shstk avx512_bf16 clzero irperf xsaveerptr rdpru wbnoinvd cppc amd_ibpb_ret arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic vgif x2avic v_spec_ctrl vnmi avx512vbmi umip pku ospke avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq rdpid overflow_recov succor smca fsrm flush_l1d
Virtualization: AMD-V
L1d cache: 512 KiB (16 instances)
L1i cache: 512 KiB (16 instances)
L2 cache: 16 MiB (16 instances)
L3 cache: 64 MiB (2 instances)
NUMA node(s): 1
NUMA node0 CPU(s): 0-31
Vulnerability Gather data sampling: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Mitigation; Safe RET
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Enhanced / Automatic IBRS; IBPB conditional; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Not affected

Versions of relevant libraries:
[pip3] numpy==2.4.3
[pip3] nvidia-cublas-cu12==12.8.4.1
[pip3] nvidia-cuda-cupti-cu12==12.8.90
[pip3] nvidia-cuda-nvrtc-cu12==12.8.93
[pip3] nvidia-cuda-runtime-cu12==12.8.90
[pip3] nvidia-cudnn-cu12==9.19.0.56
[pip3] nvidia-cufft-cu12==11.3.3.83
[pip3] nvidia-curand-cu12==10.3.9.90
[pip3] nvidia-cusolver-cu12==11.7.3.90
[pip3] nvidia-cusparse-cu12==12.5.8.93
[pip3] nvidia-cusparselt-cu12==0.7.1
[pip3] nvidia-nccl-cu12==2.28.9
[pip3] nvidia-nvjitlink-cu12==12.8.93
[pip3] nvidia-nvtx-cu12==12.8.90
[pip3] torch==2.11.0+cu128
[pip3] torchcodec==0.11.1+cu128
[pip3] torchvision==0.26.0+cu128
[pip3] triton==3.6.0
[conda] Could not collect

[mollyxu]

Hi @ladhiet, thanks for flagging this and for providing a detailed example! I was able to reproduce this issue, we will look into it