The encoding process is taking a long time.
The encoding process is taking a long time.
It appears the encoding speed depends on several factors. A high bitrate video at 10000Kbps generally requires longer processing times—around an hour, possibly closer to 45 minutes—especially when CPU is operating near full capacity. On your ROG Strix G17 with a 7940hx and 3070 processor, the GPU contributes significantly, but it's still under heavy load, which slows things down compared to more powerful systems like the 5800x3D. The main concern seems to be that the CPU is often busy, limiting performance. It might help to check if AV1 encoding is supported on your desktop model (3070) or if other codecs are being used.
The AV1 encoding support was introduced in the 40-series, meaning systems with 30-series might experience slower CPU performance.
4000 series includes both hardware and software encoding, while the 3000 series only supports decoding. This means you're handling encoding on the CPU. The performance gap will vary based on CPU settings and other configurations. Generally, the 7940hx tends to perform better in CPU encoding compared to the 5800X3D for AV1.
It seems like AV1 was promoted with 30 series as well, though decoding it can be tricky. The R9 also appears to handle encoding in CapCut, with both CPU and GPU usage around 30%. It runs very quickly, making it hard to verify. The file sizes are much smaller compared to others, which is a big plus. I think I'll switch to a different encoder on my desktop, particularly for longer videos.
The adjustments might vary a bit, but overall they remain quite alike. Capcut has many less obvious options, such as app data installations, though the core features stay similar. The main distinction usually comes down to hardware differences—your gaming laptop is significantly more powerful and quiet compared to a standard PC with basic cooling. In short, 32 threads definitely make a difference. Tldr: performance and silence are where it stands.
Your 15W 7540U in a laptop runs about 35% slower in Blender compared to your 5800X3D, even though it has fewer cores and uses significantly less power—Mobile Zen4 definitely delivers.
It seems confusing why Nvidia would handle decoding and encoding separately after two years. Wouldn't that make AV1 outdated for that timeframe? It appears to affect video quality, since CPU performance versus hardware acceleration might not match up perfectly. The actual output depends on both the algorithms and the hardware, so even if the device can't run everything smoothly, the CPU can still handle tasks if settings align.
AMD did the same. AV1 decode in RDNA2, AV1 encode in RDNA3. The most common end user use case is to play back already encoded content, which only requires decode. Encoding is a more specialised niche. It might also more be about playback getting support out there so that it is worth people encoding to AV1. You trade off speed and quality in how you encode. GPU encoders might have leaned towards speed over quality, even at their highest settings. CPU has generally been accepted as best quality, but Nvidia GPUs were good enough to use for most. AMD said RDNA4 had much improved encoding quality compared to previous, not that it is so commonly used.
Decoding has been supported for years, not as a breakthrough. Intel, AMD, Nvidia, MediaTek all contributed. It's straightforward; no hidden intent from Nvidia to skip encoding. Hardware support came when major companies were ready, and its value remains limited compared to HEVC. Once Twitch rolls out AV1 ingestion for regular users (not just the beta), it might become more practical. Most attempts to re-encode local media are likely wasting effort and reducing file quality.