Heh, "not maxing out" your CPU means nothing, except that it is not the system bottleneck; that bottleneck "could be" the source, the decoder, the pipeline engine, the pipeline filters, the cpu, the gpu, the encoder, or simply your expectations.

Intel QSV is a hardware encoder, which reduces and balances the rendering load between your CPU and GPU, depending on their respective capabilities and their roles in each of the functions above.

I have a weak CPU with a weak iGPU. My QSV encodes consistently run 95% GPU, and 22% CPU.

So 60% CPU is relatively balanced; embrace it and feel validated.

That's not how anything works. There's no render scenario in which your CPU and GPU will be pegged at 100%.

That's not how anything works. There's no render scenario in which your CPU and GPU will be pegged at 100%.

Indeed, and I am sure if it would be running at 100%, thermal throttling will kick in sooner or later regardless of any fancy cooling.

@MarkAnthony121 Some tweaks you could be trying out to speed up rendering:

1. Maybe obvious, lowering quality with Voukoder (not using the default)
2. Disable e-cores : see https://www.vegascreativesoftware.info/us/forum/rendering-2x-faster-when-restricting-to-p-cores-for-alder-lake--137188/
3. Disable igpu - the internal graphic unit of the cpu (you lose QSV, but gain CPU thermal capacity and memory)

@MarkAnthony121 Voukoder default renders CPU only, you can select GPU support also in Voukoder.

A short render test on my system: Magix AVC NV took 1:46, with Voukoder default 2:00 and with Voukoder NV 1:28.

But for max Quality at min Filesize, I prefer Voukoder default.

What's your GPU?

I'd try QSV in Voukoder for a possible speed boost. I'd leave the iGPU there as it really helps with decoding, at least for certain media and QSV encoding is fast, perfect for draft renders.

If you want to go that route you can play with voltages and core clock speeds if you have some extra thermal overhead you are willing to sacrifice for render speed.

Feel free to play around with this benchmark and see what helps or hurts your speeds:
https://forms.gle/PCD4TT1ynEXKwpD17

This one is quite taxing on GPUs so hopefully you have a decent one or the CPU will be waiting for it.

@MarkAnthony121 Voukoder default renders CPU only, you can select GPU support also in Voukoder.

A short render test on my system: Magix AVC NV took 1:46, with Voukoder default 2:00 and with Voukoder NV 1:28.

But for max Quality at min Filesize, I prefer Voukoder default.

Ahh so Voukoder isn't necessarily for speedy renders. It's for quality renders and best possible file sizes? That would be so important to me. I do full wedding renders and uploads where with my render settings a 2 hour 4K video at a crappy 14,000,000 bitrate turns out 20GB. And when I pay $1000 for my online album delivery platform, every GB counts.

What's your GPU?

I'd try QSV in Voukoder for a possible speed boost. I'd leave the iGPU there as it really helps with decoding, at least for certain media and QSV encoding is fast, perfect for draft renders.

If you want to go that route you can play with voltages and core clock speeds if you have some extra thermal overhead you are willing to sacrifice for render speed.

Feel free to play around with this benchmark and see what helps or hurts your speeds:
https://forms.gle/PCD4TT1ynEXKwpD17

This one is quite taxing on GPUs so hopefully you have a decent one or the CPU will be waiting for it.

You said for drafts. Are you saying if I'm rendering my final deliverable wedding films QSV isn't ideal? With higher bitrate it seems to looks pretty good. Unless the whole point is that my file sizes will be too big?

My understanding is that QSV is inefficient in terms of quality/file size. While very fast there's a tradeoff especially if you care about GB for uploads.

Personally I use x264 for final renders (unless it's a very long project) and while it taxes the CPU more and takes longer, gives nice crisp video at reasonable bitrates. HEVC/h265/x265 is even more space efficient but not as compatible with devices so it depends on how the delivery platform works and what the clients receive (is it re-encoded?) You can get great results with less than half the bitrate of AVC.

Anyone running an i9-13900K with any tips for system optimization for the fastest rendering possible? I've been experimenting with Voukoder as well but it usually turns out slower than Vegas' own built in rendering. The fastest along with best balance of quality that I've found for UHD 2160 is Magix AAC/AVC with Intel QSV encoding. But still the CPU is not maxing out, maybe 60%. Any other encode modes turn out slower for me. Any ideas or BIOS tweaks anyone else is using to maybe unlock more of the full potential of the CPU?

The only CPU-related bios tweaks I've found significant on my 11900k system with an Asus motherboard have to do with core affinity. I essentially monitored core temps while running prime95 and lowered affinity on those running the hottest. But I think the p- and e-core separation on 13th gen Intels probably makes that kind of tweaking unnecessary.

Of the AVC render templates, I get the best overall quality and performance from Magix Nvenc and Qsv, in that order, but only separated by the slimmest of margins. Render performance will vary significantly, however, depending on FX loading on the gpu selected in Video prefs. I often get better performance from Qsv rendering in projects with allot of FX going to a higher powered AMD gpu. Same with my laptop and it's medium powered Nvidia 3060. For the highest quality renders, I choose Magix Hevc 10-bit MainConcept. But Qsv 10-bit Hevc is significantly faster and often generates smaller uploads that are not that far off in quality compared to MainConcept.

To guarantee you're having a good look at things, I recommend using a high quality 4k monitor. I love my BenQ at home and am having good success with my 4k Zenbook display and a 4k Samsung tv on the road.

Yes, typically using CPU-only encodes produces the highest quality results, even if they take the longest. In video world, speed comes at the cost of quality.

As I've written before, despite the objective quality metrics, Which John Dennis graciously supplied, I find that bit-for-bit I achieve the best results with Render+ via HOS. I also find it easy to work with.

Anyway the reason I'm writing is that Render+ with my 13900K produces "hotspots" in the CPU. In short, a hotspots in the CPU are locations where a core or cores are very highly loaded and the concentration of heat is hard to dissipate effectively because it's such a small area. Render+ x265 really heats up some of the cores in my CPU and over 5.2GHz or so will cause throttling.

I have a 280mm AIO water cooler that can dissipate about 250W when running Cinebench but I have a feeling that loads all cores evenly but less than the fewer cores Render+ is hitting.

What I have done to mitigate this issue while still maintaining performance in other applications is in the BIOS set P core max clocks as follows. 1-2 Ps active 5.5GHz, 1-4 P's active 5.2GHz, all P's active 5.0GHz. This keeps temps under control in Render+, which isn't a huge power draw but has the hotspot issue. But when I'm working in PS or some other apps that may only use 1-4 cores allows those cores to ramp up to very high clocks for the few seconds I need to quickly render something to keep the creative juices flowing without waiting.

Just something to consider. Perhaps if you have a monster 420 AIO it isn't a problem or a really good SP rated chip you won't have the issue but I thought I'd mention my experience.

Mark

The Intel Extreme Tuning utility has a profiles feature where you can adjust performance for specific applications. Perhaps you could set some specific settings that are optimal for Vegas but don't interfere with other applications.