Nvidia released RTX/Quadro 527.27 graphics driver for the RTX 6000 Ada workstation graphics card.

On the thermals side, both 13th Gen and Ryzen 7000 can be a handful if you leave them at default settings. Both platforms auto-overclock pretty aggressively out of the box, but if you turn that stuff off, thermals are pretty much a non-issue for both as long as you have a decent cooler. We have a couple articles for that if you want to take a look: https://www.pugetsystems.com/labs/articles/amd-ryzen-7950x-impact-of-precision-boost-overdrive-pbo-on-thermals-and-content-creation-performance-2373/#Ryzen_7950X_PBO_and_CPB_CPU_Temperature and https://www.pugetsystems.com/labs/articles/intel-core-i9-13900k-impact-of-multicore-enhancement-mce-and-long-power-duration-limits-on-thermals-and-content-creation-performance-2375/#Core_i9_13900K_MCE_and_P1P2_Limits_CPU_Temperature . We aren’t looking at power usage right now, but that is something we do want to add in the future. We ideally would want to measure total system power draw from the wall, but haven’t had much success finding a good power meter that can do USB/network logging. Most of our Labs team is remote, so we really need something better than a kill-a-watt.

No real comment on the Torrent. We do all our testing in open-air test beds to remove the chassis cooling as a factor. I think the product team have looked at the Torrent, but quite operation is such a huge thing to our customers that I kind of doubt we would ever move to the Torrent as a primary chassis. Does look nice though!

As for Quick Sync, hardware encoding is slightly lower quality than software, but most people take the hit in exchange for how much faster it is. You can combat the quality difference just by turning the bitrate up a bit, but for most delivery that is going onto Youtube or sites like that, the main consensus I have seen in the pro field is that no one is going to be able to tell the difference. If you are using H.264/HEVC as a master/archive format, however, I could totally see wanting as much quality as possible for the file size. You are probably going to do things like 2-pass encoding in that situation as well, which you usually can’t do with hardware encoding.

Quality is something I actually have on my to-do list. Probably won’t get to it until early next year, but I would love to do an analysis on NVENC vs Quick Sync vs AMD encoding speed an quality. And do the same on the decoding side as well (although that is much harder to test, and not as important).

Another option is Redshift which also has GPU out-of-core rendering.

My system with GPU disabled in Lightroom
https://benchmarks.pugetsystems.com/benchmarks/view.php?id=116609
Overall Score: 937 Active: 67.6 Passive: 119.8

My system with GPU enabled in Lightroorm (RX-590
https://benchmarks.pugetsystems.com/benchmarks/view.php?id=116631
Overall score: 976 Active: 74.5 Passive: 120.7

My system with GPU enabled in Lightroorm and X.M.P. enabled (I just learned about this). This boosted my RAM from 2133MHz to 3200MHz
https://benchmarks.pugetsystems.com/benchmarks/view.php?id=116733
Overall score: 1130.5 Active: 86.1 Passive: 140

Seems that RAM speed may be important for Lightroom! I got a quite a nice little speed boost (16%) from one little change in BIOS,

Makes sense from a efficiency standpoint with a the releases back to back! Thanks!

@Matt_PugetSystems:disqus so it looks like 7950x and 13900K are neck and neck when it comes to video editing (Resolve, Premiere), but 7950x has an edge with Blender and 3D rendering in general? If so, did you guys also compare heat, power usage, and throttling, as well as which does better on air cooling (since Puget focuses on air cooling). I’ve read that 13900K will easily reach 100c on air, while the 7950x may be more easy to tame on air. Further, I’ve read you may be able to reduce the heat/throttling of 13900K by undervolting. So I’m wondering, overall, in terms of heat, power usage, and throttling, how they fare against each other.

BTW, while on the subject of cooling. Has Puget ever considered Fractal Torrent? I know you guys like the Define series for quiet systems, due to the sound dampening features, but I’ve been impressed with how quiet the torrent can be with lower RPM, larger front and bottom fans.

Lastly, speaking on the edge of Intel CPU with QuickSync, have you ever seen the articles comparing resulting video quality of the final output when using QuickSync vs dGPU vs CPU? Supposedly, CPU, while it can be slower, may produce the best final result. I read somewhere QuickSync final result quality may not be as good as dGPU or CPU (one or the other, or both). Perhaps the difference is hard to make out except upon extremely close frame-by-frame comparisons that most people won’t be doing?

Thanks!