Awesome, that’s the same card i’m getting. I was looking to undervolt it too. Are using Adrenaline for the undervolt or something else?
1 Like
In the past i did read that the ones from saphirre are best.
I’m using the Adrenaline software, it works very well and I’ve always been happy with it. My only complaint is the memory runs to hot - 85-90c on average. Not to hot that it is an issue but it would be nicer if it ran cooler. I’ve been thinking about replacing the pads on the memory with Thermal Grizzles putty pro.
1 Like
I would like to know does the benchmarks results reflect real world performance when rendering videos using Rhea or Artemus?
Have you noticed it being faster or slower.
Thanks
Here’s my benchmark comparison 4070 Ti SUPER vs 9070 XT, both at Stock>
Topaz Video AI v6.1.3
System Information
OS: Windows v11.24
CPU: AMD Ryzen 7 7800X3D 8-Core Processor 31.586 GB
GPU: NVIDIA GeForce RTX 4070 Ti SUPER 15.687 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 25.95 fps 2X: 17.47 fps 4X: 04.62 fps
Iris 1X: 26.35 fps 2X: 15.86 fps 4X: 04.34 fps
Proteus 1X: 25.25 fps 2X: 16.98 fps 4X: 05.87 fps
Gaia 1X: 08.87 fps 2X: 06.13 fps 4X: 04.09 fps
Nyx 1X: 10.44 fps 2X: 09.20 fps
Nyx Fast 1X: 19.01 fps
Rhea 4X: 03.56 fps
RXL 4X: 03.13 fps
Hyperion HDR 1X: 27.69 fps
4X Slowmo Apollo: 35.64 fps APFast: 83.15 fps Chronos: 19.73 fps CHFast: 30.14 fps
16X Slowmo Aion: 37.28 fps
Topaz Video AI v6.1.3
System Information
OS: Windows v11.24
CPU: AMD Ryzen 7 7800X3D 8-Core Processor 31.586 GB
GPU: AMD Radeon RX 9070 XT 15.374 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 30.85 fps 2X: 14.91 fps 4X: 04.33 fps
Iris 1X: 34.07 fps 2X: 17.93 fps 4X: 05.48 fps
Proteus 1X: 34.63 fps 2X: 18.79 fps 4X: 05.04 fps
Gaia 1X: 16.43 fps 2X: 10.44 fps 4X: 05.22 fps
Nyx 1X: 13.46 fps 2X: 10.81 fps
Nyx Fast 1X: 25.40 fps
Rhea 4X: 03.58 fps
RXL 4X: 03.41 fps
Hyperion HDR 1X: 30.70 fps
4X Slowmo Apollo: 33.33 fps APFast: 73.29 fps Chronos: 21.05 fps CHFast: 31.77 fps
16X Slowmo Aion: 52.92 fps
Artemis at 2x seems oddly slow compared to the rest of the results. I thought maybe memory bandwidth related, but overclocking the RAM saw little change. GPU core overclock is more beneficial. Could just need a Topaz optimized release, but out of the box this is stellar.
1 Like
faster than mine as well even with a 9900x. I was directly comparing the 7900xtx to 4070 super and the 4070 beat it by a about 1 fps in the 2x column. the 7900xtx cost more and ran hotter so I returned it.
These are incredible results, almost on par with a 4090 for gaia and nyx!
I have had trouble with the 7900xtx benchmarking better than it actually ran. Can you confirm that the 9070xt is indeed significantly faster than 4070ti super in real use? Espeically gaia? If these results are real then I am definitely going to have to get one of these at some point.
1 Like
Who really uses Artemis though? Once you learn how to properly use the different Iris models and Rhea, there is no need for Artemis. It always does a worse job and is sloiwer. Iris, Rhea and Nyx are all I really use and they excel on the 9070 XT.
1 Like
But then Iris (except for LQ V1) and Rhea both create „monster faces“ and also sometimes really ugly repetitive patterns especially for grass, trees and rocks in the background.
Artemis doesn’t do that. And e.g. on the Mac also is considerably faster than Iris.
(That doesn’t mean that I would use Artemis - I don’t- but still I can see reasons for others to still use it)
1 Like
I use Iris V1, Iris Med and Rhea 99% of the time. I don’t get monster faces with rhea unless I have tuned the settings wrong. Artemis usually strips out a lot of details like skin texture. After 1000+ renders I have never had a situation where Artemis does a better job,
I use Artemis Medium Dehalo almost exclusively in TVAI with a source blend of 75-90%. I mostly use it with noisy, edge enhanced, HD movie sources and this combo produces the most consistent natural output. I don’t have to worry about a weird result with this. I don’t normally output using the 2x model anyway, so still a nice speed boost for me.
1 Like
Well, then you’re lucky and have the right source material.
But, regardless of the settings Rhea will create one or the other monster face in low res sources with small faces in the background - not as bad as Iris MQ, though.
Plus the issue with repetitive low res patterns that can totally ruin an otherwise near perfect upscale.
As said, I don’t often use Artemis myself (while for some difficult material Gaia still can be useful) - but I don’t do that „it works for me -so it has to work for everyone…“
I don’t want to reinstall the SUPER in the same PC build again to demo a proper result, it’s a small form factor set-up, pain in the butt for time. That said, I did run GAIA on 1920x800p output recently. I don’t recall exactly how fast it was on the SUPER, but it was definitely slower than 11 FPS and the SUPER was overclocked at the time. The 9070XT on the same source settled at just over 16 FPS, fluctuating higher for simple content like credits and company logos.
I totally agree with the low res pattern issue. It seems to occur when there are blurry things not in focus. A very annoying thing I run into with some renders. In those cases I use Iris V1.
You don’t get crazy looking faces with Artemis? It was always so terrible with teeth. It drove me nuts.
1 Like
Hey, I’ve noticed on average it is around 10-20% faster real world. However, I just have the trial. I don’t know if the trial limits the speeds or not.
Device: -2 means “Auto”. This I always remember due a bug from ages ago where a particular TVAI version changed the processor device to Auto. This severely affected performance on Apple Silicon and took ages for Topaz to rectify the issue. It appears the effects are still lingering. Whenever I see device: -2 in the Benchmarks, I always wonder if the user intentionally set it to Auto or if it’s because of that bug.
Thanks.
Andy
1 Like
No, i just took it out of the box.
It showed better results on another benchmark:
Topaz Video AI v6.1.3
System Information
OS: Windows v11.24
CPU: AMD Ryzen 9 9900X 12-Core Processor 61.613 GB
GPU: AMD Radeon RX 9070 XT 15.374 GB
GPU: AMD Radeon(TM) Graphics 1.9548 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 28.66 fps 2X: 10.16 fps 4X: 02.69 fps
Iris 1X: 33.17 fps 2X: 11.62 fps 4X: 03.56 fps
Proteus 1X: 33.66 fps 2X: 13.12 fps 4X: 03.70 fps
Gaia 1X: 17.10 fps 2X: 09.85 fps 4X: 03.44 fps
Nyx 1X: 13.25 fps 2X: 10.28 fps
Nyx Fast 1X: 24.99 fps
Rhea 4X: 02.99 fps
RXL 4X: 02.93 fps
Hyperion HDR 1X: 26.99 fps
4X Slowmo Apollo: 28.01 fps APFast: 66.99 fps Chronos: 20.33 fps CHFast: 28.46 fps
16X Slowmo Aion: 45.84 fps
1 Like
What’s your RAM speed? This is almost exactly the hardware I’m looking to buy.
So, I finally got my 9070 XT and it was basically DOA. While testing it after it install it began to run worse and worse and was benchmarking slower than my 7900 XTX. Then it just starting making my entire PC start to act crazy. I put in for an RMA to have it repaired instead of returning it since I have no idea when I would be able to get another one.
