Nothing overclocked, the 4070 Ti Super is a VERY quiet GPU (MSI 3X).
Topaz Video AI v4.2.1
System Information
OS: Windows v11.23
CPU: 13th Gen Intel(R) Core(TM) i7-13700KF 31.842 GB
GPU: AMD Radeon RX 7900 XTX 23.938 GB
Processing Settings
device: 0 vram: 1 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 17.44 fps 2X: 09.50 fps 4X: 03.26 fps
Iris 1X: 21.61 fps 2X: 11.39 fps 4X: 03.79 fps
Proteus 1X: 20.49 fps 2X: 12.11 fps 4X: 04.49 fps
Gaia 1X: 09.25 fps 2X: 06.12 fps 4X: 03.61 fps
Nyx 1X: 08.17 fps 2X: 06.68 fps
Nyx Fast 1X: 15.75 fps
4X Slowmo Apollo: 26.46 fps APFast: 59.73 fps Chronos: 13.47 fps CHFast: 19.87 fps
16X Slowmo Aion: 29.45 fps
关于cpu和gpu搭配的几点建议:
1.benchmark score的影响。benchmark score只能作为参考,并不能准确反映你的电脑真实的速度。我平时使用Proteus模型,在参考user benchmarks时,4x ,13700k 搭配4090能大道6.xfps,13900k搭配 4070ti super也能达到6.xfps,我选择升级为13700k 搭配4080 super,以为能够达到比较高的fps,结果只能达到4.2fps。所以,不要相信其他用户的benchmark score,较高的fps可能是采用了某些特殊的魔法。另外,同等价位下,amd 和NVIDIA的gpu区别不大。
2.cpu 的利用率的影响。benchmarking时 cpu 的利用率很低,不到50%,功耗也只有130多w。正常使用时,cpu 利用率能达到70%左右,功耗达到200w。如果同时运行2个任务,cpu 利用率能达到90%多,功耗会达到210多w,内存占用会翻倍,但是总的fps只能提高5%。
3.超频的影响。cpu 和 gpu 超频能够提升5%—10%的fps。内存一定要开启xmp,否则性能会大幅下降。我的13700k超频到5.7g8+4.6g8使用。超频会导致稳定性的大幅下降,特别是降低cpu电压后,benchmark score提升较大,但是日常使用容易奔溃,得不偿失。所以一定要经过压力测试。
4.关闭e核和超线程的影响。关闭e核和超线程能提高benchmark score,但是不能提升日常使用时的fps。
5.视频的长短的影响。较长的视频处理速度会较慢。比如,一段30分钟的视频,前1分钟能达到3fps,前3分钟能达到2fps,第5分钟之后只能达到1.6fps。所以,似乎可以把较长的视频分割为较短的视频来提高工作效率。
Several suggestions on the combination of CPU and GPU:
- The impact of benchmark score. The benchmark score can only be used as a reference and cannot accurately reflect the true speed of your computer. I usually use the Proteus model, and when referring to user benchmarks, 4x, 13700k with 4090 can achieve 6.xfps, and 13900k with 4070ti super can also achieve 6.xfps. I chose to upgrade to 13700k with 4080 super, thinking that it could achieve a relatively high fps, but the result was only 4.2fps. So, don’t trust the benchmark score of other users, higher fps may be due to some special magic. Additionally, at the same price point, there is not much difference in GPU between AMD and NVIDIA.
- The impact of CPU utilization. During benchmarking, the CPU utilization rate is very low, less than 50%, and the power consumption is only over 130 watts. When used normally, the CPU utilization rate can reach about 70% and the power consumption can reach 200w. If two tasks are run simultaneously, the CPU utilization can reach over 90%, power consumption can reach over 210 watts, memory usage will double, but the total FPS can only be increased by 5%.
- The impact of overclocking. CPU and GPU overclocking can increase fps by 5-10%. Memory must have xmp enabled, otherwise performance will significantly decrease. My 13700k overclocked to 5.7g8+4.6g8 for use. Overfrequency can lead to a significant decrease in stability, especially when the CPU voltage is lowered, resulting in a significant improvement in benchmark score. However, it is prone to crashing during daily use and the benefits outweigh the losses. So it is necessary to undergo stress testing.
- The impact of shutting down the e-core and hyper threading. Turning off e-core and hyper threading can improve benchmark score, but it cannot improve fps for daily use.
- The impact of video length. The processing speed of longer videos will be slower. For example, a 30 minute video can reach 3fps in the first minute, 2fps in the first 3 minutes, and only 1.6fps after the 5th minute. So, it seems possible to divide longer videos into shorter ones to improve work efficiency.
The % usage of the CPU is not a good indicator of the work the CPU is doing because the software can hammer a few cores while letting the rest of the cores do little to no work. However the speed of production could still be limited by how quickly those few maxed-out cores can handle their share of processing each frame.
It’s important to look at benchmarks of CPU / GPU when they are tested together. The performance of a component should not be considered in a vacuum because it could easily be limited by the other component It’s paired with. So I try to use the benchmarks to judge the speed of CPU / GPU combos.
My 5950X / 3080TI rig had parity btween the CPU and GPU such that upgrading either component would net me no meaningful improvement because the upgraded component would be held back by the component that didn’t get upgraded.
Topaz Video AI v4.2.1
System Information
OS: Windows v11.23
CPU: AMD Ryzen 9 7950X 16-Core Processor 63.139 GB
GPU: NVIDIA GeForce RTX 4080 15.674 GB
GPU: AMD Radeon(TM) Graphics 0.47445 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 29.14 fps 2X: 15.37 fps 4X: 04.43 fps
Iris 1X: 28.53 fps 2X: 18.80 fps 4X: 04.83 fps
Proteus 1X: 28.20 fps 2X: 19.39 fps 4X: 05.42 fps
Gaia 1X: 10.34 fps 2X: 07.05 fps 4X: 04.48 fps
Nyx 1X: 11.72 fps 2X: 10.54 fps
Nyx Fast 1X: 21.26 fps
4X Slowmo Apollo: 40.41 fps APFast: 86.60 fps Chronos: 22.32 fps CHFast: 33.04 fps
16X Slowmo Aion: 44.74 fps
After a little tuning with bumping DDR4-3600 → DDR4-3800 (1.38V) and a light default overclock with AMD Ryzen Master using Precision Mode. I get this (wow @ Iris 1x):
Topaz Video AI v4.2.1
System Information
OS: Windows v11.22
CPU: AMD Ryzen 9 3950X 16-Core Processor 31.928 GB
GPU: NVIDIA GeForce RTX 4090 23.576 GB
GPU: NVIDIA GeForce RTX 4090 23.576 GB
Processing Settings
device: 0 vram: 1 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 36.03 fps 2X: 14.65 fps 4X: 04.00 fps
Iris 1X: 53.17 fps 2X: 16.09 fps 4X: 04.23 fps
Proteus 1X: 42.75 fps 2X: 15.47 fps 4X: 04.37 fps
Gaia 1X: 16.53 fps 2X: 10.87 fps 4X: 03.65 fps
Nyx 1X: 18.34 fps 2X: 14.48 fps
Nyx Fast 1X: 32.81 fps
4X Slowmo Apollo: 41.20 fps APFast: 61.26 fps Chronos: 32.00 fps CHFast: 30.29 fps
16X Slowmo Aion: 31.55 fps
AsRock X570 Phantom Gaming-ITX/TB3 Mainboard (BIOS vP5.60)
AMD Ryzen 9 3950X
16GB x2 (32GB) DDR4-3800 16/19/19/39/85/1T
MSI 4090 Suprim Liquid X AIO
nVidia Studio Driver v551.86 (03/19/2024)
Topaz Video AI v4.2.2
System Information
OS: Windows v11.23
CPU: 12th Gen Intel(R) Core(TM) i5-12500 63.745 GB
GPU: NVIDIA GeForce RTX 4060 7.7705 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 10.39 fps 2X: 07.37 fps 4X: 02.31 fps
Iris 1X: 10.78 fps 2X: 06.44 fps 4X: 01.76 fps
Proteus 1X: 10.29 fps 2X: 07.11 fps 4X: 02.47 fps
Gaia 1X: 03.45 fps 2X: 02.35 fps 4X: 01.60 fps
Nyx 1X: 04.10 fps 2X: 03.52 fps
Nyx Fast 1X: 07.44 fps
4X Slowmo Apollo: 15.82 fps APFast: 40.59 fps Chronos: 08.15 fps CHFast: 12.54 fps
16X Slowmo Aion: 05.45 fps
Topaz Video AI v4.2.2
System Information
OS: Windows v10.22
CPU: AMD Ryzen 9 5950X 16-Core Processor 31.926 GB
GPU: AMD Radeon RX 6800 XT 15.954 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 12.80 fps 2X: 07.13 fps 4X: 02.42 fps
Iris 1X: 14.81 fps 2X: 07.93 fps 4X: 02.51 fps
Proteus 1X: 14.17 fps 2X: 08.80 fps 4X: 03.06 fps
Gaia 1X: 06.04 fps 2X: 04.23 fps 4X: 02.54 fps
Nyx 1X: 05.62 fps 2X: 04.82 fps
Nyx Fast 1X: 10.66 fps
4X Slowmo Apollo: 14.55 fps APFast: 35.54 fps Chronos: 08.68 fps CHFast: 14.93 fps
16X Slowmo Aion: 35.58 fps
Topaz Video AI v4.2.2
System Information
OS: Windows v11.23
CPU: Intel(R) Xeon(R) w5-2465X 127.25 GB
GPU: NVIDIA GeForce RTX 4070 11.73 GB
GPU: NVIDIA RTX A4000 15.79 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 18.20 fps 2X: 12.63 fps 4X: 03.75 fps
Iris 1X: 18.34 fps 2X: 11.28 fps 4X: 03.24 fps
Proteus 1X: 17.85 fps 2X: 12.23 fps 4X: 04.15 fps
Gaia 1X: 06.27 fps 2X: 04.37 fps 4X: 02.97 fps
Nyx 1X: 07.31 fps 2X: 06.09 fps
Nyx Fast 1X: 13.43 fps
4X Slowmo Apollo: 25.33 fps APFast: 54.86 fps Chronos: 13.17 fps CHFast: 20.56 fps
16X Slowmo Aion: 22.95 fps
RTX4070 80%MAX TDP
Topaz Video AI v4.2.2
System Information
OS: Windows v11.23
CPU: Intel(R) Core(TM) i7-9700K CPU @ 3.60GHz 31.914 GB
GPU: NVIDIA GeForce RTX 4090 23.576 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 42.76 fps 2X: 12.25 fps 4X: 03.10 fps
Iris 1X: 41.02 fps 2X: 13.31 fps 4X: 03.34 fps
Proteus 1X: 32.85 fps 2X: 13.42 fps 4X: 03.52 fps
Gaia 1X: 13.77 fps 2X: 09.50 fps 4X: 02.91 fps
Nyx 1X: 16.04 fps 2X: 12.52 fps
Nyx Fast 1X: 27.27 fps
4X Slowmo Apollo: 28.65 fps APFast: 53.97 fps Chronos: 29.88 fps CHFast: 25.91 fps
16X Slowmo Aion: 25.07 fps
感谢分享,我目前还是老配置9700k + 4090FE + 32G (8GBx4) DDR4 3200Mhz (已开XMP),一直在看不同版本的benchmark作为对比和升级参考。我基本上用Proteus 1080p → 4k超电影,平均都1个半小时左右,FPS速度基本上是3.8 - 4.2FPS之间,9700k的使用率基本在7 80%,GPU使用不很高。之前我9700k也超过5.0Ghz,如你所述一样,跑着跑着经常蓝屏,只能用默频最高4.7Ghz使用了。
现就想看看怎么升级提高FPS同时实惠不贵,我感觉CPU的瓶颈和大小核数量有关,9700k的8C8T实在是拖后腿,不知道如果只升到13600k + 32G (16Gx2) DDR5 6400Mhz的配置能不能有接近两倍的提升,总体我看下来估计还是得13700k (14700k) 或 13900k才行,只不过价格已经上去很多了。
Topaz Video AI v4.2.2
System Information
OS: Windows v10.22
CPU: AMD Ryzen Threadripper 3970X 32-Core Processor 127.87 GB
GPU: NVIDIA GeForce RTX 3090 23.77 GB
Processing Settings
device: 0 vram: 1 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 25.79 fps 2X: 13.85 fps 4X: 03.80 fps
Iris 1X: 22.85 fps 2X: 13.94 fps 4X: 03.69 fps
Proteus 1X: 24.34 fps 2X: 14.23 fps 4X: 04.18 fps
Gaia 1X: 08.63 fps 2X: 05.88 fps 4X: 03.17 fps
Nyx 1X: 10.43 fps 2X: 08.58 fps
Nyx Fast 1X: 20.32 fps
4X Slowmo Apollo: 32.99 fps APFast: 53.13 fps Chronos: 19.38 fps CHFast: 26.71 fps
16X Slowmo Aion: 28.12 fps
我目前使用13700k超频到p核5.7g、e核4.6g,加上4080super,内存32g ddr4 3400mhz。平时主要使用proteus 从4k提升到8k,或者2k提升到4k。
我之前使用的是3900x+2060,这个搭配下cpu 和gpu基本都全速运行。后来我升级了显卡到4080super,结果cpu出现了瓶颈,总体提升只有50%左右。
后来我又升级了cpu到13700k,结果是13700k依然是瓶颈。gpu使用率依然只有一半左右。
其中,2k提升到4k分辨率对cpu的要求比2k或4k提升到8k的要求低。提升到4k时,gpu使用率会高一些。
对于您的配置,如果升级cpu,估计14900k是比较合理的,散热较好不降频的话,性能估计是9700k的4倍左右的提升。
如果您追求性价比,个人建议使用13700k。性能大概是14900k的四分之三,9700k的3倍。
如果您不着急,也可以等intel的15代cpu,据悉取消了超线程,性能提升较大。我自己测试,超线程除了提升跑分分数外,在ai处理视频时基本没有用处,还会增加功耗和温度。
Hi guys,
3090Ti is undervolted 2055 MHz @ 0,983 V
Topaz Video AI v4.2.2
System Information
OS: Windows v11.23
CPU: AMD Ryzen 7 5800X3D 8-Core Processor 31.931 GB
GPU: NVIDIA GeForce RTX 3090 Ti 23.743 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 24.18 fps 2X: 12.48 fps 4X: 03.55 fps
Iris 1X: 23.95 fps 2X: 14.38 fps 4X: 04.08 fps
Proteus 1X: 22.91 fps 2X: 15.35 fps 4X: 04.04 fps
Gaia 1X: 08.85 fps 2X: 06.17 fps 4X: 03.52 fps
Nyx 1X: 10.83 fps 2X: 08.61 fps
Nyx Fast 1X: 20.20 fps
4X Slowmo Apollo: 33.19 fps APFast: 57.27 fps Chronos: 18.67 fps CHFast: 28.73 fps
16X Slowmo Aion: 28.08 fps
Topaz Video AI v4.2.2
System Information
OS: Windows v11.22
CPU: AMD Ryzen 9 3950X 16-Core Processor 31.928 GB
GPU: NVIDIA GeForce RTX 4090 23.576 GB
GPU: NVIDIA GeForce RTX 4090 23.576 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 36.37 fps 2X: 14.16 fps 4X: 03.25 fps
Iris 1X: 40.97 fps 2X: 14.81 fps 4X: 04.03 fps
Proteus 1X: 43.85 fps 2X: 15.78 fps 4X: 04.14 fps
Gaia 1X: 16.43 fps 2X: 11.16 fps 4X: 03.59 fps
Nyx 1X: 18.37 fps 2X: 14.08 fps
Nyx Fast 1X: 32.27 fps
4X Slowmo Apollo: 34.74 fps APFast: 61.22 fps Chronos: 32.49 fps CHFast: 29.45 fps
16X Slowmo Aion: 32.54 fps
AsRock X570 Phantom Gaming-ITX/TB3 Mainboard (BIOS vP5.60)
AMD Ryzen 9 3950X
16GB x2 (32GB) DDR4-3800 16/19/19/39/85/1T
MSI 4090 Suprim Liquid X AIO
nVidia Studio Driver v551.86 (03/19/2024)
Topaz Video AI v4.2.2
System Information
OS: Windows v11.23
CPU: AMD Ryzen 3 3200G with Radeon Vega Graphics 13.926 GB
GPU: AMD Radeon(TM) Vega 8 Graphics 1.985 GB
GPU: NVIDIA GeForce RTX 3080 9.8174 GB
Processing Settings
device: 1 vram: 1 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 24.71 fps 2X: 08.91 fps 4X: 02.22 fps
Iris 1X: 22.13 fps 2X: 08.74 fps 4X: 02.32 fps
Proteus 1X: 19.51 fps 2X: 10.04 fps 4X: 02.48 fps
Gaia 1X: 06.25 fps 2X: 04.17 fps 4X: 01.93 fps
Nyx 1X: 07.77 fps 2X: 06.17 fps
Nyx Fast 1X: 13.95 fps
4X Slowmo Apollo: 19.88 fps APFast: 42.95 fps Chronos: 14.16 fps CHFast: 18.37 fps
16X Slowmo Aion: 23.75 fps
感谢回复,赞同,以上是我 PC 备用机的配置,我记得同一部电影 1080p → 4k 的 Proteus 才 1.8 FPS, 2 FPS 都很难跑上去,不是能用的状态,现在这个时间点我想性价比可能只有把备用机置换成 13600k + 32G (16Gx2) DDR5 6400Mhz + 3080FE,应该至少还能把 1.8 拉到 3.5 FPS 附近。至于 4090FE 的主力机就等 15 代好了。到时候一并再换,多忍个大半年就行。
Topaz Video AI v4.2.2
System Information
OS: Mac v14.0301
CPU: Apple M1 Pro 16 GB
GPU: Apple M1 Pro 10.667 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 05.87 fps 2X: 03.81 fps 4X: 01.03 fps
Iris 1X: 04.37 fps 2X: 02.25 fps 4X: ERR fps
Proteus 1X: 05.49 fps 2X: 03.50 fps 4X: 00.87 fps
Gaia 1X: 01.88 fps 2X: 01.34 fps 4X: 01.00 fps
Nyx 1X: 01.92 fps 2X: 01.75 fps
Nyx Fast 1X: 04.18 fps
4X Slowmo Apollo: 06.60 fps APFast: 22.86 fps Chronos: 02.02 fps CHFast: 03.35 fps
16X Slowmo Aion: 04.98 fps
然后这是我 MBP M1 Pro 丐版的跑分,和 Intel/AMD 平台的完全没有可比性。还是同一个超分设置,FPS 居然最高可以跑到 3.8 附近,稳定在 3.5 - 3.6,甩备用机一大截。
Topaz Video AI v4.2.2
System Information
OS: Mac v14.0401
CPU: Apple M2 Max 96 GB
GPU: Apple M2 Max 72 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 12.08 fps 2X: 07.18 fps 4X: 02.70 fps
Iris 1X: 11.42 fps 2X: 03.80 fps 4X: 01.51 fps
Proteus 1X: 11.15 fps 2X: 07.42 fps 4X: 02.41 fps
Gaia 1X: 03.23 fps 2X: 02.24 fps 4X: 01.70 fps
Nyx 1X: 04.42 fps 2X: 04.02 fps
Nyx Fast 1X: 10.72 fps
4X Slowmo Apollo: 13.23 fps APFast: 44.54 fps Chronos: 03.80 fps CHFast: 06.45 fps
16X Slowmo Aion: 13.80 fps
Topaz Video AI v4.2.2
System Information
OS: Windows v11.22
CPU: AMD Ryzen 9 7950X 16-Core Processor 63.213 GB
GPU: NVIDIA GeForce RTX 4090 23.576 GB
GPU: AMD Radeon(TM) Graphics 0.47446 GB
GPU: NVIDIA GeForce RTX 4090 23.576 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 40.45 fps 2X: 17.02 fps 4X: 04.82 fps
Iris 1X: 38.96 fps 2X: 20.68 fps 4X: 05.40 fps
Proteus 1X: 41.49 fps 2X: 20.15 fps 4X: 05.52 fps
Gaia 1X: 15.81 fps 2X: 10.74 fps 4X: 04.80 fps
Nyx 1X: 17.31 fps 2X: 14.76 fps
Nyx Fast 1X: 31.43 fps
4X Slowmo Apollo: 42.56 fps APFast: 85.91 fps Chronos: 31.97 fps CHFast: 37.33 fps
16X Slowmo Aion: 31.49 fps
Windows 11 debloated
No Antivirus or cleaner programs
ASUS-PRIME Z790-A WIFI (Bios 1641)
i9-13900K + Kraken AIO
4090 drivers + 551.61 studio drivers
128 Gig DDR5-4800 UDIMM 1.1V CL40
Motherboard bios all auto no OC
Corsair HX1200 PS
4 onboard Nvme 4th gen drives
- Boot drive 2 TB
- Cache drive 2 TB
- Media drive 4 TB
- Mixdown drive 2 TB
Topaz Video AI v4.2.2
System Information
OS: Windows v11.22
CPU: 13th Gen Intel(R) Core(TM) i9-13900K 127.75 GB
GPU: NVIDIA GeForce RTX 4090 23.576 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 720x480
Benchmark Results
Artemis 1X: 187.67 fps 2X: 79.41 fps 4X: 23.19 fps
Iris 1X: 167.59 fps 2X: 109.76 fps 4X: 29.61 fps
Proteus 1X: 215.88 fps 2X: 116.18 fps 4X: 29.50 fps
Gaia 1X: 84.38 fps 2X: 59.50 fps 4X: 25.99 fps
Nyx 1X: 54.54 fps 2X: 49.99 fps
Nyx Fast 1X: 90.64 fps
4X Slowmo Apollo: 206.64 fps APFast: 342.72 fps Chronos: 162.08 fps CHFast: 185.24 fps
16X Slowmo Aion: 126.21 fps
Topaz Video AI v4.2.2
System Information
OS: Windows v11.22
CPU: 13th Gen Intel(R) Core(TM) i9-13900K 127.75 GB
GPU: NVIDIA GeForce RTX 4090 23.576 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 38.75 fps 2X: 20.33 fps 4X: 04.40 fps
Iris 1X: 37.03 fps 2X: 22.15 fps 4X: 04.84 fps
Proteus 1X: 40.66 fps 2X: 20.58 fps 4X: 05.11 fps
Gaia 1X: 14.97 fps 2X: 10.07 fps 4X: 04.55 fps
Nyx 1X: 16.76 fps 2X: 13.87 fps
Nyx Fast 1X: 30.19 fps
4X Slowmo Apollo: 40.02 fps APFast: 71.97 fps Chronos: 31.32 fps CHFast: 34.39 fps
16X Slowmo Aion: 32.17 fps