hello Catherine.
I'm a HF user for every day, and I only use the raw format files to keep more detail of picture, So the processing speed of the computer impact affects my workflow.
I would like to know the utilization mechanism of CPU hyper threading. For example, like 13700KF, will the big core and small core participate in rendering at the same time? Will 24 threads simultaneously render 24 RAW files? Or will it only use 8 big cores to render 8 raw files at the same time？
In addition, I am using the RTX3090 24G version, and 128G DDR4 ram ，specifically for HF. However, when checking the system manager, I found that the utilization of graphics memory is always at a relatively low level. Is it possible to use more graphics memory for rendering in future versions？
Many thank for you
About CPU hyper threading and GPU memory
Re: About CPU hyper threading and GPU memory
how can I setup the performance to get faster speed
- Attachments
-
- 微信图片_20230707061617.png
- (114.45 KiB) Not downloaded yet
-
- 微信图片_20230707061614.png
- (618.72 KiB) Not downloaded yet
-
- 微信图片_20230707061608.png
- (77.08 KiB) Not downloaded yet
Re: About CPU hyper threading and GPU memory
Hello!
You definitely need a faster CPU to speed up decoding the raw images. All the cores will be used, big and small. 13700K(F) is a good choice.
The GPU memory consumption depends on the resolution of your images. There only has to be enough. There is no way to make processing faster by using more graphics memory. But actual rendering is quite fast, it's the raw decoding that's the bottleneck (done by the CPU).
By the way, your system is limited to PCI-E 3.0, and the GPU supports PCI-E 4.0. Simply switching to PCI-E 4.0 (which 13700K supports) will increase the rendering performance by 1.5x - but that's only the actual GPU processing part, which is only a fraction of the total time taken to process a stack.
You definitely need a faster CPU to speed up decoding the raw images. All the cores will be used, big and small. 13700K(F) is a good choice.
The GPU memory consumption depends on the resolution of your images. There only has to be enough. There is no way to make processing faster by using more graphics memory. But actual rendering is quite fast, it's the raw decoding that's the bottleneck (done by the CPU).
By the way, your system is limited to PCI-E 3.0, and the GPU supports PCI-E 4.0. Simply switching to PCI-E 4.0 (which 13700K supports) will increase the rendering performance by 1.5x - but that's only the actual GPU processing part, which is only a fraction of the total time taken to process a stack.