Fat Fritz and Nvidia graphics cards


Answer

Unlike previous versions of the program the new engine Fat Fritz has one peculiarity: this new engine makes use of the resources of the extremely high-performance graphics processors of the newer NVIDIA graphics cards. Thus these high-performance processors of the NVIDIA cards are used as co-processors and offer the user a dramatic acceleration of engine performance! The RTX mode supported by Fat Fritz as well as the CUDA mode are at the moment of writing supported uniquely by NVIDIA graphics cards and employed by us during development.

Before you become involved with the various modes, all that needs to be understood is that the Fat Fritz engine requires a very high-performing graphics card, e.g. an NVIDIA RTX 2060, in order to perform to its full capacity.

Detailinformation about NVIDIA graphic cards .......

Without an appropriate GPU (meaning the graphics processor) the engine will be palpably slower, a fact which you should take into account when experimenting with the engine. Moreover Fat Fritz can be used exclusively with the main processor, but in that case you will achieve nothing close to the playing strength reached in combination with a rapid NVIDIA GPU.

If you are weighing up whether to a acquire such an NVIDIA card, you absolutely must be aware whether your network has sufficient performance reserves. In compact and energy saving desktop PCs it is possible that there is no slot available for these graphics cards. You really need to check this out before purchasing an NVIDIA GPU (graphics card)!

The following link takes you to a performance comparison of the various NVIDIA graphics cards.

https://www.phoronix.com/scan.php?page=news_item&px=LCZero-NVIDIA-Benchmarks

The Fat Fritz Engine can work with all these cards. Since there are considerable price differences, you would have to decide for yourself which model gets your preference.


Summary: Full performance can be achieved with NVIDIA RTX cards, since these are the only ones which support the fastest mode. All other models are clearly slower.


Information on CUDA can be found under:

https://en.wikipedia.org/wiki/CUDA

Information on Open CL can be found under:

https://en.wikipedia.org/wiki/OpenCL

Tags
Created on
15.11.2019
Rating
Feedback

Back to List