Topic: How to determine suitability and efficiency of GPU mining from spec [discussion] (Read 46 times)
April 21, 2021, 02:23:40 AM
I think you should give a name of an algorithm you plan to mine, every algorithms works differently, some focus on graphic cards memory and some prefers the clock speed
April 20, 2021, 04:16:18 PM
I am looking at several factors, memory size, FP32/64. Which one is important?
I compared specs for several AMD and NVIDIA GPUs, it is rumored that AMD's GPUs are speicially suited for Cryptomining, but why?
It appears answer is in FP64 speed. (double floating point). Does mining software extensively use these calculations a lot?
Because in general AMD GPU's ratio of FP64 to FP32 is double the nvidia's.
Lot of AMD GPU's, for example, VEGA56,64 has FP64/32 ratio of 16 whereas likes of 1080/2080 Nvidia is 32.
Select AMD GPU's MI lines i.e. MI50//MI60/MI100 has ratio of whopping 2 only meaning its FP64 ratio is monstrious.
I compared specs for several AMD and NVIDIA GPUs, it is rumored that AMD's GPUs are speicially suited for Cryptomining, but why?
It appears answer is in FP64 speed. (double floating point). Does mining software extensively use these calculations a lot?
Because in general AMD GPU's ratio of FP64 to FP32 is double the nvidia's.
Lot of AMD GPU's, for example, VEGA56,64 has FP64/32 ratio of 16 whereas likes of 1080/2080 Nvidia is 32.
Select AMD GPU's MI lines i.e. MI50//MI60/MI100 has ratio of whopping 2 only meaning its FP64 ratio is monstrious.
Jump to: