Difference between revisions of "Talk:GPU"

From Chessprogramming wiki
Jump to: navigation, search
(SIMD + Scalar Unit: new section)
(SIMD + Scalar Unit)
Line 20: Line 20:
  
 
According to AMD papers every SIMD unit has one scalar unit, Nvidia seems to have SFUs, special function units.
 
According to AMD papers every SIMD unit has one scalar unit, Nvidia seems to have SFUs, special function units.
 +
 
[[User:Smatovic|Smatovic]] ([[User talk:Smatovic|talk]]) 11:47, 18 April 2021 (CEST)
 
[[User:Smatovic|Smatovic]] ([[User talk:Smatovic|talk]]) 11:47, 18 April 2021 (CEST)

Revision as of 10:47, 18 April 2021

Nvidia architectures

Afaik Nvidia did never official mention SIMD in their papers as hardware architecture, with Tesla they only referred to as SIMT.

Nevertheless, my own conclusions are:

Tesla has 8 wide SIMD, executing a Warp of 32 threads over 4 cycles.

Fermi has 16 wide SIMD, executing a Warp of 32 threads over 2 cycles.

Kepler is somehow odd, not sure how the compute units are partitioned.

Maxwell and Pascal have 32 wide SIMD, executing a Warp of 32 threads over 1 cycle.

Volta and Turing seem to have 16 wide FPU SIMDs, but my own experiments show 32 wide VALU.

Smatovic (talk) 11:46, 18 April 2021 (CEST)

SIMD + Scalar Unit

According to AMD papers every SIMD unit has one scalar unit, Nvidia seems to have SFUs, special function units.

Smatovic (talk) 11:47, 18 April 2021 (CEST)