SubsurfaceScattering.compute:
Before:
; -------- Statistics ---------------------
; SGPRs: 62 out of 104 used
; VGPRs: 42 out of 256 used
; LDS: 6416 out of 32768 bytes used
; 0 bytes scratch space used
; Instructions: 1280 ALU, 264 Control Flow, 75 TFETCH
After:
; -------- Statistics ---------------------
; SGPRs: 62 out of 104 used
; VGPRs: 41 out of 256 used
; LDS: 6416 out of 32768 bytes used
; 0 bytes scratch space used
; Instructions: 1276 ALU, 264 Control Flow, 75 TFETCH