1
0
mirror of https://github.com/GOSTSec/sgminer synced 2025-01-11 07:17:58 +00:00

Clean up use of macros in poclbm and use bitselect everywhere possible.

This commit is contained in:
Con Kolivas 2012-03-05 16:44:53 +11:00
parent 5fd7d462ab
commit 2a3b7359eb

View File

@ -36,7 +36,10 @@ __constant uint K[64] = {
#ifdef BITALIGN
#pragma OPENCL EXTENSION cl_amd_media_ops : enable
#define rotr(x, y) amd_bitalign((u)x, (u)x, (u)y)
#ifdef BFI_INT
#else
#define rotr(x, y) rotate((u)x, (u)(32 - y))
#endif
#ifdef BFI_INT
// Well, slight problem... It turns out BFI_INT isn't actually exposed to
// OpenCL (or CAL IL for that matter) in any way. However, there is
// a similar instruction, BYTE_ALIGN_INT, which is exposed to OpenCL via
@ -49,23 +52,19 @@ __constant uint K[64] = {
// Ma can also be implemented in terms of BFI_INT...
#define Ma(x, y, z) amd_bytealign( (z^x), (y), (x) )
#else // BFI_INT
// Later SDKs optimise this to BFI INT without patching and GCN
// actually fails if manually patched with BFI_INT
// AMD's KernelAnalyzer throws errors compiling the kernel if we use
// amd_bytealign on constants with vectors enabled, so we use this to avoid
// problems. (this is used 4 times, and likely optimized out by the compiler.)
#define Ma2(x, y, z) bitselect((u)x, (u)y, (u)z ^ (u)x)
#else // BFI_INT
//GCN actually fails if manually patched with BFI_INT
#define ch(x, y, z) bitselect((u)z, (u)y, (u)x)
#define Ma(x, y, z) bitselect((u)x, (u)y, (u)z ^ (u)x)
#endif
#else // BITALIGN
#define ch(x, y, z) (z ^ (x & (y ^ z)))
#define Ma(x, y, z) ((x & z) | (y & (x | z)))
#define rotr(x, y) rotate((u)x, (u)(32 - y))
#define Ma2(x, y, z) Ma(x, y, z)
#endif
// AMD's KernelAnalyzer throws errors compiling the kernel if we use
// amd_bytealign on constants with vectors enabled, so we use this to avoid
// problems. (this is used 4 times, and likely optimized out by the compiler.)
#define Ma2(x, y, z) ((y & z) | (x & (y | z)))
__kernel
__attribute__((vec_type_hint(u)))