| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
[libclc] Move cross to CLC library; add missing half overloads (#139713) The half overloads are trivially identical to the float and double ones. It didn't seem worth using 'gentype' for the OpenCL layer or CLC declarations so they're just written out explicitly. It does help avoid less trivial repetition in the CLC implementation, though. | 1 年前 | |
[libclc] Move cross to CLC library; add missing half overloads (#139713) The half overloads are trivially identical to the float and double ones. It didn't seem worth using 'gentype' for the OpenCL layer or CLC declarations so they're just written out explicitly. It does help avoid less trivial repetition in the CLC implementation, though. | 1 年前 | |
[libclc] Move (fast) length & distance to CLC library (#139701) This commit also refactors how geometric builtins are defined and declared, by sharing more helpers. It also removes an unnecessary gentype-like helper in favour of the more complete math/gentype.inc. There are no changes to the IR for any of these four builtins. The 'normalize' builtin will follow in a subsequent commit because it would involve the addition of missing halfn-type overloads for completeness. | 1 年前 | |
[libclc] Move (fast) length & distance to CLC library (#139701) This commit also refactors how geometric builtins are defined and declared, by sharing more helpers. It also removes an unnecessary gentype-like helper in favour of the more complete math/gentype.inc. There are no changes to the IR for any of these four builtins. The 'normalize' builtin will follow in a subsequent commit because it would involve the addition of missing halfn-type overloads for completeness. | 1 年前 | |
[NFC][libclc] Simplify clc_dot and dot implementation (#142922) llvm-diff shows no change to amdgcn--amdhsa.bc | 1 年前 | |
[NFC][libclc] Simplify clc_dot and dot implementation (#142922) llvm-diff shows no change to amdgcn--amdhsa.bc | 1 年前 | |
[NFC][libclc] add missing __CLC_ prefix all internal macros (#153523) This unifies naming scheme of macros to address review comment https://github.com/intel/llvm/pull/19779#discussion_r2272194357 math constant value macros are not changed, e.g. #define AU0 -9.86494292470009928597e-03 | 11 个月前 | |
[libclc] Move (fast) length & distance to CLC library (#139701) This commit also refactors how geometric builtins are defined and declared, by sharing more helpers. It also removes an unnecessary gentype-like helper in favour of the more complete math/gentype.inc. There are no changes to the IR for any of these four builtins. The 'normalize' builtin will follow in a subsequent commit because it would involve the addition of missing halfn-type overloads for completeness. | 1 年前 | |
[libclc] Move (fast) length & distance to CLC library (#139701) This commit also refactors how geometric builtins are defined and declared, by sharing more helpers. It also removes an unnecessary gentype-like helper in favour of the more complete math/gentype.inc. There are no changes to the IR for any of these four builtins. The 'normalize' builtin will follow in a subsequent commit because it would involve the addition of missing halfn-type overloads for completeness. | 1 年前 | |
[NFC][libclc] add missing __CLC_ prefix all internal macros (#153523) This unifies naming scheme of macros to address review comment https://github.com/intel/llvm/pull/19779#discussion_r2272194357 math constant value macros are not changed, e.g. #define AU0 -9.86494292470009928597e-03 | 11 个月前 | |
[libclc] Add (fast) normalize to CLC; add half overloads (#139759) For simplicity the half overloads just call into the float versions of the builtin. Otherwise there are no codegen changes to any target. | 1 年前 | |
[libclc] Move (fast) length & distance to CLC library (#139701) This commit also refactors how geometric builtins are defined and declared, by sharing more helpers. It also removes an unnecessary gentype-like helper in favour of the more complete math/gentype.inc. There are no changes to the IR for any of these four builtins. The 'normalize' builtin will follow in a subsequent commit because it would involve the addition of missing halfn-type overloads for completeness. | 1 年前 | |
[libclc][NFC] Fix return after else | 1 年前 | |
[libclc] Add (fast) normalize to CLC; add half overloads (#139759) For simplicity the half overloads just call into the float versions of the builtin. Otherwise there are no codegen changes to any target. | 1 年前 | |
[libclc] Add generic native half implementation of __clc_normalize (#150165) This is ported from https://github.com/intel/llvm/blob/sycl/libclc/libspirv/lib/generic/geometric/normalize.cl and can pass a closed-source OpenCL CTS "test_geometrics geom_normalize --half CL_DEVICE_TYPE_GPU" on intel GPU. llvm-diff amdgcn--amdhsa.bc shows fpext/fptrunc insts are now removed from normalize function. | 1 年前 |
| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
| 1 年前 | ||
| 1 年前 | ||
| 1 年前 | ||
| 1 年前 | ||
| 1 年前 | ||
| 1 年前 | ||
| 11 个月前 | ||
| 1 年前 | ||
| 1 年前 | ||
| 11 个月前 | ||
| 1 年前 | ||
| 1 年前 | ||
| 1 年前 | ||
| 1 年前 | ||
| 1 年前 |