mirror of git://gcc.gnu.org/git/gcc.git
Update ChangeLog.*
This commit is contained in:
parent
802f59b795
commit
fb9549791f
|
@ -1,3 +1,44 @@
|
||||||
|
==================== Branch work222-float, patch #314 ====================
|
||||||
|
|
||||||
|
Add bf and hf duplicate to v4sf.
|
||||||
|
|
||||||
|
2025-10-06 Michael Meissner <meissner@linux.ibm.com>
|
||||||
|
|
||||||
|
gcc/
|
||||||
|
|
||||||
|
* config/rs6000/float16.md (CVT_FP16_TO_V4SF_INSN): New mode attribute.
|
||||||
|
(FP16_VECTOR8): Likewise.
|
||||||
|
(fp16_vector8): Likewise.
|
||||||
|
(FP16_VECTOR4): Likewise.
|
||||||
|
(UNSPEC_CVT_FP16_TO_V4SF): Replace UNSPEC_XVCVBF16SPN_BF and
|
||||||
|
UNSPEC_XVCVBF16SPN_V8BF.
|
||||||
|
(extendbf<mode>2): Change xvcvbf16spn_v8bf to cvt_fp16_to_v4sf_v8bf.
|
||||||
|
(cvt_fp16_to_v4sf_<mode>): Rename from xvcvbf16spn_v8bf, and add HFmode
|
||||||
|
support.
|
||||||
|
(cvt_fp16_to_v4sf_<mode>_le): New insns.
|
||||||
|
(cvt_fp16_to_v4sf_<mode>_be): Likewise.
|
||||||
|
(dup_<mode>_to_v4sf): Likewise.
|
||||||
|
(xvcvbf16spn_bf): Change UNSPEC_XVCVBF16SPN_BF to
|
||||||
|
UNSPEC_CVT_FP16_TO_V4SF.
|
||||||
|
* config/rs6000/rs6000-modes.def (V4BFmode): New mode.
|
||||||
|
(V4HFmode): Likewise.
|
||||||
|
|
||||||
|
==================== Branch work222-float, patch #313 ====================
|
||||||
|
|
||||||
|
Iterate on 16-bit floating point.
|
||||||
|
|
||||||
|
2025-10-06 Michael Meissner <meissner@linux.ibm.com>
|
||||||
|
|
||||||
|
gcc/
|
||||||
|
|
||||||
|
* config/rs6000/float16.md (VP16): New mode iterator.
|
||||||
|
(VP16_HW): Likewise.
|
||||||
|
(FP16_BINARY): Rename from BF_OPS.
|
||||||
|
(FP16_BINARY_NAME): Rename from BF_OPS_NAME.
|
||||||
|
(UNSPEC_XXSPLTW_FP16): Rename from UNSPEC_XXSPLTW_BF.
|
||||||
|
(<FP16_BINARY_NAME>bf3): Rename from <BF_OPS_NAME>.
|
||||||
|
(xxspltw_<mode>): Rename for xxspltw_bf. Add HFmode support.
|
||||||
|
|
||||||
==================== Branch work222-float, patch #312 ====================
|
==================== Branch work222-float, patch #312 ====================
|
||||||
|
|
||||||
Add 16-bit floating point abs; Remove expand/truncate optimization.
|
Add 16-bit floating point abs; Remove expand/truncate optimization.
|
||||||
|
|
Loading…
Reference in New Issue