Fp8 block scale quantization for fmha fwd (#3330)

* add block scale parameters to kernel

* add block scale to kernel

* add smoke test

* format

* Revert "format"

This reverts commit 356c3c9706.

* only format my code

* format py

* fix auto not allowd in function prototype

* change instance tttt to ttff

* fix structured binding issue

* change s_acc elementwise op

* async pipeline add block scale

* add quantation P using shift exp2

* precompute (m - shift) once per row

* change blk scale seqstrt ptr name

* fix some name

* fix for  deduction guide

* fix some comments

* add P scale to qr_ksvs_pipeline

* add comment to idx_identity

* change the method of calculating descale block index

* unify naming style: use block_scale_ as name prefix

* unify naming style

* update the CHANGELOG.md

* Add FP8 block scale quantization support for FMHA forward kernel

---------

Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com>
Co-authored-by: Po Yen Chen <PoYen.Chen@amd.com>

example/ck_tile/01_fmha/codegen/cpp_symbol_map.py

@@ -77,11 +77,13 @@ def get_mask_cpp_check_expr(mask: str) -> str:
 QSCALE_MAP = {
     "no": "ck_tile::BlockAttentionQuantScaleEnum::NO_SCALE",
     "pertensor": "ck_tile::BlockAttentionQuantScaleEnum::PERTENSOR",
+    "blockscale": "ck_tile::BlockAttentionQuantScaleEnum::BLOCKSCALE",
 }
 
 QSCALE_CHECK_MAP = {
     "no": "quant_scale_enum::no_scale",
     "pertensor": "quant_scale_enum::pertensor",
+    "blockscale": "quant_scale_enum::blockscale",
 }
 
 BIAS_MAP = {

example/ck_tile/01_fmha/codegen/ops/fmha_fwd.py

@@ -1018,7 +1018,7 @@ class KernelComponentFactoryGfx9(CompatibilityRuleFactoryGfx9):
             # no need lse/dropout kernels
             for logits, qscale, mask, bias, sink in itertools.product(
                 ["t", "f"],
-                ["no", "pertensor"],
+                ["no", "pertensor", "blockscale"],
                 get_mask_map(mask_impl).keys(),
                 ["no"],
                 ["f", "t"],
@@ -1146,7 +1146,10 @@ class KernelComponentFactoryGfx12(CompatibilityRuleFactory):
         elif dtype in cls._DT_FP8_FP8BF16 or dtype in cls._DT_FP8FP32:
             # no need lse/dropout kernels
             for logits, qscale, mask, bias in itertools.product(
-                ["f"], ["no", "pertensor"], get_mask_map(mask_impl).keys(), ["no"]
+                ["f"],
+                ["no", "pertensor", "blockscale"],
+                get_mask_map(mask_impl).keys(),
+                ["no"],
             ):
                 pipelines.append(FmhaFwdPipeline("qr", "row", "f", "f", "f", "f", logits, bias, "f", "f", qscale, mask, "f", "f", "f"))  # fmt: skip
                 pipelines.append(FmhaFwdPipeline("qr", "row", "t", "t", "t", "t", logits, bias, "f", "f", qscale, mask, "f", "f", "f"))  # fmt: skip