Skip to content

Commit 3e14243

Browse files
xe: jit: gemm: adjust kernel for e3m0
1 parent e4ea11a commit 3e14243

File tree

1 file changed

+1
-1
lines changed

1 file changed

+1
-1
lines changed

src/gpu/intel/jit/gemm/selector/db/kernel.db

+1-1
Original file line numberDiff line numberDiff line change
@@ -601,7 +601,7 @@ auto _CATALOG_ = kcatalog::toArray({
601601
{{'F', "gemm", {"E", "E", "S"}, {"N", "N", "N"}}, {-1, -1, {-1, -1, -1}, {-1, -1, -1}, {-1, -1, -1}, {-1, -1, -1}, {8, 8, 1}, "ABI"}, "av16+m64@64 am16 aB wg 8x4 cb4x2 ks32 xaf vav hi pt sr br bk0 sn grf256 sys", {16, (LoopType) 255, 256, {(LoopType) 208, (LoopType) 255, (LoopType) 255}, {524288, 524288, 16777216}, {524288, 524288, 16777216}, {32, 32, 32}, {8, 4, 1}, 1, (WGType) 1, 257, 32768, 0, {8, 8, 4}, {true, true, true}}, {'E', 17, {1.00706e+06, 522382, 0, 0, 0, 0, 0.725659, 1.44632, 0.970408, 1.74134, 0.0067111, 0.0067111, 0, 0.90349, 1.42986, 1.13348, 2.91269e-12}}},
602602
{{'F', "gemm", {"E", "E", "S"}, {"N", "N", "N"}}, {-1, -1, {-1, -1, -1}, {-1, -1, -1}, {-1, -1, -1}, {-1, -1, -1}, {8, 8, 1}, "ABIps"}, "av32+m16@64 am32+m32@72 aB wg 8x4 xaf vav hi pt sr br sb64 bk0 sn grf256 sys kv afb", {16, (LoopType) 255, 256, {(LoopType) 208, (LoopType) 255, (LoopType) 255}, {524288, 786432, 16777216}, {524288, 786432, 64}, {32, 16, 32}, {8, 4, 1}, 1, (WGType) 1, 441, 0, 0, {8, 8, 4}, {true, true, true}}, {'E', 17, {876646, 564122, 0, 0, 6.5151e+06, 7.83974e+06, 0.629669, 0.87362, 0.885543, 1.48097, 0.00440774, 0.00440774, 0, 1, 1.66234, 1.24996, 2.85794e-12}}},
603603
{{'F', "gemm", {"E", "E", "S"}, {"N", "N", "N"}}, {-1, -1, {-1, -1, -1}, {-1, -1, -1}, {-1, -1, -1}, {-1, -1, -1}, {8, 8, 1}, "ABI"}, "av16+m16@32 am16x2 aB wg 4x4x2 kr cb4 ks16 xaf st vav hi pt sr br bk0 sn grf256 sys", {16, (LoopType) 255, 256, {(LoopType) 208, (LoopType) 255, (LoopType) 2}, {1048576, 524288, 16777216}, {1048576, 524288, 16777216}, {64, 32, 16}, {4, 4, 2}, 1, (WGType) 1, 261, 16384, 65536, {8, 8, 4}, {true, true, true}}, {'E', 17, {1.02495e+06, 13797.9, 15430.1, 758509, 0, 0, 0.535333, 1.20812, 0.912657, 1.84068, 0.00529983, 0.00529983, 0, 1, 1.60581, 1.15873, 3.51036e-12}}},
604-
{{'F', "gemm", {"E", "E", "S"}, {"N", "N", "N"}}, {-1, -1, {-1, -1, -1}, {-1, -1, -1}, {-1, -1, -1}, {-1, -1, -1}, {8, 8, 1}, "ABIqps"}, "av16+m32@72 am32+m32@64 aB wg 4x8 xaf vav hi pt sr br sb64 bk0 sn grf256 sys kv afb", {16, (LoopType) 255, 256, {(LoopType) 208, (LoopType) 255, (LoopType) 255}, {1048576, 655360, 16777216}, {1048576, 655360, 32}, {64, 24, 32}, {4, 8, 1}, 1, (WGType) 1, 441, 0, 0, {8, 8, 4}, {true, true, true}}, {'E', 17, {903365, 697556, 0, 0, 8.2903e+06, 1.21651e+07, 0.724506, 0.722081, 0.92287, 1.55416, 0.00402055, 0.00402055, 0, 0.997691, 1.6726, 1.18622, 5.18793e-12}}},
604+
{{'F', "gemm", {"E", "E", "S"}, {"N", "N", "N"}}, {-1, -1, {-1, -1, -1}, {-1, -1, -1}, {-1, -1, -1}, {-1, -1, -1}, {8, 8, 1}, "ABIqps"}, "av16+m32@72 am32+m32@64 aB wg 4x8 xaf vav hi pt sr br sb64 bk0 sn grf256 sys kv afb", {16, (LoopType) 255, 256, {(LoopType) 208, (LoopType) 255, (LoopType) 255}, {1048576, 655360, 16777216}, {1048576, 655360, 32}, {32, 16, 32}, {4, 8, 1}, 1, (WGType) 1, 441, 0, 0, {8, 8, 4}, {true, true, true}}, {'E', 17, {903365, 697556, 0, 0, 8.2903e+06, 1.21651e+07, 0.724506, 0.722081, 0.92287, 1.55416, 0.00402055, 0.00402055, 0, 0.997691, 1.6726, 1.18622, 5.18793e-12}}},
605605
{{'F', "gemm", {"E", "E", "S"}, {"N", "N", "N"}}, {-1, -1, {-1, -1, -1}, {-1, -1, -1}, {-1, -1, -1}, {-1, -1, -1}, {8, 8, 1}, "ABI"}, "av16+m32@40 am32+m32@32 aB wg 2x8x2 kr xaf st vav hi pt sr br sb64 bk0 sn grf256 sys", {16, (LoopType) 255, 256, {(LoopType) 208, (LoopType) 255, (LoopType) 2}, {1048576, 131072, 16777216}, {1048576, 131072, 16777216}, {64, 8, 32}, {2, 8, 2}, 1, (WGType) 1, 261, 0, 32768, {8, 8, 4}, {true, true, true}}, {'E', 17, {1.06324e+06, 220443, 364.907, 276934, 0, 0, 0.524524, 1.25881, 0.793843, 2.21167, 0.00974309, 0.00974309, 0, 0.984682, 1.55809, 1.03396, 4.08729e-12}}},
606606
{{'F', "gemm", {"F", "B", "S"}, {"N", "N", "N"}}, {-1, -1, {-1, -1, -1}, {-1, -1, -1}, {-1, -1, -1}, {-1, -1, -1}, {1, 1, 1}, "#I"}, "aB16x2 aB16x2 aB wg 4x8 cab4 ks64 af vav hi pt sr br bk0 sn nb 4x8 dm grf256 sys l4", {16, (LoopType) 255, 256, {(LoopType) 208, (LoopType) 255, (LoopType) 255}, {524288, 131072, 16777216}, {524288, 131072, 16777216}, {32, 8, 64}, {4, 8, 1}, 1, (WGType) 1, 257, 98304, 0, {2, 2, 4}, {true, true, true}}, {'E', 17, {1.07581e+06, 764320, 0, 0, 0, 0, 0.804535, 1.46469, 0.96438, 2.27185, 0.0120677, 0.0120677, 0, 1, 1.38109, 0.955498, 2.48341e-12}}},
607607
{{'F', "gemm", {"F", "B", "S"}, {"N", "T", "N"}}, {-1, -1, {-1, -1, -1}, {-1, -1, -1}, {-1, -1, -1}, {-1, -1, -1}, {1, 1, 1}, "#I"}, "aB16 at16 aB wg 1x2x16 kr cab3x2 ks16 af vav hi pt sr br bk0 grf256 kv afb sys sn l4 l2d", {16, (LoopType) 255, 256, {(LoopType) 208, (LoopType) 255, (LoopType) 2}, {524288, 262144, 16777216}, {524288, 262144, 32}, {32, 16, 16}, {1, 2, 16}, 1, (WGType) 1, 445, 6144, 6144, {2, 2, 4}, {true, true, true}}, {'E', 17, {1.44116e+06, -85531, -20925.3, 122459, 4.34995e+06, 2.28557e+06, 0.844362, 0.875124, 0.584811, 1.33561, 0.0318181, 0.0286306, 0.00945064, 0.972737, 1.204, 0.897885, 9.01622e-12}}},

0 commit comments

Comments
 (0)