"git@developer.sourcefind.cn:hehl2/torchaudio.git" did not exist on "720c36b1fad2980301f163b3835163c30ea7632b"
Commit fba3d780 authored by mtgu0705's avatar mtgu0705
Browse files

fix bug, function now passes.

parent c0ef46ff
......@@ -339,19 +339,21 @@ struct DeviceGemm_Xdl_CShuffleV3_BPreshuffle : public DeviceGemmV2BPreshuffle<AL
if(arg.KBatch > 1)
{
const auto kernel =
kernel_gemm_xdl_cshuffle_v3<GridwiseGemm,
kernel_gemm_xdl_cshuffle_v3_b_preshuffle<GridwiseGemm,
false,
InMemoryDataOperationEnum::AtomicAdd,
minimum_occupancy>;
minimum_occupancy,
TailNumber::Odd>;
Run(kernel);
}
else
{
const auto kernel =
kernel_gemm_xdl_cshuffle_v3<GridwiseGemm,
kernel_gemm_xdl_cshuffle_v3_b_preshuffle<GridwiseGemm,
false,
InMemoryDataOperationEnum::Set,
minimum_occupancy>;
minimum_occupancy,
TailNumber::Odd>;
Run(kernel);
}
}
......
......@@ -377,14 +377,14 @@ struct ThreadwiseTensorSliceTransfer_v2
if constexpr(InvalidElementAsNaN)
{
dst_buf(Number<dst_offset / PackedSize>{}) =
dst_buf(Number<dst_offset>{}) =
is_src_valid
? type_convert<DstData>(src_vector.template AsType<SrcData>()[i])
: NumericLimits<DstData>::QuietNaN();
}
else
{
dst_buf(Number<dst_offset / PackedSize>{}) =
dst_buf(Number<dst_offset>{}) =
type_convert<DstData>(src_vector.template AsType<SrcData>()[i]);
}
});
......@@ -1619,7 +1619,7 @@ struct ThreadwiseTensorSliceTransfer_StaticToStatic
constexpr index_t src_offset = src_desc.CalculateOffset(
src_slice_origin_idx + idx_md + i * dst_scalar_step_in_vector);
src_tmp_vector.template AsType<SrcData>()(i) = src_buf[Number<src_offset / PackedSize>{}];
src_tmp_vector.template AsType<SrcData>()(i) = src_buf[Number<src_offset>{}];
});
// copy data from src_tmp_vector to dst_tmp_vector (data cast data from SrcData to
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment