Skip to content

Add tests for fma #953

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
NeKon69 wants to merge 13 commits into
base: main
Choose a base branch
from
+247 −0
Open

Add tests for fma #953

Show file tree
Hide file tree
Changes from 12 commits
Commits
Show all changes
13 commits
Select commit Hold shift + click to select a range
f2ba5a6
[Matrix] Add tests for fma
NeKon69 Mar 11, 2026
3983163
upgrade tests to use real double values + test some edge cases
NeKon69 Mar 11, 2026
d1b8197
update variable name
NeKon69 Mar 12, 2026
8440721
implement suggestions
NeKon69 Mar 18, 2026
3d2df24
update values
NeKon69 Mar 18, 2026
fde3513
Merge branch 'main' into fma-scalar-vector-matrix-tests
NeKon69 Mar 21, 2026
383306e
add XFAILS for one of the targets, add tests for more precise values
NeKon69 Mar 21, 2026
4ceba74
update tests
NeKon69 Mar 25, 2026
a482934
delete the comment in main, update buffer values to use the same valu…
NeKon69 Mar 25, 2026
0110647
add xfail for intel vulkan
NeKon69 Mar 27, 2026
53871f1
delete clang xfail
NeKon69 Mar 28, 2026
35c626f
xfail arm64 && warp on clang
NeKon69 Mar 28, 2026
df1ccbc
switch to ULP tolerance and update the link to the issue
NeKon69 Apr 4, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
141 changes: 141 additions & 0 deletions test/Feature/HLSLLib/fma.matrix.test
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,141 @@
#--- source.hlsl

StructuredBuffer<double4> A : register(t0);
StructuredBuffer<double4> B : register(t1);
StructuredBuffer<double4> C : register(t2);

RWStructuredBuffer<double4> Out : register(u3);

[numthreads(1,1,1)]
void main() {
double3x2 r32 = fma(double3x2(A[0].xyz, A[1].xyz),
double3x2(B[0].xyz, B[1].xyz),
double3x2(C[0].xyz, C[1].xyz));

double2x4 r24 = fma(double2x4(A[2], A[3]),
double2x4(B[2], B[3]),
double2x4(C[2], C[3]));

double3x1 r31 = fma(double3x1(A[4].xyz),
double3x1(B[4].xyz),
double3x1(C[4].xyz));

double4x4 r44 = fma(double4x4(A[5], A[6], A[7], A[8]),
double4x4(B[5], B[6], B[7], B[8]),
double4x4(C[5], C[6], C[7], C[8]));

Out[0] = double4(r32[0], r32[1]);
Out[1] = double4(r32[2], 0.0, 0.0);
Out[2] = r24[0];
Out[3] = r24[1];
Out[4] = double4(r31[0][0], r31[1][0], r31[2][0], 0.0);
Out[5] = r44[0];
Out[6] = r44[1];
Out[7] = r44[2];
Out[8] = r44[3];
}

//--- pipeline.yaml

---
Shaders:
- Stage: Compute
Entry: main
DispatchSize: [1, 1, 1]
Buffers:
- Name: A
Format: Float64
Stride: 32
Data: [ 1.5, -2.5, 0.5, 0.0,
-0.5, 2.0, 3.0, 0.0,
0.25, -0.25, 10.4, -10.6,
0.0, 10.0, -10.0, 10.5,
1.25, -4.0, 2.5, 0.0,
1.0, 2.0, 3.0, 4.0,
-1.0, -2.0, -3.0, -4.0,
0.5, 1.5, -0.5, -1.5,
10.0, -10.0, 2.25, -2.25 ]
- Name: B
Format: Float64
Stride: 32
Data: [ 2.0, -1.0, 4.0, 0.0,
4.0, -0.5, 1.5, 0.0,
1.0, 1.0, 1.0, 1.0,
1.0, 2.0, -1.0, 1.0,
2.0, -0.5, 3.0, 0.0,
2.0, -1.0, 0.5, 3.0,
4.0, 0.5, -2.0, -1.0,
1.0, 2.0, 3.0, 4.0,
-0.5, 1.5, -1.0, 2.0 ]
- Name: C
Format: Float64
Stride: 32
Data: [ 0.25, 0.75, -1.5, 0.0,
1.5, 1.0, -2.0, 0.0,
0.0, 0.0, 0.0, 0.0,
0.0, 0.5, 0.5, 0.0,
0.5, 1.0, -1.5, 0.0,
0.5, 1.0, -1.5, 0.0,
-0.5, 2.0, 1.0, 3.0,
1.5, -2.0, 0.5, -4.0,
5.0, -5.0, 0.25, -0.25 ]
- Name: Out
Format: Float64
Stride: 32
FillSize: 288
- Name: Expected
Format: Float64
Stride: 32
Data: [ 3.25, 3.25, 0.5, -0.5,
0.0, 2.5, 0.0, 0.0,
0.25, -0.25, 10.4, -10.6,
0.0, 20.5, 10.5, 10.5,
3.0, 3.0, 6.0, 0.0,
2.5, -1.0, 0.0, 12.0,
-4.5, 1.0, 7.0, 7.0,
2.0, 1.0, -1.0, -10.0,
0.0, -20.0, -2.0, -4.75 ]
Results:
- Result: Result
Rule: BufferExact
Actual: Out
Expected: Expected
DescriptorSets:
- Resources:
- Name: A
Kind: StructuredBuffer
DirectXBinding:
Register: 0
Space: 0
VulkanBinding:
Binding: 0
- Name: B
Kind: StructuredBuffer
DirectXBinding:
Register: 1
Space: 0
VulkanBinding:
Binding: 1
- Name: C
Kind: StructuredBuffer
DirectXBinding:
Register: 2
Space: 0
VulkanBinding:
Binding: 2
- Name: Out
Kind: RWStructuredBuffer
DirectXBinding:
Register: 3
Space: 0
VulkanBinding:
Binding: 3
#--- end

# Bug https://github.com/llvm/offload-test-suite/issues/1000
# XFAIL: arm64 && WARP

# REQUIRES: Double
# RUN: split-file %s %t
# RUN: %dxc_target -Gis -HV 202x -T cs_6_5 -Fo %t.o %t/source.hlsl
# RUN: %offloader %t/pipeline.yaml %t.o
104 changes: 104 additions & 0 deletions test/Feature/HLSLLib/fma.test
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,104 @@
#--- source.hlsl

StructuredBuffer<double4> A : register(t0);
StructuredBuffer<double4> B : register(t1);
StructuredBuffer<double4> C : register(t2);

RWStructuredBuffer<double4> Out : register(u3);

[numthreads(1,1,1)]
void main() {
Out[0] = fma(A[0], B[0], C[0]);
Out[1] = double4(fma(A[1].xyz, B[1].xyz, C[1].xyz), fma(A[1].w, B[1].w, C[1].w));
Out[2] = double4(fma(A[2].xy, B[2].xy, C[2].xy), fma(A[2].zw, B[2].zw, C[2].zw));
Out[3] = fma(double4(1.0000000149011612l, -1.0000000149011612l, 1.0000000000000002l, 0.5l), double4(67108865.0l, 67108865.0l, -0.9999999999999998l, -8.0l), double4(-67108866.0l, 67108866.0l, 1.0l, 1.0l));
}

//--- pipeline.yaml

---
Shaders:
- Stage: Compute
Entry: main
DispatchSize: [1, 1, 1]
Buffers:
- Name: A
Format: Float64
Stride: 32
Data: [ 1.5, -2.0, 0.75, -3.25,
2.5, -4.0, 1.25, 3.5,
1.0000000149011612, -1.0000000149011612, 1.0000000000000002, 0.5 ]
# last row: (1 + 2^-26), -(1 + 2^-26), (1 + 2^-52), 0.5
- Name: B
Format: Float64
Stride: 32
Data: [ 2.25, 4.5, -8.0, -2.0,
-1.5, -2.0, 8.0, -2.0,
67108865.0, 67108865.0, -0.9999999999999998, -8.0 ]
# last row: (2^26 + 1), (2^26 + 1), -(1 - 2^-52), -8
- Name: C
Format: Float64
Stride: 32
Data: [ 0.125, 1.25, 2.5, -0.5,
0.25, -1.5, -0.75, 0.5,
-67108866.0, 67108866.0, 1.0, 1.0 ]
# last row: -(2^26 + 2), (2^26 + 2), 1, 1
- Name: Out
Format: Float64
Stride: 32
FillSize: 128
- Name: Expected
Format: Float64
Stride: 32
Data: [ 3.5, -7.75, -3.5, 6.0,
-3.5, 6.5, 9.25, -6.5,
0x1.0000000000000p-26, -0x1.0000000000000p-26, 0x1.0000000000000p-104, -3.0,
0x1.0000000000000p-26, -0x1.0000000000000p-26, 0x1.0000000000000p-104, -3.0 ]
# last two rows: 2^-26, -(2^-26), 2^-104, -3
Results:
- Result: Result
Rule: BufferExact
Actual: Out
Expected: Expected
DescriptorSets:
- Resources:
- Name: A
Kind: StructuredBuffer
DirectXBinding:
Register: 0
Space: 0
VulkanBinding:
Binding: 0
- Name: B
Kind: StructuredBuffer
DirectXBinding:
Register: 1
Space: 0
VulkanBinding:
Binding: 1
- Name: C
Kind: StructuredBuffer
DirectXBinding:
Register: 2
Space: 0
VulkanBinding:
Binding: 2
- Name: Out
Kind: RWStructuredBuffer
DirectXBinding:
Register: 3
Space: 0
VulkanBinding:
Binding: 3
#--- end

# Bug https://github.com/llvm/offload-test-suite/issues/1000
# XFAIL: arm64 && WARP

# Bug https://github.com/llvm/offload-test-suite/issues/1026
# XFAIL: Intel && Vulkan && DXC

# REQUIRES: Double
# RUN: split-file %s %t
# RUN: %dxc_target -Gis -HV 202x -T cs_6_5 -Fo %t.o %t/source.hlsl
# RUN: %offloader %t/pipeline.yaml %t.o
Loading