Support per field matrix layout (#3101)

* Support per field matrix layout

* Fix warnings.

* Fix.

* Fix tests.

* Fix spiv gen.

* Fix.

* More test fixes.

* Fix.

* Run only GPU tests on self-hosted servers.

* Remove -use-glsl-matrix-layout-modifier.

* Fix.

---------

Co-authored-by: Yong He <yhe@nvidia.com>

3 files changed, 12 insertions, 16 deletions

diff --git a/tests/cross-compile/array-of-buffers.slang.glsl b/tests/cross-compile/array-of-buffers.slang.glsl
index 62990033e..a198a5277 100644
--- a/tests/cross-compile/array-of-buffers.slang.glsl
+++ b/tests/cross-compile/array-of-buffers.slang.glsl
@@ -11,6 +11,7 @@ layout(std140) uniform _S1
 {
     uint index_0;
 } C_0;
+
 struct S_0
 {
     vec4 f_0;
@@ -36,9 +37,7 @@ out vec4 _S6;
 
 void main()
 {
-    S_0 _S7 = ((sb1_0[C_0.index_0])._data[(C_0.index_0)]);
-    vec4 _S8 = cb_0[C_0.index_0].f_0 + _S7.f_0;
-    uint _S9 = ((bb_0[C_0.index_0])._data[(int(C_0.index_0 * 4U))/4]);
-    _S6 = _S8 + ((sb2_0[C_0.index_0])._data[(C_0.index_0)]) + vec4(float(_S9));
+    uint _S7 = ((bb_0[C_0.index_0])._data[(int(C_0.index_0 * 4U))/4]);
+    _S6 = cb_0[C_0.index_0].f_0 + sb1_0[C_0.index_0]._data[C_0.index_0].f_0 + sb2_0[C_0.index_0]._data[C_0.index_0] + vec4(float(_S7));
     return;
 }
diff --git a/tests/cross-compile/array-of-buffers.slang.hlsl b/tests/cross-compile/array-of-buffers.slang.hlsl
index e709d323a..ca4b76bbe 100644
--- a/tests/cross-compile/array-of-buffers.slang.hlsl
+++ b/tests/cross-compile/array-of-buffers.slang.hlsl
@@ -13,6 +13,7 @@ cbuffer C_0 : register(b0)
 {
     SLANG_ParameterGroup_C_0 C_0;
 }
+
 struct S_0
 {
     float4 f_0;
@@ -25,10 +26,9 @@ StructuredBuffer<S_0 >  sb1_0[int(4)] : register(t0);
 RWStructuredBuffer<float4 >  sb2_0[int(5)] : register(u0);
 
 ByteAddressBuffer  bb_0[int(6)] : register(t4);
+
 float4 main() : SV_TARGET
 {
-    S_0 _S1 = sb1_0[C_0.index_0][C_0.index_0];
-    float4 _S2 = cb_0[C_0.index_0].f_0 + _S1.f_0;
-    uint _S3 = bb_0[C_0.index_0].Load(int(C_0.index_0 * 4U));
-    return _S2 + sb2_0[C_0.index_0][C_0.index_0] + (float4)float(_S3);
+    uint _S1 = bb_0[C_0.index_0].Load(int(C_0.index_0 * 4U));
+    return cb_0[C_0.index_0].f_0 + sb1_0[C_0.index_0].Load(C_0.index_0).f_0 + sb2_0[C_0.index_0][C_0.index_0] + (float4)float(_S1);
 }
diff --git a/tests/cross-compile/func-resource-param-array.slang.glsl b/tests/cross-compile/func-resource-param-array.slang.glsl
index d5d1bd08c..9e396e55f 100644
--- a/tests/cross-compile/func-resource-param-array.slang.glsl
+++ b/tests/cross-compile/func-resource-param-array.slang.glsl
@@ -16,22 +16,22 @@ layout(std430, binding = 2) buffer _S3 {
 
 int f_0(uint _S4)
 {
-    return ((a_0)._data[(_S4)]);
+    return a_0._data[_S4];
 }
 
 int f_1(uint _S5, uint _S6)
 {
-    return ((b_0[_S5])._data[(_S6)]);
+    return b_0[_S5]._data[_S6];
 }
 
 int g_0(uint _S7, uint _S8)
 {
-    return ((b_0[_S7])._data[(_S8)]);
+    return b_0[_S7]._data[_S8];
 }
 
 int g_1(uint _S9, uint _S10, uint _S11)
 {
-    return ((c_0[_S9][_S10])._data[(_S11)]);
+    return c_0[_S9][_S10]._data[_S11];
 }
 
 layout(local_size_x = 1, local_size_y = 1, local_size_z = 1) in;
@@ -39,10 +39,7 @@ void main()
 {
     uint ii_0 = gl_GlobalInvocationID.x;
     uint jj_0 = gl_GlobalInvocationID.y;
-
-    int tmp_0 = f_0(ii_0) + f_0(jj_0) + f_1(ii_0, jj_0) + g_0(ii_0, jj_0) + g_1(ii_0, jj_0, gl_GlobalInvocationID.z);
-
-    ((a_0)._data[(ii_0)]) = tmp_0;
+    a_0._data[ii_0] = f_0(ii_0) + f_0(jj_0) + f_1(ii_0, jj_0) + g_0(ii_0, jj_0) + g_1(ii_0, jj_0, gl_GlobalInvocationID.z);
     return;
 }