CUDA half comparison support (#1834)

* #include an absolute path didn't work - because paths were taken to always be relative.

* Split out StringEscapeUtil.

* Added StringEscapeUtil.

* Fix typo in unix quoting type.

* Small comment improvements.

* Try to fix linux linking issue.

* Fix typo.

* Attempt to fix linux link issue.

* Update VS proj even though nothing really changed.

* Fix another typo issue.

* Fix for windows issue.
Fixed bug.

* Make separate Utils for escaping.

* Fix typo.

* Split out into StringEscapeHandler.

* Windows shell does handle removing quotes (so remove code to remove them).

* Handle unescaping if not initiating using the shell.

* Slight improvement around shell like decoding.

* Simplify command extraction.

* Add shared-library category type.

* Fix bug in command extraction.

* Typo in transcendental category.

* Enable unit-test on in smoke test category.

* Make parsing failing output as a failing test.

* Fixes for transcendental tests. Disable tests that do not work.

* Changed category parsing.

* Removed the TestResult parameter from _gatherTestsForFile.
Made testsList only output.

* Remove testing if all tests were disabled.

* Make args of CommandLine always unescaped.

* Add category.

* Don't need escaping on unix/linux.

* Remove some no longer used functions.

* Add requireSMVersion to CUDAExtensionTracker.

* half-calc.slang now works for CUDA.

* bit-cast-16-bit works on CUDA.

* WIP handling of CUDA vector<half> types.

* Half swizzle CUDA.

* Half vector test.

* Fix swizzle half bug.

* Fix compilation issue with narrowing to Index.

* Add unary ops.

* Add some vector scalar maths ops.

* Add half vector conversions for CUDA.

* Fix erroneous comment.

* Support for half comparisons.

* First pass test for half compare.

* Fix bug in CUDA specialized emit control.
Updated tests to have pre and post inc/dec.

* Removed unneeded parts of the cuda prelude.

* Half structured buffer works on CUDA.

Co-authored-by: Tim Foley <tfoleyNV@users.noreply.github.com>

7 files changed, 121 insertions, 8 deletions

diff --git a/tests/compute/half-calc.slang b/tests/compute/half-calc.slang
index e0dd01315..0f321ef98 100644
--- a/tests/compute/half-calc.slang
+++ b/tests/compute/half-calc.slang
@@ -29,5 +29,9 @@ void computeMain(uint3 dispatchThreadID : SV_DispatchThreadID)
     v += half(1.0f);
     v += offset;
     
+    v++;
+    --v;
+    v--;
+    
     outputBuffer[tid] = v;
 }
\ No newline at end of file
diff --git a/tests/compute/half-calc.slang.expected.txt b/tests/compute/half-calc.slang.expected.txt
index 2915a0dbc..389e44adf 100644
--- a/tests/compute/half-calc.slang.expected.txt
+++ b/tests/compute/half-calc.slang.expected.txt
@@ -1,4 +1,4 @@
-3F800000
-40800000
-40E00000
-41200000
+0
+40400000
+40C00000
+41100000
diff --git a/tests/compute/half-structured-buffer.slang b/tests/compute/half-structured-buffer.slang
index db0837d53..e701bb0fa 100644
--- a/tests/compute/half-structured-buffer.slang
+++ b/tests/compute/half-structured-buffer.slang
@@ -1,4 +1,6 @@
 //TEST(compute):COMPARE_COMPUTE:-vk -compute -profile cs_6_2 -render-features half -shaderobj
+//TEST(compute):COMPARE_COMPUTE:-cuda -compute -render-features half -shaderobj
+
 //Disable on Dx12 for now - because writing to structured buffer produces unexpected results
 //TEST_DISABLED(compute):COMPARE_COMPUTE:-dx12 -compute -use-dxil -profile cs_6_2 -render-features half -shaderobj
 
diff --git a/tests/compute/half-vector-calc.slang b/tests/compute/half-vector-calc.slang
index 3ae204796..b145e27ec 100644
--- a/tests/compute/half-vector-calc.slang
+++ b/tests/compute/half-vector-calc.slang
@@ -23,6 +23,10 @@ void computeMain(uint3 dispatchThreadID : SV_DispatchThreadID)
     v1 += v2.wzy;
     v2 += v0.xyxy;
     
+    v1 ++;
+    --v2;
+    v3++;
+    
     // Unary
     v2 = +v2.yxwz;
     v2 = -v2.zwxy;
diff --git a/tests/compute/half-vector-calc.slang.expected.txt b/tests/compute/half-vector-calc.slang.expected.txt
index 49c339529..2e80e4e2a 100644
--- a/tests/compute/half-vector-calc.slang.expected.txt
+++ b/tests/compute/half-vector-calc.slang.expected.txt
@@ -1,5 +1,5 @@
 type: float
-30.000000
-161.500000
-492.000000
-1021.500000
+73.000000
+206.500000
+539.000000
+1070.000000
diff --git a/tests/compute/half-vector-compare.slang b/tests/compute/half-vector-compare.slang
new file mode 100644
index 000000000..5f4670456
--- /dev/null
+++ b/tests/compute/half-vector-compare.slang
@@ -0,0 +1,98 @@
+//DISABLE_TEST(compute):COMPARE_COMPUTE:-dx12 -compute -output-using-type -use-dxil -profile cs_6_2 -render-features half -shaderobj
+//TEST(compute):COMPARE_COMPUTE:-vk -compute -output-using-type -profile cs_6_2 -render-features half -shaderobj
+//TEST(compute):COMPARE_COMPUTE:-cuda -compute -output-using-type -render-features half -shaderobj
+
+// Test for doing a calculation using half
+
+//TEST_INPUT:ubuffer(data=[0.2 10.0 12.0 16.0], stride=4):name=inputBuffer
+RWStructuredBuffer<int> inputBuffer;
+
+//TEST_INPUT:ubuffer(data=[0 0 0 0], stride=4):out,name outputBuffer
+RWStructuredBuffer<float> outputBuffer;
+
+struct Values
+{
+    __init(int index)
+    {
+        m_index = index;
+    }
+    
+    [mutating] half next()
+    {
+        float v = inputBuffer[m_index & 3];
+        m_index++;
+        return half(v);
+    }
+    
+    int m_index = 0;
+};
+
+[numthreads(4, 1, 1)]
+void computeMain(uint3 dispatchThreadID : SV_DispatchThreadID)
+{
+    uint tid = dispatchThreadID.x;
+    
+    Values values = Values(int(tid));
+    
+    int r = 0;
+    
+    half s0 = values.next();
+    half s1 = values.next();
+    
+    if (s0 < s1)
+    {
+        r += 0x1;
+    }
+    
+    half2 h2_0 = half2(values.next(), values.next());
+    half2 h2_1 = half2(values.next(), values.next());
+    
+    if (any(h2_0 < h2_1))
+    {
+        r += 0x2;
+    }
+    
+    if (all(h2_0 < h2_1))
+    {
+        r += 0x4;
+    }
+    
+    half3 h3_0 = half3(values.next(), values.next(), values.next());
+    half3 h3_1 = half3(values.next(), values.next(), values.next());
+    
+    if (any(h3_0 > h3_1))
+    {
+        r += 0x8;
+    }
+    
+    if (all(h3_0 <= h3_1))
+    {
+        r += 0x10;
+    }
+    
+    half4 h4_0 = half4(values.next(), values.next(), values.next(), values.next());
+    half4 h4_1 = half4(values.next(), values.next(), values.next(), values.next());
+    
+    
+    if (any(h4_0 > h4_1))
+    {
+        r += 0x8;
+    }
+    
+    if (all(h4_0 <= h4_1))
+    {
+        r += 0x10;
+    }
+    
+    if (any(!(h4_0 == h4_1)))
+    {
+        r += 0x20;
+    }
+    
+    if (all(h4_0 != h4_1))
+    {
+        r += 0x40;
+    }
+    
+    outputBuffer[tid] = r;
+}
diff --git a/tests/compute/half-vector-compare.slang.expected.txt b/tests/compute/half-vector-compare.slang.expected.txt
new file mode 100644
index 000000000..51c83b301
--- /dev/null
+++ b/tests/compute/half-vector-compare.slang.expected.txt
@@ -0,0 +1,5 @@
+type: float
+32.000000
+32.000000
+32.000000
+32.000000