[BACKEND] Minor fixes found when building triton with LLVM 17/main branches (#2089)

- These minor fixes are not specific to interface changes from LLVM main
or official llvm-17 branch and can be applied on triton main branch.
- https://github.com/darkbuck/triton/tree/darkbuck/main/llvm-main-branch
has extra changes to build again LLVM main branch build to enable me to
work on other backends on the main branch only. That's the hobby effort
and just FYR.

include/triton/Tools/Sys/GetEnv.hpp

@@ -24,7 +24,9 @@
 
 #include <algorithm>
 #include <cstdlib>
+#include <set>
 #include <string>
+
 namespace triton {
 
 const std::set<std::string> ENV_VARS = {

lib/Dialect/TritonGPU/IR/Dialect.cpp

@@ -1460,9 +1460,10 @@ struct TritonGPUInferLayoutInterface
     return success();
   }
 
-  LogicalResult inferDotOpEncoding(Attribute operandEncoding, unsigned opIdx,
-                                   Attribute retEncoding,
-                                   Optional<Location> location) const override {
+  LogicalResult
+  inferDotOpEncoding(Attribute operandEncoding, unsigned opIdx,
+                     Attribute retEncoding,
+                     std::optional<Location> location) const override {
     auto mmaRetEncoding = retEncoding.dyn_cast<MmaEncodingAttr>();
     if (mmaRetEncoding && mmaRetEncoding.isHopper()) {
       // TODO: support gmma when A/B does not reside in shared memory

lib/Dialect/TritonNvidiaGPU/Transforms/WSMaterialization.cpp

@@ -30,6 +30,8 @@
 #include "triton/Dialect/TritonNvidiaGPU/IR/Dialect.h"
 #include "triton/Dialect/TritonNvidiaGPU/Transforms/Utility.h"
 
+#include <set>
+
 using namespace mlir;
 namespace ttg = triton::gpu;
 namespace ttng = triton::nvidia_gpu;

python/setup.py

@@ -335,6 +335,7 @@ setup(
             "matplotlib",
             "pandas",
             "tabulate",
+            "torch",
         ],
     },
 )