Fix native code generation in test_aot.py (#4127)

Signed-off-by: Anatoly Myachev <[email protected]>

Author: anmyachev

python/test/unit/tools/test_aot.py

@@ -388,8 +388,7 @@ def _compile_kernel(dir, signature, kernel_name, out_name, out_path, num_warps,
             str(num_warps),
             "-g",
             grid,
-            "-gspv",
-            str(not generate_native_code),
+            *(["-gnc"] if generate_native_code else []),
             kernel_path,
         ],
         check=True,

python/triton/tools/compile.py

@@ -53,7 +53,8 @@
     parser.add_argument("--signature", "-s", type=str, help="Signature of the kernel", required=True)
     parser.add_argument("--grid", "-g", type=str, help="Launch grid of the kernel", required=True)
     parser.add_argument("--grf-mode", "-gm", type=str, default="large", help="Detemine spv build flags")
-    parser.add_argument("--generate-spv", "-gspv", type=bool, default=True, help="Cache SPV or native binary for XPU")
+    parser.add_argument("--generate-native-code", "-gnc", action="store_true",
+                        help="Generate native binary instead of SPV for XPU")
     args = parser.parse_args()
 
     out_name = args.out_name if args.out_name else args.kernel_name
@@ -115,7 +116,7 @@ def constexpr(s):
     if is_xpu():
         opts = {
             "num_warps": args.num_warps, "num_stages": args.num_stages, "threads_per_warp": args.threads_per_warp,
-            "grf_mode": args.grf_mode, "generate_native_code": not args.generate_spv
+            "grf_mode": args.grf_mode, "generate_native_code": args.generate_native_code
         }
     ccinfo = triton.compile(src, options=opts)
     if is_cuda():
@@ -195,7 +196,7 @@ def constexpr(s):
             "gridX": grid[0],
             "gridY": grid[1],
             "gridZ": grid[2],
-            "is_spv": "true" if args.generate_spv else "false",
+            "is_spv": "false" if args.generate_native_code else "true",
             "_placeholder": "",
         }
         for ext in ['h', 'cpp']: