add float_args benchmark (pytorch#143143)

71% improvement with automatic dynamic float arguments

with specialize_float=False
```
float_args,compile_time_instruction_count,346293869
```

with specialize_float=True
```
float_args,compile_time_instruction_count,1198546486
```

Pull Request resolved: pytorch#143143
Approved by: https://github.com/laithsakka
ghstack dependencies: pytorch#141517

Author: bobrenjc93

benchmarks/dynamo/pr_time_benchmarks/benchmarks/float_args.py

@@ -0,0 +1,47 @@
+import sys
+
+from benchmark_base import BenchmarkBase
+
+import torch
+from torch._inductor.utils import fresh_inductor_cache
+
+
+class Benchmark(BenchmarkBase):
+    def __init__(self):
+        super().__init__(
+            category="float_args",
+            backend="inductor",
+            device="cpu",
+        )
+
+    def name(self):
+        return f"{self.category()}"
+
+    def description(self):
+        return "Benchmark to measure recompilations with float arguments."
+
+    def _prepare_once(self):
+        torch.manual_seed(0)
+
+    def _prepare(self):
+        torch._dynamo.reset()
+
+    def _work(self):
+        @torch.compile(backend="inductor")
+        def f(x, y):
+            return x + y
+
+        with fresh_inductor_cache():
+            for i in range(8):
+                f(torch.arange(3), i * 2.5)
+
+
+def main():
+    result_path = sys.argv[1]
+    Benchmark().enable_compile_time_instruction_count().collect_all().append_results(
+        result_path
+    )
+
+
+if __name__ == "__main__":
+    main()