[libc-commits] [libc] [libc] Polish GPU benchmarking (PR #153900)

[Joseph Huber via libc-commits]

================
@@ -66,7 +64,7 @@ template <typename F, typename T>
   uint64_t stop = gpu::processor_clock();
   cpp::atomic_thread_fence(cpp::MemoryOrder::ACQ_REL);
   asm("" ::"r"(stop));
-  volatile T output = result;
+  volatile auto output = result;
----------------
jhuber6 wrote:

I vaguely remember needing that because the PTX optimizer would defy the inline assembly so I still needed it. That's fine it needed.

https://github.com/llvm/llvm-project/pull/153900