[clang] [llvm] [TargetParser] Add a list of Intel GPUs, and use it in offload-arch (PR #222072)

[Nikita Kornev via cfe-commits]

================
@@ -0,0 +1,60 @@
+//===-- IntelGPUTargetParser - Parser for Intel GPU targets ----*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+// This file implements a target parser for the Intel GPU list.
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/TargetParser/IntelGPUTargetParser.h"
+#include "llvm/ADT/Twine.h"
+
+using namespace llvm;
+using namespace IntelGPU;
+
+// A GMDID packs the architecture, release and revision of the GPU IP.
+static constexpr uint32_t GMDIDArchitectureShift = 22;
+static constexpr uint32_t GMDIDReleaseShift = 14;
+static constexpr uint32_t GMDIDReleaseMask = 0xff;
+static constexpr uint32_t GMDIDRevisionMask = 0x3f;
+
+GMDID llvm::IntelGPU::decodeGMDID(uint32_t IPVersion) {
+  return {IPVersion >> GMDIDArchitectureShift,
+          (IPVersion >> GMDIDReleaseShift) & GMDIDReleaseMask,
+          IPVersion & GMDIDRevisionMask};
----------------
KornevNikita wrote:

cde1dcac57babc6c86443e757d485e407d0769a8

https://github.com/llvm/llvm-project/pull/222072