File: AMDGPUFeatures.td

package info (click to toggle)
llvm-toolchain-20 1%3A20.1.8-1~exp1
  • links: PTS, VCS
  • area: main
  • in suites: experimental
  • size: 2,111,388 kB
  • sloc: cpp: 7,438,767; ansic: 1,393,871; asm: 1,012,926; python: 241,728; f90: 86,635; objc: 75,411; lisp: 42,144; pascal: 17,286; sh: 10,027; ml: 5,082; perl: 4,730; awk: 3,523; makefile: 3,349; javascript: 2,251; xml: 892; fortran: 672
file content (56 lines) | stat: -rw-r--r-- 2,081 bytes parent folder | download | duplicates (4)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
//===-- AMDGPUFeatures.td - AMDGPU Feature Definitions -----*- tablegen -*-===//
//
// 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
//
//===----------------------------------------------------------------------===//

def FeatureFP64 : SubtargetFeature<"fp64",
  "FP64",
  "true",
  "Enable double precision operations"
>;

def FeatureFMA : SubtargetFeature<"fmaf",
  "FMA",
  "true",
  "Enable single precision FMA (not as fast as mul+add, but fused)"
>;

// Addressable local memory size is the maximum number of bytes of LDS that can
// be allocated to a single workgroup.
class SubtargetFeatureAddressableLocalMemorySize <int Value> : SubtargetFeature<
  "addressablelocalmemorysize"#Value,
  "AddressableLocalMemorySize",
  !cast<string>(Value),
  "The size of local memory in bytes"
>;

def FeatureAddressableLocalMemorySize32768 : SubtargetFeatureAddressableLocalMemorySize<32768>;
def FeatureAddressableLocalMemorySize65536 : SubtargetFeatureAddressableLocalMemorySize<65536>;
def FeatureAddressableLocalMemorySize163840 : SubtargetFeatureAddressableLocalMemorySize<163840>;

class SubtargetFeatureWavefrontSize <int ValueLog2> : SubtargetFeature<
  "wavefrontsize"#!shl(1, ValueLog2),
  "WavefrontSizeLog2",
  !cast<string>(ValueLog2),
  "The number of threads per wavefront"
>;

def FeatureWavefrontSize16 : SubtargetFeatureWavefrontSize<4>;
def FeatureWavefrontSize32 : SubtargetFeatureWavefrontSize<5>;
def FeatureWavefrontSize64 : SubtargetFeatureWavefrontSize<6>;

class SubtargetFeatureGeneration <string Value, string FeatureName,
                                 string Subtarget,
                                  list<SubtargetFeature> Implies> :
        SubtargetFeature <FeatureName, "Gen", Subtarget#"::"#Value,
                          Value#" GPU generation", Implies>;

def FeaturePromoteAlloca : SubtargetFeature <"promote-alloca",
  "EnablePromoteAlloca",
  "true",
  "Enable promote alloca pass"
>;