triton-windows 3.3.1.post19__cp312-cp312-win_amd64.whl → 3.4.0.post20__cp312-cp312-win_amd64.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Potentially problematic release.


This version of triton-windows might be problematic. Click here for more details.

Files changed (166) hide show
  1. triton/_C/libtriton.pyd +0 -0
  2. triton/__init__.py +4 -1
  3. triton/_filecheck.py +87 -0
  4. triton/_internal_testing.py +26 -15
  5. triton/_utils.py +110 -21
  6. triton/backends/__init__.py +20 -23
  7. triton/backends/amd/__init__.py +0 -0
  8. triton/backends/amd/compiler.py +112 -78
  9. triton/backends/amd/driver.c +5 -2
  10. triton/backends/amd/driver.py +149 -47
  11. triton/backends/compiler.py +7 -21
  12. triton/backends/nvidia/bin/ptxas.exe +0 -0
  13. triton/backends/nvidia/compiler.py +92 -93
  14. triton/backends/nvidia/driver.c +90 -98
  15. triton/backends/nvidia/driver.py +303 -128
  16. triton/compiler/code_generator.py +212 -111
  17. triton/compiler/compiler.py +110 -25
  18. triton/experimental/__init__.py +0 -0
  19. triton/experimental/gluon/__init__.py +4 -0
  20. triton/experimental/gluon/_compiler.py +0 -0
  21. triton/experimental/gluon/_runtime.py +99 -0
  22. triton/experimental/gluon/language/__init__.py +18 -0
  23. triton/experimental/gluon/language/_core.py +312 -0
  24. triton/experimental/gluon/language/_layouts.py +230 -0
  25. triton/experimental/gluon/language/_math.py +12 -0
  26. triton/experimental/gluon/language/_semantic.py +287 -0
  27. triton/experimental/gluon/language/_standard.py +47 -0
  28. triton/experimental/gluon/language/nvidia/__init__.py +4 -0
  29. triton/experimental/gluon/language/nvidia/blackwell/__init__.py +202 -0
  30. triton/experimental/gluon/language/nvidia/blackwell/tma.py +32 -0
  31. triton/experimental/gluon/language/nvidia/hopper/__init__.py +11 -0
  32. triton/experimental/gluon/language/nvidia/hopper/mbarrier.py +51 -0
  33. triton/experimental/gluon/language/nvidia/hopper/tma.py +96 -0
  34. triton/experimental/gluon/nvidia/__init__.py +4 -0
  35. triton/experimental/gluon/nvidia/blackwell.py +3 -0
  36. triton/experimental/gluon/nvidia/hopper.py +40 -0
  37. triton/knobs.py +481 -0
  38. triton/language/__init__.py +39 -14
  39. triton/language/core.py +794 -537
  40. triton/language/extra/cuda/__init__.py +10 -7
  41. triton/language/extra/cuda/gdc.py +42 -0
  42. triton/language/extra/cuda/libdevice.py +394 -394
  43. triton/language/extra/cuda/utils.py +21 -21
  44. triton/language/extra/hip/libdevice.py +113 -104
  45. triton/language/math.py +65 -66
  46. triton/language/random.py +12 -2
  47. triton/language/semantic.py +1706 -1770
  48. triton/language/standard.py +116 -51
  49. triton/runtime/autotuner.py +117 -59
  50. triton/runtime/build.py +76 -12
  51. triton/runtime/cache.py +18 -47
  52. triton/runtime/driver.py +32 -29
  53. triton/runtime/interpreter.py +72 -35
  54. triton/runtime/jit.py +146 -110
  55. triton/testing.py +16 -12
  56. triton/tools/disasm.py +3 -4
  57. triton/tools/tensor_descriptor.py +36 -0
  58. triton/windows_utils.py +14 -6
  59. {triton_windows-3.3.1.post19.dist-info → triton_windows-3.4.0.post20.dist-info}/METADATA +7 -2
  60. triton_windows-3.4.0.post20.dist-info/RECORD +186 -0
  61. triton_windows-3.4.0.post20.dist-info/entry_points.txt +3 -0
  62. triton_windows-3.4.0.post20.dist-info/licenses/LICENSE +23 -0
  63. triton_windows-3.4.0.post20.dist-info/top_level.txt +1 -0
  64. triton/backends/amd/include/hip/amd_detail/amd_channel_descriptor.h +0 -358
  65. triton/backends/amd/include/hip/amd_detail/amd_device_functions.h +0 -1010
  66. triton/backends/amd/include/hip/amd_detail/amd_hip_atomic.h +0 -1638
  67. triton/backends/amd/include/hip/amd_detail/amd_hip_bf16.h +0 -1814
  68. triton/backends/amd/include/hip/amd_detail/amd_hip_bfloat16.h +0 -293
  69. triton/backends/amd/include/hip/amd_detail/amd_hip_common.h +0 -32
  70. triton/backends/amd/include/hip/amd_detail/amd_hip_complex.h +0 -174
  71. triton/backends/amd/include/hip/amd_detail/amd_hip_cooperative_groups.h +0 -835
  72. triton/backends/amd/include/hip/amd_detail/amd_hip_fp16.h +0 -1809
  73. triton/backends/amd/include/hip/amd_detail/amd_hip_fp8.h +0 -1391
  74. triton/backends/amd/include/hip/amd_detail/amd_hip_gl_interop.h +0 -108
  75. triton/backends/amd/include/hip/amd_detail/amd_hip_math_constants.h +0 -124
  76. triton/backends/amd/include/hip/amd_detail/amd_hip_runtime.h +0 -405
  77. triton/backends/amd/include/hip/amd_detail/amd_hip_runtime_pt_api.h +0 -196
  78. triton/backends/amd/include/hip/amd_detail/amd_hip_unsafe_atomics.h +0 -565
  79. triton/backends/amd/include/hip/amd_detail/amd_hip_vector_types.h +0 -2226
  80. triton/backends/amd/include/hip/amd_detail/amd_math_functions.h +0 -104
  81. triton/backends/amd/include/hip/amd_detail/amd_surface_functions.h +0 -244
  82. triton/backends/amd/include/hip/amd_detail/amd_warp_functions.h +0 -538
  83. triton/backends/amd/include/hip/amd_detail/amd_warp_sync_functions.h +0 -288
  84. triton/backends/amd/include/hip/amd_detail/concepts.hpp +0 -30
  85. triton/backends/amd/include/hip/amd_detail/device_library_decls.h +0 -133
  86. triton/backends/amd/include/hip/amd_detail/functional_grid_launch.hpp +0 -218
  87. triton/backends/amd/include/hip/amd_detail/grid_launch.h +0 -67
  88. triton/backends/amd/include/hip/amd_detail/grid_launch.hpp +0 -50
  89. triton/backends/amd/include/hip/amd_detail/grid_launch_GGL.hpp +0 -26
  90. triton/backends/amd/include/hip/amd_detail/helpers.hpp +0 -137
  91. triton/backends/amd/include/hip/amd_detail/hip_api_trace.hpp +0 -1446
  92. triton/backends/amd/include/hip/amd_detail/hip_assert.h +0 -101
  93. triton/backends/amd/include/hip/amd_detail/hip_cooperative_groups_helper.h +0 -242
  94. triton/backends/amd/include/hip/amd_detail/hip_fp16_gcc.h +0 -254
  95. triton/backends/amd/include/hip/amd_detail/hip_fp16_math_fwd.h +0 -96
  96. triton/backends/amd/include/hip/amd_detail/hip_ldg.h +0 -100
  97. triton/backends/amd/include/hip/amd_detail/hip_prof_str.h +0 -10570
  98. triton/backends/amd/include/hip/amd_detail/hip_runtime_prof.h +0 -78
  99. triton/backends/amd/include/hip/amd_detail/host_defines.h +0 -184
  100. triton/backends/amd/include/hip/amd_detail/hsa_helpers.hpp +0 -102
  101. triton/backends/amd/include/hip/amd_detail/macro_based_grid_launch.hpp +0 -798
  102. triton/backends/amd/include/hip/amd_detail/math_fwd.h +0 -698
  103. triton/backends/amd/include/hip/amd_detail/ockl_image.h +0 -177
  104. triton/backends/amd/include/hip/amd_detail/program_state.hpp +0 -107
  105. triton/backends/amd/include/hip/amd_detail/texture_fetch_functions.h +0 -491
  106. triton/backends/amd/include/hip/amd_detail/texture_indirect_functions.h +0 -478
  107. triton/backends/amd/include/hip/channel_descriptor.h +0 -39
  108. triton/backends/amd/include/hip/device_functions.h +0 -38
  109. triton/backends/amd/include/hip/driver_types.h +0 -468
  110. triton/backends/amd/include/hip/hip_bf16.h +0 -36
  111. triton/backends/amd/include/hip/hip_bfloat16.h +0 -44
  112. triton/backends/amd/include/hip/hip_common.h +0 -100
  113. triton/backends/amd/include/hip/hip_complex.h +0 -38
  114. triton/backends/amd/include/hip/hip_cooperative_groups.h +0 -46
  115. triton/backends/amd/include/hip/hip_deprecated.h +0 -95
  116. triton/backends/amd/include/hip/hip_ext.h +0 -161
  117. triton/backends/amd/include/hip/hip_fp16.h +0 -36
  118. triton/backends/amd/include/hip/hip_fp8.h +0 -33
  119. triton/backends/amd/include/hip/hip_gl_interop.h +0 -32
  120. triton/backends/amd/include/hip/hip_hcc.h +0 -24
  121. triton/backends/amd/include/hip/hip_math_constants.h +0 -36
  122. triton/backends/amd/include/hip/hip_profile.h +0 -27
  123. triton/backends/amd/include/hip/hip_runtime.h +0 -75
  124. triton/backends/amd/include/hip/hip_runtime_api.h +0 -9261
  125. triton/backends/amd/include/hip/hip_texture_types.h +0 -29
  126. triton/backends/amd/include/hip/hip_vector_types.h +0 -41
  127. triton/backends/amd/include/hip/hip_version.h +0 -17
  128. triton/backends/amd/include/hip/hiprtc.h +0 -421
  129. triton/backends/amd/include/hip/library_types.h +0 -78
  130. triton/backends/amd/include/hip/math_functions.h +0 -42
  131. triton/backends/amd/include/hip/surface_types.h +0 -63
  132. triton/backends/amd/include/hip/texture_types.h +0 -194
  133. triton/backends/amd/include/hsa/Brig.h +0 -1131
  134. triton/backends/amd/include/hsa/amd_hsa_common.h +0 -91
  135. triton/backends/amd/include/hsa/amd_hsa_elf.h +0 -462
  136. triton/backends/amd/include/hsa/amd_hsa_kernel_code.h +0 -269
  137. triton/backends/amd/include/hsa/amd_hsa_queue.h +0 -109
  138. triton/backends/amd/include/hsa/amd_hsa_signal.h +0 -80
  139. triton/backends/amd/include/hsa/hsa.h +0 -5738
  140. triton/backends/amd/include/hsa/hsa_amd_tool.h +0 -91
  141. triton/backends/amd/include/hsa/hsa_api_trace.h +0 -579
  142. triton/backends/amd/include/hsa/hsa_api_trace_version.h +0 -68
  143. triton/backends/amd/include/hsa/hsa_ext_amd.h +0 -3146
  144. triton/backends/amd/include/hsa/hsa_ext_finalize.h +0 -531
  145. triton/backends/amd/include/hsa/hsa_ext_image.h +0 -1454
  146. triton/backends/amd/include/hsa/hsa_ven_amd_aqlprofile.h +0 -488
  147. triton/backends/amd/include/hsa/hsa_ven_amd_loader.h +0 -667
  148. triton/backends/amd/include/hsa/hsa_ven_amd_pc_sampling.h +0 -416
  149. triton/backends/amd/include/roctracer/ext/prof_protocol.h +0 -107
  150. triton/backends/amd/include/roctracer/hip_ostream_ops.h +0 -4515
  151. triton/backends/amd/include/roctracer/hsa_ostream_ops.h +0 -1727
  152. triton/backends/amd/include/roctracer/hsa_prof_str.h +0 -3059
  153. triton/backends/amd/include/roctracer/roctracer.h +0 -779
  154. triton/backends/amd/include/roctracer/roctracer_ext.h +0 -81
  155. triton/backends/amd/include/roctracer/roctracer_hcc.h +0 -24
  156. triton/backends/amd/include/roctracer/roctracer_hip.h +0 -37
  157. triton/backends/amd/include/roctracer/roctracer_hsa.h +0 -112
  158. triton/backends/amd/include/roctracer/roctracer_plugin.h +0 -137
  159. triton/backends/amd/include/roctracer/roctracer_roctx.h +0 -67
  160. triton/backends/amd/include/roctracer/roctx.h +0 -229
  161. triton/language/_utils.py +0 -21
  162. triton/language/extra/cuda/_experimental_tma.py +0 -106
  163. triton/tools/experimental_descriptor.py +0 -32
  164. triton_windows-3.3.1.post19.dist-info/RECORD +0 -260
  165. triton_windows-3.3.1.post19.dist-info/top_level.txt +0 -14
  166. {triton_windows-3.3.1.post19.dist-info → triton_windows-3.4.0.post20.dist-info}/WHEEL +0 -0
@@ -1,194 +0,0 @@
1
- /*
2
- Copyright (c) 2015 - 2023 Advanced Micro Devices, Inc. All rights reserved.
3
-
4
- Permission is hereby granted, free of charge, to any person obtaining a copy
5
- of this software and associated documentation files (the "Software"), to deal
6
- in the Software without restriction, including without limitation the rights
7
- to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
8
- copies of the Software, and to permit persons to whom the Software is
9
- furnished to do so, subject to the following conditions:
10
-
11
- The above copyright notice and this permission notice shall be included in
12
- all copies or substantial portions of the Software.
13
-
14
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
15
- IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
16
- FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
17
- AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
18
- LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
19
- OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
20
- THE SOFTWARE.
21
- */
22
-
23
- #ifndef HIP_INCLUDE_HIP_TEXTURE_TYPES_H
24
- #define HIP_INCLUDE_HIP_TEXTURE_TYPES_H
25
-
26
- #if defined(__clang__)
27
- #pragma clang diagnostic push
28
- #pragma clang diagnostic ignored "-Wreserved-identifier"
29
- #pragma clang diagnostic ignored "-Wreserved-macro-identifier"
30
- #pragma clang diagnostic ignored "-Wc++98-compat"
31
- #endif
32
-
33
- #if !defined(__HIPCC_RTC__)
34
- #include <hip/hip_common.h>
35
- #endif
36
-
37
- #if !defined(__HIP_PLATFORM_AMD__) && defined(__HIP_PLATFORM_NVIDIA__)
38
- #include "texture_types.h"
39
- #elif defined(__HIP_PLATFORM_AMD__) && !defined(__HIP_PLATFORM_NVIDIA__)
40
- /*******************************************************************************
41
- * *
42
- * *
43
- * *
44
- *******************************************************************************/
45
- #if !defined(__HIPCC_RTC__)
46
- #include <limits.h>
47
- #include <hip/channel_descriptor.h>
48
- #include <hip/driver_types.h>
49
- #endif // !defined(__HIPCC_RTC__)
50
-
51
- #define hipTextureType1D 0x01
52
- #define hipTextureType2D 0x02
53
- #define hipTextureType3D 0x03
54
- #define hipTextureTypeCubemap 0x0C
55
- #define hipTextureType1DLayered 0xF1
56
- #define hipTextureType2DLayered 0xF2
57
- #define hipTextureTypeCubemapLayered 0xFC
58
-
59
- /**
60
- * Should be same as HSA_IMAGE_OBJECT_SIZE_DWORD/HSA_SAMPLER_OBJECT_SIZE_DWORD
61
- */
62
- #define HIP_IMAGE_OBJECT_SIZE_DWORD 12
63
- #define HIP_SAMPLER_OBJECT_SIZE_DWORD 8
64
- #define HIP_SAMPLER_OBJECT_OFFSET_DWORD HIP_IMAGE_OBJECT_SIZE_DWORD
65
- #define HIP_TEXTURE_OBJECT_SIZE_DWORD (HIP_IMAGE_OBJECT_SIZE_DWORD + HIP_SAMPLER_OBJECT_SIZE_DWORD)
66
-
67
- /**
68
- * An opaque value that represents a hip texture object
69
- */
70
- struct __hip_texture;
71
- typedef struct __hip_texture* hipTextureObject_t;
72
-
73
- /**
74
- * hip texture address modes
75
- */
76
- enum hipTextureAddressMode {
77
- hipAddressModeWrap = 0,
78
- hipAddressModeClamp = 1,
79
- hipAddressModeMirror = 2,
80
- hipAddressModeBorder = 3
81
- };
82
-
83
- /**
84
- * hip texture filter modes
85
- */
86
- enum hipTextureFilterMode { hipFilterModePoint = 0, hipFilterModeLinear = 1 };
87
-
88
- /**
89
- * hip texture read modes
90
- */
91
- enum hipTextureReadMode { hipReadModeElementType = 0, hipReadModeNormalizedFloat = 1 };
92
-
93
- /**
94
- * hip texture reference
95
- */
96
- typedef struct textureReference {
97
- int normalized;
98
- enum hipTextureReadMode readMode;// used only for driver API's
99
- enum hipTextureFilterMode filterMode;
100
- enum hipTextureAddressMode addressMode[3]; // Texture address mode for up to 3 dimensions
101
- struct hipChannelFormatDesc channelDesc;
102
- int sRGB; // Perform sRGB->linear conversion during texture read
103
- unsigned int maxAnisotropy; // Limit to the anisotropy ratio
104
- enum hipTextureFilterMode mipmapFilterMode;
105
- float mipmapLevelBias;
106
- float minMipmapLevelClamp;
107
- float maxMipmapLevelClamp;
108
-
109
- hipTextureObject_t textureObject;
110
- int numChannels;
111
- enum hipArray_Format format;
112
- }textureReference;
113
-
114
- /**
115
- * hip texture descriptor
116
- */
117
- typedef struct hipTextureDesc {
118
- enum hipTextureAddressMode addressMode[3]; // Texture address mode for up to 3 dimensions
119
- enum hipTextureFilterMode filterMode;
120
- enum hipTextureReadMode readMode;
121
- int sRGB; // Perform sRGB->linear conversion during texture read
122
- float borderColor[4];
123
- int normalizedCoords;
124
- unsigned int maxAnisotropy;
125
- enum hipTextureFilterMode mipmapFilterMode;
126
- float mipmapLevelBias;
127
- float minMipmapLevelClamp;
128
- float maxMipmapLevelClamp;
129
- }hipTextureDesc;
130
-
131
- #if __cplusplus
132
-
133
- /*******************************************************************************
134
- * *
135
- * *
136
- * *
137
- *******************************************************************************/
138
- #if __HIP__
139
- #define __HIP_TEXTURE_ATTRIB __attribute__((device_builtin_texture_type))
140
- #else
141
- #define __HIP_TEXTURE_ATTRIB
142
- #endif
143
-
144
- typedef textureReference* hipTexRef;
145
-
146
- template <class T, int texType = hipTextureType1D,
147
- enum hipTextureReadMode mode = hipReadModeElementType>
148
- struct __HIP_TEXTURE_ATTRIB texture : public textureReference {
149
- texture(int norm = 0, enum hipTextureFilterMode fMode = hipFilterModePoint,
150
- enum hipTextureAddressMode aMode = hipAddressModeClamp) {
151
- normalized = norm;
152
- readMode = mode;
153
- filterMode = fMode;
154
- addressMode[0] = aMode;
155
- addressMode[1] = aMode;
156
- addressMode[2] = aMode;
157
- channelDesc = hipCreateChannelDesc<T>();
158
- sRGB = 0;
159
- textureObject = nullptr;
160
- maxAnisotropy = 0;
161
- mipmapLevelBias = 0;
162
- minMipmapLevelClamp = 0;
163
- maxMipmapLevelClamp = 0;
164
- }
165
-
166
- texture(int norm, enum hipTextureFilterMode fMode, enum hipTextureAddressMode aMode,
167
- struct hipChannelFormatDesc desc) {
168
- normalized = norm;
169
- readMode = mode;
170
- filterMode = fMode;
171
- addressMode[0] = aMode;
172
- addressMode[1] = aMode;
173
- addressMode[2] = aMode;
174
- channelDesc = desc;
175
- sRGB = 0;
176
- textureObject = nullptr;
177
- maxAnisotropy = 0;
178
- mipmapLevelBias = 0;
179
- minMipmapLevelClamp = 0;
180
- maxMipmapLevelClamp = 0;
181
- }
182
- };
183
-
184
- #endif /* __cplusplus */
185
-
186
- #else
187
- #error("Must define exactly one of __HIP_PLATFORM_AMD__ or __HIP_PLATFORM_NVIDIA__");
188
- #endif
189
-
190
- #if defined(__clang__)
191
- #pragma clang diagnostic pop
192
- #endif
193
-
194
- #endif