tinygrad 0.10.2__py3-none-any.whl → 0.11.0__py3-none-any.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.
Files changed (131) hide show
  1. tinygrad/__init__.py +1 -1
  2. tinygrad/apps/llm.py +206 -0
  3. tinygrad/codegen/__init__.py +116 -0
  4. tinygrad/codegen/devectorizer.py +315 -172
  5. tinygrad/codegen/expander.py +8 -16
  6. tinygrad/codegen/gpudims.py +89 -0
  7. tinygrad/codegen/linearize.py +205 -203
  8. tinygrad/codegen/lowerer.py +92 -139
  9. tinygrad/codegen/opt/__init__.py +38 -0
  10. tinygrad/codegen/opt/heuristic.py +125 -0
  11. tinygrad/codegen/opt/kernel.py +510 -0
  12. tinygrad/{engine → codegen/opt}/search.py +51 -35
  13. tinygrad/codegen/opt/swizzler.py +134 -0
  14. tinygrad/codegen/opt/tc.py +127 -0
  15. tinygrad/codegen/quantize.py +67 -0
  16. tinygrad/device.py +122 -132
  17. tinygrad/dtype.py +152 -35
  18. tinygrad/engine/jit.py +81 -54
  19. tinygrad/engine/memory.py +46 -27
  20. tinygrad/engine/realize.py +82 -41
  21. tinygrad/engine/schedule.py +70 -445
  22. tinygrad/frontend/__init__.py +0 -0
  23. tinygrad/frontend/onnx.py +1253 -0
  24. tinygrad/frontend/torch.py +5 -0
  25. tinygrad/gradient.py +19 -27
  26. tinygrad/helpers.py +95 -47
  27. tinygrad/nn/__init__.py +7 -8
  28. tinygrad/nn/optim.py +72 -41
  29. tinygrad/nn/state.py +37 -23
  30. tinygrad/renderer/__init__.py +40 -60
  31. tinygrad/renderer/cstyle.py +143 -128
  32. tinygrad/renderer/llvmir.py +113 -62
  33. tinygrad/renderer/ptx.py +50 -32
  34. tinygrad/renderer/wgsl.py +27 -23
  35. tinygrad/runtime/autogen/am/am.py +5861 -0
  36. tinygrad/runtime/autogen/am/pm4_nv.py +962 -0
  37. tinygrad/runtime/autogen/am/pm4_soc15.py +931 -0
  38. tinygrad/runtime/autogen/am/sdma_4_0_0.py +5209 -0
  39. tinygrad/runtime/autogen/am/sdma_4_4_2.py +5209 -0
  40. tinygrad/runtime/autogen/am/sdma_5_0_0.py +7103 -0
  41. tinygrad/runtime/autogen/am/sdma_6_0_0.py +8085 -0
  42. tinygrad/runtime/autogen/am/smu_v13_0_0.py +3068 -0
  43. tinygrad/runtime/autogen/am/smu_v14_0_2.py +3605 -0
  44. tinygrad/runtime/autogen/amd_gpu.py +1433 -67197
  45. tinygrad/runtime/autogen/comgr.py +35 -9
  46. tinygrad/runtime/autogen/comgr_3.py +906 -0
  47. tinygrad/runtime/autogen/cuda.py +2419 -494
  48. tinygrad/runtime/autogen/hsa.py +57 -16
  49. tinygrad/runtime/autogen/ib.py +7171 -0
  50. tinygrad/runtime/autogen/io_uring.py +917 -118
  51. tinygrad/runtime/autogen/kfd.py +748 -26
  52. tinygrad/runtime/autogen/libc.py +613 -218
  53. tinygrad/runtime/autogen/libusb.py +1643 -0
  54. tinygrad/runtime/autogen/nv/nv.py +8602 -0
  55. tinygrad/runtime/autogen/nv_gpu.py +7218 -2072
  56. tinygrad/runtime/autogen/opencl.py +2 -4
  57. tinygrad/runtime/autogen/sqtt.py +1789 -0
  58. tinygrad/runtime/autogen/vfio.py +3 -3
  59. tinygrad/runtime/autogen/webgpu.py +273 -264
  60. tinygrad/runtime/graph/cuda.py +3 -3
  61. tinygrad/runtime/graph/hcq.py +68 -29
  62. tinygrad/runtime/graph/metal.py +29 -13
  63. tinygrad/runtime/graph/remote.py +114 -0
  64. tinygrad/runtime/ops_amd.py +537 -320
  65. tinygrad/runtime/ops_cpu.py +108 -7
  66. tinygrad/runtime/ops_cuda.py +12 -14
  67. tinygrad/runtime/ops_disk.py +13 -10
  68. tinygrad/runtime/ops_dsp.py +47 -40
  69. tinygrad/runtime/ops_gpu.py +13 -11
  70. tinygrad/runtime/ops_hip.py +6 -9
  71. tinygrad/runtime/ops_llvm.py +35 -15
  72. tinygrad/runtime/ops_metal.py +29 -19
  73. tinygrad/runtime/ops_npy.py +5 -3
  74. tinygrad/runtime/ops_null.py +28 -0
  75. tinygrad/runtime/ops_nv.py +306 -234
  76. tinygrad/runtime/ops_python.py +62 -52
  77. tinygrad/runtime/ops_qcom.py +28 -39
  78. tinygrad/runtime/ops_remote.py +482 -0
  79. tinygrad/runtime/ops_webgpu.py +28 -28
  80. tinygrad/runtime/support/am/amdev.py +114 -249
  81. tinygrad/runtime/support/am/ip.py +211 -172
  82. tinygrad/runtime/support/amd.py +138 -0
  83. tinygrad/runtime/support/{compiler_hip.py → compiler_amd.py} +40 -8
  84. tinygrad/runtime/support/compiler_cuda.py +8 -11
  85. tinygrad/runtime/support/elf.py +2 -1
  86. tinygrad/runtime/support/hcq.py +184 -97
  87. tinygrad/runtime/support/ib.py +172 -0
  88. tinygrad/runtime/support/llvm.py +3 -4
  89. tinygrad/runtime/support/memory.py +251 -0
  90. tinygrad/runtime/support/nv/__init__.py +0 -0
  91. tinygrad/runtime/support/nv/ip.py +581 -0
  92. tinygrad/runtime/support/nv/nvdev.py +183 -0
  93. tinygrad/runtime/support/system.py +170 -0
  94. tinygrad/runtime/support/usb.py +268 -0
  95. tinygrad/runtime/support/webgpu.py +18 -0
  96. tinygrad/schedule/__init__.py +0 -0
  97. tinygrad/schedule/grouper.py +119 -0
  98. tinygrad/schedule/kernelize.py +368 -0
  99. tinygrad/schedule/multi.py +231 -0
  100. tinygrad/shape/shapetracker.py +40 -46
  101. tinygrad/shape/view.py +88 -52
  102. tinygrad/tensor.py +968 -542
  103. tinygrad/uop/__init__.py +117 -0
  104. tinygrad/{codegen/transcendental.py → uop/decompositions.py} +125 -38
  105. tinygrad/uop/mathtraits.py +169 -0
  106. tinygrad/uop/ops.py +1021 -0
  107. tinygrad/uop/spec.py +228 -0
  108. tinygrad/{codegen → uop}/symbolic.py +239 -216
  109. tinygrad/uop/upat.py +163 -0
  110. tinygrad/viz/assets/cdnjs.cloudflare.com/ajax/libs/highlight.js/11.10.0/languages/x86asm.min.js +19 -0
  111. tinygrad/viz/assets/d3js.org/d3.v7.min.js +2 -0
  112. tinygrad/viz/assets/dagrejs.github.io/project/dagre/latest/dagre.min.js +801 -0
  113. tinygrad/viz/index.html +203 -403
  114. tinygrad/viz/js/index.js +718 -0
  115. tinygrad/viz/js/worker.js +29 -0
  116. tinygrad/viz/serve.py +224 -102
  117. {tinygrad-0.10.2.dist-info → tinygrad-0.11.0.dist-info}/METADATA +24 -16
  118. tinygrad-0.11.0.dist-info/RECORD +141 -0
  119. {tinygrad-0.10.2.dist-info → tinygrad-0.11.0.dist-info}/WHEEL +1 -1
  120. tinygrad/codegen/kernel.py +0 -693
  121. tinygrad/engine/multi.py +0 -161
  122. tinygrad/ops.py +0 -1003
  123. tinygrad/runtime/ops_cloud.py +0 -220
  124. tinygrad/runtime/support/allocator.py +0 -94
  125. tinygrad/spec.py +0 -155
  126. tinygrad/viz/assets/d3js.org/d3.v5.min.js +0 -2
  127. tinygrad/viz/assets/dagrejs.github.io/project/dagre-d3/latest/dagre-d3.min.js +0 -4816
  128. tinygrad/viz/perfetto.html +0 -178
  129. tinygrad-0.10.2.dist-info/RECORD +0 -99
  130. {tinygrad-0.10.2.dist-info → tinygrad-0.11.0.dist-info/licenses}/LICENSE +0 -0
  131. {tinygrad-0.10.2.dist-info → tinygrad-0.11.0.dist-info}/top_level.txt +0 -0
@@ -414,7 +414,6 @@ CL_Rx = 0x10BA # macro
414
414
  CL_RGx = 0x10BB # macro
415
415
  CL_RGBx = 0x10BC # macro
416
416
  CL_DEPTH = 0x10BD # macro
417
- CL_DEPTH_STENCIL = 0x10BE # macro
418
417
  CL_sRGB = 0x10BF # macro
419
418
  CL_sRGBx = 0x10C0 # macro
420
419
  CL_sRGBA = 0x10C1 # macro
@@ -435,7 +434,6 @@ CL_UNSIGNED_INT16 = 0x10DB # macro
435
434
  CL_UNSIGNED_INT32 = 0x10DC # macro
436
435
  CL_HALF_FLOAT = 0x10DD # macro
437
436
  CL_FLOAT = 0x10DE # macro
438
- CL_UNORM_INT24 = 0x10DF # macro
439
437
  CL_UNORM_INT_101010_2 = 0x10E0 # macro
440
438
  CL_MEM_OBJECT_BUFFER = 0x10F0 # macro
441
439
  CL_MEM_OBJECT_IMAGE2D = 0x10F1 # macro
@@ -1477,7 +1475,7 @@ __all__ = \
1477
1475
  'CL_COMPILE_PROGRAM_FAILURE', 'CL_COMPLETE', 'CL_CONTEXT_DEVICES',
1478
1476
  'CL_CONTEXT_INTEROP_USER_SYNC', 'CL_CONTEXT_NUM_DEVICES',
1479
1477
  'CL_CONTEXT_PLATFORM', 'CL_CONTEXT_PROPERTIES',
1480
- 'CL_CONTEXT_REFERENCE_COUNT', 'CL_DEPTH', 'CL_DEPTH_STENCIL',
1478
+ 'CL_CONTEXT_REFERENCE_COUNT', 'CL_DEPTH',
1481
1479
  'CL_DEVICE_ADDRESS_BITS', 'CL_DEVICE_AFFINITY_DOMAIN_L1_CACHE',
1482
1480
  'CL_DEVICE_AFFINITY_DOMAIN_L2_CACHE',
1483
1481
  'CL_DEVICE_AFFINITY_DOMAIN_L3_CACHE',
@@ -1704,7 +1702,7 @@ __all__ = \
1704
1702
  'CL_SAMPLER_PROPERTIES', 'CL_SAMPLER_REFERENCE_COUNT',
1705
1703
  'CL_SIGNED_INT16', 'CL_SIGNED_INT32', 'CL_SIGNED_INT8',
1706
1704
  'CL_SNORM_INT16', 'CL_SNORM_INT8', 'CL_SUBMITTED', 'CL_SUCCESS',
1707
- 'CL_TRUE', 'CL_UNORM_INT16', 'CL_UNORM_INT24', 'CL_UNORM_INT8',
1705
+ 'CL_TRUE', 'CL_UNORM_INT16', 'CL_UNORM_INT8',
1708
1706
  'CL_UNORM_INT_101010', 'CL_UNORM_INT_101010_2',
1709
1707
  'CL_UNORM_SHORT_555', 'CL_UNORM_SHORT_565', 'CL_UNSIGNED_INT16',
1710
1708
  'CL_UNSIGNED_INT32', 'CL_UNSIGNED_INT8', 'CL_VERSION_MAJOR_BITS',