1
0
mirror of https://github.com/em-winterschon/ZLUDA-CUDA-on-Intel-GPU synced 2026-10-24 00:49:02 +03:00

Commit Graph

  • 872054ae40 Fix linguist instructions master Andrzej Janik 2024-08-07 13:29:03 +02:00
  • 90a1f77891 Update README Andrzej Janik 2024-08-06 16:32:23 +02:00
  • 164c172236 Clean up ZLUDA redirection helper Andrzej Janik 2022-02-04 14:14:51 +01:00
  • 2753d956df Overhaul DLL injection Andrzej Janik 2022-02-04 00:50:25 +01:00
  • c869a0d611 Add tests for injecting into CLR process Andrzej Janik 2022-02-03 12:28:42 +01:00
  • 9923a36b76 Redo DLL injection Andrzej Janik 2022-02-01 23:57:36 +01:00
  • 89bc40618b Implement static typing for dynamically-loaded CUDA DLLs Andrzej Janik 2022-01-28 16:44:46 +01:00
  • 07aa1103aa Add OGL interop to cuda proc macros Andrzej Janik 2022-01-26 11:32:20 +01:00
  • 6f76c8b34c Fix crash when printing arrays Andrzej Janik 2022-01-08 18:44:59 +01:00
  • 2e56871643 Fix luid printing Andrzej Janik 2022-01-08 00:33:26 +01:00
  • 869efbe0e2 Move zluda_dump to the new CUDA infrastructure Andrzej Janik 2022-01-07 04:20:33 +01:00
  • 9390db962b Start converting everything to the new log formatting Andrzej Janik 2021-12-20 09:35:13 +01:00
  • bdcef897cc Start converting zluda_dump logging to provide more detailed Andrzej Janik 2021-12-19 01:18:03 +01:00
  • 971951bc9e Improve reporting of recovered unrecognized statement/directive Andrzej Janik 2021-12-14 00:02:23 +01:00
  • 0ca14d740f Better reporting of unrecognized tokens Andrzej Janik 2021-12-13 22:25:26 +01:00
  • 7ba1586d6c Make all user errors recoverable Andrzej Janik 2021-12-13 17:20:06 +01:00
  • 04dbafaf4a Fix typo Andrzej Janik 2021-12-08 00:39:08 +01:00
  • 50e793e869 Improve logging of dumps Andrzej Janik 2021-12-08 00:29:00 +01:00
  • dd915688bd Dump function names and allow overriding result of cuDeviceComputeCapability Andrzej Janik 2021-12-06 23:54:47 +01:00
  • 2c6d7ffb7a Fix remaining issues with detouring nvcuda Andrzej Janik 2021-12-05 23:01:46 +01:00
  • 26bf0eeaf2 Add more tests for CUDA redirection Andrzej Janik 2021-12-02 23:47:37 +01:00
  • 400feaf015 Add test for injecting app that directly uses nvcuda Andrzej Janik 2021-12-01 23:08:07 +01:00
  • fd1c13560f Add failing test for overriding directly linked CUDA dll Andrzej Janik 2021-11-30 01:25:51 +01:00
  • 3558a0a65c Add integration tests for injector Andrzej Janik 2021-11-28 23:51:41 +01:00
  • 0104814ac3 Refactor ZLUDA redirection Andrzej Janik 2021-11-27 14:33:04 +01:00
  • a125b0746f Do full dumping from dark api module creation Andrzej Janik 2021-11-21 02:23:01 +01:00
  • 2cb5960a18 Add missing file Andrzej Janik 2021-11-16 02:08:14 +01:00
  • 24e100cb9c Start converting cuModuleLoad to the new tracing infrastructure Andrzej Janik 2021-11-16 02:07:50 +01:00
  • e459086c5b Now dump function calls Andrzej Janik 2021-11-14 02:25:51 +01:00
  • dd7ced8b37 Create proper logging infrastructure Andrzej Janik 2021-11-12 00:31:40 +01:00
  • 816365e7df Fix shared munging pass and add fix cuModuleLoadData Andrzej Janik 2021-09-29 21:49:47 +00:00
  • 0172dc58e5 Redo shared memory transformation Andrzej Janik 2021-09-29 02:24:32 +02:00
  • b763415006 Add CUDA tests showing problems with .shared unification Andrzej Janik 2021-09-27 00:42:10 +02:00
  • c23be576e8 Finish fixing shared memory pass Andrzej Janik 2021-09-26 01:24:14 +02:00
  • 370c0bd09e Start implementing .shared unification Andrzej Janik 2021-09-24 01:31:50 +02:00
  • 9609f86033 Fix minor bugs Andrzej Janik 2021-09-19 00:39:43 +00:00
  • afe9120868 Fix linkage Andrzej Janik 2021-09-18 22:49:00 +00:00
  • 04a411fe22 Have an implementation for vprintf Andrzej Janik 2021-09-18 20:22:47 +00:00
  • ccf3c02ac1 Minor fixes Andrzej Janik 2021-09-18 01:36:12 +00:00
  • 3de01b3f8b Handle ld.volatile/st.volatile Andrzej Janik 2021-09-17 21:26:15 +00:00
  • d5a4b068dd Redo handling of sregs Andrzej Janik 2021-09-17 20:53:44 +00:00
  • 6ef19d6501 Add early support for more sregs Andrzej Janik 2021-09-17 18:31:12 +00:00
  • 5b2352723f Implement function pointers and activemask Andrzej Janik 2021-09-17 16:24:25 +00:00
  • c37223fe67 Set linking output Andrzej Janik 2021-09-17 01:25:34 +00:00
  • 62ce1fd3a9 Implement linking Andrzej Janik 2021-09-16 23:26:02 +00:00
  • 04394dbb04 Oops Andrzej Janik 2021-09-16 23:13:10 +00:00
  • 314e3dcb49 Add missing V-RAY host functions Andrzej Janik 2021-09-16 23:11:34 +00:00
  • ca0d8ec666 Add missing vray instructions Andrzej Janik 2021-09-16 01:25:09 +02:00
  • 467782b1d0 Fix some unhandled cases in cvt instruction Andrzej Janik 2021-09-14 23:38:06 +00:00
  • 2cd0fcb650 Parse and test const buffers Andrzej Janik 2021-09-14 22:41:46 +02:00
  • 986fa49097 Zero out buffer on creation Andrzej Janik 2021-09-13 23:43:50 +00:00
  • dbb6f09ffa Continue HIP conversion Andrzej Janik 2021-09-13 17:59:40 +00:00
  • e248a2c9a9 Dump modules passed for linking Andrzej Janik 2021-09-13 18:11:47 +02:00
  • 2f951fa04c Support more export table functions Andrzej Janik 2021-09-12 22:10:07 +02:00
  • 5290190727 Add support for a new export table Andrzej Janik 2021-09-12 15:12:26 +02:00
  • ab67cd46fc Fix dumping on x64 Linux (and possibly Windows) Andrzej Janik 2021-09-10 19:21:25 +02:00
  • da9cf4d583 Update tests, disable OpenCL-style shared mem conversion, emit linking information Andrzej Janik 2021-09-09 00:17:39 +00:00
  • a27d1e119f Merge commit 'e2fbdf7d7b78ee360fe5fcc13ed1a7cca727c921' into amd Andrzej Janik 2021-09-08 23:07:54 +02:00
  • e2fbdf7d7b Emit Execution mode for denorm flush Andrzej Janik 2021-09-08 23:05:52 +02:00
  • 18245be7d5 Make ptx unit tests run on AMD (except denormals) Andrzej Janik 2021-09-07 23:24:49 +00:00
  • 82510ce8fd Convert unit tests to HIP Andrzej Janik 2021-09-06 22:58:12 +02:00
  • a71cd44104 HIP conversion part #3 Andrzej Janik 2021-09-05 17:29:33 +02:00
  • a63f004540 HIP conversion part #2 Andrzej Janik 2021-09-05 14:58:15 +02:00
  • 5ec18f14a1 HIP conversion part #1 Andrzej Janik 2021-08-27 23:04:52 +00:00
  • 4ae7feb93a Start converting host code to HIP Andrzej Janik 2021-08-27 17:27:48 +00:00
  • 9631a8d242 Don't wait on a blocking event Andrzej Janik 2021-08-08 17:43:14 +02:00
  • e2432d0df1 Improve AMD compatibility Andrzej Janik 2021-08-08 14:26:56 +02:00
  • 043172bd9b Additional options to clang Andrzej Janik 2021-08-08 02:42:38 +02:00
  • 5969e59aae Explicitly mark input to AMD as bitcode Andrzej Janik 2021-08-08 02:35:17 +02:00
  • 4b4ba90219 Persist AMD kernels for later debug Andrzej Janik 2021-08-07 18:10:53 +02:00
  • 44decaf396 Use raw interop for building programs Andrzej Janik 2021-08-07 16:53:26 +02:00
  • 407664600a Hack to read clang output Andrzej Janik 2021-08-07 16:16:59 +02:00
  • 3ce6aee65d Try seeking before reading Andrzej Janik 2021-08-07 16:06:15 +02:00
  • 3070d983ab Take path to llvm-spirv from environment Andrzej Janik 2021-08-07 15:29:19 +02:00
  • fefdd528d5 Handle xnack suffix in device name Andrzej Janik 2021-08-07 15:21:45 +02:00
  • 20c9aa4f02 Fix warnings Andrzej Janik 2021-08-07 15:13:33 +02:00
  • 8d79c5d0f8 Addd missing file Andrzej Janik 2021-08-07 15:11:18 +02:00
  • a610136e17 Remove L0 from nvml Andrzej Janik 2021-08-07 15:11:08 +02:00
  • 479014a783 Wire up AMD compilation Andrzej Janik 2021-08-06 13:19:55 +02:00
  • 5bfc2a56b9 Remove all use of L0 Andrzej Janik 2021-08-06 02:15:57 +02:00
  • becda31524 Convert OpenCL host code to SVM Andrzej Janik 2021-08-04 19:34:56 +02:00
  • 638786b0ec Hack enough functionality that AMD GPU code builds Andrzej Janik 2021-08-03 00:22:47 +02:00
  • b4de21fbc5 Use calls to OpenCL builtins when translating sregs, do SPIRV->LLVM conversion on every build Andrzej Janik 2021-08-02 01:04:05 +02:00
  • 4a71fefb8a Change codegen for mul.wide Andrzej Janik 2021-08-01 19:20:08 +02:00
  • 8f68287b18 Tune generated code, add a workaround for geekbench Andrzej Janik 2021-07-25 15:19:43 +02:00
  • 9d4f26bd07 Finish converting to OpenCL Andrzej Janik 2021-07-22 00:20:33 +02:00
  • 3d2024bf62 Start converting to OpenCL Andrzej Janik 2021-07-21 01:46:50 +02:00
  • 58fb8a234c Synchronize through barrier Andrzej Janik 2021-07-06 00:16:46 +02:00
  • 2c6bee4955 Fix overzealus check Andrzej Janik 2021-07-05 01:14:46 +02:00
  • 1897d33916 Fix typo Andrzej Janik 2021-07-05 01:04:19 +02:00
  • 5b593ec185 Implement stream-wide event reuse Andrzej Janik 2021-07-04 19:06:37 +02:00
  • ce25035051 Use immediate command lists Andrzej Janik 2021-07-04 16:46:07 +02:00
  • ecc33f7b10 Make everything async Andrzej Janik 2021-07-04 15:40:08 +02:00
  • d76ffd691c Remember to actually submit workload Andrzej Janik 2021-07-04 13:24:35 +02:00
  • b460e359ae First attempt at async host side Andrzej Janik 2021-07-04 12:54:27 +02:00
  • ad2059872a Regenerate SPIR-V for ptx_impl and fix weird handling of ptr-ptr add or sub Andrzej Janik 2021-07-03 02:13:38 +02:00
  • e328ecc550 Be more correct when emitting brev, refactor inst->func call pass Andrzej Janik 2021-07-02 22:45:09 +02:00
  • 7d4fbedfcf Allow to set range of dump kernels Andrzej Janik 2021-06-30 23:56:15 +02:00
  • b2a455e12e Bunch of tiny fixes and improvements Andrzej Janik 2021-06-28 01:40:35 +02:00
  • 196242b410 Revert "Fix offset calculation in kernel launch" Andrzej Janik 2021-06-27 21:55:08 +02:00