Commit Graph

  • 5b31fbb270 benchmarking: changes to not fill up all of the RAM main Daniel 2025-05-10 15:10:26 +0200
  • 2ba1fef5ba benchmarking: removed caches to get initial performance measurement. still some problems Daniel 2025-05-10 13:11:27 +0200
  • 6d3c3164cf expression processing: added support for inverse/reciprocal Daniel 2025-05-10 09:36:02 +0200
  • 7121329a17 benchmarking: started tuning benchmarking results. found some errors that need fixing Daniel 2025-05-09 19:19:53 +0200
  • 327e4ebf1b benchmarking: prepared tests for using actual data Daniel 2025-05-09 13:58:10 +0200
  • 2c8a9cd2d8 added support for variables and parameters as array. also improved conversion of variables and parameters into Expressionelement Daniel 2025-05-09 11:04:10 +0200
  • aaa3f2c7c0 interpreter: reverted previous change to make interpreter work correctly again Daniel 2025-05-09 09:45:43 +0200
  • ae03ebdf1d implementation: finished re-reading chapter and fixed suboptimal wording Daniel 2025-05-08 16:32:05 +0200
  • 6b7205e026 implementation: Finished chapter. checking for errors and textual improvements not finished Daniel 2025-05-08 12:34:08 +0200
  • b69a3efe96 implementation: continued writing transpiler section Daniel 2025-05-04 13:54:05 +0200
  • 18d89e27ca implementation: started transpiler section Daniel 2025-05-03 12:25:34 +0200
  • e8e457eae9 implementation: finished interpreter section and started transpiler section Daniel 2025-05-02 12:35:03 +0200
  • c4187a131e implementation: finished interpreter section Daniel 2025-05-01 14:36:14 +0200
  • 101b13e7e7 implementation: started writing interpreter section; finished CPU-side part; also found error with interpreter Daniel 2025-04-27 13:17:35 +0200
  • e571fa5bd6 implementation: finished pre-processing section; updated code Daniel 2025-04-26 13:46:23 +0200
  • ad2eab2e0a implementation: continued pre-processing section; added cache to ExpressionProcessing.jl to improve performance Daniel 2025-04-25 16:22:57 +0200
  • b40a06af3f implementation: started describing frontend Daniel 2025-04-21 11:58:48 +0200
  • 210831146a implementation: started writing impl; finished technology section Daniel 2025-04-20 12:18:40 +0200
  • 90a4194283 expression processing: fixed error if expression contained nested unary operators such as log(sqrt(4)) Daniel 2025-04-19 10:54:41 +0200
  • 0d888edc52
    Merge pull request #2 from daniwipes/expression-file daniwipes 2025-04-19 08:43:35 +0200
  • 742a544e1a concept and design: finished reworking sections Daniel 2025-04-18 12:41:54 +0200
  • 4fe9040a6f Make @btime work Gabriel Kronberger 2025-04-18 11:41:25 +0200
  • bc49b33149 Bugfix Gabriel Kronberger 2025-04-18 11:41:11 +0200
  • 293c5f13a4 Add files with expressions, parser, Nikuradse dataset and a new test case. Gabriel Kronberger 2025-04-18 08:19:22 +0200
  • 6ab826cc42 concept and design: rewrote requirements section Daniel 2025-04-17 15:12:58 +0200
  • c3f19d807a relwork: implemented Kronberger feedback Daniel 2025-04-17 13:31:11 +0200
  • 6880c1ceb5 benchmarking: added uni performance test Daniel 2025-04-16 20:54:12 +0200
  • c62aff806a small updates and notes for further writing Daniel 2025-04-15 19:32:39 +0200
  • ef721b13e0 evaluation: updated notes for chapter Daniel 2025-04-13 14:20:16 +0200
  • a5c34a53b7 benchmarking: reverted previous; made interpreter use fast math Daniel 2025-04-13 13:26:35 +0200
  • 6d6874c7ba benchmarking: added results for test 4 on uni-pc Daniel 2025-04-13 11:45:52 +0200
  • 278a493595 benchmarking: tested int32 also on uni pc 4-interpreter-using-int32 Daniel 2025-04-13 11:43:17 +0200
  • af3b72f196 benchmarking: used int32 wherever possible; resulted in noticeable performance drop Daniel 2025-04-13 11:32:54 +0200
  • 4c60331288 evaluation: added introduction text and made plan for additional text Daniel 2025-04-12 16:22:14 +0200
  • 101ccef67b benchmarking: tuned blocksizes; slightly improved performance; mostly improved standard deviation 3-tuned-blocksize-i128-t96 Wiplinger Daniel - s2310454043 2025-04-12 13:20:50 +0200
  • c6e2ce47aa benchmarking: redid performance tests on uni pc Wiplinger Daniel - s2310454043 2025-04-12 12:13:28 +0200
  • 9746db56c0 benchmarking: added tests for performance tuning with Nsight compute and systems 2-performance-inbounds Daniel 2025-04-12 10:24:45 +0200
  • 5a9760d221 concept and design: minor improvements Daniel 2025-04-10 11:17:50 +0200
  • c68e0d04a0 concept and design: improved wording and added overview diagram of kernel usage Daniel 2025-04-10 10:21:01 +0200
  • 258d33c338 benchmarking: redid inbounds tests because first test run did not use inbounds everywhere Daniel 2025-04-06 14:29:07 +0200
  • 20fcbab4ca concept and design: added transpiler section Daniel 2025-04-06 13:59:14 +0200
  • 9e1094ac43 benchmarking: added reminder to redo one of the performance tests Daniel 2025-04-05 12:56:06 +0200
  • 2a8de064a6 concept & design: improved existing sections Daniel 2025-04-05 12:54:26 +0200
  • 8afc3a5e3b concept and design: continued with architecture Daniel 2025-04-04 14:18:56 +0200
  • d8f5454e9c concept and design: started writing this chapter Daniel 2025-04-03 13:43:23 +0200
  • 2b9c394f1b benchmarking: improved performance with @inbounds. still slower in most cases Daniel 2025-04-01 21:48:59 +0200
  • d9c83caad9 benchmarking: updated blocksize to 256 with moderate improvements 1-performance-256-blocksize Daniel 2025-03-30 13:56:25 +0200
  • 1dc0c1898d benchmarking: fixed bugs; took initial_benchmark initial-implementation Daniel 2025-03-30 12:54:50 +0200
  • ad175abac0 benchmarking: added gpu evaluators to performance testing. getting execution errors still Daniel 2025-03-29 13:35:59 +0100
  • 690ee33db1 benchmarks: started preparing benchmarks Daniel 2025-03-29 12:01:06 +0100
  • effd477558 transpiler: generates valid PTX and evaluates expressions correctly Daniel 2025-03-28 19:32:48 +0100
  • 9df78ca72e transpiler: invalid memory access error finally fixed Daniel 2025-03-27 22:32:24 +0100
  • 561b37160b transpiler: trying to fix problem with writing to global memory; not yet fixed Daniel 2025-03-27 18:00:31 +0100
  • eaee21ca75 transpiler: results are now written in results array; preperation for performance testing Daniel 2025-03-27 09:55:29 +0100
  • baa37ea183 code: started finalising transpilation process and preparing for performance testing and tuning Daniel 2025-03-23 13:38:22 +0100
  • db02e9f90f
    Merge pull request #1 from daniwipes/cpu-interpreter daniwipes 2025-03-23 10:23:54 +0100
  • e33be8f59e relwork: finished second read to improve wording and correct mistakes Daniel 2025-03-23 09:57:23 +0100
  • 3c289f13d1 relwork: slight rewording and fixing errors Daniel 2025-03-21 17:49:55 +0100
  • a718a3572e relwork: finished relwork Daniel 2025-03-21 14:35:55 +0100
  • db3ea32b66 relwork: continuation of compilers section Daniel 2025-03-20 13:31:45 +0100
  • d514b07434 relwork: started compilers section Daniel 2025-03-18 10:32:14 +0100
  • de5493ca3e relwork: finished occupancy and ptx sections Daniel 2025-03-16 17:43:32 +0100
  • 84fdf5c9ca relwork: continuation of programing gpus Daniel 2025-03-15 14:33:33 +0100
  • f3446a2b11 finished implementing thesis feedback Daniel 2025-03-14 16:11:25 +0100
  • ed9d8766be started implementing feedback Daniel 2025-03-13 13:18:50 +0100
  • fddfa23b4f related work: slight restructuring; continued with section programming gpus Daniel 2025-03-09 13:04:12 +0100
  • 4e48686b62 related work: small continuation of explaining SIMT Daniel 2025-03-08 14:12:50 +0100
  • b683f3ae96 relwork: continued with 'programming GPUs' Daniel 2025-03-08 12:28:46 +0100
  • 203e157f11 Related work: continuation of GPGPU Daniel 2025-03-02 12:23:59 +0100
  • 34d98f9997 Related Work: finished equation learning section; started GPGPU section Daniel 2025-03-01 13:14:37 +0100
  • 28ef6b121e related work: continuation of equation learning section Daniel 2025-02-27 11:41:01 +0100
  • 99ed6a1cca Related Work: started with equation learning section Daniel 2025-02-26 13:34:46 +0100
  • 52b5407b5c Introduction: slight improvements Daniel 2025-02-23 11:11:00 +0100
  • 433e69fff5 Introduction: finished first version of chapter Daniel 2025-02-22 11:42:18 +0100
  • f4f39ec47c Improvements / fixes. Gabriel Kronberger 2025-02-19 17:18:43 +0100
  • 942adb8612 Add CPU Interpreter and a test case. Gabriel Kronberger 2025-02-19 16:38:11 +0100
  • 8bad911585 introduction: updated background + research question Daniel 2025-02-15 11:12:44 +0100
  • 250da02353 updated introduction: background and motivation Daniel 2025-02-14 12:42:42 +0100
  • 4afc15a737 small cleanup Daniel 2025-01-26 10:16:23 +0100
  • 7598c51df8 finished generating parameter loading code Daniel 2025-01-25 11:15:54 +0100
  • b2774322a1 added loading of variables from memory into registers. Note: Needed to leave, so code currently not compiling Daniel 2025-01-24 13:13:53 +0100
  • 85464083c3 aded symtable for loading vars and params to local memory Daniel 2025-01-19 11:00:29 +0100
  • 219c0bb14e started implementing parameter loading Daniel 2025-01-06 14:01:55 +0100
  • f7926c3438 finished implementing operators Daniel 2025-01-05 11:19:03 +0100
  • 094f8c9499 Merge branch 'main' of https://github.com/daniwipes/master-thesis Daniel 2025-01-04 10:40:26 +0100
  • 9702fe2343 added thesis structure Daniel 2025-01-04 10:38:27 +0100
  • a97b804530 testing interpreter because loop seems to do everything. Can't test right now because I don't have a nvidia gpu in my laptop Daniel Wiplinger 2024-12-11 11:29:54 +0100
  • 8d129dbfcc rewrote function for generating code for operators. now the entire operation will be returned and not just the operator Daniel 2024-12-10 22:58:18 +0100
  • 67ef9a5139 reworked code to use new 'register manager' Daniel 2024-12-08 10:50:09 +0100
  • 203900bb26 nevermind, improved register management worked Daniel 2024-12-07 10:23:04 +0100
  • 1e7f6e9010 tried streamlining register management Daniel 2024-12-07 10:12:53 +0100
  • 68cedd75fc updated all to 32-bit to save registers and boost performance Daniel 2024-11-01 11:23:58 +0100
  • 9fc55c4c15 started implementing transpilation of expression Daniel 2024-10-27 11:48:11 +0100
  • 0e24d74e54 small increment and fixes Daniel 2024-10-26 11:41:00 +0200
  • ee3c5001bd added information on how to best approach register assignment Daniel 2024-10-20 12:24:18 +0200
  • 1f6b40b750 added possibility to create registers Daniel 2024-10-19 13:44:38 +0200
  • de73d83d9e updated compat helper to not always run Daniel Wiplinger 2024-10-02 13:19:14 +0200
  • 7283082699 added guard clause generation Daniel 2024-09-28 11:41:13 +0200
  • d875fc7325 first steps towards ptx generation Daniel 2024-09-25 12:45:46 +0200
  • d60cba7e4d continued understanding given PTX file and made plan on how to approach the transpiler part Daniel 2024-09-22 09:32:39 +0200