summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
* Add MMC5, not in a working state, but can start castlevania iii, this is a ↵Peter Fors2025-11-028-35/+379
| | | | horrible mapper to implement.
* add new mapper, 004_0 MMC3Peter Fors2025-11-025-1/+238
|
* add sdl version for simpler graphical debuggingPeter Fors2025-11-021-0/+361
|
* fix multiple mappersPeter Fors2025-11-0212-168/+203
|
* partial rewrite of ppu_read/write gained 1.27% performancePeter Fors2025-10-311-37/+24
|
* rewrote memory_read/write to use if/else chain for readability/understandabilityPeter Fors2025-10-311-36/+20
|
* Add script to build local gcc, build.sh already updatedPeter Fors2025-10-311-0/+102
|
* some tas tests, desync is happeningPeter Fors2025-10-318-4343/+83
|
* some optimization tests that didn't pan outPeter Fors2025-10-302-90/+2
|
* smb 3041fps, life force 2058fpsPeter Fors2025-10-291-115/+49
|
* Finally back above 3000fps - this was a lot of workPeter Fors2025-10-292-13/+100
| | | | | | | | | | | | | | | | | | | | vital@claybabble:/work/current/mknes(master*|u=)$ ./Bench.sh Performance counter coverage: 100% (no multiplexing - full precision) 352509230343 instructions # 4.81 insn per cycle # 0.01 stalled cycles per insn 73277044838 cycles # 5.416 GHz 2957685039 stalled-cycles-frontend # 4.04% frontend cycles idle 70065301653 branches # 5.179 G/sec 297927451 branch-misses # 0.43% of all branches Throughput: 26056.16 MIPS, 5416.36 Mcycles/sec cycles/frame mean= 1788990 sd= 2803 relSD=0.157% n=10 insn/frame mean= 8606182 sd= 0 relSD=0.000% n=10 time (ms) mean= 1352.883 sd= 2.140 relSD=0.158% n=10 FPS (frames/second) = 3027.61 ms/frame = 0.330294
* reached 2944fps againPeter Fors2025-10-287-27/+18
|
* update .gitignorePeter Fors2025-10-281-0/+1
|
* add bootstrap.sh to build a local version of gcc 5.2.0 for nowPeter Fors2025-10-281-0/+102
|
* small optimizations of sprite evaluation in ppu_render_pixelPeter Fors2025-10-287-145/+227
|
* Changes to layoutPeter Fors2025-10-263-18/+12
|
* Refactor benchmarking to self-contained C implementationPeter Fors2025-10-254-70/+360
| | | | | | | | | | | | | - Add mknes_bench.c with direct PMC access via perf_event_open() - Remove dependency on external perf/awk for statistics - Add RT priority and CPU affinity control in C code - Use static BSS allocation (page-aligned) instead of malloc - Add stalled-cycles-backend counter (gracefully handles AMD unavailability) - Add throughput metrics (MIPS, Mcycles/sec) - Optimize Bench.sh to only regenerate profile data when needed - Add -n and -f flags for configurable runs and frames - Suppress mapper messages during benchmark - ~6x faster benchmark workflow (20s first run, 16s subsequent)
* Forgot .gitignorePeter Fors2025-10-251-0/+1
|
* Better benchmarking, some small optimizationsPeter Fors2025-10-259-150/+211
|
* optimized, but bug with rasterdemosPeter Fors2025-10-1610-283/+174
|
* updated gitignorePeter Fors2025-10-091-0/+5
|
* missed onePeter Fors2025-10-091-0/+67
|
* Rearrangement and refactoring and optimizations and more accuracyPeter Fors2025-10-0952-1763/+11454
|
* testsPeter Fors2025-06-0810-123/+116
|
* added scaffolding for apuPeter Fors2025-06-045-23/+226
|
* optimization of ppuPeter Fors2025-05-293-4/+5
|
* rewrite of memory_read()Peter Fors2025-05-292-9/+11
|
* added a Bench.sh script to do benchmarking for mePeter Fors2025-05-291-0/+9
|
* added mapper_tick() functionality, regressed 200fps, and optimized it back ↵Peter Fors2025-05-299-84/+63
| | | | up to ~1940fps
* renamingPeter Fors2025-05-2718-185/+216
|
* 3011fpsPeter Fors2025-05-2423-366/+328
|
* new and changed mappersPeter Fors2025-05-0819-151/+338
|
* forgot the renamed mappersPeter Fors2025-05-032-0/+31
|
* cleanup and rewrite of memory_read()Peter Fors2025-05-0310-163/+61
|
* cleanup and rewrite of ppu_registers.cPeter Fors2025-05-0235-351/+517
|
* add a few python scripts for mapper generation and setupPeter Fors2025-04-283-0/+121
|
* reverted rewrite of ppu, optimized what functions should be forced inline, ↵Peter Fors2025-04-2826-4028/+25
| | | | gained ~2.5% performance
* before ppu rearrangement and fix of scanline 261Peter Fors2025-04-287-36/+75
|
* add -mtune=genericPeter Fors2025-04-212-7/+7
|
* back to 2560fps after wrangling the ppu_state into two cachelinesPeter Fors2025-04-217-27/+52
|
* Battletoads workingPeter Fors2025-04-167-50/+88
|
* 2592fps again...Peter Fors2025-04-111-3/+0
|
* Before partial rewrite of ppu_tick()Peter Fors2025-04-113-26/+42
|
* Last few optimizations to ppu_render_pixel, down to 1.59ns per pixelPeter Fors2025-04-093-8/+9
|
* 2690fpsPeter Fors2025-04-086-47/+47
|
* Back at 2618fpsPeter Fors2025-04-082-27/+30
|
* cleanup and optimization tests, unfortunately nothing great.Peter Fors2025-04-075-16/+23
|
* almost back to normal after bytestream excursionPeter Fors2025-04-074-159/+165
|
* cleanupPeter Fors2025-04-065-44/+28
|
* optimized away an if(dot == 256) in the hot path ~5% improvementPeter Fors2025-04-063-44/+59
|
generated by cgit v1.2.3 (git 2.46.0) at 2026-01-25 20:35:53 +0000