Commit Graph

395 Commits

Author SHA1 Message Date
Zihao Yu 9307402f18 apps,microbench: update doc 2019-05-25 16:28:54 +08:00
Zihao Yu 4544fccc4c apps,microbench: add train input of medium size 2019-05-25 16:20:52 +08:00
Zihao Yu ed0f10a778 Revert "apps,litenes: use uint16_t for word memory accessing"
This reverts commit 5ae9a76c1e.

* it may cause unaligned memory accessing on riscv32 and mips32
2019-05-09 20:24:46 +08:00
Yanyan Jiang 4f186f5da9 am: merge changes 2019-05-07 05:02:21 +00:00
Yanyan Jiang aea00f757d litenes: lock FPS to 60 2019-05-06 18:36:36 +00:00
Yanyan Jiang e39f4f4d39 x86-qemu, makefiles: fixes, remove x86 tracing (need redesign) 2019-05-06 17:45:53 +08:00
Yanyan Jiang 6102652dc8 x86-qemu, makefiles: simplify ld commands 2019-05-06 17:11:43 +08:00
Zihao Yu 8ac4d491b8 apps,litenes,fce: do not wait for next frame if the current frame is skipped 2019-05-02 20:14:48 +08:00
Zihao Yu 6d9672f6e8 apps,litenes,ppu: process one frame in ppu_frame()
* this makes it possible to use more aggressive code for skip frame
2019-05-02 19:40:53 +08:00
Zihao Yu aa54af854c apps,litenes,ppu: fix buffer overflow bug when drawing sprite with y > H 2019-05-02 18:18:35 +08:00
Zihao Yu b7af5518f6 apps,litenes: smaller memory consumption 2019-05-02 15:55:15 +08:00
Zihao Yu c0524c8890 apps,litenes,ppu: simply sprite0 hit checking 2019-05-02 15:28:08 +08:00
Zihao Yu b9161a8568 apps,litenes,fce: use smaller screen array 2019-05-02 13:30:55 +08:00
Zihao Yu 4c772f1a1e apps,litenes,ppu: refactor scanline drawing 2019-05-02 11:30:10 +08:00
Zihao Yu 86476dc148 apps,litenes,ppu: only draw the scanline covered by sprite0 for hit check 2019-05-01 16:15:07 +08:00
Zihao Yu 5da7efea86 apps,litenes,ppu: re-organize pattern table to fetch XHLidx friendly 2019-05-01 13:46:02 +08:00
Zihao Yu 14422979c5 apps,litenes,ppu: split scanline updating from drawing 2019-05-01 11:55:45 +08:00
Zihao Yu 4e18c1bff3 apps,litenes,ppu: store XHLidx to check sprite0 hit, instead of XHL value 2019-05-01 11:06:00 +08:00
Zihao Yu c4abe72189 apps,litenes,ppu: directly set color16 to background for all pass
* Since last commit, the non-mirror pass and the mirror pass do not
  overlap. Combining them will give the complete screen.
2019-05-01 10:17:57 +08:00
Zihao Yu c4b65cfa51 apps,litenes,ppu: skip tile_x which is on the left of the screen 2019-04-30 22:16:26 +08:00
Zihao Yu 34bd77711a apps,litenes,ppu: draw background scanline with bbg color directly 2019-04-30 22:14:26 +08:00
Zihao Yu 2469881c61 apps,litenes,ppu: directly set color16 to background for the non-mirror pass 2019-04-30 18:29:34 +08:00
Zihao Yu 5ae9a76c1e apps,litenes: use uint16_t for word memory accessing 2019-04-30 11:55:48 +08:00
Zihao Yu 342e5ec844 apps,litenes,cpu: skip spin 2019-04-30 11:26:46 +08:00
Zihao Yu 4095581af4 apps,litenes,ppu: refactor profiling code 2019-04-29 16:20:01 +08:00
Zihao Yu ac0249005a apps,litenes,ppu: do not examine every sprite
* also put the checking of sprite0 hit out of the inner loop
2019-04-29 16:08:42 +08:00
Zihao Yu 83f10b8cc6 apps,litenes,fce: fix frame_cnt 2019-04-29 14:22:01 +08:00
Zihao Yu e6c8d1952f apps,litenes,ppu: use memcpy() for DMA 2019-04-29 13:39:00 +08:00
Zihao Yu 93b25bf7f2 apps,litenes,memory: inline mmc and psg accessing 2019-04-29 13:24:13 +08:00
Zihao Yu 1a98d2009d apps,litenes,cpu: optimize lazy flag for zf 2019-04-29 12:48:49 +08:00
Zihao Yu e98b03d54d apps,litenes,cpu: remove extra cycle statistic during addressing
* it seems still work
2019-04-29 12:41:02 +08:00
Zihao Yu aebf0f2211 apps,litenes,cpu: let flag be an integer instead of a bit
* this can simplify some flag computation
* to use the flag as a bit, use cpu_flag_setbit(flag)
2019-04-29 11:45:59 +08:00
Zihao Yu e8dfb2fecc apps,litenes,cpu: do not compute target if the branch is not taken 2019-04-29 11:43:58 +08:00
Zihao Yu d506f5e8fe apps,litenes,ppu: change palette cache to color cache 2019-04-29 10:08:45 +08:00
Zihao Yu a3a875bb7a apps,litenes: clean up 2019-04-29 02:10:41 +08:00
Zihao Yu b6ccd9e998 apps,litenes,ppu: optimize sprite drawing 2019-04-29 02:08:22 +08:00
Zihao Yu ad21d7b8e8 apps,litenes,ppu: misc optimization 2019-04-29 00:15:41 +08:00
Zihao Yu 1255a78591 apps,litenes,ppu: skip transparent tiles 2019-04-28 22:13:43 +08:00
Zihao Yu 3d67310923 apps,litenes,ppu: use one dimension array for XHL tables
* this can reduce the address calculation
2019-04-28 21:57:32 +08:00
Zihao Yu 797d39291f apps,litenes,fce: read key during the processing of scanlines
* this will reduce the latency of key respones when the host is very
slow
2019-04-28 21:54:46 +08:00
Zihao Yu 0218408031 apps,litenes,ppu: use struct for sprite 2019-04-28 11:34:08 +08:00
Zihao Yu 8b496feda8 apps,litenes,ppu: optimize loop 2019-04-27 21:54:57 +08:00
Zihao Yu 236abcd412 apps,litenes,ppu,draw: remove boundary check 2019-04-27 21:16:33 +08:00
Zihao Yu 6aa80a43c6 apps,litenes,ppu: optimize loop boundary 2019-04-27 20:05:44 +08:00
Zihao Yu 7e645c1382 apps,litenes,ppu: use uint16_t for XHL table, instead of uint64_t 2019-04-27 20:04:38 +08:00
Zihao Yu 107eaf04b1 apps,litenes,ppu: fix scanline bug 2019-04-27 16:08:24 +08:00
Zihao Yu 4f623fc794 apps,litenes,ppu: use preprocess table to obtain attribute 2019-04-25 23:11:28 +08:00
Zihao Yu 331d001b84 apps,litenes,fce: add PROFILE macro 2019-04-24 22:25:55 +08:00
Zihao Yu 4f3092cb8a apps,litenes: add non-stretch display mode 2019-04-24 22:18:47 +08:00
Zihao Yu b38213b4cd apps,litenes,ppu: only guard ppu.PPUADDR when it is updated 2019-04-24 22:17:35 +08:00