14866 Commits
Author SHA1 Message Date
Henrik Rydgard 466cdb8ddf x86 Jit: Basic implementation of SIMD vmmul. Can be improved. 2014-11-30 19:27:43 +01:00
Henrik Rydgard 74e70f1159 Fix silly typo 2014-11-30 17:24:56 +01:00
Henrik Rydgard ac772f25ff x86 JIT: Join adjacent vrot calls together to avoid redundant sin/cos calls. Add a prototype, fix minor issues. 2014-11-30 11:04:13 +01:00
Henrik Rydgård 7deb8055ee Merge pull request #7117 from unknownbrackets/jit-simd
x86jit: Implement vmmov using SIMD
2014-11-30 09:24:22 +01:00
Unknown W. Brackets 679a824487 x86jit: when reusing a reg, don't zap slot.
If we returned the same reg it was in before, don't zap the slots because
slot #1 may be valid.
2014-11-29 23:11:50 -08:00
Unknown W. Brackets c3578bd8eb x86jit: Find the correct available regs.
If there aren't n available regs, they may not of course be in slots 0 and
1.  Have to go and find them.
2014-11-29 22:12:40 -08:00
Unknown W. Brackets df6a07f24e x86jit: Allow only one reg for sequential load. 2014-11-29 22:11:52 -08:00
Unknown W. Brackets bb26e4f7d0 x86jit: Implement vmmov using SIMD.
4x -> 87x in microbenchmarking.
2014-11-29 18:46:38 -08:00
Kingcom 6b29fe9d60 Change VROT disassembly 2014-11-29 18:33:33 +01:00
Unknown W. Brackets 67608639c5 x86jit: Oops, can't have loaded regs here yet. 2014-11-29 09:07:25 -08:00
Henrik Rydgard 8bd20ed8d1 x86 jit: Implement matrix init ops in SIMD. Turn off SIMD again by default (oops) 2014-11-29 12:30:21 +01:00
Henrik Rydgard 8f016d3e48 Merge some matrix utils and stuff from the NEON branch 2014-11-29 11:37:45 +01:00
Henrik Rydgård ae15722a2e Merge pull request #7112 from unknownbrackets/jit-simd
jit: MAP_NOINIT should always mean MAP_DIRTY
2014-11-29 10:19:33 +01:00
Unknown W. Brackets f6f943de63 jit: MAP_NOINIT should always mean MAP_DIRTY. 2014-11-29 00:14:08 -08:00
Henrik Rydgard 32c81c3265 x86 jit vcrsp.t: Oops, don't "SimpleReg" before doing the SIMD solution.. 2014-11-28 01:06:32 +01:00
Henrik Rydgard 344f71b092 x86 jit: Commit commented-out haddps-based vdot.q as reminder not to use haddps... 2014-11-28 00:19:11 +01:00
Henrik Rydgård 6ea5f6ab02 Merge pull request #7104 from Kingcom/Assembler
Replace assembler with a submodule
2014-11-27 23:56:49 +01:00
Henrik Rydgard 8f4d322dc6 Another oops... 2014-11-27 23:33:03 +01:00
Henrik Rydgard bcdfb496a0 Oops, bad merge 2014-11-27 23:12:57 +01:00
Henrik Rydgard c5bf3adec0 x86 jit: use the correct fp move instruction, minor optimization in vdot 2014-11-27 23:08:15 +01:00
Unknown W. Brackets 0839ce59e1 x86jit: Apply dirty when reusing 1-reg VS. 2014-11-27 01:05:11 -08:00
Unknown W. Brackets bbeb5758b7 x86jit: Simplify VS() / VSX() usage. 2014-11-27 00:07:17 -08:00
Unknown W. Brackets 039510a3e3 x86jit: Respect dirty even for non-simd. 2014-11-26 23:29:20 -08:00
Unknown W. Brackets f63c165f64 x86jit: Fix several cases of missing dirty checks. 2014-11-26 23:28:14 -08:00
Henrik Rydgard acb711007f x86 jit: SIMD-ify cross product 2014-11-27 00:18:19 +01:00
Henrik Rydgard 5033babb10 x86 Jit: SIMD-ify vdot 2014-11-26 23:47:18 +01:00
Henrik Rydgard bbd0afd148 x86 jit: Disable SIMD by default, needs just a little more debugging.. 2014-11-26 22:30:41 +01:00
Henrik Rydgard 4b25afb7b4 x86 Jit: SIMD some more instructions 2014-11-26 22:30:06 +01:00
Henrik Rydgard becdb14a3f x86 jit: Fix a couple issues in simd storefromregister 2014-11-26 22:25:18 +01:00
Henrik Rydgård 80797dc723 Merge pull request #7108 from unknownbrackets/jit-simd
Fix the simd issue at least for Wipeout
2014-11-26 21:42:02 +01:00
Unknown W. Brackets abd425d9f1 x86jit: Re-enable the simd optimization. 2014-11-26 09:21:15 -08:00
Unknown W. Brackets e5dabaabe2 x86jit: Optimize simd->non for 1-lane a little. 2014-11-26 09:20:50 -08:00
Unknown W. Brackets 5d0c32d1e6 x86jit: Assume non-simd regs are dirty. 2014-11-26 09:19:50 -08:00
Henrik Rydgard 804de50711 x86 jit: SIMD-ify VFPU register file writebacks where possible 2014-11-26 01:33:05 +01:00
Henrik Rydgard b3c8a82c49 x86 jit: SIMD-ify some more 2014-11-25 23:56:46 +01:00
Henrik Rydgard b5ee47a80c x86 jit: SIMD-ify lv.q and sv.q 2014-11-25 23:28:29 +01:00
Henrik Rydgård 4db6b7f3e2 SIMD-ify a couple instructions a bit 2014-11-25 22:47:26 +01:00
Unknown W. Brackets ced555ee47 Oops, fix a mistake in umd replacement. 2014-11-25 12:03:16 -08:00
Unknown W. Brackets efd7f1cc36 Fix a few warnings, cleanup. 2014-11-25 11:58:03 -08:00
Unknown W. Brackets 53c3841698 Buildfix for Linux and Symbian. 2014-11-25 11:48:08 -08:00
Unknown W. Brackets aeb3a327f0 Oops, typo. 2014-11-25 11:39:34 -08:00
Unknown W. Brackets 0d8dc48200 Don't try to close invalid files. 2014-11-25 11:34:39 -08:00
Unknown W. Brackets 0551773599 Add an interface to retry HTTP requests.
They could fail / be spotty in some situations.  A little retry can go a
long way.
2014-11-25 09:16:47 -08:00
Unknown W. Brackets 53e6d49568 Shortcut a read outside the file.
This happened to me.
2014-11-25 09:16:17 -08:00
Unknown W. Brackets 99b422aa0a Add more verbose logging in case of error. 2014-11-25 08:59:29 -08:00
Unknown W. Brackets 6d9ff8fac4 Fix disconnect order. 2014-11-25 00:33:23 -08:00
Unknown W. Brackets 219d546b79 Handle case in HTTP headers better. 2014-11-25 00:30:48 -08:00
Unknown W. Brackets df5020c439 Treat generation 0 a bit specially.
I guess we could always check gen 0 first, that'd be readahead/backfill.
2014-11-24 14:11:24 -08:00
Unknown W. Brackets 27eb38df8d Fix a possible leak.
Can definitely see this happening.
2014-11-24 14:08:21 -08:00
Unknown W. Brackets 824e8cf17c Add a readahead cache for HTTP.
So the only other thing it could do is idle backfill...
2014-11-24 01:08:30 -08:00