Ben Vanik
|
ad9472e60b
|
Better sign extend, thanks to @inolen.
|
2015-07-27 20:25:20 -07:00 |
|
gibbed
|
42ef3f224a
|
Oops. Fixing bad asserts.
|
2015-07-26 11:12:54 -05:00 |
|
gibbed
|
b713f99e42
|
Constant support for VECTOR_CONVERT_I2F.
|
2015-07-26 10:55:41 -05:00 |
|
gibbed
|
20d79911f0
|
Constant support for Emit16_IN_32.
|
2015-07-26 10:54:47 -05:00 |
|
gibbed
|
599686f68d
|
Unsigned Emit16_IN_32.
|
2015-07-26 09:57:52 -05:00 |
|
Ben Vanik
|
91d7acf59e
|
Fixing some particularly bad /W4.
|
2015-07-18 23:04:21 -07:00 |
|
Ben Vanik
|
f520d3a2a4
|
Run with latest clang-format.
|
2015-07-15 22:44:30 -07:00 |
|
Ben Vanik
|
e05cc3ba37
|
x64_sequences can now be run through clang-format.
Progress on #291.
|
2015-06-28 15:04:24 -07:00 |
|
Ben Vanik
|
0359184bc0
|
Simplifying the sequences code to start refactoring.
Progress on #291.
|
2015-06-28 12:01:47 -07:00 |
|
Ben Vanik
|
0630b524fa
|
Removing unused HIR opcodes.
Progress on #291.
|
2015-06-28 12:01:46 -07:00 |
|
Ben Vanik
|
5e950cb066
|
Fixing VS intellisense for xbyak.
Progress on #291.
|
2015-06-28 12:01:45 -07:00 |
|
Ben Vanik
|
fb1f4906d9
|
xb format --all (we are now format clean). Buildbot will yell at you.
|
2015-06-22 22:26:51 -07:00 |
|
Ben Vanik
|
fb591cbadb
|
Disabling use of FMA until I can figure out the rounding issues.
|
2015-06-20 11:21:32 -07:00 |
|
Drew DeVault
|
a064ba053e
|
Add missing includes for cstring
|
2015-06-16 23:16:18 -06:00 |
|
Ben Vanik
|
c34db170f5
|
Adding fallback for pre-MOVBE-supporting processors.
|
2015-06-16 20:06:50 -07:00 |
|
Ben Vanik
|
ddaf08ca8d
|
Adding flag to make LOAD/STORE also perform a byte swap.
|
2015-06-15 18:59:28 -07:00 |
|
gibbed
|
c949c1d303
|
Support src1 and src2 being a constant in OPCODE_EXTRACT.
|
2015-06-14 02:08:41 -05:00 |
|
gibbed
|
6d933c18f7
|
SHL_V128.
|
2015-06-09 19:55:54 -05:00 |
|
Ben Vanik
|
f0c47e036b
|
Fixing SHR_V128.
|
2015-06-09 17:42:09 -07:00 |
|
Ben Vanik
|
cd62d4e461
|
PACK to 2101010.
|
2015-06-07 19:44:19 -07:00 |
|
Ben Vanik
|
429698c2ec
|
Replacing DidCarry with manual calculation and fixing compares.
|
2015-06-06 15:24:46 -07:00 |
|
gibbed
|
54e8868160
|
Some (hopefully correct) constant fixes for MUL_ADD_V128, MUL_SUB_V128, and SWIZZLE V128. Needs extra work on AVX2 side.
|
2015-06-05 23:29:31 -05:00 |
|
Dr. Chat
|
809c327041
|
Emulate PACK/UNPACK for non-F16C CPUs
|
2015-06-03 22:33:45 -05:00 |
|
Ben Vanik
|
40a6a12800
|
Attempting to inline a lot of MMIO operations. Works for GPU stuff mainly.
|
2015-06-02 20:15:43 -07:00 |
|
Ben Vanik
|
8244409501
|
Refactoring to enable future time scaling, coordinated clocks, etc.
|
2015-05-26 22:20:58 -07:00 |
|
gibbed
|
f2685dbb1d
|
Fixed SPLAT_F32 for non-AVX2 branch.
|
2015-05-25 14:32:38 -05:00 |
|
gibbed
|
05081fa030
|
Fixed some constant loads.
|
2015-05-24 16:18:06 -05:00 |
|
Dr. Chat
|
f7a8dc3748
|
Change OPCODE_SPLAT instructions from SSE->AVX
|
2015-05-23 17:27:26 -05:00 |
|
Ben Vanik
|
c1a0f46ea8
|
Implementing dcbz.
|
2015-05-17 16:40:38 -07:00 |
|
Dr. Chat
|
5dbec09818
|
Change SSE opcodes back to AVX (mixing is bad)
|
2015-05-16 14:10:28 -05:00 |
|
Dr. Chat
|
65ff862495
|
Whoops (typo)
|
2015-05-16 13:41:33 -05:00 |
|
Dr. Chat
|
8e35ec0620
|
Rewrite FMA mul sub/add to avoid register shuffling
Also downgrade some AVX opcodes to SSE
|
2015-05-16 13:31:42 -05:00 |
|
Rick Gibbed
|
7546cca936
|
Merge pull request #207 from DrChat/mul_8_fix
Fix 8bit multiply instructions
|
2015-05-15 10:22:30 -05:00 |
|
Dr. Chat
|
dbdf3271fb
|
Fix 8bit multiply instructions
|
2015-05-15 10:21:07 -05:00 |
|
Dr. Chat
|
a46b63558e
|
Clean up mul sub/mul add code and fix a minor bug in mulsub f64
|
2015-05-15 10:09:19 -05:00 |
|
Ben Vanik
|
0d3e79ad2c
|
Supporting const src1 mul_hi.
|
2015-05-13 00:19:07 -07:00 |
|
Ben Vanik
|
91c6ad8715
|
Fixing adde_constant.
|
2015-05-12 22:53:20 -07:00 |
|
Dr. Chat
|
82eecb752b
|
Fix 8bit and 16bit mul instructions and clean up some code
|
2015-05-12 13:35:13 -05:00 |
|
gibbed
|
24b13b03f5
|
Fixing more cpu feature guards.
|
2015-05-12 00:26:45 -05:00 |
|
Ben Vanik
|
497e5a9f69
|
Fixing cpu feature guards.
|
2015-05-11 22:13:55 -07:00 |
|
Ben Vanik
|
ad7efa964f
|
Adding --enable_haswell_instructions=false to disable modern instructions.
|
2015-05-11 21:42:10 -07:00 |
|
Ben Vanik
|
94c62b91d0
|
--trace_functions and --trace_function_coverage
|
2015-05-05 22:44:36 -07:00 |
|
Dr. Chat
|
2e9aec194f
|
Ctrl+C Ctrl+V
|
2015-05-05 16:00:41 -05:00 |
|
Dr. Chat
|
26d8858684
|
Emulated vector shr
|
2015-05-05 15:58:51 -05:00 |
|
Dr. Chat
|
cb127ae9ba
|
Emulate some vector arithmetic opcodes (until we get a native implementation)
|
2015-05-05 15:39:37 -05:00 |
|
Dr. Chat
|
cff09a4509
|
OPCODE_VECTOR_SHA and SPLAT_I16 for non-AVX2 CPUs
|
2015-05-05 14:35:37 -05:00 |
|
gibbed
|
3249f84700
|
It's vpblendw.
|
2015-05-05 13:58:21 -05:00 |
|
gibbed
|
6694461698
|
Fixed ecx restore in EmitShlXX/EmitShrXX/EmitShaXX.
|
2015-05-05 13:58:03 -05:00 |
|
Ben Vanik
|
743e01209b
|
Update x64_sequences.cc
|
2015-05-05 10:23:04 -07:00 |
|
Dr. Chat
|
c3f1e74814
|
And actually use vpblendw
|
2015-05-05 12:16:38 -05:00 |
|