| Commit message (Expand) | Author | Age | Files | Lines |
|---|
| * | * converted register_maybe_adjust_setbase() to the high level code generator | jonas | 2015-12-05 | 1 | -1/+1 |
| * | * Removed unused vars for x86 compiler. | yury | 2015-09-17 | 1 | -1/+0 |
| * | * remove implicit typecast to extended automatically inserted for trunc/round | jonas | 2015-07-03 | 1 | -0/+11 |
| * | * don't give an internalerror when trying to prefetch a regvar or even | jonas | 2015-04-08 | 1 | -1/+1 |
| * | + cpu capability CPUX86_HAS_CMOV | florian | 2015-02-21 | 1 | -1/+1 |
| * | + make use of vfnmsub*/vfmsub*/vfnmadd* instructions if possible | florian | 2014-05-02 | 1 | -9/+67 |
| * | * try to registers more cleverly during fma code generation | florian | 2014-04-26 | 1 | -4/+23 |
| * | * fix compilation of i8086 compiler | florian | 2014-04-13 | 1 | -0/+4 |
| * | + support for FMA intrinsic: if there is no hardware support, the compiler th... | florian | 2014-04-13 | 1 | -1/+97 |
| * | * moved pbestrealtype from symdef to symcpu | jonas | 2014-04-01 | 1 | -1/+1 |
| * | * fixed compilation for i8086 with 2.6.x (it gave an unreachable code | jonas | 2014-03-29 | 1 | -11/+8 |
| * | + Consider target_info.cprefix in symbol names. | sergei | 2014-03-28 | 1 | -2/+2 |
| * | * completed thlcgobj.location_force_fpureg(), use it everywhere and removed | jonas | 2014-03-10 | 1 | -1/+1 |
| * | + create avx code in abs as well | florian | 2014-02-22 | 1 | -6/+27 |
| * | * i8086 has no cpu_capabilites | florian | 2014-02-16 | 1 | -1/+6 |
| * | * popcnt has a separate cpuid flag so we should separate it from sse | florian | 2014-02-16 | 1 | -1/+1 |
| * | + proper (i.e. not broken) implementation of arctan() for the 8087 and 80287 | nickysn | 2013-12-12 | 1 | -0/+9 |
| * | * use the fwait prefixed FSTCW, followed by FWAIT, instead of FNSTCW in | nickysn | 2013-12-05 | 1 | -2/+13 |
| * | * Don't generate x87 instructions for sin and cos on win64, somehow managed t... | sergei | 2013-11-15 | 1 | -4/+14 |
| * | * Mantis #17273: don't generate x87 instructions on win64 target. | sergei | 2013-11-08 | 1 | -8/+29 |
| * | * tx86inlinenode.second_popcnt: use emit_ref_reg instead of appending instruc... | sergei | 2013-10-28 | 1 | -2/+2 |
| * | * handle LOC_*FPUREGISTER correctly when using sse2 sqr, resolve #26408 | florian | 2013-06-16 | 1 | -1/+3 |
| * | * basic avx support for floating point operations (use -Cfavx to activate) | florian | 2013-06-14 | 1 | -28/+68 |
| * | * renamed thlcgobj.tcgsize2orddef to defutil.cgsize_orddef | jonas | 2013-06-01 | 1 | -1/+1 |
| * | + support for LOC_(C)MMREGISTER in hlcg | jonas | 2013-05-31 | 1 | -5/+5 |
| * | * do not use fsin and fcos on 286-, since they're 387+ | nickysn | 2013-05-25 | 1 | -0/+32 |
| * | * use the generic implementation of second_IncludeExclude in case the cpu is ... | nickysn | 2013-05-25 | 1 | -0/+8 |
| * | * fixed round() on i8086 after r24502 | nickysn | 2013-05-15 | 1 | -6/+0 |
| * | + enabled returning int64 in ax:bx:cx:dx for the i8086 | nickysn | 2013-05-14 | 1 | -10/+0 |
| * | * fixed bugs in tx86inlinenode, introduced in r24237 | nickysn | 2013-04-14 | 1 | -21/+28 |
| * | - removed ti8086inlinenode.second_round_real, second_trunc_real and load_fpu_... | nickysn | 2013-04-14 | 1 | -7/+16 |
| * | * use the generic implementation of abs for the i8086 | nickysn | 2013-03-25 | 1 | -0/+4 |
| * | * do not generate prefetchnt instructions on the i8086 | nickysn | 2013-03-25 | 1 | -2/+2 |
| * | * fix compilation on x86-64 | florian | 2012-09-03 | 1 | -1/+2 |
| * | * x86-64 and i386 use the popcnt instruction of possible | florian | 2012-09-02 | 1 | -1/+9 |
| * | + generic popcnt support | florian | 2012-09-02 | 1 | -0/+6 |
| * | * first draft to support the popcnt instruction, works so far for x86 with a ... | florian | 2012-09-02 | 1 | -0/+24 |
| * | * decrease the fpu stack position counter after explicitly inserted | jonas | 2012-06-29 | 1 | -0/+3 |
| * | * moved subsetref/reg and bit_set/test support from cgobj to hlcgobj for | jonas | 2012-05-13 | 1 | -7/+15 |
| * | o added ARM VPFv2/VFPv3 support: | jonas | 2009-12-03 | 1 | -9/+9 |
| * | * Always use RIP relative symbol addressing for x86_64 Windows targets. Bug #... | yury | 2009-05-13 | 1 | -0/+1 |
| * | * the compiler now explicitly keeps track of the minimally guaranteed | jonas | 2009-02-08 | 1 | -6/+6 |
| * | * always specify an explicit alignment for tgobj.gettemp (so e.g. | jonas | 2008-11-08 | 1 | -5/+5 |
| * | * fixed wrong location.size (sign) for x86's in_abs_long + test | jonas | 2008-05-01 | 1 | -2/+2 |
| * | * cleanup and simplify the set type handling | peter | 2008-03-02 | 1 | -1/+1 |
| * | * fixed include/exclude for byte-sized sets in registers | jonas | 2008-01-23 | 1 | -4/+15 |
| * | * typecast fix | peter | 2007-11-29 | 1 | -6/+2 |
| * | + inline abs(<longint>) on x86 | florian | 2007-11-25 | 1 | -0/+40 |
| * | * x86-64 fixed | florian | 2007-10-28 | 1 | -1/+1 |
| * | * fixed assembling of fisttp of sse3 instruction set | florian | 2007-10-27 | 1 | -15/+26 |
