273 commits

Author	SHA1	Message	Date
unknown	2b073a276d	all shuffling for multiplies now in SSE2	2013-09-18 03:02:36 -04:00
unknown	fce5402ba1	started SSE2 intrinsics dynamic inclusion for shuffling	2013-09-18 01:41:36 -04:00
unknown	12aa9cf331	forgot to in-line SIGNED_CLAMP, other tech fixes	2013-09-17 03:40:15 -04:00
unknown	76f8358287	try to amend vector intelligence for GCC 4.8.1 upgrade	2013-09-17 02:26:32 -04:00
unknown	af2c59b4b9	quieten integer overflow warning with gcc -pedantic	2013-09-16 22:45:00 -04:00
unknown	b176173f55	reconstructed scalar 64-bit segmentations as 16-bit SSE steps	2013-09-16 22:33:30 -04:00
unknown	9e260a641a	restored old static vector clamping for VADD/VSUB	2013-09-16 03:01:24 -04:00
unknown	25cdf537a6	applied remaining vector microoptimizations	2013-09-16 02:38:14 -04:00
unknown	8b9712aefd	More stupid wrong assumptions about signedness. Die.	2013-09-16 02:23:07 -04:00
unknown	d8d3c7ce90	in-depth vector microoptimizations to VMUDN SSE	2013-09-16 02:18:54 -04:00
unknown	4a003f47c2	fix wrong assumption on sign if corner case 0x8000*0x8000	2013-09-16 02:15:46 -04:00
unknown	3f370bf8ea	vectorization boost and corner case fix to VMULF	2013-09-16 01:52:17 -04:00
unknown	28e1081ef3	corrected sign mask set in VACC hi (Mario, SSB gfx)	2013-09-15 21:53:00 -04:00
unknown	7c791ae5ba	$vco is now vectorized. (And that's the last of them.)	2013-09-15 02:19:59 -04:00
unknown	b81fb83d53	$vcc is now vectorized.	2013-09-14 19:31:22 -04:00
unknown	ea759d0db6	$vce is now vectorized.	2013-09-14 14:14:50 -04:00
unknown	ca343c7900	compress signed clamping vectorizer	2013-09-14 13:31:49 -04:00
unknown	1192748b33	amend storage ambiguities to the compiler's vectorizer	2013-09-14 03:35:32 -04:00
unknown	675bddbb76	fix almost every possible remaining warning	2013-09-13 17:51:16 -04:00
unknown	29d2c8f278	useless warnings about call-parameter type mismatches	2013-09-13 17:38:23 -04:00
unknown	8e69fa7b8b	fix unused function/structure warnings	2013-09-13 17:27:37 -04:00
unknown	dfb3b6371a	obvious bug is obvious.	2013-09-13 17:22:42 -04:00
unknown	49bd94cd9f	completely vectorized all accumulator R/W	2013-09-13 15:04:06 -04:00
unknown	0f98ed926f	refactored old buggy VCH writeback to VCC	2013-09-12 03:06:38 -04:00
unknown	52dc51b015	forgot to include sign when checking le/ge, reversed ternary	2013-09-11 22:55:13 -04:00
unknown	4d4773175b	accidentally wrote VCO instead of VCC	2013-09-11 11:48:23 -04:00
unknown	2341bd39fc	potential debugger hazard when testing old functions	2013-09-11 09:19:29 -04:00
unknown	1aa5dfe5fd	fixed misprioritization of 64-bit type conversion	2013-09-11 08:36:01 -04:00
unknown	68bd431a63	full static rewrite of vector unit	2013-09-11 06:58:09 -04:00
unknown	64311fb516	more Notepad++ typo suggestions	2013-09-02 19:21:45 -04:00
RJ Swedlow	46082660bc	support for more pedantic, ANSI C rules	2013-08-26 23:03:02 -04:00
RJ Swedlow	be355e4190	ANSI GNU /Wall strict compliance updates	2013-08-26 21:19:12 -04:00
RJ Swedlow	c55651c805	finished scalar unit jump table prototype	2013-08-23 02:01:43 -04:00
RJ Swedlow	a122360a7e	static reduction of vector control register jump tables	2013-08-17 02:10:13 -04:00
RJ Swedlow	e9481816f4	no more USER32/SHELL32 dependencies, plain CRT for messages	2013-08-14 15:48:35 -04:00
RJ Swedlow	7d6c1417c6	safer intrinsics to guard against bugged MinGW arch code gen	2013-08-09 14:42:19 -04:00
RJ Swedlow	36c6d2d031	fixed a corner case for overflow clamps	2013-08-08 23:28:53 -04:00
RJ Swedlow	e87066933e	uh, forgot the sperm	2013-08-08 22:52:19 -04:00
RJ Swedlow	410e27d2fc	found vector ops where new static clamp can apply	2013-08-08 22:30:53 -04:00
RJ Swedlow	16ee596312	C inline functions compatible with Visual Studio	2013-08-08 20:23:22 -04:00
RJ Swedlow	a635542af6	discovered an old SSE intrinsic for corner case	2013-08-08 18:07:56 -04:00
RJ Swedlow	2f8751e15b	faster vector sign clamps	2013-08-08 17:27:52 -04:00
RJ Swedlow	d98b5aac64	new macro option EMULATE_VECTOR_RESULT_BUFFER	2013-05-15 11:19:18 -04:00
RJ Swedlow	34dcc48a53	repairs to manual dynamic vector parallelization mode	2013-05-13 21:35:17 -04:00
RJ Swedlow	8a6aa39a18	finished implementing parallel shuffle macros	2013-05-13 20:12:37 -04:00
RJ Swedlow	bebd498375	starting to implement shuffling in master template	2013-05-10 12:17:37 -04:00
RJ Swedlow	b54d1785d5	prototype solution to shuffle vector scalars to coefficient	2013-05-09 23:18:40 -04:00
RJ Swedlow	f1ff598a14	clean up the crap	2013-04-13 22:53:28 -05:00
RJ Swedlow	7c8c574509	momentary reversion to bitwise search method, seems better	2013-04-11 21:54:11 -05:00
RJ Swedlow	0aad6e82d5	inline function method to maintain clamping functions	2013-04-11 18:50:05 -05:00