SPARC assembly optimizations from David Miller.
diff --git a/src/mesa/math/m_debug_util.h b/src/mesa/math/m_debug_util.h
index dc8024a..efedda9 100644
--- a/src/mesa/math/m_debug_util.h
+++ b/src/mesa/math/m_debug_util.h
@@ -1,4 +1,4 @@
-/* $Id: m_debug_util.h,v 1.3 2001/03/30 14:44:43 gareth Exp $ */
+/* $Id: m_debug_util.h,v 1.4 2001/05/23 14:27:03 brianp Exp $ */
/*
* Mesa 3-D graphics library
@@ -38,7 +38,9 @@
* NOTE: it works only on CPUs which know the 'rdtsc' command (586 or higher)
* (hope, you don't try to debug Mesa on a 386 ;)
*/
-#if defined(__GNUC__) && defined(__i386__) && defined(USE_X86_ASM)
+#if defined(__GNUC__) && \
+ ((defined(__i386__) && defined(USE_X86_ASM)) || \
+ (defined(__sparc__) && defined(USE_SPARC_ASM)))
#define RUN_DEBUG_BENCHMARK
#endif
@@ -67,6 +69,8 @@
* It is assumed that all calculations are done in the cache.
*/
+#if defined(__i386__)
+
#if 1 /* PPro, PII, PIII version */
/* Profiling on the P6 architecture requires a little more work, due to
@@ -183,6 +187,30 @@
#endif
+#elif defined(__sparc__)
+
+#define INIT_COUNTER() \
+ do { counter_overhead = 5; } while(0)
+
+#define BEGIN_RACE(x) \
+x = LONG_MAX; \
+for (cycle_i = 0; cycle_i <10; cycle_i++) { \
+ register long cycle_tmp1 asm("l0"); \
+ register long cycle_tmp2 asm("l1"); \
+ /* rd %tick, %l0 */ \
+ __asm__ __volatile__ (".word 0xa1410000" : "=r" (cycle_tmp1)); /* save timestamp */
+
+#define END_RACE(x) \
+ /* rd %tick, %l1 */ \
+ __asm__ __volatile__ (".word 0xa3410000" : "=r" (cycle_tmp2)); \
+ if (x > (cycle_tmp2-cycle_tmp1)) x = cycle_tmp2 - cycle_tmp1; \
+} \
+x -= counter_overhead;
+
+#else
+#error Your processor is not supported for RUN_XFORM_BENCHMARK
+#endif
+
#else
#define BEGIN_RACE(x)
diff --git a/src/mesa/math/m_xform.c b/src/mesa/math/m_xform.c
index 6943084..c043b09 100644
--- a/src/mesa/math/m_xform.c
+++ b/src/mesa/math/m_xform.c
@@ -1,4 +1,4 @@
-/* $Id: m_xform.c,v 1.13 2001/05/21 16:33:41 gareth Exp $ */
+/* $Id: m_xform.c,v 1.14 2001/05/23 14:27:03 brianp Exp $ */
/*
* Mesa 3-D graphics library
@@ -56,6 +56,10 @@
#include "X86/common_x86_asm.h"
#endif
+#ifdef USE_SPARC_ASM
+#include "SPARC/sparc.h"
+#endif
+
clip_func _mesa_clip_tab[5];
clip_func _mesa_clip_np_tab[5];
dotprod_func _mesa_dotprod_tab[5];
@@ -206,6 +210,9 @@
#ifdef USE_X86_ASM
_mesa_init_all_x86_transform_asm();
#endif
+#ifdef USE_SPARC_ASM
+ _mesa_init_all_sparc_transform_asm();
+#endif
}
void
diff --git a/src/mesa/sparc/clip.S b/src/mesa/sparc/clip.S
new file mode 100644
index 0000000..a569428
--- /dev/null
+++ b/src/mesa/sparc/clip.S
@@ -0,0 +1,234 @@
+/* $Id: clip.S,v 1.1 2001/05/23 14:27:03 brianp Exp $ */
+
+#ifdef __sparc_v9__
+#define LDPTR ldx
+#define V4F_DATA 0x00
+#define V4F_START 0x08
+#define V4F_COUNT 0x10
+#define V4F_STRIDE 0x14
+#define V4F_SIZE 0x18
+#define V4F_FLAGS 0x1c
+#else
+#define LDPTR ld
+#define V4F_DATA 0x00
+#define V4F_START 0x04
+#define V4F_COUNT 0x08
+#define V4F_STRIDE 0x0c
+#define V4F_SIZE 0x10
+#define V4F_FLAGS 0x14
+#endif
+
+#define VEC_SIZE_1 1
+#define VEC_SIZE_2 3
+#define VEC_SIZE_3 7
+#define VEC_SIZE_4 15
+
+ .text
+ .align 64
+
+one_dot_zero:
+ .word 0x3f800000 /* 1.0f */
+
+ /* This trick is shamelessly stolen from the x86
+ * Mesa asm. Very clever, and we can do it too
+ * since we have the necessary add with carry
+ * instructions on Sparc.
+ */
+clip_table:
+ .byte 0, 1, 0, 2, 4, 5, 4, 6
+ .byte 0, 1, 0, 2, 8, 9, 8, 10
+ .byte 32, 33, 32, 34, 36, 37, 36, 38
+ .byte 32, 33, 32, 34, 40, 41, 40, 42
+ .byte 0, 1, 0, 2, 4, 5, 4, 6
+ .byte 0, 1, 0, 2, 8, 9, 8, 10
+ .byte 16, 17, 16, 18, 20, 21, 20, 22
+ .byte 16, 17, 16, 18, 24, 25, 24, 26
+ .byte 63, 61, 63, 62, 55, 53, 55, 54
+ .byte 63, 61, 63, 62, 59, 57, 59, 58
+ .byte 47, 45, 47, 46, 39, 37, 39, 38
+ .byte 47, 45, 47, 46, 43, 41, 43, 42
+ .byte 63, 61, 63, 62, 55, 53, 55, 54
+ .byte 63, 61, 63, 62, 59, 57, 59, 58
+ .byte 31, 29, 31, 30, 23, 21, 23, 22
+ .byte 31, 29, 31, 30, 27, 25, 27, 26
+
+/* GLvector4f *clip_vec, GLvector4f *proj_vec,
+ GLubyte clipMask[], GLubyte *orMask, GLubyte *andMask */
+
+ .align 64
+__pc_tramp:
+ retl
+ nop
+
+ .globl _mesa_sparc_cliptest_points4
+_mesa_sparc_cliptest_points4:
+ save %sp, -64, %sp
+ call __pc_tramp
+ sub %o7, (. - one_dot_zero - 4), %g1
+ ld [%g1 + 0x0], %f4
+ add %g1, 0x4, %g1
+
+ ld [%i0 + V4F_STRIDE], %l1
+ ld [%i0 + V4F_COUNT], %g7
+ LDPTR [%i0 + V4F_START], %i0
+ LDPTR [%i1 + V4F_START], %i5
+ ldub [%i3], %g2
+ ldub [%i4], %g3
+ sll %g3, 8, %g3
+ or %g2, %g3, %g2
+
+ ld [%i1 + V4F_FLAGS], %g3
+ or %g3, VEC_SIZE_4, %g3
+ st %g3, [%i1 + V4F_FLAGS]
+ mov 3, %g3
+ st %g3, [%i1 + V4F_SIZE]
+ st %g7, [%i1 + V4F_COUNT]
+ clr %l2
+ clr %l0
+
+ /* l0: i
+ * g7: count
+ * l1: stride
+ * l2: c
+ * g2: (tmpAndMask << 8) | tmpOrMask
+ * g1: clip_table
+ * i0: from[stride][i]
+ * i2: clipMask
+ * i5: vProj[4][i]
+ */
+
+1: ld [%i0 + 0x0c], %f3 ! LSU Group
+ ld [%i0 + 0x0c], %g5 ! LSU Group
+ ld [%i0 + 0x08], %g4 ! LSU Group
+ fdivs %f4, %f3, %f8 ! FGM
+ addcc %g5, %g5, %g5 ! IEU1 Group
+ addx %g0, 0x0, %g3 ! IEU1 Group
+ addcc %g4, %g4, %g4 ! IEU1 Group
+ addx %g3, %g3, %g3 ! IEU1 Group
+ subcc %g5, %g4, %g0 ! IEU1 Group
+ ld [%i0 + 0x04], %g4 ! LSU Group
+ addx %g3, %g3, %g3 ! IEU1 Group
+ addcc %g4, %g4, %g4 ! IEU1 Group
+ addx %g3, %g3, %g3 ! IEU1 Group
+ subcc %g5, %g4, %g0 ! IEU1 Group
+ ld [%i0 + 0x00], %g4 ! LSU Group
+ addx %g3, %g3, %g3 ! IEU1 Group
+ addcc %g4, %g4, %g4 ! IEU1 Group
+ addx %g3, %g3, %g3 ! IEU1 Group
+ subcc %g5, %g4, %g0 ! IEU1 Group
+ addx %g3, %g3, %g3 ! IEU1 Group
+ ldub [%g1 + %g3], %g3 ! LSU Group
+ cmp %g3, 0 ! IEU1 Group, stall
+ be 2f ! CTI
+ stb %g3, [%i2] ! LSU
+ sll %g3, 8, %g4 ! IEU1 Group
+ add %l2, 1, %l2 ! IEU0
+ st %g0, [%i5 + 0x00] ! LSU
+ or %g4, 0xff, %g4 ! IEU0 Group
+ or %g2, %g3, %g2 ! IEU1
+ st %g0, [%i5 + 0x04] ! LSU
+ and %g2, %g4, %g2 ! IEU0 Group
+ st %g0, [%i5 + 0x08] ! LSU
+ b 3f ! CTI
+ st %f4, [%i5 + 0x0c] ! LSU Group
+2: ld [%i0 + 0x00], %f0 ! LSU Group
+ ld [%i0 + 0x04], %f1 ! LSU Group
+ ld [%i0 + 0x08], %f2 ! LSU Group
+ fmuls %f0, %f8, %f0 ! FGM
+ st %f0, [%i5 + 0x00] ! LSU Group
+ fmuls %f1, %f8, %f1 ! FGM
+ st %f1, [%i5 + 0x04] ! LSU Group
+ fmuls %f2, %f8, %f2 ! FGM
+ st %f2, [%i5 + 0x08] ! LSU Group
+ st %f8, [%i5 + 0x0c] ! LSU Group
+3: add %i5, 0x10, %i5 ! IEU1
+ add %l0, 1, %l0 ! IEU0 Group
+ add %i2, 1, %i2 ! IEU0 Group
+ cmp %l0, %g7 ! IEU1 Group
+ bne 1b ! CTI
+ add %i0, %l1, %i0 ! IEU0 Group
+ stb %g2, [%i3] ! LSU
+ srl %g2, 8, %g3 ! IEU0 Group
+ cmp %l2, %g7 ! IEU1 Group
+ bl,a 1f ! CTI
+ clr %g3 ! IEU0
+1: stb %g3, [%i4] ! LSU Group
+ ret ! CTI Group
+ restore %i1, 0x0, %o0
+
+ .globl _mesa_sparc_cliptest_points4_np
+_mesa_sparc_cliptest_points4_np:
+ save %sp, -64, %sp
+
+ call __pc_tramp
+ sub %o7, (. - one_dot_zero - 4), %g1
+ add %g1, 0x4, %g1
+
+ ld [%i0 + V4F_STRIDE], %l1
+ ld [%i0 + V4F_COUNT], %g7
+ LDPTR [%i0 + V4F_START], %i0
+ LDPTR [%i1 + V4F_START], %i5
+ ldub [%i3], %g2
+ ldub [%i4], %g3
+ sll %g3, 8, %g3
+ or %g2, %g3, %g2
+
+ ld [%i1 + V4F_FLAGS], %g3
+ or %g3, VEC_SIZE_4, %g3
+ st %g3, [%i1 + V4F_FLAGS]
+ mov 3, %g3
+ st %g3, [%i1 + V4F_SIZE]
+ st %g7, [%i1 + V4F_COUNT]
+ clr %l2
+ clr %l0
+
+ /* l0: i
+ * g7: count
+ * l1: stride
+ * l2: c
+ * g2: (tmpAndMask << 8) | tmpOrMask
+ * g1: clip_table
+ * i0: from[stride][i]
+ * i2: clipMask
+ */
+
+1: ld [%i0 + 0x0c], %g5 ! LSU Group
+ ld [%i0 + 0x08], %g4 ! LSU Group
+ addcc %g5, %g5, %g5 ! IEU1 Group
+ addx %g0, 0x0, %g3 ! IEU1 Group
+ addcc %g4, %g4, %g4 ! IEU1 Group
+ addx %g3, %g3, %g3 ! IEU1 Group
+ subcc %g5, %g4, %g0 ! IEU1 Group
+ ld [%i0 + 0x04], %g4 ! LSU Group
+ addx %g3, %g3, %g3 ! IEU1 Group
+ addcc %g4, %g4, %g4 ! IEU1 Group
+ addx %g3, %g3, %g3 ! IEU1 Group
+ subcc %g5, %g4, %g0 ! IEU1 Group
+ ld [%i0 + 0x00], %g4 ! LSU Group
+ addx %g3, %g3, %g3 ! IEU1 Group
+ addcc %g4, %g4, %g4 ! IEU1 Group
+ addx %g3, %g3, %g3 ! IEU1 Group
+ subcc %g5, %g4, %g0 ! IEU1 Group
+ addx %g3, %g3, %g3 ! IEU1 Group
+ ldub [%g1 + %g3], %g3 ! LSU Group
+ cmp %g3, 0 ! IEU1 Group, stall
+ be 2f ! CTI
+ stb %g3, [%i2] ! LSU
+ sll %g3, 8, %g4 ! IEU1 Group
+ add %l2, 1, %l2 ! IEU0
+ or %g4, 0xff, %g4 ! IEU0 Group
+ or %g2, %g3, %g2 ! IEU1
+ and %g2, %g4, %g2 ! IEU0 Group
+2: add %l0, 1, %l0 ! IEU0 Group
+ add %i2, 1, %i2 ! IEU0 Group
+ cmp %l0, %g7 ! IEU1 Group
+ bne 1b ! CTI
+ add %i0, %l1, %i0 ! IEU0 Group
+ stb %g2, [%i3] ! LSU
+ srl %g2, 8, %g3 ! IEU0 Group
+ cmp %l2, %g7 ! IEU1 Group
+ bl,a 1f ! CTI
+ clr %g3 ! IEU0
+1: stb %g3, [%i4] ! LSU Group
+ ret ! CTI Group
+ restore %i1, 0x0, %o0
diff --git a/src/mesa/sparc/sparc.c b/src/mesa/sparc/sparc.c
new file mode 100644
index 0000000..83741cf
--- /dev/null
+++ b/src/mesa/sparc/sparc.c
@@ -0,0 +1,109 @@
+/* $Id: sparc.c,v 1.1 2001/05/23 14:27:03 brianp Exp $ */
+
+/*
+ * Mesa 3-D graphics library
+ * Version: 3.1
+ *
+ * Copyright (C) 1999 Brian Paul All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/*
+ * Sparc assembly code by David S. Miller
+ */
+
+
+#include "glheader.h"
+#include "context.h"
+#include "math/m_vertices.h"
+#include "math/m_xform.h"
+#include "tnl/t_context.h"
+#include "sparc.h"
+
+#ifdef DEBUG
+#include "math/m_debug.h"
+#endif
+
+#define XFORM_ARGS GLvector4f *to_vec, \
+ const GLfloat m[16], \
+ const GLvector4f *from_vec
+
+#define DECLARE_XFORM_GROUP(pfx, sz) \
+ extern void _mesa_##pfx##_transform_points##sz##_general(XFORM_ARGS); \
+ extern void _mesa_##pfx##_transform_points##sz##_identity(XFORM_ARGS); \
+ extern void _mesa_##pfx##_transform_points##sz##_3d_no_rot(XFORM_ARGS); \
+ extern void _mesa_##pfx##_transform_points##sz##_perspective(XFORM_ARGS); \
+ extern void _mesa_##pfx##_transform_points##sz##_2d(XFORM_ARGS); \
+ extern void _mesa_##pfx##_transform_points##sz##_2d_no_rot(XFORM_ARGS); \
+ extern void _mesa_##pfx##_transform_points##sz##_3d(XFORM_ARGS);
+
+#define ASSIGN_XFORM_GROUP(pfx, sz) \
+ _mesa_transform_tab[sz][MATRIX_GENERAL] = \
+ _mesa_##pfx##_transform_points##sz##_general; \
+ _mesa_transform_tab[sz][MATRIX_IDENTITY] = \
+ _mesa_##pfx##_transform_points##sz##_identity; \
+ _mesa_transform_tab[sz][MATRIX_3D_NO_ROT] = \
+ _mesa_##pfx##_transform_points##sz##_3d_no_rot; \
+ _mesa_transform_tab[sz][MATRIX_PERSPECTIVE] = \
+ _mesa_##pfx##_transform_points##sz##_perspective; \
+ _mesa_transform_tab[sz][MATRIX_2D] = \
+ _mesa_##pfx##_transform_points##sz##_2d; \
+ _mesa_transform_tab[sz][MATRIX_2D_NO_ROT] = \
+ _mesa_##pfx##_transform_points##sz##_2d_no_rot; \
+ _mesa_transform_tab[sz][MATRIX_3D] = \
+ _mesa_##pfx##_transform_points##sz##_3d;
+
+
+#ifdef USE_SPARC_ASM
+DECLARE_XFORM_GROUP(sparc, 1)
+DECLARE_XFORM_GROUP(sparc, 2)
+DECLARE_XFORM_GROUP(sparc, 3)
+DECLARE_XFORM_GROUP(sparc, 4)
+
+extern GLvector4f *_mesa_sparc_cliptest_points4(GLvector4f *clip_vec,
+ GLvector4f *proj_vec,
+ GLubyte clipMask[],
+ GLubyte *orMask,
+ GLubyte *andMask);
+
+extern GLvector4f *_mesa_sparc_cliptest_points4_np(GLvector4f *clip_vec,
+ GLvector4f *proj_vec,
+ GLubyte clipMask[],
+ GLubyte *orMask,
+ GLubyte *andMask);
+#endif
+
+void _mesa_init_all_sparc_transform_asm(void)
+{
+#ifdef USE_SPARC_ASM
+ ASSIGN_XFORM_GROUP(sparc, 1)
+ ASSIGN_XFORM_GROUP(sparc, 2)
+ ASSIGN_XFORM_GROUP(sparc, 3)
+ ASSIGN_XFORM_GROUP(sparc, 4)
+
+ _mesa_clip_tab[4] = _mesa_sparc_cliptest_points4;
+ _mesa_clip_np_tab[4] = _mesa_sparc_cliptest_points4_np;
+
+#ifdef DEBUG
+ _math_test_all_transform_functions("sparc");
+ _math_test_all_cliptest_functions("sparc");
+#endif
+
+#endif
+}
diff --git a/src/mesa/sparc/sparc.h b/src/mesa/sparc/sparc.h
new file mode 100644
index 0000000..64422a3
--- /dev/null
+++ b/src/mesa/sparc/sparc.h
@@ -0,0 +1,37 @@
+/* $Id: sparc.h,v 1.1 2001/05/23 14:27:03 brianp Exp $ */
+
+/*
+ * Mesa 3-D graphics library
+ * Version: 3.1
+ *
+ * Copyright (C) 1999 Brian Paul All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/*
+ * Sparc assembly code by David S. Miller
+ */
+
+
+#ifndef SPARC_H
+#define SPARC_H
+
+extern void _mesa_init_all_sparc_transform_asm(void);
+
+#endif /* !(SPARC_H) */
diff --git a/src/mesa/sparc/sparc_matrix.h b/src/mesa/sparc/sparc_matrix.h
new file mode 100644
index 0000000..4b452fd
--- /dev/null
+++ b/src/mesa/sparc/sparc_matrix.h
@@ -0,0 +1,277 @@
+/* $Id: sparc_matrix.h,v 1.1 2001/05/23 14:27:03 brianp Exp $ */
+
+#define M0 %f16
+#define M1 %f17
+#define M2 %f18
+#define M3 %f19
+#define M4 %f20
+#define M5 %f21
+#define M6 %f22
+#define M7 %f23
+#define M8 %f24
+#define M9 %f25
+#define M10 %f26
+#define M11 %f27
+#define M12 %f28
+#define M13 %f29
+#define M14 %f30
+#define M15 %f31
+
+/* Seems to work, disable if unaligned traps begin to appear... -DaveM */
+#define USE_LD_DOUBLE
+
+#ifndef USE_LD_DOUBLE
+
+#define LDMATRIX_0_1_2_3_12_13_14_15(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 1 * 0x4)], M1; \
+ ld [BASE + ( 2 * 0x4)], M2; \
+ ld [BASE + ( 3 * 0x4)], M3; \
+ ld [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (13 * 0x4)], M13; \
+ ld [BASE + (14 * 0x4)], M14; \
+ ld [BASE + (15 * 0x4)], M15
+
+#define LDMATRIX_0_1_12_13(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 1 * 0x4)], M1; \
+ ld [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (13 * 0x4)], M13
+
+#define LDMATRIX_0_12_13(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (13 * 0x4)], M13
+
+#define LDMATRIX_0_1_2_12_13_14(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 1 * 0x4)], M1; \
+ ld [BASE + ( 2 * 0x4)], M2; \
+ ld [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (13 * 0x4)], M13; \
+ ld [BASE + (14 * 0x4)], M14
+
+#define LDMATRIX_0_12_13_14(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (13 * 0x4)], M13; \
+ ld [BASE + (14 * 0x4)], M14
+
+#define LDMATRIX_0_14(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + (14 * 0x4)], M14
+
+#define LDMATRIX_0_1_2_3_4_5_6_7_12_13_14_15(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 1 * 0x4)], M1; \
+ ld [BASE + ( 2 * 0x4)], M2; \
+ ld [BASE + ( 3 * 0x4)], M3; \
+ ld [BASE + ( 4 * 0x4)], M4; \
+ ld [BASE + ( 5 * 0x4)], M5; \
+ ld [BASE + ( 6 * 0x4)], M6; \
+ ld [BASE + ( 7 * 0x4)], M7; \
+ ld [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (13 * 0x4)], M13; \
+ ld [BASE + (14 * 0x4)], M14; \
+ ld [BASE + (15 * 0x4)], M15
+
+#define LDMATRIX_0_1_4_5_12_13(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 1 * 0x4)], M1; \
+ ld [BASE + ( 4 * 0x4)], M4; \
+ ld [BASE + ( 5 * 0x4)], M5; \
+ ld [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (13 * 0x4)], M13
+
+#define LDMATRIX_0_5_12_13(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 5 * 0x4)], M5; \
+ ld [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (13 * 0x4)], M13
+
+#define LDMATRIX_0_1_2_3_4_5_6_12_13_14(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 1 * 0x4)], M1; \
+ ld [BASE + ( 2 * 0x4)], M2; \
+ ld [BASE + ( 3 * 0x4)], M3; \
+ ld [BASE + ( 4 * 0x4)], M4; \
+ ld [BASE + ( 5 * 0x4)], M5; \
+ ld [BASE + ( 6 * 0x4)], M6; \
+ ld [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (13 * 0x4)], M13; \
+ ld [BASE + (14 * 0x4)], M14
+
+#define LDMATRIX_0_5_12_13_14(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 5 * 0x4)], M5; \
+ ld [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (13 * 0x4)], M13; \
+ ld [BASE + (14 * 0x4)], M14
+
+#define LDMATRIX_0_5_14(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 5 * 0x4)], M5; \
+ ld [BASE + (14 * 0x4)], M14
+
+#define LDMATRIX_0_1_2_3_4_5_6_7_8_9_10_11_12_13_14_15(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 1 * 0x4)], M1; \
+ ld [BASE + ( 2 * 0x4)], M2; \
+ ld [BASE + ( 3 * 0x4)], M3; \
+ ld [BASE + ( 4 * 0x4)], M4; \
+ ld [BASE + ( 5 * 0x4)], M5; \
+ ld [BASE + ( 6 * 0x4)], M6; \
+ ld [BASE + ( 7 * 0x4)], M7; \
+ ld [BASE + ( 8 * 0x4)], M8; \
+ ld [BASE + ( 9 * 0x4)], M9; \
+ ld [BASE + (10 * 0x4)], M10; \
+ ld [BASE + (11 * 0x4)], M11; \
+ ld [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (13 * 0x4)], M13; \
+ ld [BASE + (14 * 0x4)], M14; \
+ ld [BASE + (15 * 0x4)], M15
+
+#define LDMATRIX_0_5_12_13(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 5 * 0x4)], M5; \
+ ld [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (13 * 0x4)], M13
+
+#define LDMATRIX_0_1_2_4_5_6_8_9_10_12_13_14(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 1 * 0x4)], M1; \
+ ld [BASE + ( 2 * 0x4)], M2; \
+ ld [BASE + ( 4 * 0x4)], M4; \
+ ld [BASE + ( 5 * 0x4)], M5; \
+ ld [BASE + ( 6 * 0x4)], M6; \
+ ld [BASE + ( 8 * 0x4)], M8; \
+ ld [BASE + ( 9 * 0x4)], M9; \
+ ld [BASE + (10 * 0x4)], M10; \
+ ld [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (13 * 0x4)], M13; \
+ ld [BASE + (14 * 0x4)], M14
+
+#define LDMATRIX_0_5_10_12_13_14(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 5 * 0x4)], M5; \
+ ld [BASE + (10 * 0x4)], M10; \
+ ld [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (13 * 0x4)], M13; \
+ ld [BASE + (14 * 0x4)], M14
+
+#define LDMATRIX_0_5_8_9_10_14(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 5 * 0x4)], M5; \
+ ld [BASE + ( 8 * 0x4)], M8; \
+ ld [BASE + ( 9 * 0x4)], M9; \
+ ld [BASE + (10 * 0x4)], M10; \
+ ld [BASE + (14 * 0x4)], M14
+
+#else /* !(USE_LD_DOUBLE) */
+
+#define LDMATRIX_0_1_2_3_12_13_14_15(BASE) \
+ ldd [BASE + ( 0 * 0x4)], M0; \
+ ldd [BASE + ( 2 * 0x4)], M2; \
+ ldd [BASE + (12 * 0x4)], M12; \
+ ldd [BASE + (14 * 0x4)], M14
+
+#define LDMATRIX_0_1_12_13(BASE) \
+ ldd [BASE + ( 0 * 0x4)], M0; \
+ ldd [BASE + (12 * 0x4)], M12
+
+#define LDMATRIX_0_12_13(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ldd [BASE + (12 * 0x4)], M12
+
+#define LDMATRIX_0_1_2_12_13_14(BASE) \
+ ldd [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 2 * 0x4)], M2; \
+ ldd [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (14 * 0x4)], M14
+
+#define LDMATRIX_0_12_13_14(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ldd [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (14 * 0x4)], M14
+
+#define LDMATRIX_0_14(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + (14 * 0x4)], M14
+
+#define LDMATRIX_0_1_2_3_4_5_6_7_12_13_14_15(BASE) \
+ ldd [BASE + ( 0 * 0x4)], M0; \
+ ldd [BASE + ( 2 * 0x4)], M2; \
+ ldd [BASE + ( 4 * 0x4)], M4; \
+ ldd [BASE + ( 6 * 0x4)], M6; \
+ ldd [BASE + (12 * 0x4)], M12; \
+ ldd [BASE + (14 * 0x4)], M14
+
+#define LDMATRIX_0_5_12_13(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 5 * 0x4)], M5; \
+ ldd [BASE + (12 * 0x4)], M12
+
+#define LDMATRIX_0_1_2_3_4_5_6_12_13_14(BASE) \
+ ldd [BASE + ( 0 * 0x4)], M0; \
+ ldd [BASE + ( 2 * 0x4)], M2; \
+ ldd [BASE + ( 4 * 0x4)], M4; \
+ ld [BASE + ( 6 * 0x4)], M6; \
+ ldd [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (14 * 0x4)], M14
+
+#define LDMATRIX_0_5_12_13_14(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 5 * 0x4)], M5; \
+ ldd [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (14 * 0x4)], M14
+
+#define LDMATRIX_0_5_14(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 5 * 0x4)], M5; \
+ ld [BASE + (14 * 0x4)], M14
+
+#define LDMATRIX_0_1_2_3_4_5_6_7_8_9_10_11_12_13_14_15(BASE) \
+ ldd [BASE + ( 0 * 0x4)], M0; \
+ ldd [BASE + ( 2 * 0x4)], M2; \
+ ldd [BASE + ( 4 * 0x4)], M4; \
+ ldd [BASE + ( 6 * 0x4)], M6; \
+ ldd [BASE + ( 8 * 0x4)], M8; \
+ ldd [BASE + (10 * 0x4)], M10; \
+ ldd [BASE + (12 * 0x4)], M12; \
+ ldd [BASE + (14 * 0x4)], M14
+
+#define LDMATRIX_0_1_4_5_12_13(BASE) \
+ ldd [BASE + ( 0 * 0x4)], M0; \
+ ldd [BASE + ( 4 * 0x4)], M4; \
+ ldd [BASE + (12 * 0x4)], M12
+
+#define LDMATRIX_0_5_12_13(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 5 * 0x4)], M5; \
+ ldd [BASE + (12 * 0x4)], M12
+
+#define LDMATRIX_0_1_2_4_5_6_8_9_10_12_13_14(BASE) \
+ ldd [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 2 * 0x4)], M2; \
+ ldd [BASE + ( 4 * 0x4)], M4; \
+ ld [BASE + ( 6 * 0x4)], M6; \
+ ldd [BASE + ( 8 * 0x4)], M8; \
+ ld [BASE + (10 * 0x4)], M10; \
+ ldd [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (14 * 0x4)], M14
+
+#define LDMATRIX_0_5_10_12_13_14(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 5 * 0x4)], M5; \
+ ld [BASE + (10 * 0x4)], M10; \
+ ldd [BASE + (12 * 0x4)], M12; \
+ ld [BASE + (14 * 0x4)], M14
+
+#define LDMATRIX_0_5_8_9_10_14(BASE) \
+ ld [BASE + ( 0 * 0x4)], M0; \
+ ld [BASE + ( 5 * 0x4)], M5; \
+ ldd [BASE + ( 8 * 0x4)], M8; \
+ ld [BASE + (10 * 0x4)], M10; \
+ ld [BASE + (14 * 0x4)], M14
+
+#endif /* USE_LD_DOUBLE */
diff --git a/src/mesa/sparc/xform.S b/src/mesa/sparc/xform.S
new file mode 100644
index 0000000..368fdd9
--- /dev/null
+++ b/src/mesa/sparc/xform.S
@@ -0,0 +1,1410 @@
+/* $Id: xform.S,v 1.1 2001/05/23 14:27:03 brianp Exp $ */
+
+ /* TODO
+ *
+ * 1) It would be nice if load/store double could be used
+ * at least for the matrix parts. I think for the matrices
+ * it is safe, but for the vertices it probably is not due to
+ * things like glInterleavedArrays etc.
+ *
+ * UPDATE: Trying this now in sparc_matrix.h -DaveM_990624
+ *
+ * 2) One extremely slick trick would be if we could enclose
+ * groups of xform calls on the same vertices such that
+ * we just load the matrix into f16-->f31 before the calls
+ * and then we would not have to do them here. This may be
+ * tricky and not much of a gain though.
+ */
+
+#ifdef __sparc_v9__
+#define LDPTR ldx
+#define V4F_DATA 0x00
+#define V4F_START 0x08
+#define V4F_COUNT 0x10
+#define V4F_STRIDE 0x14
+#define V4F_SIZE 0x18
+#define V4F_FLAGS 0x1c
+#else
+#define LDPTR ld
+#define V4F_DATA 0x00
+#define V4F_START 0x04
+#define V4F_COUNT 0x08
+#define V4F_STRIDE 0x0c
+#define V4F_SIZE 0x10
+#define V4F_FLAGS 0x14
+#endif
+
+#define VEC_SIZE_1 1
+#define VEC_SIZE_2 3
+#define VEC_SIZE_3 7
+#define VEC_SIZE_4 15
+
+ .text
+ .align 64
+
+__set_v4f_1:
+ ld [%o0 + V4F_FLAGS], %g2
+ mov 1, %g1
+ st %g1, [%o0 + V4F_SIZE]
+ or %g2, VEC_SIZE_1, %g2
+ retl
+ st %g2, [%o0 + V4F_FLAGS]
+__set_v4f_2:
+ ld [%o0 + V4F_FLAGS], %g2
+ mov 2, %g1
+ st %g1, [%o0 + V4F_SIZE]
+ or %g2, VEC_SIZE_2, %g2
+ retl
+ st %g2, [%o0 + V4F_FLAGS]
+__set_v4f_3:
+ ld [%o0 + V4F_FLAGS], %g2
+ mov 3, %g1
+ st %g1, [%o0 + V4F_SIZE]
+ or %g2, VEC_SIZE_3, %g2
+ retl
+ st %g2, [%o0 + V4F_FLAGS]
+__set_v4f_4:
+ ld [%o0 + V4F_FLAGS], %g2
+ mov 4, %g1
+ st %g1, [%o0 + V4F_SIZE]
+ or %g2, VEC_SIZE_4, %g2
+ retl
+ st %g2, [%o0 + V4F_FLAGS]
+
+#include "sparc_matrix.h"
+
+ /* First the raw versions. */
+
+ .globl _mesa_sparc_transform_points1_general
+_mesa_sparc_transform_points1_general:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_1_2_3_12_13_14_15(%o1)
+
+ cmp %g3, 1
+ st %g3, [%o0 + V4F_COUNT]
+ bl 3f
+ clr %o1
+
+ be 2f
+ andn %g3, 1, %o2
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ add %g1, %o5, %g1 ! IEU0
+ ld [%g1 + 0x00], %f8 ! LSU Group
+ add %o1, 2, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ fmuls %f0, M0, %f1 ! FGM Group 1-cycle stall on %f0
+ fmuls %f0, M1, %f2 ! FGM Group
+ fmuls %f0, M2, %f3 ! FGM Group
+ fmuls %f0, M3, %f4 ! FGM Group
+ fmuls %f8, M0, %f9 ! FGM Group f1 available
+ fadds %f1, M12, %f1 ! FGA
+ st %f1, [%g2 + 0x00] ! LSU
+ fmuls %f8, M1, %f10 ! FGM Group f2 available
+ fadds %f2, M13, %f2 ! FGA
+ st %f2, [%g2 + 0x04] ! LSU
+ fmuls %f8, M2, %f11 ! FGM Group f3 available
+ fadds %f3, M14, %f3 ! FGA
+ st %f3, [%g2 + 0x08] ! LSU
+ fmuls %f8, M3, %f12 ! FGM Group f4 available
+ fadds %f4, M15, %f4 ! FGA
+ st %f4, [%g2 + 0x0c] ! LSU
+ fadds %f9, M12, %f9 ! FGA Group f9 available
+ st %f9, [%g2 + 0x10] ! LSU
+ fadds %f10, M13, %f10 ! FGA Group f10 available
+ st %f10, [%g2 + 0x14] ! LSU
+ fadds %f11, M14, %f11 ! FGA Group f11 available
+ st %f11, [%g2 + 0x18] ! LSU
+ fadds %f12, M15, %f12 ! FGA Group f12 available
+ st %f12, [%g2 + 0x1c] ! LSU
+ cmp %o1, %o2 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x20, %g2 ! IEU0 Group
+
+ cmp %o1, %g3
+ be 3f
+ nop
+
+2: ld [%g1 + 0x00], %f0 ! LSU Group
+ fmuls %f0, M0, %f1 ! FGM Group 1-cycle stall on %f0
+ fmuls %f0, M1, %f2 ! FGM Group
+ fmuls %f0, M2, %f3 ! FGM Group
+ fmuls %f0, M3, %f4 ! FGM Group
+ fadds %f1, M12, %f1 ! FGA Group
+ st %f1, [%g2 + 0x00] ! LSU
+ fadds %f2, M13, %f2 ! FGA Group
+ st %f2, [%g2 + 0x04] ! LSU
+ fadds %f3, M14, %f3 ! FGA Group
+ st %f3, [%g2 + 0x08] ! LSU
+ fadds %f4, M15, %f4 ! FGA Group
+ st %f4, [%g2 + 0x0c] ! LSU
+
+3:
+ ba __set_v4f_4
+ nop
+
+ .globl _mesa_sparc_transform_points1_identity
+_mesa_sparc_transform_points1_identity:
+ cmp %o0, %o2
+ be 4f
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ cmp %g3, 1
+ st %g3, [%o0 + V4F_COUNT]
+ bl 3f
+ clr %o1
+
+ be 2f
+ andn %g3, 1, %o2
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ add %g1, %o5, %g1 ! IEU0
+ ld [%g1 + 0x00], %f1 ! LSU Group
+ add %o1, 2, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ st %f0, [%g2 + 0x00] ! LSU Group
+ cmp %o1, %o2 ! IEU1
+ st %f1, [%g2 + 0x10] ! LSU Group
+ bne 1b ! CTI
+ add %g2, 0x20, %g2 ! IEU0
+
+ cmp %o1, %g3
+ be 3f
+ nop
+
+2: ld [%g1 + 0x00], %f0
+ addx %g0, %g0, %g0
+ st %f0, [%g2 + 0x00]
+
+3:
+ ba __set_v4f_1
+ nop
+
+4: retl
+ nop
+
+ .globl _mesa_sparc_transform_points1_2d
+_mesa_sparc_transform_points1_2d:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_1_12_13(%o1)
+
+ cmp %g3, 1
+ st %g3, [%o0 + V4F_COUNT]
+ bl 3f
+ clr %o1
+
+ be 2f
+ andn %g3, 1, %o2
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ add %g1, %o5, %g1 ! IEU0
+ ld [%g1 + 0x00], %f8 ! LSU Group
+ add %o1, 2, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ fmuls %f0, M0, %f1 ! FGM Group
+ fmuls %f0, M1, %f2 ! FGM Group
+ fmuls %f8, M0, %f9 ! FGM Group
+ fmuls %f8, M1, %f10 ! FGM Group
+ fadds %f1, M12, %f3 ! FGA Group f1 available
+ st %f3, [%g2 + 0x00] ! LSU
+ fadds %f2, M13, %f4 ! FGA Group f2 available
+ st %f4, [%g2 + 0x04] ! LSU
+ fadds %f9, M12, %f11 ! FGA Group f9 available
+ st %f11, [%g2 + 0x10] ! LSU
+ fadds %f10, M13, %f12 ! FGA Group f10 available
+ st %f12, [%g2 + 0x14] ! LSU
+ cmp %o1, %o2 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x20, %g2 ! IEU0 Group
+
+ cmp %o1, %g3
+ be 3f
+ nop
+
+2: ld [%g1 + 0x00], %f0
+ fmuls %f0, M0, %f1
+ fmuls %f0, M1, %f2
+ fadds %f1, M12, %f3
+ st %f3, [%g2 + 0x00]
+ fadds %f2, M13, %f4
+ st %f4, [%g2 + 0x04]
+
+3:
+ ba __set_v4f_2
+ nop
+
+ .globl _mesa_sparc_transform_points1_2d_no_rot
+_mesa_sparc_transform_points1_2d_no_rot:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_12_13(%o1)
+
+ cmp %g3, 1
+ st %g3, [%o0 + V4F_COUNT]
+ bl 3f
+ clr %o1
+
+ be 2f
+ andn %g3, 1, %o2
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ add %g1, %o5, %g1 ! IEU0
+ ld [%g1 + 0x00], %f4 ! LSU Group
+ add %o1, 2, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ fmuls %f0, M0, %f1 ! FGM Group
+ fmuls %f4, M0, %f5 ! FGM Group
+ fadds %f1, M12, %f3 ! FGA Group, 2 cycle stall, f1 available
+ st %f3, [%g2 + 0x00] ! LSU
+ st M13, [%g2 + 0x04] ! LSU Group, f5 available
+ fadds %f5, M12, %f6 ! FGA
+ st %f6, [%g2 + 0x10] ! LSU Group
+ st M13, [%g2 + 0x14] ! LSU Group
+ cmp %o1, %o2 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x20, %g2 ! IEU0 Group
+
+ cmp %o1, %g3
+ be 3f
+ nop
+
+2: ld [%g1 + 0x00], %f0
+ fmuls %f0, M0, %f1
+ fadds %f1, M12, %f3
+ st %f3, [%g2 + 0x00]
+ st M13, [%g2 + 0x04]
+
+3:
+ ba __set_v4f_2
+ nop
+
+ .globl _mesa_sparc_transform_points1_3d
+_mesa_sparc_transform_points1_3d:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_1_2_12_13_14(%o1)
+
+ cmp %g3, 1
+ st %g3, [%o0 + V4F_COUNT]
+ bl 3f
+ clr %o1
+
+ be 2f
+ andn %g3, 1, %o2
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ add %g1, %o5, %g1 ! IEU0
+ ld [%g1 + 0x00], %f4 ! LSU Group
+ add %o1, 2, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ fmuls %f0, M0, %f1 ! FGM Group
+ fmuls %f0, M1, %f2 ! FGM Group
+ fmuls %f0, M2, %f3 ! FGM Group
+ fmuls %f4, M0, %f5 ! FGM Group
+ fadds %f1, M12, %f1 ! FGA Group, f1 available
+ st %f1, [%g2 + 0x00] ! LSU
+ fmuls %f4, M1, %f6 ! FGM
+ fadds %f2, M13, %f2 ! FGA Group, f2 available
+ st %f2, [%g2 + 0x04] ! LSU
+ fmuls %f4, M2, %f7 ! FGM
+ fadds %f3, M14, %f3 ! FGA Group, f3 available
+ st %f3, [%g2 + 0x08] ! LSU
+ fadds %f5, M12, %f5 ! FGA Group, f5 available
+ st %f5, [%g2 + 0x10] ! LSU
+ fadds %f6, M13, %f6 ! FGA Group, f6 available
+ st %f6, [%g2 + 0x14] ! LSU
+ fadds %f7, M14, %f7 ! FGA Group, f7 available
+ st %f7, [%g2 + 0x18] ! LSU
+ cmp %o1, %o2 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x20, %g2 ! IEU0 Group
+
+ cmp %o1, %g3
+ be 3f
+ nop
+
+2: ld [%g1 + 0x00], %f0
+ fmuls %f0, M0, %f1
+ fmuls %f0, M1, %f2
+ fmuls %f0, M2, %f3
+ fadds %f1, M12, %f1
+ st %f1, [%g2 + 0x00]
+ fadds %f2, M13, %f2
+ st %f2, [%g2 + 0x04]
+ fadds %f3, M14, %f3
+ st %f3, [%g2 + 0x08]
+
+3:
+ ba __set_v4f_3
+ nop
+
+ .globl _mesa_sparc_transform_points1_3d_no_rot
+_mesa_sparc_transform_points1_3d_no_rot:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_12_13_14(%o1)
+
+ cmp %g3, 1
+ st %g3, [%o0 + V4F_COUNT]
+ bl 3f
+ clr %o1
+
+ be 2f
+ andn %g3, 1, %o2
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ add %g1, %o5, %g1 ! IEU0
+ ld [%g1 + 0x00], %f2 ! LSU Group
+ add %o1, 2, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ fmuls %f0, M0, %f1 ! FGM Group
+ fmuls %f2, M0, %f3 ! FGM Group
+ fadds %f1, M12, %f1 ! FGA Group, 2 cycle stall, f1 available
+ st %f1, [%g2 + 0x00] ! LSU
+ fadds %f3, M12, %f3 ! FGA Group, f3 available
+ st M13, [%g2 + 0x04] ! LSU
+ st M14, [%g2 + 0x08] ! LSU Group
+ st %f3, [%g2 + 0x10] ! LSU Group
+ st M13, [%g2 + 0x14] ! LSU Group
+ st M14, [%g2 + 0x18] ! LSU Group
+ cmp %o1, %o2 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x20, %g2 ! IEU0 Group
+
+ cmp %o1, %g3
+ be 3f
+ nop
+
+2: ld [%g1 + 0x00], %f0
+ fmuls %f0, M0, %f1
+ fadds %f1, M12, %f1
+ st %f1, [%g2 + 0x00]
+ st M13, [%g2 + 0x04]
+ st M14, [%g2 + 0x08]
+
+3:
+ ba __set_v4f_3
+ nop
+
+ .globl _mesa_sparc_transform_points1_perspective
+_mesa_sparc_transform_points1_perspective:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_14(%o1)
+
+ cmp %g3, 1
+ st %g3, [%o0 + V4F_COUNT]
+ bl 3f
+ clr %o1
+
+ be 2f
+ andn %g3, 1, %o2
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ add %g1, %o5, %g1 ! IEU0
+ ld [%g1 + 0x00], %f2 ! LSU Group
+ add %o1, 2, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ fmuls %f0, M0, %f1 ! FGM Group
+ st %f1, [%g2 + 0x00] ! LSU
+ fmuls %f2, M0, %f3 ! FGM Group
+ st %g0, [%g2 + 0x04] ! LSU
+ st M14, [%g2 + 0x08] ! LSU Group
+ st %g0, [%g2 + 0x0c] ! LSU Group
+ st %f3, [%g2 + 0x10] ! LSU Group
+ st %g0, [%g2 + 0x14] ! LSU Group
+ st M14, [%g2 + 0x18] ! LSU Group
+ st %g0, [%g2 + 0x1c] ! LSU Group
+ cmp %o1, %o2 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x20, %g2 ! IEU0 Group
+
+ cmp %o1, %g3
+ be 3f
+ nop
+
+2: ld [%g1 + 0x00], %f0
+ fmuls %f0, M0, %f1
+ st %f1, [%g2 + 0x00]
+ st %g0, [%g2 + 0x04]
+ st M14, [%g2 + 0x08]
+ st %g0, [%g2 + 0x0c]
+
+3:
+ ba __set_v4f_4
+ nop
+
+ .globl _mesa_sparc_transform_points2_general
+_mesa_sparc_transform_points2_general:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_1_2_3_4_5_6_7_12_13_14_15(%o1)
+
+ cmp %g3, 0
+ st %g3, [%o0 + V4F_COUNT]
+ be 2f
+ clr %o1
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ add %o1, 1, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ fmuls %f0, M0, %f2 ! FGM Group
+ fmuls %f0, M1, %f3 ! FGM Group
+ fmuls %f0, M2, %f4 ! FGM Group
+ fmuls %f0, M3, %f5 ! FGM Group
+ fadds %f2, M12, %f2 ! FGA Group f2 available
+ fmuls %f1, M4, %f6 ! FGM
+ fadds %f3, M13, %f3 ! FGA Group f3 available
+ fmuls %f1, M5, %f7 ! FGM
+ fadds %f4, M14, %f4 ! FGA Group f4 available
+ fmuls %f1, M6, %f8 ! FGM
+ fadds %f5, M15, %f5 ! FGA Group f5 available
+ fmuls %f1, M7, %f9 ! FGM
+ fadds %f2, %f6, %f2 ! FGA Group f6 available
+ st %f2, [%g2 + 0x00] ! LSU
+ fadds %f3, %f7, %f3 ! FGA Group f7 available
+ st %f3, [%g2 + 0x04] ! LSU
+ fadds %f4, %f8, %f4 ! FGA Group f8 available
+ st %f4, [%g2 + 0x08] ! LSU
+ fadds %f5, %f9, %f5 ! FGA Group f9 available
+ st %f5, [%g2 + 0x0c] ! LSU
+ cmp %o1, %g3 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x10, %g2 ! IEU0 Group
+2:
+ ba __set_v4f_4
+ nop
+
+ .globl _mesa_sparc_transform_points2_identity
+_mesa_sparc_transform_points2_identity:
+ cmp %o2, %o0
+ be 3f
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ cmp %g3, 0
+ st %g3, [%o0 + V4F_COUNT]
+ be 2f
+ clr %o1
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ add %o1, 1, %o1 ! IEU0
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ add %g1, %o5, %g1 ! IEU0
+ cmp %o1, %g3 ! IEU1
+ st %f0, [%g2 + 0x00] ! LSU Group
+ st %f1, [%g2 + 0x04] ! LSU Group
+ bne 1b ! CTI
+ add %g2, 0x10, %g2 ! IEU0
+2:
+ ba __set_v4f_2
+ nop
+
+3: retl
+ nop
+
+ .globl _mesa_sparc_transform_points2_2d
+_mesa_sparc_transform_points2_2d:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_1_4_5_12_13(%o1)
+
+ cmp %g3, 1
+ st %g3, [%o0 + V4F_COUNT]
+ bl 3f
+ clr %o1
+
+ be 2f
+ andn %g3, 1, %o2
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ add %o1, 2, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ fmuls %f0, M0, %f2 ! FGM
+ ld [%g1 + 0x00], %f8 ! LSU Group
+ fmuls %f0, M1, %f3 ! FGM
+ ld [%g1 + 0x04], %f9 ! LSU Group
+ fmuls %f1, M4, %f6 ! FGM
+ fmuls %f1, M5, %f7 ! FGM Group
+ add %g1, %o5, %g1 ! IEU0
+ fmuls %f8, M0, %f10 ! FGM Group f2 available
+ fadds %f2, M12, %f2 ! FGA
+ fmuls %f8, M1, %f11 ! FGM Group f3 available
+ fadds %f3, M13, %f3 ! FGA
+ fmuls %f9, M4, %f12 ! FGM Group
+ fmuls %f9, M5, %f13 ! FGM Group
+ fadds %f10, M12, %f10 ! FGA Group f2, f10 available
+ fadds %f2, %f6, %f2 ! FGA Group f3, f11 available
+ st %f2, [%g2 + 0x00] ! LSU
+ fadds %f11, M13, %f11 ! FGA Group f12 available
+ fadds %f3, %f7, %f3 ! FGA Group f13 available
+ st %f3, [%g2 + 0x04] ! LSU
+ fadds %f10, %f12, %f10 ! FGA Group f10 available
+ st %f10, [%g2 + 0x10] ! LSU
+ fadds %f11, %f13, %f11 ! FGA Group f11 available
+ st %f11, [%g2 + 0x14] ! LSU
+ cmp %o1, %o2 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x20, %g2 ! IEU0 Group
+
+ cmp %o1, %g3
+ be 3f
+ nop
+
+2: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ fmuls %f0, M0, %f2 ! FGM Group
+ fmuls %f0, M1, %f3 ! FGM Group
+ fmuls %f1, M4, %f6 ! FGM Group
+ fmuls %f1, M5, %f7 ! FGM Group
+ fadds %f2, M12, %f2 ! FGA Group f2 available
+ fadds %f3, M13, %f3 ! FGA Group f3 available
+ fadds %f2, %f6, %f2 ! FGA Group 2 cycle stall, f2 available
+ st %f2, [%g2 + 0x00] ! LSU
+ fadds %f3, %f7, %f3 ! FGA Group f3 available
+ st %f3, [%g2 + 0x04] ! LSU
+
+3:
+ ba __set_v4f_2
+ nop
+
+ .globl _mesa_sparc_transform_points2_2d_no_rot
+_mesa_sparc_transform_points2_2d_no_rot:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_5_12_13(%o1)
+
+ cmp %g3, 1
+ st %g3, [%o0 + V4F_COUNT]
+ bl 3f
+ clr %o1
+
+ be 2f
+ andn %g3, 1, %o2
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ add %o1, 2, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ ld [%g1 + 0x00], %f4 ! LSU Group
+ fmuls %f0, M0, %f2 ! FGM
+ ld [%g1 + 0x04], %f5 ! LSU Group
+ fmuls %f1, M5, %f3 ! FGM
+ fmuls %f4, M0, %f6 ! FGM Group
+ add %g1, %o5, %g1 ! IEU0
+ fmuls %f5, M5, %f7 ! FGM Group
+ fadds %f2, M12, %f2 ! FGA Group f2 available
+ st %f2, [%g2 + 0x00] ! LSU
+ fadds %f3, M13, %f3 ! FGA Group f3 available
+ st %f3, [%g2 + 0x04] ! LSU
+ fadds %f6, M12, %f6 ! FGA Group f6 available
+ st %f6, [%g2 + 0x10] ! LSU
+ fadds %f7, M13, %f7 ! FGA Group f7 available
+ st %f7, [%g2 + 0x14] ! LSU
+ cmp %o1, %o2 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x20, %g2 ! IEU0 Group
+
+ cmp %o1, %g3
+ be 3f
+ nop
+
+2: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ fmuls %f0, M0, %f2 ! FGM Group
+ fmuls %f1, M5, %f3 ! FGM Group
+ fadds %f2, M12, %f2 ! FGA Group, 2 cycle stall, f2 available
+ st %f2, [%g2 + 0x00] ! LSU
+ fadds %f3, M13, %f3 ! FGA Group f3 available
+ st %f3, [%g2 + 0x04] ! LSU
+
+3:
+ ba __set_v4f_2
+ nop
+
+ /* orig: 12 cycles */
+ .globl _mesa_sparc_transform_points2_3d
+_mesa_sparc_transform_points2_3d:
+ ld [%o2 + V4F_STRIDE], %o5
+ ld [%o2 + V4F_START], %g1
+ ld [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_1_2_3_4_5_6_12_13_14(%o1)
+
+ cmp %g3, 1
+ st %g3, [%o0 + V4F_COUNT]
+ bl 3f
+ clr %o1
+
+ be 2f
+ andn %g3, 1, %o2
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ add %o1, 2, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ ld [%g1 + 0x00], %f9 ! LSU Group
+ fmuls %f0, M0, %f2 ! FGM
+ ld [%g1 + 0x04], %f10 ! LSU Group
+ fmuls %f0, M1, %f3 ! FGM
+ fmuls %f0, M2, %f4 ! FGM Group
+ add %g1, %o5, %g1 ! IEU0
+ fmuls %f1, M4, %f6 ! FGM Group
+ fmuls %f1, M5, %f7 ! FGM Group f2 available
+ fadds %f2, M12, %f2 ! FGA
+ fmuls %f1, M6, %f8 ! FGM Group f3 available
+ fadds %f3, M13, %f3 ! FGA
+ fmuls %f9, M0, %f11 ! FGM Group f4 available
+ fadds %f4, M14, %f4 ! FGA
+ fmuls %f9, M1, %f12 ! FGM Group f6 available
+ fmuls %f9, M2, %f13 ! FGM Group f2, f7 available
+ fadds %f2, %f6, %f2 ! FGA
+ st %f2, [%g2 + 0x00] ! LSU
+ fmuls %f10, M4, %f14 ! FGM Group f3, f8 available
+ fadds %f3, %f7, %f3 ! FGA
+ st %f3, [%g2 + 0x04] ! LSU
+ fmuls %f10, M5, %f15 ! FGM Group f4, f11 available
+ fadds %f11, M12, %f11 ! FGA
+ fmuls %f10, M6, %f0 ! FGM Group f12 available
+ fadds %f12, M13, %f12 ! FGA
+ fadds %f13, M14, %f13 ! FGA Group f13 available
+ fadds %f4, %f8, %f4 ! FGA Group f14 available
+ st %f4, [%g2 + 0x08] ! LSU
+ fadds %f11, %f14, %f11 ! FGA Group f15, f11 available
+ st %f11, [%g2 + 0x10] ! LSU
+ fadds %f12, %f15, %f12 ! FGA Group f0, f12 available
+ st %f12, [%g2 + 0x14] ! LSU
+ fadds %f13, %f0, %f13 ! FGA Group f13 available
+ st %f13, [%g2 + 0x18] ! LSU
+
+ cmp %o1, %o2 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x20, %g2 ! IEU0 Group
+
+ cmp %o1, %g3
+ be 3f
+ nop
+
+2: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ fmuls %f0, M0, %f2 ! FGM Group
+ fmuls %f0, M1, %f3 ! FGM Group
+ fmuls %f0, M2, %f4 ! FGM Group
+ fmuls %f1, M4, %f6 ! FGM Group
+ fmuls %f1, M5, %f7 ! FGM Group f2 available
+ fadds %f2, M12, %f2 ! FGA
+ fmuls %f1, M6, %f8 ! FGM Group f3 available
+ fadds %f3, M13, %f3 ! FGA
+ fadds %f4, M14, %f4 ! FGA Group f4 available
+ fadds %f2, %f6, %f2 ! FGA Group stall, f2, f6, f7 available
+ st %f2, [%g2 + 0x00] ! LSU
+ fadds %f3, %f7, %f3 ! FGA Group f3, f8 available
+ st %f3, [%g2 + 0x04] ! LSU
+ fadds %f4, %f8, %f4 ! FGA Group f4 available
+ st %f4, [%g2 + 0x08] ! LSU
+
+3:
+ ba __set_v4f_3
+ nop
+
+ .globl _mesa_sparc_transform_points2_3d_no_rot
+_mesa_sparc_transform_points2_3d_no_rot:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_5_12_13_14(%o1)
+
+ cmp %g3, 1
+ st %g3, [%o0 + V4F_COUNT]
+ bl 3f
+ clr %o3
+
+ be 2f
+ andn %g3, 1, %o2
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ add %o3, 2, %o3 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ ld [%g1 + 0x00], %f4 ! LSU Group
+ fmuls %f0, M0, %f2 ! FGM
+ ld [%g1 + 0x04], %f5 ! LSU Group
+ fmuls %f1, M5, %f3 ! FGM
+ fmuls %f4, M0, %f6 ! FGM Group
+ add %g1, %o5, %g1 ! IEU0
+ fmuls %f5, M5, %f7 ! FGM Group
+ fadds %f2, M12, %f2 ! FGA Group f2 available
+ st %f2, [%g2 + 0x00] ! LSU
+ fadds %f3, M13, %f3 ! FGA Group f3 available
+ st %f3, [%g2 + 0x04] ! LSU
+ fadds %f6, M12, %f6 ! FGA Group f6 available
+ st M14, [%g2 + 0x08] ! LSU
+ fadds %f7, M13, %f7 ! FGA Group f7 available
+ st %f6, [%g2 + 0x10] ! LSU
+ st %f7, [%g2 + 0x14] ! LSU Group
+ st M14, [%g2 + 0x18] ! LSU Group
+ cmp %o3, %o2 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x20, %g2 ! IEU0 Group
+
+ cmp %o3, %g3
+ be 3f
+ nop
+
+2: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ fmuls %f0, M0, %f2 ! FGM Group
+ fmuls %f1, M5, %f3 ! FGM Group
+ fadds %f2, M12, %f2 ! FGA Group, 2 cycle stall, f2 available
+ st %f2, [%g2 + 0x00] ! LSU
+ fadds %f3, M13, %f3 ! FGA Group f3 available
+ st %f3, [%g2 + 0x04] ! LSU
+ st M14, [%g2 + 0x08] ! LSU Group
+
+3: ld [%o1 + (14 * 0x4)], %g3
+ cmp %g3, 0
+ bne __set_v4f_3
+ nop
+ ba __set_v4f_2
+ nop
+
+ .globl _mesa_sparc_transform_points2_perspective
+_mesa_sparc_transform_points2_perspective:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_5_14(%o1)
+
+ cmp %g3, 0
+ st %g3, [%o0 + V4F_COUNT]
+ be 2f
+ clr %o1
+
+1: ld [%g1 + 0x00], %f0
+ ld [%g1 + 0x04], %f1
+ add %o1, 1, %o1
+ add %g1, %o5, %g1
+ fmuls %f0, M0, %f2
+ st %f2, [%g2 + 0x00]
+ fmuls %f1, M5, %f3
+ st %f3, [%g2 + 0x04]
+ st M14, [%g2 + 0x08]
+ st %g0, [%g2 + 0x0c]
+ cmp %o1, %g3
+ bne 1b
+ add %g2, 0x10, %g2
+2:
+ ba __set_v4f_4
+ nop
+
+ .globl _mesa_sparc_transform_points3_general
+_mesa_sparc_transform_points3_general:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_1_2_3_4_5_6_7_8_9_10_11_12_13_14_15(%o1)
+
+ cmp %g3, 0
+ st %g3, [%o0 + V4F_COUNT]
+ be 2f
+ clr %o1
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ ld [%g1 + 0x08], %f2 ! LSU Group
+ add %o1, 1, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ fmuls %f0, M0, %f3 ! FGM
+ fmuls %f1, M4, %f7 ! FGM Group
+ fmuls %f0, M1, %f4 ! FGM Group
+ fmuls %f1, M5, %f8 ! FGM Group
+ fmuls %f0, M2, %f5 ! FGM Group f3 available
+ fmuls %f1, M6, %f9 ! FGM Group f7 available
+ fadds %f3, %f7, %f3 ! FGA
+ fmuls %f0, M3, %f6 ! FGM Group f4 available
+ fmuls %f1, M7, %f10 ! FGM Group f8 available
+ fadds %f4, %f8, %f4 ! FGA
+ fmuls %f2, M8, %f7 ! FGM Group f5 available
+ fmuls %f2, M9, %f8 ! FGM Group f9,f3 available
+ fadds %f5, %f9, %f5 ! FGA
+ fmuls %f2, M10, %f9 ! FGM Group f6 available
+ fadds %f6, %f10, %f6 ! FGA Group f10,f4 available
+ fmuls %f2, M11, %f10 ! FGM
+ fadds %f3, M12, %f3 ! FGA Group f7 available
+ fadds %f4, M13, %f4 ! FGA Group f8,f5 available
+ fadds %f5, M14, %f5 ! FGA Group f9 available
+ fadds %f6, M15, %f6 ! FGA Group f10,f6 available
+ fadds %f3, %f7, %f3 ! FGA Group f3 available
+ st %f3, [%g2 + 0x00] ! LSU
+ fadds %f4, %f8, %f4 ! FGA Group f4 available
+ st %f4, [%g2 + 0x04] ! LSU
+ fadds %f5, %f9, %f5 ! FGA Group f5 available
+ st %f5, [%g2 + 0x08] ! LSU
+ fadds %f6, %f10, %f6 ! FGA Group f6 available
+ st %f6, [%g2 + 0x0c] ! LSU
+ cmp %o1, %g3 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x10, %g2 ! IEU0 Group
+2:
+ ba __set_v4f_4
+ nop
+
+ .globl _mesa_sparc_transform_points3_identity
+_mesa_sparc_transform_points3_identity:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ cmp %g3, 0
+ st %g3, [%o0 + V4F_COUNT]
+ be 2f
+ clr %o1
+
+1: ld [%g1 + 0x00], %f0
+ ld [%g1 + 0x04], %f1
+ ld [%g1 + 0x08], %f2
+ add %o1, 1, %o1
+ add %g1, %o5, %g1
+ cmp %o1, %g3
+ st %f0, [%g2 + 0x00]
+ st %f1, [%g2 + 0x04]
+ st %f2, [%g2 + 0x08]
+ bne 1b
+ add %g2, 0x10, %g2
+2:
+ ba __set_v4f_3
+ nop
+
+ .globl _mesa_sparc_transform_points3_2d
+_mesa_sparc_transform_points3_2d:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_1_4_5_12_13(%o1)
+
+ cmp %g3, 0
+ st %g3, [%o0 + V4F_COUNT]
+ be 2f
+ clr %o1
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ ld [%g1 + 0x08], %f2 ! LSU Group
+ add %o1, 1, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ fmuls %f0, M0, %f3 ! FGM
+ fmuls %f0, M1, %f4 ! FGM Group
+ fmuls %f1, M4, %f6 ! FGM Group
+ fmuls %f1, M5, %f7 ! FGM Group
+ fadds %f3, M12, %f3 ! FGA Group f3 available
+ fadds %f4, M13, %f4 ! FGA Group f4 available
+ fadds %f3, %f6, %f3 ! FGA Group f6 available
+ st %f3, [%g2 + 0x00] ! LSU
+ fadds %f4, %f7, %f4 ! FGA Group f7 available
+ st %f4, [%g2 + 0x04] ! LSU
+ st %f2, [%g2 + 0x08] ! LSU Group
+ cmp %o1, %g3 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x10, %g2 ! IEU0 Group
+2:
+ ba __set_v4f_3
+ nop
+
+ .globl _mesa_sparc_transform_points3_2d_no_rot
+_mesa_sparc_transform_points3_2d_no_rot:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_5_12_13(%o1)
+
+ cmp %g3, 0
+ st %g3, [%o0 + V4F_COUNT]
+ be 2f
+ clr %o1
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ ld [%g1 + 0x08], %f2 ! LSU Group
+ add %o1, 1, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ fmuls %f0, M0, %f3 ! FGM
+ fmuls %f1, M5, %f4 ! FGM Group
+ st %f2, [%g2 + 0x08] ! LSU
+ fadds %f3, M12, %f3 ! FGA Group
+ st %f3, [%g2 + 0x00] ! LSU
+ fadds %f4, M13, %f4 ! FGA Group
+ st %f4, [%g2 + 0x04] ! LSU
+ cmp %o1, %g3 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x10, %g2 ! IEU0 Group
+2:
+ ba __set_v4f_3
+ nop
+
+ .globl _mesa_sparc_transform_points3_3d
+_mesa_sparc_transform_points3_3d:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_1_2_4_5_6_8_9_10_12_13_14(%o1)
+
+ cmp %g3, 0
+ st %g3, [%o0 + V4F_COUNT]
+ be 2f
+ clr %o1
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ ld [%g1 + 0x08], %f2 ! LSU Group
+ add %o1, 1, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ fmuls %f0, M0, %f3 ! FGM
+ fmuls %f1, M4, %f6 ! FGM Group
+ fmuls %f0, M1, %f4 ! FGM Group
+ fmuls %f1, M5, %f7 ! FGM Group
+ fmuls %f0, M2, %f5 ! FGM Group f3 available
+ fmuls %f1, M6, %f8 ! FGM Group f6 available
+ fadds %f3, %f6, %f3 ! FGA
+ fmuls %f2, M8, %f9 ! FGM Group f4 available
+ fmuls %f2, M9, %f10 ! FGM Group f7 available
+ fadds %f4, %f7, %f4 ! FGA
+ fmuls %f2, M10, %f11 ! FGM Group f5 available
+ fadds %f5, %f8, %f5 ! FGA Group f8, f3 available
+ fadds %f3, %f9, %f3 ! FGA Group f9 available
+ fadds %f4, %f10, %f4 ! FGA Group f10, f4 available
+ fadds %f5, %f11, %f5 ! FGA Group stall, f11, f5 available
+ fadds %f3, M12, %f3 ! FGA Group f3 available
+ st %f3, [%g2 + 0x00] ! LSU
+ fadds %f4, M13, %f4 ! FGA Group f4 available
+ st %f4, [%g2 + 0x04] ! LSU
+ fadds %f5, M14, %f5 ! FGA Group f5 available
+ st %f5, [%g2 + 0x08] ! LSU
+ cmp %o1, %g3 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x10, %g2 ! IEU0 Group
+2:
+ ba __set_v4f_3
+ nop
+
+ .globl _mesa_sparc_transform_points3_3d_no_rot
+_mesa_sparc_transform_points3_3d_no_rot:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_5_10_12_13_14(%o1)
+
+ cmp %g3, 0
+ st %g3, [%o0 + V4F_COUNT]
+ be 2f
+ clr %o1
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ ld [%g1 + 0x08], %f2 ! LSU Group
+ add %o1, 1, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ cmp %o1, %g3 ! IEU1 Group
+ fmuls %f0, M0, %f3 ! FGM
+ fmuls %f1, M5, %f4 ! FGM Group
+ fmuls %f2, M10, %f5 ! FGM Group
+ fadds %f3, M12, %f3 ! FGA Group, stall, f3 available
+ st %f3, [%g2 + 0x00] ! LSU
+ fadds %f4, M13, %f4 ! FGA Group, f4 available
+ st %f4, [%g2 + 0x04] ! LSU
+ fadds %f5, M14, %f5 ! FGA Group, f5 available
+ st %f5, [%g2 + 0x08] ! LEU
+ bne 1b ! CTI
+ add %g2, 0x10, %g2 ! IEU0 Group
+2:
+ ba __set_v4f_3
+ nop
+
+ .globl _mesa_sparc_transform_points3_perspective
+_mesa_sparc_transform_points3_perspective:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_5_8_9_10_14(%o1)
+
+ cmp %g3, 0
+ st %g3, [%o0 + V4F_COUNT]
+ be 2f
+ clr %o1
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ ld [%g1 + 0x08], %f2 ! LSU Group
+ add %o1, 1, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ fmuls %f0, M0, %f3 ! FGM
+ fmuls %f2, M8, %f6 ! FGM Group
+ fmuls %f1, M5, %f4 ! FGM Group
+ fmuls %f2, M9, %f7 ! FGM Group
+ fmuls %f2, M10, %f5 ! FGM Group f3 available
+ fadds %f3, %f6, %f3 ! FGA Group f6 available
+ st %f3, [%g2 + 0x00] ! LSU
+ fadds %f4, %f7, %f4 ! FGA Group stall, f4, f7 available
+ st %f4, [%g2 + 0x04] ! LSU
+ fadds %f5, M14, %f5 ! FGA Group
+ st %f5, [%g2 + 0x08] ! LSU
+ fnegs %f2, %f6 ! FGA Group
+ st %f6, [%g2 + 0x0c] ! LSU
+ cmp %o1, %g3 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x10, %g2 ! IEU0 Group
+2:
+ ba __set_v4f_4
+ nop
+
+ .globl _mesa_sparc_transform_points4_general
+_mesa_sparc_transform_points4_general:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_1_2_3_4_5_6_7_8_9_10_11_12_13_14_15(%o1)
+
+ cmp %g3, 0
+ st %g3, [%o0 + V4F_COUNT]
+ be 2f
+ clr %o1
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ ld [%g1 + 0x08], %f2 ! LSU Group
+ ld [%g1 + 0x0c], %f3 ! LSU Group
+ add %o1, 1, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ fmuls %f0, M0, %f4 ! FGM Group
+ fmuls %f1, M4, %f8 ! FGM Group
+ fmuls %f0, M1, %f5 ! FGM Group
+ fmuls %f1, M5, %f9 ! FGM Group
+ fmuls %f0, M2, %f6 ! FGM Group f4 available
+ fmuls %f1, M6, %f10 ! FGM Group f8 available
+ fadds %f4, %f8, %f4 ! FGA
+ fmuls %f0, M3, %f7 ! FGM Group f5 available
+ fmuls %f1, M7, %f11 ! FGM Group f9 available
+ fadds %f5, %f9, %f5 ! FGA
+ fmuls %f2, M8, %f12 ! FGM Group f6 available
+ fmuls %f2, M9, %f13 ! FGM Group f10, f4 available
+ fadds %f6, %f10, %f6 ! FGA
+ fmuls %f2, M10, %f14 ! FGM Group f7 available
+ fmuls %f2, M11, %f15 ! FGM Group f11, f5 available
+ fadds %f7, %f11, %f7 ! FGA
+ fmuls %f3, M12, %f8 ! FGM Group f12 available
+ fadds %f4, %f12, %f4 ! FGA
+ fmuls %f3, M13, %f9 ! FGM Group f13, f6 available
+ fadds %f5, %f13, %f5 ! FGA
+ fmuls %f3, M14, %f10 ! FGM Group f14 available
+ fadds %f6, %f14, %f6 ! FGA
+ fmuls %f3, M15, %f11 ! FGM Group f15, f7 available
+ fadds %f7, %f15, %f7 ! FGA
+ fadds %f4, %f8, %f4 ! FGA Group f8, f4 available
+ st %f4, [%g2 + 0x00] ! LSU
+ fadds %f5, %f9, %f5 ! FGA Group f9, f5 available
+ st %f5, [%g2 + 0x04] ! LSU
+ fadds %f6, %f10, %f6 ! FGA Group f10, f6 available
+ st %f6, [%g2 + 0x08] ! LSU
+ fadds %f7, %f11, %f7 ! FGA Group f11, f7 available
+ st %f7, [%g2 + 0x0c] ! LSU
+ cmp %o1, %g3 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x10, %g2 ! IEU0 Group
+2:
+ ba __set_v4f_4
+ nop
+
+ .globl _mesa_sparc_transform_points4_identity
+_mesa_sparc_transform_points4_identity:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ cmp %g3, 0
+ st %g3, [%o0 + V4F_COUNT]
+ be 2f
+ clr %o1
+
+1: ld [%g1 + 0x00], %f0
+ ld [%g1 + 0x04], %f1
+ ld [%g1 + 0x08], %f2
+ add %o1, 1, %o1
+ ld [%g1 + 0x0c], %f3
+ add %g1, %o5, %g1
+ st %f0, [%g2 + 0x00]
+ st %f1, [%g2 + 0x04]
+ st %f2, [%g2 + 0x08]
+ cmp %o1, %g3
+ st %f3, [%g2 + 0x0c]
+ bne 1b
+ add %g2, 0x10, %g2
+2:
+ ba __set_v4f_4
+ nop
+
+ .globl _mesa_sparc_transform_points4_2d
+_mesa_sparc_transform_points4_2d:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_1_4_5_12_13(%o1)
+
+ cmp %g3, 0
+ st %g3, [%o0 + V4F_COUNT]
+ be 2f
+ clr %o1
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ ld [%g1 + 0x08], %f2 ! LSU Group
+ ld [%g1 + 0x0c], %f3 ! LSU Group
+ add %o1, 1, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ fmuls %f0, M0, %f4 ! FGM
+ fmuls %f1, M4, %f8 ! FGM Group
+ fmuls %f0, M1, %f5 ! FGM Group
+ fmuls %f1, M5, %f9 ! FGM Group f4 available
+ fmuls %f3, M12, %f12 ! FGM Group
+ fmuls %f3, M13, %f13 ! FGM Group f8 available
+ fadds %f4, %f8, %f4 ! FGA
+ fadds %f5, %f9, %f5 ! FGA Group stall, f5, f9 available
+ fadds %f4, %f12, %f4 ! FGA Group 2 cycle stall, f4, f12, f13 avail
+ st %f4, [%g2 + 0x00] ! LSU
+ fadds %f5, %f13, %f5 ! FGA Group f5 available
+ st %f5, [%g2 + 0x04] ! LSU
+ st %f2, [%g2 + 0x08] ! LSU Group
+ st %f3, [%g2 + 0x0c] ! LSU Group
+ cmp %o1, %g3 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x10, %g2 ! IEU0 Group
+2:
+ ba __set_v4f_4
+ nop
+
+ .globl _mesa_sparc_transform_points4_2d_no_rot
+_mesa_sparc_transform_points4_2d_no_rot:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_1_4_5_12_13(%o1)
+
+ cmp %g3, 0
+ st %g3, [%o0 + V4F_COUNT]
+ be 2f
+ clr %o1
+
+1: ld [%g1 + 0x00], %f0
+ ld [%g1 + 0x04], %f1
+ ld [%g1 + 0x08], %f2
+ ld [%g1 + 0x0c], %f3
+ add %o1, 1, %o1
+ add %g1, %o5, %g1
+ fmuls %f0, M0, %f4
+ fmuls %f3, M12, %f8
+ fmuls %f1, M5, %f5
+ fmuls %f3, M13, %f9
+ fadds %f4, %f8, %f4
+ st %f4, [%g2 + 0x00]
+ fadds %f5, %f9, %f5
+ st %f5, [%g2 + 0x04]
+ st %f2, [%g2 + 0x08]
+ st %f3, [%g2 + 0x0c]
+ cmp %o1, %g3
+ bne 1b
+ add %g2, 0x10, %g2
+2:
+ ba __set_v4f_4
+ nop
+
+ .globl _mesa_sparc_transform_points4_3d
+_mesa_sparc_transform_points4_3d:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_1_2_4_5_6_8_9_10_12_13_14(%o1)
+
+ cmp %g3, 0
+ st %g3, [%o0 + V4F_COUNT]
+ be 2f
+ clr %o1
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ ld [%g1 + 0x08], %f2 ! LSU Group
+ ld [%g1 + 0x0c], %f3 ! LSU Group
+ add %o1, 1, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ fmuls %f0, M0, %f4 ! FGM
+ fmuls %f1, M4, %f7 ! FGM Group
+ fmuls %f0, M1, %f5 ! FGM Group
+ fmuls %f1, M5, %f8 ! FGM Group
+ fmuls %f0, M2, %f6 ! FGM Group f4 available
+ fmuls %f1, M6, %f9 ! FGM Group f7 available
+ fadds %f4, %f7, %f4 ! FGA
+ fmuls %f2, M8, %f10 ! FGM Group f5 available
+ fmuls %f2, M9, %f11 ! FGM Group f8 available
+ fadds %f5, %f8, %f5 ! FGA
+ fmuls %f2, M10, %f12 ! FGM Group f6 available
+ fmuls %f3, M12, %f13 ! FGM Group f9, f4 available
+ fadds %f6, %f9, %f6 ! FGA
+ fmuls %f3, M13, %f14 ! FGM Group f10 available
+ fadds %f4, %f10, %f4 ! FGA
+ fmuls %f3, M14, %f15 ! FGM Group f11, f5 available
+ fadds %f5, %f11, %f5 ! FGA
+ fadds %f6, %f12, %f6 ! FGA Group stall, f12, f13, f6 available
+ fadds %f4, %f13, %f4 ! FGA Group f14, f4 available
+ st %f4, [%g2 + 0x00] ! LSU
+ fadds %f5, %f14, %f5 ! FGA Group f15, f5 available
+ st %f5, [%g2 + 0x04] ! LSU
+ fadds %f6, %f15, %f6 ! FGA Group f6 available
+ st %f6, [%g2 + 0x08] ! LSU
+ st %f3, [%g2 + 0x0c] ! LSU Group
+ cmp %o1, %g3 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x10, %g2 ! IEU0 Group
+2:
+ ba __set_v4f_4
+ nop
+
+ .globl _mesa_sparc_transform_points4_3d_no_rot
+_mesa_sparc_transform_points4_3d_no_rot:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_5_10_12_13_14(%o1)
+
+ cmp %g3, 0
+ st %g3, [%o0 + V4F_COUNT]
+ be 2f
+ clr %o1
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ ld [%g1 + 0x08], %f2 ! LSU Group
+ ld [%g1 + 0x0c], %f3 ! LSU Group
+ add %o1, 1, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ fmuls %f0, M0, %f4 ! FGM
+ fmuls %f3, M12, %f7 ! FGM Group
+ fmuls %f1, M5, %f5 ! FGM Group
+ fmuls %f3, M13, %f8 ! FGM Group
+ fmuls %f2, M10, %f6 ! FGM Group f4 available
+ fmuls %f3, M14, %f9 ! FGM Group f7 available
+ fadds %f4, %f7, %f4 ! FGA
+ st %f4, [%g2 + 0x00] ! LSU
+ fadds %f5, %f8, %f5 ! FGA Group stall, f5, f8 available
+ st %f5, [%g2 + 0x04] ! LSU
+ fadds %f6, %f9, %f6 ! FGA Group stall, f6, f9 available
+ st %f6, [%g2 + 0x08] ! LSU
+ st %f3, [%g2 + 0x0c] ! LSU Group
+ cmp %o1, %g3 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x10, %g2 ! IEU0 Group
+2:
+ ba __set_v4f_4
+ nop
+
+ .globl _mesa_sparc_transform_points4_perspective
+_mesa_sparc_transform_points4_perspective:
+ ld [%o2 + V4F_STRIDE], %o5
+ LDPTR [%o2 + V4F_START], %g1
+ LDPTR [%o0 + V4F_START], %g2
+ ld [%o2 + V4F_COUNT], %g3
+
+ LDMATRIX_0_5_8_9_10_14(%o1)
+
+ cmp %g3, 0
+ st %g3, [%o0 + V4F_COUNT]
+ be 2f
+ clr %o1
+
+1: ld [%g1 + 0x00], %f0 ! LSU Group
+ ld [%g1 + 0x04], %f1 ! LSU Group
+ ld [%g1 + 0x08], %f2 ! LSU Group
+ ld [%g1 + 0x0c], %f3 ! LSU Group
+ add %o1, 1, %o1 ! IEU0
+ add %g1, %o5, %g1 ! IEU1
+ fmuls %f0, M0, %f4 ! FGM
+ fmuls %f2, M8, %f7 ! FGM Group
+ fmuls %f1, M5, %f5 ! FGM Group
+ fmuls %f2, M9, %f8 ! FGM Group
+ fmuls %f2, M10, %f6 ! FGM Group f4 available
+ fmuls %f3, M14, %f9 ! FGM Group f7 available
+ fadds %f4, %f7, %f4 ! FGA
+ st %f4, [%g2 + 0x00] ! LSU
+ fadds %f5, %f8, %f5 ! FGA Group stall, f5, f8 available
+ st %f5, [%g2 + 0x04] ! LSU
+ fadds %f6, %f9, %f6 ! FGA Group stall, f6, f9 available
+ st %f6, [%g2 + 0x08] ! LSU
+ fnegs %f2, %f7 ! FGA Group
+ st %f7, [%g2 + 0x0c] ! LSU
+ cmp %o1, %g3 ! IEU1
+ bne 1b ! CTI
+ add %g2, 0x10, %g2 ! IEU0 Group
+2:
+ ba __set_v4f_4
+ nop