aboutsummaryrefslogtreecommitdiff
path: root/mesalib/src/mesa/x86/sse_normal.S
diff options
context:
space:
mode:
Diffstat (limited to 'mesalib/src/mesa/x86/sse_normal.S')
-rw-r--r--mesalib/src/mesa/x86/sse_normal.S261
1 files changed, 261 insertions, 0 deletions
diff --git a/mesalib/src/mesa/x86/sse_normal.S b/mesalib/src/mesa/x86/sse_normal.S
new file mode 100644
index 000000000..b6d5ef965
--- /dev/null
+++ b/mesalib/src/mesa/x86/sse_normal.S
@@ -0,0 +1,261 @@
+
+/*
+ * Mesa 3-D graphics library
+ * Version: 3.5
+ *
+ * Copyright (C) 1999-2001 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.
+ */
+
+/** TODO:
+ * - insert PREFETCH instructions to avoid cache-misses !
+ * - some more optimizations are possible...
+ * - for 40-50% more performance in the SSE-functions, the
+ * data (trans-matrix, src_vert, dst_vert) needs to be 16byte aligned !
+ */
+
+#ifdef USE_SSE_ASM
+#include "assyntax.h"
+#include "matypes.h"
+#include "norm_args.h"
+
+ SEG_TEXT
+
+#define M(i) REGOFF(i * 4, EDX)
+#define S(i) REGOFF(i * 4, ESI)
+#define D(i) REGOFF(i * 4, EDI)
+#define STRIDE REGOFF(12, ESI)
+
+
+ALIGNTEXT16
+GLOBL GLNAME(_mesa_sse_transform_rescale_normals_no_rot)
+HIDDEN(_mesa_sse_transform_rescale_normals_no_rot)
+GLNAME(_mesa_sse_transform_rescale_normals_no_rot):
+
+#define FRAME_OFFSET 8
+ PUSH_L ( ESI )
+ PUSH_L ( EDI )
+
+ MOV_L ( ARG_IN, ESI ) /* ptr to source GLvector3f */
+ MOV_L ( ARG_DEST, EDI ) /* ptr to dest GLvector3f */
+
+ MOV_L ( ARG_MAT, EDX ) /* ptr to matrix */
+ MOV_L ( REGOFF(MATRIX_INV, EDX), EDX) /* matrix->inv */
+
+ MOV_L ( REGOFF(V4F_COUNT, ESI), ECX ) /* source count */
+
+ TEST_L ( ECX, ECX )
+ JZ( LLBL(K_G3TRNNRR_finish) ) /* count was zero; go to finish */
+
+ MOV_L ( STRIDE, EAX ) /* stride */
+ MOV_L ( ECX, REGOFF(V4F_COUNT, EDI) ) /* set dest-count */
+
+ IMUL_L( CONST(16), ECX ) /* count *= 16 */
+ MOV_L( REGOFF(V4F_START, ESI), ESI ) /* ptr to first source vertex */
+
+ MOV_L( REGOFF(V4F_START, EDI), EDI ) /* ptr to first dest vertex */
+ ADD_L( EDI, ECX ) /* count += dest ptr */
+
+ALIGNTEXT32
+ MOVSS ( M(0), XMM1 ) /* m0 */
+ MOVSS ( M(5), XMM2 ) /* m5 */
+ UNPCKLPS( XMM2, XMM1 ) /* m5 | m0 */
+ MOVSS ( ARG_SCALE, XMM0 ) /* scale */
+ SHUFPS ( CONST(0x0), XMM0, XMM0 ) /* scale | scale */
+ MULPS ( XMM0, XMM1 ) /* m5*scale | m0*scale */
+ MULSS ( M(10), XMM0 ) /* m10*scale */
+
+ALIGNTEXT32
+LLBL(K_G3TRNNRR_top):
+ MOVLPS ( S(0), XMM2 ) /* uy | ux */
+ MULPS ( XMM1, XMM2 ) /* uy*m5*scale | ux*m0*scale */
+ MOVLPS ( XMM2, D(0) ) /* ->D(1) | D(0) */
+
+ MOVSS ( S(2), XMM2 ) /* uz */
+ MULSS ( XMM0, XMM2 ) /* uz*m10*scale */
+ MOVSS ( XMM2, D(2) ) /* ->D(2) */
+
+LLBL(K_G3TRNNRR_skip):
+ ADD_L ( CONST(16), EDI )
+ ADD_L ( EAX, ESI )
+ CMP_L ( ECX, EDI )
+ JNE ( LLBL(K_G3TRNNRR_top) )
+
+LLBL(K_G3TRNNRR_finish):
+ POP_L ( EDI )
+ POP_L ( ESI )
+ RET
+#undef FRAME_OFFSET
+
+
+
+ALIGNTEXT16
+GLOBL GLNAME(_mesa_sse_transform_rescale_normals)
+HIDDEN(_mesa_sse_transform_rescale_normals)
+GLNAME(_mesa_sse_transform_rescale_normals):
+
+#define FRAME_OFFSET 8
+ PUSH_L ( ESI )
+ PUSH_L ( EDI )
+
+ MOV_L ( ARG_IN, ESI ) /* ptr to source GLvector3f */
+ MOV_L ( ARG_DEST, EDI ) /* ptr to dest GLvector3f */
+
+ MOV_L ( ARG_MAT, EDX ) /* ptr to matrix */
+ MOV_L ( REGOFF(MATRIX_INV, EDX), EDX) /* matrix->inv */
+
+ MOV_L ( REGOFF(V4F_COUNT, ESI), ECX ) /* source count */
+
+ TEST_L ( ECX, ECX )
+ JZ( LLBL(K_G3TRNR_finish) ) /* count was zero; go to finish */
+
+ MOV_L ( STRIDE, EAX ) /* stride */
+ MOV_L ( ECX, REGOFF(V4F_COUNT, EDI) ) /* set dest-count */
+
+ IMUL_L( CONST(16), ECX ) /* count *= 16 */
+ MOV_L( REGOFF(V4F_START, ESI), ESI ) /* ptr to first source vertex */
+
+ MOV_L( REGOFF(V4F_START, EDI), EDI ) /* ptr to first dest vertex */
+ ADD_L( EDI, ECX ) /* count += dest ptr */
+
+ALIGNTEXT32
+ MOVSS ( M(0), XMM0 ) /* m0 */
+ MOVSS ( M(4), XMM1 ) /* m4 */
+ UNPCKLPS( XMM1, XMM0 ) /* m4 | m0 */
+
+ MOVSS ( ARG_SCALE, XMM4 ) /* scale */
+ SHUFPS ( CONST(0x0), XMM4, XMM4 ) /* scale | scale */
+
+ MULPS ( XMM4, XMM0 ) /* m4*scale | m0*scale */
+ MOVSS ( M(1), XMM1 ) /* m1 */
+ MOVSS ( M(5), XMM2 ) /* m5 */
+ UNPCKLPS( XMM2, XMM1 ) /* m5 | m1 */
+ MULPS ( XMM4, XMM1 ) /* m5*scale | m1*scale */
+ MOVSS ( M(2), XMM2 ) /* m2 */
+ MOVSS ( M(6), XMM3 ) /* m6 */
+ UNPCKLPS( XMM3, XMM2 ) /* m6 | m2 */
+ MULPS ( XMM4, XMM2 ) /* m6*scale | m2*scale */
+
+ MOVSS ( M(8), XMM6 ) /* m8 */
+ MULSS ( ARG_SCALE, XMM6 ) /* m8*scale */
+ MOVSS ( M(9), XMM7 ) /* m9 */
+ MULSS ( ARG_SCALE, XMM7 ) /* m9*scale */
+
+ALIGNTEXT32
+LLBL(K_G3TRNR_top):
+ MOVSS ( S(0), XMM3 ) /* ux */
+ SHUFPS ( CONST(0x0), XMM3, XMM3 ) /* ux | ux */
+ MULPS ( XMM0, XMM3 ) /* ux*m4 | ux*m0 */
+ MOVSS ( S(1), XMM4 ) /* uy */
+ SHUFPS ( CONST(0x0), XMM4, XMM4 ) /* uy | uy */
+ MULPS ( XMM1, XMM4 ) /* uy*m5 | uy*m1 */
+ MOVSS ( S(2), XMM5 ) /* uz */
+ SHUFPS ( CONST(0x0), XMM5, XMM5 ) /* uz | uz */
+ MULPS ( XMM2, XMM5 ) /* uz*m6 | uz*m2 */
+
+ ADDPS ( XMM4, XMM3 )
+ ADDPS ( XMM5, XMM3 )
+ MOVLPS ( XMM3, D(0) )
+
+ MOVSS ( M(10), XMM3 ) /* m10 */
+ MULSS ( ARG_SCALE, XMM3 ) /* m10*scale */
+ MULSS ( S(2), XMM3 ) /* m10*scale*uz */
+ MOVSS ( S(1), XMM4 ) /* uy */
+ MULSS ( XMM7, XMM4 ) /* uy*m9*scale */
+ MOVSS ( S(0), XMM5 ) /* ux */
+ MULSS ( XMM6, XMM5 ) /* ux*m8*scale */
+
+ ADDSS ( XMM4, XMM3 )
+ ADDSS ( XMM5, XMM3 )
+ MOVSS ( XMM3, D(2) )
+
+LLBL(K_G3TRNR_skip):
+ ADD_L ( CONST(16), EDI )
+ ADD_L ( EAX, ESI )
+ CMP_L ( ECX, EDI )
+ JNE ( LLBL(K_G3TRNR_top) )
+
+LLBL(K_G3TRNR_finish):
+ POP_L ( EDI )
+ POP_L ( ESI )
+ RET
+#undef FRAME_OFFSET
+
+
+ALIGNTEXT16
+GLOBL GLNAME(_mesa_sse_transform_normals_no_rot)
+HIDDEN(_mesa_sse_transform_normals_no_rot)
+GLNAME(_mesa_sse_transform_normals_no_rot):
+
+#define FRAME_OFFSET 8
+ PUSH_L ( ESI )
+ PUSH_L ( EDI )
+
+ MOV_L ( ARG_IN, ESI ) /* ptr to source GLvector3f */
+ MOV_L ( ARG_DEST, EDI ) /* ptr to dest GLvector3f */
+
+ MOV_L ( ARG_MAT, EDX ) /* ptr to matrix */
+ MOV_L ( REGOFF(MATRIX_INV, EDX), EDX) /* matrix->inv */
+
+ MOV_L ( REGOFF(V4F_COUNT, ESI), ECX ) /* source count */
+
+ TEST_L ( ECX, ECX )
+ JZ( LLBL(K_G3TNNRR_finish) ) /* count was zero; go to finish */
+
+ MOV_L ( STRIDE, EAX ) /* stride */
+ MOV_L ( ECX, REGOFF(V4F_COUNT, EDI) ) /* set dest-count */
+
+ IMUL_L( CONST(16), ECX ) /* count *= 16 */
+ MOV_L( REGOFF(V4F_START, ESI), ESI ) /* ptr to first source vertex */
+
+ MOV_L( REGOFF(V4F_START, EDI), EDI ) /* ptr to first dest vertex */
+ ADD_L( EDI, ECX ) /* count += dest ptr */
+
+ALIGNTEXT32
+ MOVSS( M(0), XMM0 ) /* m0 */
+ MOVSS( M(5), XMM1 ) /* m5 */
+ UNPCKLPS( XMM1, XMM0 ) /* m5 | m0 */
+ MOVSS( M(10), XMM1 ) /* m10 */
+
+ALIGNTEXT32
+LLBL(K_G3TNNRR_top):
+ MOVLPS( S(0), XMM2 ) /* uy | ux */
+ MULPS( XMM0, XMM2 ) /* uy*m5 | ux*m0 */
+ MOVLPS( XMM2, D(0) )
+
+ MOVSS( S(2), XMM2 ) /* uz */
+ MULSS( XMM1, XMM2 ) /* uz*m10 */
+ MOVSS( XMM2, D(2) )
+
+LLBL(K_G3TNNRR_skip):
+ ADD_L ( CONST(16), EDI )
+ ADD_L ( EAX, ESI )
+ CMP_L ( ECX, EDI )
+ JNE ( LLBL(K_G3TNNRR_top) )
+
+LLBL(K_G3TNNRR_finish):
+ POP_L ( EDI )
+ POP_L ( ESI )
+ RET
+#undef FRAME_OFFSET
+#endif
+
+#if defined (__ELF__) && defined (__linux__)
+ .section .note.GNU-stack,"",%progbits
+#endif