Mesa (master): mesa: added OPCODE_NRM3/ NRM4 instructions for vector normalization.

[Brian Paul]

Module: Mesa
Branch: master
Commit: f6ead50827c03017e6b730313c361b39190da92f
URL:    http://cgit.freedesktop.org/mesa/mesa/commit/?id=f6ead50827c03017e6b730313c361b39190da92f

Author: Brian Paul <brian.paul at tungstengraphics.com>
Date:   Fri Nov  7 08:51:31 2008 -0700

mesa: added OPCODE_NRM3/NRM4 instructions for vector normalization.

We may emit these instructions from GLSL instead of DP3/RCP/MUL.

Also, implement SSG (set sign) instruction in the interpreter.

---

 src/mesa/shader/prog_execute.c     |   41 ++++++++++++++++++++++++++++++++++++
 src/mesa/shader/prog_instruction.c |    2 +
 src/mesa/shader/prog_instruction.h |    2 +
 3 files changed, 45 insertions(+), 0 deletions(-)

diff --git a/src/mesa/shader/prog_execute.c b/src/mesa/shader/prog_execute.c
index d843761..c0173d3 100644
--- a/src/mesa/shader/prog_execute.c
+++ b/src/mesa/shader/prog_execute.c
@@ -1019,6 +1019,36 @@ _mesa_execute_program(GLcontext * ctx,
          break;
       case OPCODE_NOP:
          break;
+      case OPCODE_NRM3:        /* 3-component normalization */
+         {
+            GLfloat a[4], result[4];
+            GLfloat tmp;
+            fetch_vector4(&inst->SrcReg[0], machine, a);
+            tmp = a[0] * a[0] + a[1] * a[1] + a[2] * a[2];
+            if (tmp != 0.0F)
+               tmp = 1.0F / tmp;
+            result[0] = tmp * a[0];
+            result[1] = tmp * a[1];
+            result[2] = tmp * a[2];
+            result[3] = 0.0;  /* undefined, but prevent valgrind warnings */
+            store_vector4(inst, machine, result);
+         }
+         break;
+      case OPCODE_NRM4:        /* 4-component normalization */
+         {
+            GLfloat a[4], result[4];
+            GLfloat tmp;
+            fetch_vector4(&inst->SrcReg[0], machine, a);
+            tmp = a[0] * a[0] + a[1] * a[1] + a[2] * a[2] + a[3] * a[3];
+            if (tmp != 0.0F)
+               tmp = 1.0F / tmp;
+            result[0] = tmp * a[0];
+            result[1] = tmp * a[1];
+            result[2] = tmp * a[2];
+            result[3] = tmp * a[3];
+            store_vector4(inst, machine, result);
+         }
+         break;
       case OPCODE_PK2H:        /* pack two 16-bit floats in one 32-bit float */
          {
             GLfloat a[4], result[4];
@@ -1277,6 +1307,17 @@ _mesa_execute_program(GLcontext * ctx,
             }
          }
          break;
+      case OPCODE_SSG:         /* set sign (-1, 0 or +1) */
+         {
+            GLfloat a[4], result[4];
+            fetch_vector4(&inst->SrcReg[0], machine, a);
+            result[0] = (GLfloat) ((a[0] > 0.0F) - (a[0] < 0.0F));
+            result[1] = (GLfloat) ((a[1] > 0.0F) - (a[1] < 0.0F));
+            result[2] = (GLfloat) ((a[2] > 0.0F) - (a[2] < 0.0F));
+            result[3] = (GLfloat) ((a[3] > 0.0F) - (a[3] < 0.0F));
+            store_vector4(inst, machine, result);
+         }
+         break;
       case OPCODE_STR:         /* set true, operands ignored */
          {
             static const GLfloat result[4] = { 1.0F, 1.0F, 1.0F, 1.0F };
diff --git a/src/mesa/shader/prog_instruction.c b/src/mesa/shader/prog_instruction.c
index 7e340ce..d4f3bcb 100644
--- a/src/mesa/shader/prog_instruction.c
+++ b/src/mesa/shader/prog_instruction.c
@@ -197,6 +197,8 @@ static const struct instruction_info InstInfo[MAX_OPCODE] = {
    { OPCODE_NOISE2, "NOISE2",  1, 1 },
    { OPCODE_NOISE3, "NOISE3",  1, 1 },
    { OPCODE_NOISE4, "NOISE4",  1, 1 },
+   { OPCODE_NRM3,   "NRM3",    1, 1 },
+   { OPCODE_NRM4,   "NRM4",    1, 1 },
    { OPCODE_PK2H,   "PK2H",    1, 1 },
    { OPCODE_PK2US,  "PK2US",   1, 1 },
    { OPCODE_PK4B,   "PK4B",    1, 1 },
diff --git a/src/mesa/shader/prog_instruction.h b/src/mesa/shader/prog_instruction.h
index 16701e4..3bcd082 100644
--- a/src/mesa/shader/prog_instruction.h
+++ b/src/mesa/shader/prog_instruction.h
@@ -188,6 +188,8 @@ typedef enum prog_opcode {
    OPCODE_NOISE2,    /*                                      X   */
    OPCODE_NOISE3,    /*                                      X   */
    OPCODE_NOISE4,    /*                                      X   */
+   OPCODE_NRM3,      /*                                          */
+   OPCODE_NRM4,      /*                                          */
    OPCODE_PK2H,      /*                            X             */
    OPCODE_PK2US,     /*                            X             */
    OPCODE_PK4B,      /*                            X             */