From aa932364fec79079429fc267f9fe49db43584514 Mon Sep 17 00:00:00 2001
From: koolsmoky <>
Date: Sat, 22 Jan 2005 14:52:02 +0000
Subject: [PATCH] enabled packed argb for cmd packet type 3

---
 glide2x/cvg/glide/src/Makefile.win32 |   2 +-
 glide2x/cvg/glide/src/ddgump.c       |  13 +-
 glide2x/cvg/glide/src/diglide.c      |  39 ++-
 glide2x/cvg/glide/src/fxglide.h      |  68 ++---
 glide2x/cvg/glide/src/gaa.c          | 132 ++++++++-
 glide2x/cvg/glide/src/gdraw.c        |  38 +--
 glide2x/cvg/glide/src/gglide.c       |  51 ++--
 glide2x/cvg/glide/src/glide.rc       |   4 +
 glide2x/cvg/glide/src/gpci.c         |  35 ++-
 glide2x/cvg/glide/src/gsst.c         |   9 +-
 glide2x/cvg/glide/src/gtexdl.c       |  14 +-
 glide2x/cvg/glide/src/guclip.c       |   7 +-
 glide2x/cvg/glide/src/gxdraw.c       |   9 +-
 glide2x/cvg/glide/src/xdraw2.asm     |   7 +-
 glide2x/cvg/glide/src/xdraw2.inc     | 416 +++++++++++++--------------
 15 files changed, 487 insertions(+), 357 deletions(-)

diff --git a/glide2x/cvg/glide/src/Makefile.win32 b/glide2x/cvg/glide/src/Makefile.win32
index f5eef77..7d688ed 100644
--- a/glide2x/cvg/glide/src/Makefile.win32
+++ b/glide2x/cvg/glide/src/Makefile.win32
@@ -84,7 +84,7 @@ endif
 CDEFS = -D__WIN32__ -DDIRECTX -DFX_DLL_ENABLE -D__3Dfx_PCI_CFG__
 
 # general
-CDEFS += -DGLIDE_HW_TRI_SETUP=1 -DGLIDE_PACKED_RGB=0 -DGLIDE_TRI_CULLING=1 -DGLIDE_DEFAULT_GAMMA=1.3f -DGLIDE_LIB=1
+CDEFS += -DGLIDE_HW_TRI_SETUP=1 -DGLIDE_PACKED_RGB=1 -DGLIDE_TRI_CULLING=1 -DGLIDE_DEFAULT_GAMMA=1.3f -DGLIDE_LIB=1
 #CDEFS += -DGLIDE3 -DGLIDE3_ALPHA -DGLIDE3_SCALER
 
 # special sli buffer clears
diff --git a/glide2x/cvg/glide/src/ddgump.c b/glide2x/cvg/glide/src/ddgump.c
index f52e8b4..045b9e1 100644
--- a/glide2x/cvg/glide/src/ddgump.c
+++ b/glide2x/cvg/glide/src/ddgump.c
@@ -19,6 +19,9 @@
 **
 ** $Header$
 ** $Log$
+** Revision 1.2  2000/10/03 18:28:33  mercury
+** 003-clean_up_cvg-000, cvg tree cleanup.
+**
 ** Revision 1.1.1.1  1999/12/07 21:49:08  joseph
 ** Initial checkin into SourceForge.
 **
@@ -296,7 +299,7 @@ GR_ENTRY(guMPDrawTriangle, void, (const GrVertex *a, const GrVertex *b, const Gr
       _gumpTexCombineFunction(0);
 
       /* render first pass */
-      grDrawTriangle(a, b, c);
+      TRISETUP(a, b, c);
 
       /* second pass */
 
@@ -362,7 +365,7 @@ GR_ENTRY(guMPDrawTriangle, void, (const GrVertex *a, const GrVertex *b, const Gr
         REG_GROUP_END();
 
         /* render other pass */
-        grDrawTriangle(a, b, c);
+        TRISETUP(a, b, c);
 
         /* restore */
         REG_GROUP_BEGIN(BROADCAST_ID, fbzColorPath, regCount, regMask);
@@ -451,7 +454,7 @@ GR_ENTRY(guMPDrawTriangle, void, (const GrVertex *a, const GrVertex *b, const Gr
         REG_GROUP_END();
         
         /* render first pass */
-        grDrawTriangle(a, b, c);
+        TRISETUP(a, b, c);
         
         /* second pass */
         /* xxx may sometimes need to copy texture coordinates */
@@ -518,7 +521,7 @@ GR_ENTRY(guMPDrawTriangle, void, (const GrVertex *a, const GrVertex *b, const Gr
         REG_GROUP_END();
 
         /* render second pass */
-        grDrawTriangle(a, b, c);
+        TRISETUP(a, b, c);
         
         /* if bias, third pass */
         if (fogP) {
@@ -542,7 +545,7 @@ GR_ENTRY(guMPDrawTriangle, void, (const GrVertex *a, const GrVertex *b, const Gr
           REG_GROUP_END();
           
           /* render third pass */
-          grDrawTriangle(a, b, c);
+          TRISETUP(a, b, c);
         }
 
         REG_GROUP_BEGIN(BROADCAST_ID, fbzColorPath, regCount, regMask);
diff --git a/glide2x/cvg/glide/src/diglide.c b/glide2x/cvg/glide/src/diglide.c
index 3a09ae9..c617b8e 100644
--- a/glide2x/cvg/glide/src/diglide.c
+++ b/glide2x/cvg/glide/src/diglide.c
@@ -19,6 +19,10 @@
 **
 ** $Header$
 ** $Log$
+** Revision 1.1.1.1.2.2  2004/12/23 20:45:56  koolsmoky
+** converted to nasm syntax
+** added x86 asm, 3dnow! triangle and mmx, 3dnow! texture download optimizations
+**
 ** Revision 1.1.1.1.2.1  2004/12/12 15:18:58  koolsmoky
 ** changes to support new cpuid; moved single_precision_asm(), double_precision_asm() from cpudetect.asm
 **
@@ -148,6 +152,9 @@ void single_precision_asm()
     fldcw WORD PTR [esp]
     pop   eax
     }
+#elif __GNUC__
+  asm("push %eax \n fnclex \n fstcw (%esp) \n movl (%esp), %eax \n "
+      "and $0x0000fcff, %eax \n movl %eax, (%esp) \n fldcw (%esp) \n pop %eax");
 #else
 #error "Need to implement single_precision_asm() for this compiler"
 #endif
@@ -171,6 +178,10 @@ void double_precision_asm()
     fldcw WORD PTR [esp]
     pop   eax
     }
+#elif __GNUC__
+  asm("push %eax \n fnclex \n fstcw (%esp) \n movw (%esp), %eax \n "
+      "and $0x0000fcff, %eax \n or $0x000002ff, %eax \n mov %eax, (%esp) \n "
+      "fldcw (%esp) \n pop %eax");
 #else
 #error "Need to implement double_precision_asm() for this compiler"
 #endif
@@ -263,26 +274,26 @@ _grSwizzleColor(GrColor_t *color)
     break;
 
   case GR_COLORFORMAT_ABGR:
-    red     = *color & 0x00ff;
-    blue    = (*color >> 16) & 0xff;
-    *color &= 0xff00ff00;
-    *color |= ((red << 16) | blue);
+    red     = (*color & 0x000000ff) << 16; /* 00RR0000 */
+    blue    = (*color & 0x00ff0000) >> 16; /* 000000BB */
+    *color &= 0xff00ff00;                  /* AA00GG00 */
+    *color |= (red | blue);                /* AARRGGBB */
     break;
 
   case GR_COLORFORMAT_RGBA:
-    blue   = (*color & 0x0000ff00) >> 8;
-    green  = (*color & 0x00ff0000) >> 16;
-    red    = (*color & 0xff000000) >> 24;
-    alpha  = (*color & 0x000000ff);
-    *color = (alpha << 24) | (red << 16) | (green << 8) | blue;
+    blue   = (*color & 0x0000ff00) >> 8;  /* 000000BB */
+    green  = (*color & 0x00ff0000) >> 8;  /* 0000GG00 */
+    red    = (*color & 0xff000000) >> 8;  /* 00RR0000 */
+    alpha  = (*color & 0x000000ff) << 24; /* AA000000 */
+    *color = alpha | red | green | blue;  /* AARRGGBB */
     break;
 
   case GR_COLORFORMAT_BGRA:
-    blue   = (*color & 0xff000000) >> 24;
-    green  = (*color & 0x00ff0000) >> 16;
-    red    = (*color & 0x0000ff00) >> 8;
-    alpha  = (*color & 0x000000ff);
-    *color = (alpha << 24) | (red << 16) | (green << 8) | blue;
+    blue   = (*color & 0xff000000) >> 24; /* 000000BB */
+    green  = (*color & 0x00ff0000) >> 8;  /* 0000GG00 */
+    red    = (*color & 0x0000ff00) << 8;  /* 00RR0000 */
+    alpha  = (*color & 0x000000ff) << 24; /* AA000000 */
+    *color = alpha | red | green | blue;  /* AARRGGBB */
     break;
 
   default:
diff --git a/glide2x/cvg/glide/src/fxglide.h b/glide2x/cvg/glide/src/fxglide.h
index 005edc0..3058ee5 100644
--- a/glide2x/cvg/glide/src/fxglide.h
+++ b/glide2x/cvg/glide/src/fxglide.h
@@ -19,6 +19,10 @@
 **
 ** $Header$
 ** $Log$
+** Revision 1.2.2.2  2004/12/23 20:45:56  koolsmoky
+** converted to nasm syntax
+** added x86 asm, 3dnow! triangle and mmx, 3dnow! texture download optimizations
+**
 ** Revision 1.2.2.1  2004/12/12 15:17:18  koolsmoky
 ** support new cpuid
 **
@@ -519,12 +523,12 @@ struct _GrState_s
  */
 
 typedef FxI32 (FX_CALL* GrTriSetupProc)(const void*, const void*, const void*);
-#if GLIDE_PACKED_RBG
-typedef GrTriSetupProc GrTriSetupProcVector[6];
-#else
 typedef GrTriSetupProc GrTriSetupProcVector[2];
+#if GLIDE_PACKED_RGB
+typedef GrTriSetupProcVector GrTriSetupProcArchVector[3];
+#else
+typedef GrTriSetupProcVector GrTriSetupProcArchVector[1];
 #endif
-typedef GrTriSetupProcVector GrTriSetupProcArchVector;
 
 /* Decalrations of the dispatchable procs found in xdraw2.asm and
  * xtexdl.c for teh triangle and texture download procs respectively.  
@@ -631,30 +635,15 @@ typedef struct GrGC_s
   GrState
     state;                      /* state of Glide/SST */
 
-#if GLIDE_DISPATCH_SETUP || GLIDE_DISPATCH_DOWNLOAD
-  struct {
 #if GLIDE_DISPATCH_SETUP
+  struct {
     /* Current triangle rendering proc specialized for culling/no
      * culling and viewport/window coordinates.
      */
-    GrTriSetupProc      triSetupProc;
-    
-    /* Vector to choose triangle rendering proc from based
-     * on culling or no-cull this vector should be specialized
-     * on viewport vs window coordinates.
-     */
-    GrTriSetupProcVector* coorTriSetupVector;
-#endif /* GLIDE_DISPATCH_SETUP */
-    
-#if GLIDE_DISPATCH_DOWNLOAD
-    /* Vector of texture download procs specialized by size
-     * and processor vendor type.
-     */
-    GrTexDownloadProcVector* texDownloadProcs;
-#endif /* GLIDE_DISPATCH_DOWNLOAD */
+    GrTriSetupProc triSetupProc;
   } curArchProcs;
-#endif /* GLIDE_DISPATCH_SETUP || GLIDE_DISPATCH_DOWNLOAD */
-
+#endif /* GLIDE_DISPATCH_SETUP */
+  
   struct cmdTransportInfo {
     FxU32  triPacketHdr; /* Pre-computed packet header for
                           * independent triangles. 
@@ -1017,21 +1006,12 @@ extern GrGCFuncs _curGCFuncs;
 void _grMipMapInit(void);
 
 #if GLIDE_DISPATCH_SETUP
-#define TRISETUP_NORGB(__cullMode) (((__cullMode) == GR_CULL_DISABLE) \
-                                    ? (*gc->curArchProcs.coorTriSetupVector)[0] \
-                                    : (*gc->curArchProcs.coorTriSetupVector)[1])
-#if GLIDE_PACKED_RBG
-#define TRISETUP_RGB(__cullMode) (((__cullMode) == GR_CULL_DISABLE) \
-                                  ? (*gc->curArchProcs.coorTriSetupVector)[2] \
-                                  : (*gc->curArchProcs.coorTriSetupVector)[3])
-#define TRISETUP_ARGB(__cullMode) (((__cullMode) == GR_CULL_DISABLE) \
-                                   ? (*gc->curArchProcs.coorTriSetupVector)[4] \
-                                   : (*gc->curArchProcs.coorTriSetupVector)[5])
-
-#else /* !GLIDE_PACKED_RGB */
-#define TRISETUP_RGB(__cullMode)   TRISETUP_NORGB(__cullMode)
-#define TRISETUP_ARGB(__cullMode)  TRISETUP_NORGB(__cullMode)
+#define TRISETUP_NORGB (*_GlideRoot.deviceArchProcs.curTriProcs + 0)
+#if GLIDE_PACKED_RGB
+#define TRISETUP_RGB   (*_GlideRoot.deviceArchProcs.curTriProcs + 1)
+#define TRISETUP_ARGB  (*_GlideRoot.deviceArchProcs.curTriProcs + 2)
 #endif /* !GLIDE_PACKED_RGB */
+#define PROC_SELECT_TRISETUP(__procVector, __cullMode) (__procVector)[(__cullMode) != GR_CULL_DISABLE]
 #define TRISETUP (*gc->curArchProcs.triSetupProc)
 #else /* !GLIDE_DISPATCH_SETUP */
 FxI32 FX_CSTYLE
@@ -2027,11 +2007,21 @@ _grCVGFifoDump_Linear(const FxU32* const linearPacketAddr);
  * probably do something silly like wrap around zero.
  */
 #if GLIDE_PACKED_RGB
-#define RGBA_COMP(__fpVal, __fpBias, __fpShift, __fpMask) \
+/*#define RGBA_COMP(__fpVal, __fpBias, __fpShift, __fpMask) \
 ((_GlideRoot.pool.ftemp1 = (float)((float)(__fpVal) + (float)(__fpBias))), \
  GR_ASSERT((__fpVal) >= 0.0f), \
  GR_ASSERT((__fpVal) < 256.0f), \
- (((*(const FxU32*)&_GlideRoot.pool.ftemp1) & (__fpMask)) << (__fpShift)))
+ (((*(const FxU32*)&_GlideRoot.pool.ftemp1) & (__fpMask)) << (__fpShift)))*/
+/* dpc - 10 feb 1998 -
+ * Some apps send color values outside of the range [0..255]
+ */
+#define RGBA_COMP(__fpVal, __fpBias, __fpShift, __fpMask) \
+((_GlideRoot.pool.ftemp1 = (float)((float)(__fpVal) + (float)(__fpBias))), \
+ ((((float)(__fpVal)) > 255.0f) \
+  ? (__fpMask) \
+  : ((((float)(__fpVal)) < 0.0f) \
+     ? 0x00UL \
+     : ((*(const FxU32*)&_GlideRoot.pool.ftemp1) & (__fpMask)))) << (__fpShift))
                                                   
 #define RGBA_COMP_CLAMP(__fpVal, __compToken) \
    RGBA_COMP(__fpVal, kPackBias##__compToken, kPackShift##__compToken, kPackMask##__compToken)
diff --git a/glide2x/cvg/glide/src/gaa.c b/glide2x/cvg/glide/src/gaa.c
index d91c87b..23bfab7 100644
--- a/glide2x/cvg/glide/src/gaa.c
+++ b/glide2x/cvg/glide/src/gaa.c
@@ -19,6 +19,9 @@
 **
 ** $Header$
 ** $Log$
+** Revision 1.1.1.1  1999/12/07 21:49:09  joseph
+** Initial checkin into SourceForge.
+**
 ** 
 ** 64    5/18/98 12:15p Peter
 ** crybaby check for alpha enable
@@ -366,19 +369,19 @@ GR_ENTRY(grAADrawPoint, void, (const GrVertex *e))
     a.a =  
       b.a =  _GlideRoot.pool.f0;
       
-    grDrawTriangle(&a, &b, e);    /* A B E */
+    TRISETUP(&a, &b, e);    /* A B E */
       
     b.x -= 2.0F;                  /* compute point D */
     b.y += 2.0F;
-    grDrawTriangle(&a, e, &b);    /* A E D */
+    TRISETUP(&a, e, &b);    /* A E D */
       
     a.x += 2.0F;                  /* compute point C */
     a.y += 2.0F;
-    grDrawTriangle(&b, e, &a);    /* D E C */
+    TRISETUP(&b, e, &a);    /* D E C */
       
     b.x += 2.0F;
     b.y -= 2.0F;
-    grDrawTriangle(&a, e, &b);    /* C E B */
+    TRISETUP(&a, e, &b);    /* C E B */
   }
 #endif /* GLIDE_HW_TRI_SETUP && GLIDE_PACKET3_TRI_SETUP */
   
@@ -663,6 +666,23 @@ GR_ENTRY(grAADrawLine, void, (const GrVertex *v1, const GrVertex *v2))
     v2->a = 0.0f;
     datalist = gc->tsuDataList;
 
+#if GLIDE_PACKED_RGB
+    if ((gc->cmdTransportInfo.paramMask & SSTCP_PKT3_PACKEDCOLOR) != 0) {
+      FxU32 packedColor = 0x00;
+      
+      if (*datalist == (GR_VERTEX_R_OFFSET << 2)) {
+        packedColor = (RGBA_COMP_CLAMP(FARRAY(v2, (GR_VERTEX_B_OFFSET << 2)), B) | 
+                       RGBA_COMP_CLAMP(FARRAY(v2, (GR_VERTEX_G_OFFSET << 2)), G) |
+                       RGBA_COMP_CLAMP(FARRAY(v2, (GR_VERTEX_R_OFFSET << 2)), R));
+        
+        datalist++;
+      }
+      
+      TRI_SET(packedColor);
+      datalist++;
+    }
+#endif
+
     while( *datalist != 0 ) {
         TRI_SETF(FARRAY(v2, *datalist));
         datalist++;
@@ -680,6 +700,23 @@ GR_ENTRY(grAADrawLine, void, (const GrVertex *v1, const GrVertex *v2))
     v1->a = 0.0f;
     datalist = gc->tsuDataList;
 
+#if GLIDE_PACKED_RGB
+    if ((gc->cmdTransportInfo.paramMask & SSTCP_PKT3_PACKEDCOLOR) != 0) {
+      FxU32 packedColor = 0x00;
+      
+      if (*datalist == (GR_VERTEX_R_OFFSET << 2)) {
+        packedColor = (RGBA_COMP_CLAMP(FARRAY(v1, (GR_VERTEX_B_OFFSET << 2)), B) | 
+                       RGBA_COMP_CLAMP(FARRAY(v1, (GR_VERTEX_G_OFFSET << 2)), G) |
+                       RGBA_COMP_CLAMP(FARRAY(v1, (GR_VERTEX_R_OFFSET << 2)), R));
+        
+        datalist++;
+      }
+      
+      TRI_SET(packedColor);
+      datalist++;
+    }
+#endif /* GLIDE_PACKED_RGB */
+
     while( *datalist != 0 ) {
         TRI_SETF(FARRAY(v1, *datalist));
         datalist++;
@@ -692,6 +729,25 @@ GR_ENTRY(grAADrawLine, void, (const GrVertex *v1, const GrVertex *v2))
 
     TRI_SETF( v2->x );
     TRI_SETF( v2->y );
+
+#if GLIDE_PACKED_RGB
+    if ((gc->cmdTransportInfo.paramMask & SSTCP_PKT3_PACKEDCOLOR) != 0) {
+      FxU32 packedColor = 0x00;
+      
+      if (*datalist == (GR_VERTEX_R_OFFSET << 2)) {
+        packedColor = (RGBA_COMP_CLAMP(FARRAY(v2, (GR_VERTEX_B_OFFSET << 2)), B) | 
+                       RGBA_COMP_CLAMP(FARRAY(v2, (GR_VERTEX_G_OFFSET << 2)), G) |
+                       RGBA_COMP_CLAMP(FARRAY(v2, (GR_VERTEX_R_OFFSET << 2)), R));
+        
+        datalist++;
+      }
+      
+      packedColor |= RGBA_COMP_CLAMP(FARRAY(v2, (GR_VERTEX_A_OFFSET << 2)), A);
+      datalist++;
+      
+      TRI_SET(packedColor);
+    }
+#endif /* GLIDE_PACKED_RGB */
       
     while( *datalist != 0 ) {
         TRI_SETF(FARRAY(v2, *datalist));
@@ -703,6 +759,25 @@ GR_ENTRY(grAADrawLine, void, (const GrVertex *v1, const GrVertex *v2))
 
     TRI_SETF( v1->x );
     TRI_SETF( v1->y );
+
+#if GLIDE_PACKED_RGB
+    if ((gc->cmdTransportInfo.paramMask & SSTCP_PKT3_PACKEDCOLOR) != 0) {
+      FxU32 packedColor = 0x00;
+      
+      if (*datalist == (GR_VERTEX_R_OFFSET << 2)) {
+        packedColor = (RGBA_COMP_CLAMP(FARRAY(v1, (GR_VERTEX_B_OFFSET << 2)), B) | 
+                       RGBA_COMP_CLAMP(FARRAY(v1, (GR_VERTEX_G_OFFSET << 2)), G) |
+                       RGBA_COMP_CLAMP(FARRAY(v1, (GR_VERTEX_R_OFFSET << 2)), R));
+        
+        datalist++;
+      }
+      
+      packedColor |= RGBA_COMP_CLAMP(FARRAY(v1, (GR_VERTEX_A_OFFSET << 2)), A);
+      datalist++;
+      
+      TRI_SET(packedColor);
+    }
+#endif /* GLIDE_PACKED_RGB */
       
     while( *datalist != 0 ) {
         TRI_SETF(FARRAY(v1, *datalist));
@@ -719,6 +794,23 @@ GR_ENTRY(grAADrawLine, void, (const GrVertex *v1, const GrVertex *v2))
     v2->a = 0.0f;
     datalist = gc->tsuDataList;
 
+#if GLIDE_PACKED_RGB
+    if ((gc->cmdTransportInfo.paramMask & SSTCP_PKT3_PACKEDCOLOR) != 0) {
+      FxU32 packedColor = 0x00;
+      
+      if (*datalist == (GR_VERTEX_R_OFFSET << 2)) {
+        packedColor = (RGBA_COMP_CLAMP(FARRAY(v2, (GR_VERTEX_B_OFFSET << 2)), B) | 
+                       RGBA_COMP_CLAMP(FARRAY(v2, (GR_VERTEX_G_OFFSET << 2)), G) |
+                       RGBA_COMP_CLAMP(FARRAY(v2, (GR_VERTEX_R_OFFSET << 2)), R));
+        
+        datalist++;
+      }
+      
+      TRI_SET(packedColor);
+      datalist++;
+    }
+#endif /* GLIDE_PACKED_RGB */
+
     while( *datalist != 0 ) {
         TRI_SETF(FARRAY(v2, *datalist));
         datalist++;
@@ -736,6 +828,22 @@ GR_ENTRY(grAADrawLine, void, (const GrVertex *v1, const GrVertex *v2))
     v1->a = 0.0f;
     datalist = gc->tsuDataList;
 
+#if GLIDE_PACKED_RGB
+    if ((gc->cmdTransportInfo.paramMask & SSTCP_PKT3_PACKEDCOLOR) != 0) {
+      FxU32 packedColor = 0x00;
+      
+      if (*datalist == (GR_VERTEX_R_OFFSET << 2)) {
+        packedColor = (RGBA_COMP_CLAMP(FARRAY(v1, (GR_VERTEX_B_OFFSET << 2)), B) | 
+                       RGBA_COMP_CLAMP(FARRAY(v1, (GR_VERTEX_G_OFFSET << 2)), G) |
+                       RGBA_COMP_CLAMP(FARRAY(v1, (GR_VERTEX_R_OFFSET << 2)), R));
+        
+        datalist++;
+      }
+      
+      TRI_SET(packedColor);
+      datalist++;
+    }
+#endif /* GLIDE_PACKED_RGB */
 
     while( *datalist != 0 ) {
         TRI_SETF(FARRAY(v1, *datalist));
@@ -797,10 +905,10 @@ GR_ENTRY(grAADrawLine, void, (const GrVertex *v1, const GrVertex *v2))
     f.y += _GlideRoot.pool.f1;
     f.a = 0.F;
 
-    grDrawTriangle(&a, v2, &b);
-    grDrawTriangle(&a, v2, v1);
-    grDrawTriangle(v1, &f, &e);
-    grDrawTriangle(v1, v2, &f);
+    TRISETUP(&a, v2, &b);
+    TRISETUP(&a, v2, v1);
+    TRISETUP(v1, &f, &e);
+    TRISETUP(v1, v2, &f);
   } else {                      /* Y major line */
     a.x += _GlideRoot.pool.f1;
     a.a = 0.F;
@@ -811,10 +919,10 @@ GR_ENTRY(grAADrawLine, void, (const GrVertex *v1, const GrVertex *v2))
     f.x -= _GlideRoot.pool.f1;
     f.a = 0.F;
 
-    grDrawTriangle(&a, &b, v2);
-    grDrawTriangle(v1, &a, v2);
-    grDrawTriangle(v1, &f, &e);
-    grDrawTriangle(v1, v2, &f);
+    TRISETUP(&a, &b, v2);
+    TRISETUP(v1, &a, v2);
+    TRISETUP(v1, &f, &e);
+    TRISETUP(v1, v2, &f);
   }
 
   gc->state.cull_mode = cullSave;
diff --git a/glide2x/cvg/glide/src/gdraw.c b/glide2x/cvg/glide/src/gdraw.c
index 12d3cf4..36213f5 100644
--- a/glide2x/cvg/glide/src/gdraw.c
+++ b/glide2x/cvg/glide/src/gdraw.c
@@ -19,6 +19,10 @@
  **
  ** $Header$
  ** $Log$
+ ** Revision 1.4.2.1  2004/12/23 20:45:56  koolsmoky
+ ** converted to nasm syntax
+ ** added x86 asm, 3dnow! triangle and mmx, 3dnow! texture download optimizations
+ **
  ** Revision 1.4  2000/01/28 20:52:17  joseph
  ** Changes to support building shared libraries with PIC support.
  **
@@ -781,7 +785,8 @@ GR_ENTRY(grDrawPlanarPolygon,
     /* now all the gradients are loaded into the chip, so we just have to */
     /* draw all the rest of the triangles */
     for (i = i+1; i < nVerts - 1; i++) {
-      _trisetup_nogradients(firstv, &vList[iList[i]], &vList[iList[i+1]]);
+      //_trisetup_nogradients(firstv, &vList[iList[i]], &vList[iList[i+1]]);
+      TRISETUP(firstv, &vList[iList[i]], &vList[iList[i+1]]);
     }
   }
 all_done:
@@ -819,7 +824,8 @@ GR_ENTRY(grDrawPlanarPolygonVertexList, void, (int nVerts, const GrVertex vList[
     /* now all the gradients are loaded into the chip, so we just have to */
     /* draw all the rest of the triangles */
     for (i = i+1; i < nVerts - 1; i++) {
-      _trisetup_nogradients(firstv, &vList[i], &vList[i+1]);
+      //_trisetup_nogradients(firstv, &vList[i], &vList[i+1]);
+      TRISETUP(firstv, &vList[i], &vList[i+1]);
     }
       
 all_done:
@@ -894,25 +900,23 @@ __doPolyVertexSend:
           const int* dataList = gc->tsuDataList;
 
 #if GLIDE_PACKED_RGB
+          if ((gc->cmdTransportInfo.paramMask & SSTCP_PKT3_PACKEDCOLOR) != 0)
           {
-            FxBool doColorP = FXFALSE;
             FxU32 packedColor = 0x00;
           
             if (*dataList == (GR_VERTEX_R_OFFSET << 2)) {
               packedColor = (RGBA_COMP_CLAMP(FARRAY(vertex, (GR_VERTEX_B_OFFSET << 2)), B) | 
                              RGBA_COMP_CLAMP(FARRAY(vertex, (GR_VERTEX_G_OFFSET << 2)), G) |
                              RGBA_COMP_CLAMP(FARRAY(vertex, (GR_VERTEX_R_OFFSET << 2)), R));
-              doColorP = FXTRUE;
               dataList++;
             }
             
             if (*dataList == (GR_VERTEX_A_OFFSET << 2)) {
               packedColor |= RGBA_COMP_CLAMP(FARRAY(vertex, (GR_VERTEX_A_OFFSET << 2)), A);
-              doColorP = FXTRUE;
               dataList++;
             }
             
-            if (doColorP) TRI_SET(packedColor);
+            TRI_SET(packedColor);
           }
 #endif /* GLIDE_PACKED_RGB */
 
@@ -957,7 +961,7 @@ __doPolyVertexSend:
   {
     int i;
     for (i = 1; i < nVerts - 1; i++) {
-      grDrawTriangle(&vList[iList[0]], &vList[iList[i]], &vList[iList[i+1]]);
+      TRISETUP(&vList[iList[0]], &vList[iList[i]], &vList[iList[i+1]]);
     }
   }
 #endif /* !(GLIDE_HW_TRI_SETUP && GLIDE_PACKET3_TRI_SETUP) */
@@ -1016,37 +1020,23 @@ __doPolyVertexSend:
           const int* dataList = gc->tsuDataList;
 
 #if GLIDE_PACKED_RGB
-          /* dpc - 10 feb 1998 -
-           * Some apps send color values outside of the range [0..255]
-           */
-#undef RGBA_COMP
-#define RGBA_COMP(__fpVal, __fpBias, __fpShift, __fpMask) \
-((_GlideRoot.pool.ftemp1 = (float)((float)(__fpVal) + (float)(__fpBias))), \
- ((((float)(__fpVal)) > 255.0f) \
-  ? (__fpMask) \
-  : ((((float)(__fpVal)) < 0.0f) \
-     ? 0x00UL \
-     : ((*(const FxU32*)&_GlideRoot.pool.ftemp1) & (__fpMask)))) << (__fpShift))
-
+          if ((gc->cmdTransportInfo.paramMask & SSTCP_PKT3_PACKEDCOLOR) != 0)
           {
-            FxBool doColorP = FXFALSE;
             FxU32 packedColor = 0x00;
           
             if (*dataList == (GR_VERTEX_R_OFFSET << 2)) {
               packedColor = (RGBA_COMP_CLAMP(FARRAY(vertex, (GR_VERTEX_B_OFFSET << 2)), B) | 
                              RGBA_COMP_CLAMP(FARRAY(vertex, (GR_VERTEX_G_OFFSET << 2)), G) |
                              RGBA_COMP_CLAMP(FARRAY(vertex, (GR_VERTEX_R_OFFSET << 2)), R));
-              doColorP = FXTRUE;
               dataList++;
             }
             
             if (*dataList == (GR_VERTEX_A_OFFSET << 2)) {
               packedColor |= RGBA_COMP_CLAMP(FARRAY(vertex, (GR_VERTEX_A_OFFSET << 2)), A);
-              doColorP = FXTRUE;
               dataList++;
             }
             
-            if (doColorP) TRI_SET(packedColor);
+            TRI_SET(packedColor);
           }
 #endif /* GLIDE_PACKED_RGB */
 
@@ -1091,7 +1081,7 @@ __doPolyVertexSend:
     int i;
     
     for (i = 1; i < nVerts - 1; i++) {
-      grDrawTriangle(&vList[0], &vList[i], &vList[i+1]);
+      TRISETUP(&vList[0], &vList[i], &vList[i+1]);
     }
   }
 #endif /* !(GLIDE_HW_TRI_SETUP && GLIDE_PACKET3_TRI_SETUP) */
diff --git a/glide2x/cvg/glide/src/gglide.c b/glide2x/cvg/glide/src/gglide.c
index 94f9e49..7d3107a 100644
--- a/glide2x/cvg/glide/src/gglide.c
+++ b/glide2x/cvg/glide/src/gglide.c
@@ -19,6 +19,9 @@
 **
 ** $Header$
 ** $Log$
+** Revision 1.2.2.4  2004/12/27 20:46:37  koolsmoky
+** added dll entry point to call grGlideShutdown when a process is detached
+**
 ** Revision 1.2.2.3  2004/12/23 21:03:14  koolsmoky
 ** swapinterval
 **
@@ -797,10 +800,10 @@ GR_ENTRY(grBufferSwap, void, (int swapInterval))
   vSync = (swapInterval > 0);
 
   /* when triple buffering, vsync must be enabled and swapbuffer interval must be 0 */
-  if (gc->grColBuf >= 3) {
+  /*if (gc->grColBuf >= 3) {
     vSync = FXTRUE;
     swapInterval = 0;
-  }
+  }*/
   
   if (swapInterval > 0) swapInterval--;
   
@@ -2149,29 +2152,35 @@ _grUpdateTriPacketHdr(FxU32 paramMask,
    * select on cpu type as well to determine if we should
    * do sw culling or not.
    */
-  if ((paramMask & SSTCP_PKT3_PACKEDCOLOR) == SSTCP_PKT3_PACKEDCOLOR) {
-    const FxU32 colorComp = paramMask & COLOR_COMP_MASK;
+  {
+    GrTriSetupProcVector* curTriProcVector = TRISETUP_NORGB;
 
-    switch(colorComp) {
-    case COLOR_COMP_ARGB:
-      gc->curArchProcs.triSetupProc = TRISETUP_ARGB(cullMode);
-      break;
+#if GLIDE_PACKED_RGB
+    if ((paramMask & SSTCP_PKT3_PACKEDCOLOR) == SSTCP_PKT3_PACKEDCOLOR) {
+      const FxU32 colorComp = paramMask & COLOR_COMP_MASK;
 
-    case COLOR_COMP_RGB:
-      gc->curArchProcs.triSetupProc = TRISETUP_RGB(cullMode);
-      break;
+      switch(colorComp) {
+      case COLOR_COMP_ARGB:
+        curTriProcVector = TRISETUP_ARGB;
+        break;
 
-      /* If no rgb data then it is not worthwhile to pack
-       * just alpha so just mask off the packed color bit
-       * and just use the looping proc.
-       */
-    default:
-      gc->curArchProcs.triSetupProc = TRISETUP_NORGB(cullMode);
-      paramMask &= ~SSTCP_PKT3_PACKEDCOLOR;
-      break;
+      case COLOR_COMP_RGB:
+        curTriProcVector = TRISETUP_RGB;
+        break;
+
+        /* If no rgb data then it is not worthwhile to pack
+         * just alpha so just mask off the packed color bit
+         * and just use the looping proc.
+         */
+      default:
+        curTriProcVector = TRISETUP_NORGB;
+        paramMask &= ~SSTCP_PKT3_PACKEDCOLOR;
+        break;
+      }
     }
-  } else {
-    gc->curArchProcs.triSetupProc = TRISETUP_NORGB(cullMode);
+#endif
+
+    gc->curArchProcs.triSetupProc = PROC_SELECT_TRISETUP(*curTriProcVector, cullMode);
   }
 #endif /* GLIDE_DISPATCH_SETUP */
 
diff --git a/glide2x/cvg/glide/src/glide.rc b/glide2x/cvg/glide/src/glide.rc
index d5562f5..7adc3f8 100644
--- a/glide2x/cvg/glide/src/glide.rc
+++ b/glide2x/cvg/glide/src/glide.rc
@@ -19,6 +19,9 @@
 **
 ** $Header$
 ** $Log$
+** Revision 1.1.1.1.2.1  2004/12/12 15:29:44  koolsmoky
+** cosmetics
+**
 ** Revision 1.1.1.1  1999/12/07 21:49:10  joseph
 ** Initial checkin into SourceForge.
 **
@@ -64,6 +67,7 @@ BEGIN
             VALUE "ProductName", PRODNAME
             VALUE "ProductVersion", VERSIONSTR
             VALUE "Graphics Subsystem", HWSTR
+            VALUE "Contact", "The Glide Open Source Project http://glide.sourceforge.net/"
         END
     END
     BLOCK "VarFileInfo"
diff --git a/glide2x/cvg/glide/src/gpci.c b/glide2x/cvg/glide/src/gpci.c
index 57544aa..91e65db 100644
--- a/glide2x/cvg/glide/src/gpci.c
+++ b/glide2x/cvg/glide/src/gpci.c
@@ -19,6 +19,9 @@
 **
 ** $Header$
 ** $Log$
+** Revision 1.1.1.1.2.5  2005/01/13 16:09:05  koolsmoky
+** Restict calls to pciOpen() pciClose() when compiled with DIRECTX option. this fixes problems with the win32 miniport opened in exclusive mode.
+**
 ** Revision 1.1.1.1.2.4  2005/01/02 04:15:53  koolsmoky
 ** disabled mtrr's on sli slave devices
 **
@@ -245,24 +248,28 @@
 
 #if GLIDE_DISPATCH_SETUP 
 /* Collection of all of the known procs for a given system */
-#if GLIDE_PACKED_RBG
-static GrTriSetupProc _triSetupProcs[][6] =
-#else /* !GLIDE_PACKED_RBG */
-static GrTriSetupProc _triSetupProcs[][2] =
-#endif /* !GLIDE_PACKED_RBG */
+#if GLIDE_PACKED_RGB
+static GrTriSetupProc _triSetupProcs[][3][2] =
+#else /* !GLIDE_PACKED_RGB */
+static GrTriSetupProc _triSetupProcs[][1][2] =
+#endif /* !GLIDE_PACKED_RGB */
 {
   /* Default Procs */
-  { _trisetup, _trisetup_cull
-#if GLIDE_PACKED_RBG
-    , _trisetup_rgb, _trisetup_cull_rgb, _trisetup_argb, _trisetup_cull_argb
-#endif /* GLIDE_PACKED_RBG */
+  {
+    {_trisetup, _trisetup_cull}
+#if GLIDE_PACKED_RGB
+   ,{_trisetup_rgb, _trisetup_cull_rgb},
+    {_trisetup_argb, _trisetup_cull_argb}
+#endif /* GLIDE_PACKED_RGB */
   }
 #if GL_AMD3D
   /* 3DNow!(tm) Procs */
-  ,{ _trisetup_3DNow, _trisetup_cull_3DNow
-#if GLIDE_PACKED_RBG
-    , _trisetup_rgb_3DNow, _trisetup_cull_rgb_3DNow, _trisetup_argb_3DNow, _trisetup_cull_argb_3DNow
-#endif /* GLIDE_PACKED_RBG */
+ ,{
+    {_trisetup_3DNow, _trisetup_cull_3DNow}
+#if GLIDE_PACKED_RGB
+   ,{_trisetup_rgb_3DNow, _trisetup_cull_rgb_3DNow},
+    {_trisetup_argb_3DNow, _trisetup_cull_argb_3DNow}
+#endif /* GLIDE_PACKED_RGB */
   }
 #endif /* GL_AMD3D */
 };
@@ -336,7 +343,7 @@ DllMain(HANDLE hInst, ULONG  ul_reason_for_call, LPVOID lpReserved)
   case DLL_PROCESS_ATTACH:
     if (!pciOpen()) {
       GDBG_INFO(80, "pci bus could not be opened\n");
-      GR_RETURN(FXFALSE);
+      return FALSE;
     }
     GDBG_INFO(80, "DllMain: DLL_PROCESS_ATTACH\n");
     break;
diff --git a/glide2x/cvg/glide/src/gsst.c b/glide2x/cvg/glide/src/gsst.c
index f1bf243..7c449c6 100644
--- a/glide2x/cvg/glide/src/gsst.c
+++ b/glide2x/cvg/glide/src/gsst.c
@@ -19,6 +19,9 @@
 **
 ** $Header$
 ** $Log$
+** Revision 1.2.2.3  2005/01/13 16:09:06  koolsmoky
+** Restict calls to pciOpen() pciClose() when compiled with DIRECTX option. this fixes problems with the win32 miniport opened in exclusive mode.
+**
 ** Revision 1.2.2.2  2004/12/23 20:45:56  koolsmoky
 ** converted to nasm syntax
 ** added x86 asm, 3dnow! triangle and mmx, 3dnow! texture download optimizations
@@ -1362,12 +1365,6 @@ __errSliExit:
   /* We're effectively open now */
   gc->open = FXTRUE;
 
-  /* Setup the procs that we can do w/o any mode knowledge */
-#if GLIDE_DISPATCH_SETUP || GLIDE_DISPATCH_DOWNLOAD
-  gc->curArchProcs.texDownloadProcs  = _GlideRoot.deviceArchProcs.curTexProcs;
-  gc->curArchProcs.coorTriSetupVector = _GlideRoot.deviceArchProcs.curTriProcs;
-#endif /* GLIDE_DISPATCH_SETUP || GLIDE_DISPATCH_DOWNLOAD */
-
   /*------------------------------------------------------
     GC Init
     ------------------------------------------------------*/
diff --git a/glide2x/cvg/glide/src/gtexdl.c b/glide2x/cvg/glide/src/gtexdl.c
index 54546a6..2d9781c 100644
--- a/glide2x/cvg/glide/src/gtexdl.c
+++ b/glide2x/cvg/glide/src/gtexdl.c
@@ -19,6 +19,10 @@
 **
 ** $Header$
 ** $Log$
+** Revision 1.2.2.1  2004/12/23 20:45:56  koolsmoky
+** converted to nasm syntax
+** added x86 asm, 3dnow! triangle and mmx, 3dnow! texture download optimizations
+**
 ** Revision 1.2  2000/10/03 18:28:33  mercury
 ** 003-clean_up_cvg-000, cvg tree cleanup.
 **
@@ -505,10 +509,10 @@ GR_ENTRY(grTexDownloadMipMapLevelPartial,
     
     _GlideRoot.stats.texBytes += max_s * (max_t - t + 1) * 4;
     
-    (*((*gc->curArchProcs.texDownloadProcs)[formatSel][widthSel]))(gc, 
-								   tmuBaseAddr,
-								   max_s, t, max_t,
-								   data);
+    (*((*_GlideRoot.deviceArchProcs.curTexProcs)[formatSel][widthSel]))(gc, 
+                                                                        tmuBaseAddr,
+                                                                        max_s, t, max_t,
+                                                                        data);
   }
 #else
   /*------------------------------------------------------------
@@ -546,7 +550,7 @@ GR_ENTRY(grTexDownloadMipMapLevelPartial,
         LINEAR_WRITE_END();
       
         src8 += width;
-        tex_address += TEX_ROW_ADDR_INCR(1, thisLod);
+        tex_address += TEX_ROW_ADDR_INCR(1);
       }
       break;
 
diff --git a/glide2x/cvg/glide/src/guclip.c b/glide2x/cvg/glide/src/guclip.c
index 8b5d156..0469e67 100644
--- a/glide2x/cvg/glide/src/guclip.c
+++ b/glide2x/cvg/glide/src/guclip.c
@@ -19,6 +19,9 @@
 **
 ** $Header$
 ** $Log$
+** Revision 1.1.1.1  1999/12/07 21:49:11  joseph
+** Initial checkin into SourceForge.
+**
  * 
  * 12    12/15/97 5:52p Atai
  * disable obsolete glide2 api for glide3
@@ -293,9 +296,9 @@ GR_DIENTRY(guDrawTriangleWithClip, void,
       )
   {
 #ifdef GLIDE3
-    grDrawTriangle( (void *)a, (void *)b, (void *)c );
+    TRISETUP( (void *)a, (void *)b, (void *)c );
 #else
-    grDrawTriangle( a, b, c );
+    TRISETUP( a, b, c );
 #endif
     return;
   }
diff --git a/glide2x/cvg/glide/src/gxdraw.c b/glide2x/cvg/glide/src/gxdraw.c
index b208cda..4721c99 100644
--- a/glide2x/cvg/glide/src/gxdraw.c
+++ b/glide2x/cvg/glide/src/gxdraw.c
@@ -19,6 +19,9 @@
  **
  ** $Header$
  ** $Log$
+ ** Revision 1.1.1.1  1999/12/07 21:49:11  joseph
+ ** Initial checkin into SourceForge.
+ **
 ** 
 ** 61    3/17/98 3:00p Peter
 ** removed vertex sorting
@@ -250,8 +253,8 @@ GR_DDFUNC(_trisetup_nogradients,
           const int* dataList = gc->tsuDataList;
 
 #if GLIDE_PACKED_RGB
+          if ((gc->cmdTransportInfo.paramMask & SSTCP_PKT3_PACKEDCOLOR) != 0)
           {
-            FxBool doColorP = FXFALSE;
             FxU32 packedColor = 0x00;
             
             if (*dataList == (GR_VERTEX_R_OFFSET << 2)) {
@@ -259,16 +262,14 @@ GR_DDFUNC(_trisetup_nogradients,
                              RGBA_COMP_CLAMP(FARRAY(vector, (GR_VERTEX_G_OFFSET << 2)), G) |
                              RGBA_COMP_CLAMP(FARRAY(vector, (GR_VERTEX_R_OFFSET << 2)), R));
               dataList++;
-              doColorP = FXTRUE;
             }
 
             if (*dataList == (GR_VERTEX_A_OFFSET << 2)) {
               packedColor |= RGBA_COMP_CLAMP(FARRAY(vector, (GR_VERTEX_A_OFFSET << 2)), A);
               dataList++;
-              doColorP = FXTRUE;
             }
             
-            if (doColorP) TRI_SET(packedColor);
+            TRI_SET(packedColor);
           }
 #endif /* GLIDE_PACKED_RGB */
 
diff --git a/glide2x/cvg/glide/src/xdraw2.asm b/glide2x/cvg/glide/src/xdraw2.asm
index 37828d0..801ee0e 100644
--- a/glide2x/cvg/glide/src/xdraw2.asm
+++ b/glide2x/cvg/glide/src/xdraw2.asm
@@ -19,6 +19,9 @@
 ;; $Header$
 ;; $Revision$
 ;; $Log$
+;; Revision 1.1.1.1.2.2  2005/01/13 16:11:39  koolsmoky
+;; prepare for packed rgb
+;;
 ;; Revision 1.1.1.1.2.1  2004/12/23 20:45:56  koolsmoky
 ;; converted to nasm syntax
 ;; added x86 asm, 3dnow! triangle and mmx, 3dnow! texture download optimizations
@@ -65,8 +68,8 @@ extrn   _FifoMakeRoom
 
 %ifdef GL_AMD3D
 ;; 3dnow!
-%MACRO WRITE_MM1_FIFO_ALIGNED 1
-    movq      [fifo+%1], mm1        ; store current param | previous param
+%MACRO WRITE_MM1_FIFO_ALIGNED 0
+    movq      [fifo], mm1           ; store current param | previous param
 %ENDMACRO ; WRITE_MM1_FIFO_ALIGNED
 
 %MACRO WRITE_MM1LOW_FIFO 0
diff --git a/glide2x/cvg/glide/src/xdraw2.inc b/glide2x/cvg/glide/src/xdraw2.inc
index 6f606d8..27620a8 100644
--- a/glide2x/cvg/glide/src/xdraw2.inc
+++ b/glide2x/cvg/glide/src/xdraw2.inc
@@ -20,6 +20,9 @@
 ;; $Header$
 ;; $Revision$
 ;; $Log$
+;; Revision 1.1.1.1.2.2  2005/01/13 16:11:39  koolsmoky
+;; prepare for packed rgb
+;;
 ;; Revision 1.1.1.1.2.1  2004/12/23 20:45:56  koolsmoky
 ;; converted to nasm syntax
 ;; added x86 asm, 3dnow! triangle and mmx, 3dnow! texture download optimizations
@@ -45,39 +48,36 @@
 ;;--------------------------------------------------------------------------
 ;; start AMD3D version
 ;;--------------------------------------------------------------------------
-%define gc      edi           ; points to graphics context
-%define fifo    ebp           ; points to fifo entries
+%define gc      edi                 ; points to graphics context
+%define fifo    ebp                 ; points to fifo entries
 %define tempVal esi
 
+%IF GLIDE_CULLING
+%define fa      eax                 ; vtx a from caller
+%define fb      ebx                 ; vtx b from caller
+%define fc      ecx                 ; vtx c from caller
+%define cull    edx                 ; cull mode
+%define intArea ecx                 ; area temp storage
+
     ;; Prologue stuff
     push      edi                   ; save caller's register variable
-    mov       gc,[_GlideRoot+curGC]; GR_DCL_GC
+    mov       gc, [_GlideRoot+curGC]; GR_DCL_GC
+
+    push      esi                   ; save caller's register variable
+    mov       fa, [esp + _va$]      ; get base address of vertex A
+
+    push      ebx                   ; save caller's register variable        
+    mov       fb, [esp + _vb$]      ; get base address of vertex B
 
     push      ebp                   ; save frame pointer
-    push      ebx                   ; save caller's register variable        
-    
-%IF GLIDE_CULLING
-%define fa      eax           ; vtx a from caller
-%define fb      ebx           ; vtx b from caller
-%define fc      ecx           ; vtx c from caller
-    
-%define cull    edx           ; cull mode
-%define intArea ecx           ; area temp storage
-    
-    mov       fb, [esp + _vb$ - 4]  ; get base address of vertex B
-    push      esi                   ; save caller's register variable
-    
     mov       cull, [gc + cull_mode]; get cull mode
+
     mov       fc, [esp + _vc$]      ; get base address of vertex C
+    mov       tempVal, [_GlideRoot+curTriSize]
 
     femms                           ; will use AMD3D, clear FPU/MMX registers
 
-    cmp       cull, 0               ; culling enabled ?
-    mov       tempVal, [_GlideRoot + curTriSize]
-
     ;; Cull Check
-    jz        .nocull                ; nope, no culling
-    mov       fa, [esp + _va$]      ; get base address of vertex A
 
     movq      mm2, [fc + X]         ; yc | xc
     shl       cull, 31              ; culltest << 31
@@ -115,63 +115,45 @@
     ; Zero Area Triangle Check
 
     test      intArea, 7fffffffh    ; if ((j & 0x7FFFFFFF) == 0)
-    jz        .__cullFail            ; area zero, triangle culled
+    jz        .__cullFail           ; area zero, triangle culled
 
     xor       intArea, cull         ; if (j ^ (culltest << 31))
-    jge       .__cullFail            ; triangle facing away from viewer, culled
+    jge       .__cullFail           ; triangle facing away from viewer, culled
 
     cmp       ebx, tempVal          ; fifo space required >= space available ?
-    jge       .__triBegin            ; yup, push out triangle data to Voodoo
+    jge       .__triBegin           ; yup, push out triangle data to Voodoo
 
     push      __LINE__              ; line number inside this function
     push      0h                    ; pointer to function name = NULL
 
     push      tempVal               ; fifo space required
-    call      _FifoMakeRoom        ; note: updates fifoPtr
-
-    add       esp, 12               ; remove 3 DWORD arguments from stack
-    jmp       .__triBegin            ; merge back with short path
-
-    ;; culling disabled
-
-.nocull:
-    ;; Check to make sure that we have enough room for
-    ;; the complete triangle packet.
-
-    add       tempVal, 4            ; fifo space needed
-    mov       ebx, [gc + fifoRoom]  ; fifo space available
-
-    cmp       ebx, tempVal          ; fifo spce available >= space needed ?
-    jge       .__triBegin            ; yup, ready to draw triangle
-
-    push      __LINE__              ; line number inside this function
-    push      0h                    ; pointer to function name = NULL
-
-    push      tempVal               ; fifo space needed
-    call      _FifoMakeRoom        ; note: updates fifoPtr
+    call      _FifoMakeRoom         ; note: updates fifoPtr
 
     add       esp, 12               ; remove 3 DWORD arguments from stack
     nop                             ; filler
 
 %ELSE   ; !GLIDE_CULLING
 
-    ;;;;;;;;;;;;;;;lea       eax, [esp+ _va$]      ; pointer to vertex pointers    
+    push      edi                   ; save caller's register variable
+    mov       gc, [_GlideRoot+curGC]; GR_DCL_GC
+
     push      esi                   ; save caller's register variable
-    
-    mov       tempVal, [_GlideRoot + curTriSize] ; data for whole triangle in bytes
+    mov       tempVal, [_GlideRoot+curTriSize] ; data for whole triangle in bytes
+
+    push      ebx                   ; save caller's register variable
     mov       ebx, [gc + fifoRoom]  ; fifo space available
 
+    push      ebp                   ; save frame pointer
     add       tempVal, 4            ; fifo space needed (include 4-byte header)
-    femms                           ; will use AMD3D, clear FPU/MMX registers
 
     cmp       ebx, tempVal          ; fifo spce available >= space needed ?
-    jge       .__triBegin            ; yup, ready to draw triangle
+    jge       .__triBegin           ; yup, ready to draw triangle
 
     push      __LINE__              ; line number inside this function
     push      0h                    ; pointer to function name = NULL
 
     push      tempVal               ; fifo space needed
-    call      _FifoMakeRoom        ; note: updates fifoPtr
+    call      _FifoMakeRoom         ; note: updates fifoPtr
 
     add       esp, 12               ; remove 3 DWORD arguments from stack
     nop                             ; filler
@@ -181,7 +163,8 @@
 %define dlp     ebx           ; points to dataList structure
 %define dlpstrt ecx           ; points to begin of dataList structure
 %define vertex  edx           ; the current vertex
-%define packCol esi
+
+   ALIGN 32
 
 .__triBegin:
     mov       eax, [gc+triPacketHdr]; Packet 3 header
@@ -193,14 +176,14 @@
     mov       dlpstrt, dlp          ; save pointer to start of dataList
     test      fifo, 4               ; is fifo pointer qword aligned ?
 
-    jz        .__fifo_aligned        ; yes, it is qword aligned
+    jz        .__fifo_aligned       ; yes, it is qword aligned
     movq      mm1, [vertex+X]       ; y | x
 
     GR_FIFO_WRITE fifo, 0, eax      ; write header to fifo; now qword aligned
-    add       fifo, 12              ; fifoPtr += 3*sizeof(FxU32)
+    add       fifo, 4               ; advance fifo for hdr; now qword aligned
 
-    WRITE_MM1_FIFO_ALIGNED -8       ; PCI write y | x
-    nop                             ; filler
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write y | x
+    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
 
 %IF GLIDE_PACK_RGB
 %IF GLIDE_PACK_ALPHA
@@ -244,35 +227,37 @@
     add       dlp, 4                ; dlp++
 
     test      eax, eax              ; end of list ?
-    jz        .__paramLoopDoneWBone1 ; yes, one DWORD in "write buffer"
+    jz        .__paramLoopDoneWBone1; yes, one DWORD in "write buffer"
           
 .__paramLoop1a:
     movd      mm2, [eax+vertex]     ; get next parameter
+    mov       eax, [dlp]            ; offset = *(dlp + 1)
+
+    add       dlp, 4                ; dlp++
+    punpckldq mm1, mm2              ; current param | previous param
+
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write current param | previous param
     add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
 
-    mov       eax, [dlp]            ; offset = *(dlp + 1)
-    add       dlp, 8                ; dlp += 2
-
-    punpckldq mm1, mm2              ; current param | previous param
     test      eax, eax              ; at end of offset list (offset == 0) ?
-  
-    WRITE_MM1_FIFO_ALIGNED -8       ; PCI write current param | previous param
     jz        .__paramLoopDoneWBzero1; exit, "write buffer" empty
 
     movd      mm1, [eax+vertex]     ; get next parameter
-    mov       eax, [dlp-4]          ; offset = *(dlp + 1)
+    mov       eax, [dlp]            ; offset = *(dlp + 1)
 
+    add       dlp, 4                ; dlp++
     test      eax, eax              ; at end of offset list (offset == 0) ?
-    jnz       .__paramLoop1a         ; nope, copy next parameter
+
+    jnz       .__paramLoop1a        ; nope, copy next parameter
 
     nop                             ; filler
-    jmp       .__paramLoopDoneWBone1 ; merge back into common stream
+    jmp       .__paramLoopDoneWBone1; merge back into common stream
 
 %ELSE ; ! GLIDE_PACK_RGB
 
     ;; here: "write buffer" empty
 
-    mov       eax, [dlp]            ; Get first offset from the data list
+    mov       eax,[dlp]             ; Get first offset from the data list
     test      eax, eax              ; at end of list ?
 
     lea       dlp, [dlp+4]          ; dlp++
@@ -282,20 +267,22 @@
     movd      mm1, [eax+vertex]     ; get next parameter
     mov       eax, [dlp]            ; offset = *(dlp + 1)
 
+    add       dlp, 4                ; dlp++
     test      eax, eax              ; at end of offset list (offset == 0) ?
-    jz        .__paramLoopDoneWBone1 ; exit, write buffer contains one DWORD
 
+    jz        .__paramLoopDoneWBone1; exit, write buffer contains one DWORD
     movd      mm2, [eax+vertex]     ; get next parameter
-    add       dlp, 8                ; dlp += 2
 
-    mov       eax, [dlp-4]          ; offset = *(dlp + 1)
-    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
+    mov       eax, [dlp]            ; offset = *(dlp + 1)
+    add       dlp, 4                ; dlp++
 
     punpckldq mm1, mm2              ; current param | previous param
-    test      eax, eax              ; at end of offset list (offset == 0) ?
 
-    WRITE_MM1_FIFO_ALIGNED -8       ; PCI write current param | previous param
-    jnz       .__paramLoop1a         ; nope, copy next parameter
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write current param | previous param
+    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
+
+    test      eax, eax              ; at end of offset list (offset == 0) ?
+    jnz       .__paramLoop1a        ; nope, copy next parameter
 
     nop                             ; filler
     jmp       .__paramLoopDoneWBzero1; write buffer empty
@@ -304,12 +291,12 @@
 
 .__fifo_aligned:
     movd      mm2, [vertex+X]       ; y | x of vertex A
-    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
-
     movd      mm1, [gc+triPacketHdr]; Packet 3 header
-    punpckldq mm1, mm2              ; x | header
 
-    WRITE_MM1_FIFO_ALIGNED -8       ; PCI write x | header
+    punpckldq mm1, mm2              ; x | header
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write x | header
+
+    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
     movd      mm1, [vertex+Y]       ; 0 | y of vertex A
 
 %IF GLIDE_PACK_RGB
@@ -351,7 +338,7 @@
     punpckldq mm1, mm4              ; RGB(A) | y
     mov       eax, [dlp]            ; get first offset from the data list
 
-    WRITE_MM1_FIFO_ALIGNED 0        ; PCI write y | RGB(A)
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write y | RGB(A)
     add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
 
     add       dlp, 4                ; dlp++
@@ -368,15 +355,17 @@
     jz        .__paramLoopDoneWBone1 ; exit, write buffer contains one DWORD
 
     movd      mm2, [eax+vertex]     ; get next parameter
-    add       dlp, 8                ; dlp += 2
+    add       dlp, 4                ; dlp++
 
-    mov       eax, [dlp-4]          ; offset = *(dlp + 1)
-    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
+    mov       eax, [dlp]            ; offset = *(dlp + 1)
+    add       dlp, 4                ; dlp++
 
     punpckldq mm1, mm2              ; current param | previous param
     test      eax, eax              ; at end of offset list (offset == 0) ?
 
-    WRITE_MM1_FIFO_ALIGNED -8       ; PCI write current param | previous param
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write current param | previous param
+    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
+
     jnz       .__paramLoop1b         ; nope, copy next parameter
 
     nop                             ; filler
@@ -386,8 +375,8 @@
     mov       eax, [dlp]            ; get first offset from the data list
     add       dlp, 4                ; dlp++
 
-    cmp       eax, 0                ; end of list ?
-    jz        .__paramLoopDoneWBone1 ; yes, "write buffer" has y data
+    test      eax, eax              ; end of list ?
+    jz        .__paramLoopDoneWBone1; yes, "write buffer" has y data
 
 .__paramLoop1b:
     movd      mm2, [eax+vertex]     ; get next parameter
@@ -398,15 +387,20 @@
 
     punpckldq mm1, mm2              ; current param | previous param
     test      eax, eax              ; at end of offset list (offset == 0) ?
-  
-    WRITE_MM1_FIFO_ALIGNED -8       ; PCI write current param | previous param
+
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write current param | previous param
+    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
+
+    test      eax, eax              ; at end of offset list (offset == 0) ?
     jz        .__paramLoopDoneWBzero1; exit, "write buffer" empty
 
     movd      mm1, [eax+vertex]     ; get next parameter
-    mov       eax, [dlp-4]          ; offset = *(dlp + 1)
+    mov       eax, [dlp]            ; offset = *(dlp + 1)
+
+    add       dlp, 4                ; dlp++
 
     test      eax, eax              ; at end of offset list (offset == 0) ?
-    jnz       .__paramLoop1b         ; nope, copy next parameter
+    jnz       .__paramLoop1b        ; nope, copy next parameter
 %ENDIF
 
 .__paramLoopDoneWBone1:
@@ -419,7 +413,7 @@
     movd      mm2, [vertex+X]       ; 0 | x if vertex B
     punpckldq mm1, mm2              ; x | old param
 
-    WRITE_MM1_FIFO_ALIGNED 0        ; PCI write: x | old param
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write: x | old param
     add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
 
     movd      mm1, [vertex+Y]       ; 0 | y of vertex B
@@ -464,7 +458,7 @@
     punpckldq mm1, mm4              ; RGB(A) | y
     mov       eax, [dlp]            ; get first offset from the data list
 
-    WRITE_MM1_FIFO_ALIGNED 0        ; PCI write y | RGB(A)
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write y | RGB(A)
     add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
 
     add       dlp, 4                ; dlp++
@@ -478,19 +472,21 @@
     mov       eax, [dlp]            ; offset = *(dlp + 1)
 
     test      eax, eax              ; at end of offset list (offset == 0) ?
-    jz        .__paramLoopDoneWBone2 ; exit, write buffer contains one DWORD
+    jz        .__paramLoopDoneWBone2; exit, write buffer contains one DWORD
 
     movd      mm2, [eax+vertex]     ; get next parameter
-    add       dlp, 8                ; dlp += 2
+    add       dlp, 4                ; dlp++
 
-    mov       eax, [dlp-4]          ; offset = *(dlp + 1)
-    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
+    mov       eax, [dlp]            ; offset = *(dlp + 1)
+    add       dlp, 4                ; dlp++
 
     punpckldq mm1, mm2              ; current param | previous param
     test      eax, eax              ; at end of offset list (offset == 0) ?
 
-    WRITE_MM1_FIFO_ALIGNED -8       ; PCI write current param | previous param
-    jnz       .__paramLoop2b         ; nope, copy next parameter
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write current param | previous param
+    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
+
+    jnz       .__paramLoop2b        ; nope, copy next parameter
 
     nop                             ; filler
     jmp       .__paramLoopDoneWBzero2; write buffer empty
@@ -504,25 +500,26 @@
 
 .__paramLoop2b:
     movd      mm2, [eax+vertex]     ; get next parameter
+    mov       eax, [dlp]            ; offset = *(dlp + 1)
+
+    add       dlp, 4                ; dlp++
+    punpckldq mm1, mm2              ; current param | previous param
+
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write current param | previous param
     add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
 
-    mov       eax, [dlp]            ; offset = *(dlp + 1)
-    add       dlp, 8                ; dlp += 2
-
-    punpckldq mm1, mm2              ; current param | previous param
     test      eax, eax              ; at end of offset list (offset == 0) ?
-  
-    WRITE_MM1_FIFO_ALIGNED -8       ; PCI write current param | previous param
     jz        .__paramLoopDoneWBzero2; exit, "write buffer" empty
 
     movd      mm1, [eax+vertex]     ; get next parameter
-    mov       eax, [dlp-4]          ; offset = *(dlp + 1)
+    mov       eax, [dlp]            ; offset = *(dlp + 1)
+
+    add       dlp, 4                ; dlp++
 
     test      eax, eax              ; at end of offset list (offset == 0) ?
-    jnz       .__paramLoop2b         ; nope, copy next parameter
+    jnz       .__paramLoop2b        ; nope, copy next parameter
 
-    nop                             ; filler
-    jmp       .__paramLoopDoneWBone2 ; write buffer contains one DWORD
+    jmp       .__paramLoopDoneWBone2; write buffer contains one DWORD
 %ENDIF
 
 
@@ -532,10 +529,9 @@
     mov       dlp, dlpstrt          ; Reset the dataList
 
     movq      mm1, [vertex+X]       ; y | x of vertex B
-    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write y | x of vertex B
 
-    WRITE_MM1_FIFO_ALIGNED -8       ; PCI write y | x of vertex B
-    nop                             ; filler
+    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
 
 %IF GLIDE_PACK_RGB
 %IF GLIDE_PACK_ALPHA
@@ -583,21 +579,23 @@
           
 .__paramLoop2a:
     movd      mm2, [eax+vertex]     ; get next parameter
+    mov       eax, [dlp]            ; offset = *(dlp + 1)
+
+    add       dlp, 4                ; dlp++
+    punpckldq mm1, mm2              ; current param | previous param
+
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write current param | previous param
     add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
 
-    mov       eax, [dlp]            ; offset = *(dlp + 1)
-    add       dlp, 8                ; dlp += 8
-
-    punpckldq mm1, mm2              ; current param | previous param
     test      eax, eax              ; at end of offset list (offset == 0) ?
-
-    WRITE_MM1_FIFO_ALIGNED -8       ; PCI write current param | previous param
     jz        .__paramLoopDoneWBzero2; exit, "write buffer" empty
 
     movd      mm1, [eax+vertex]     ; get next parameter
-    mov       eax, [dlp-4]          ; offset = *(dlp + 1)
+    mov       eax, [dlp]            ; offset = *(dlp + 1)
 
+    add       dlp, 4                ; dlp++
     test      eax, eax              ; at end of offset list (offset == 0) ?
+
     jnz       .__paramLoop2a         ; nope, copy next parameter
 
     nop                             ; filler
@@ -617,20 +615,22 @@
     movd      mm1, [eax+vertex]     ; get next parameter
     mov       eax, [dlp]            ; offset = *(dlp + 1)
 
+    add       dlp, 4                ; dlp++
     test      eax, eax              ; at end of offset list (offset == 0) ?
+
     jz        .__paramLoopDoneWBone2 ; exit, write buffer contains one DWORD
-
     movd      mm2, [eax+vertex]     ; get next parameter
-    add       dlp, 8                ; dlp += 2
 
-    mov       eax, [dlp-4]          ; offset = *(dlp + 1)
-    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
+    mov       eax, [dlp]            ; offset = *(dlp + 1)
+    add       dlp, 4                ; dlp++
 
     punpckldq mm1, mm2              ; current param | previous param
-    test      eax, eax              ; at end of offset list (offset == 0) ?
 
-    WRITE_MM1_FIFO_ALIGNED -8       ; PCI write current param | previous param
-    jnz       .__paramLoop2a         ; nope, copy next parameter
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write current param | previous param
+    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
+
+    test      eax, eax              ; at end of offset list (offset == 0) ?
+    jnz       .__paramLoop2a        ; nope, copy next parameter
 
 %ENDIF ; GLIDE_PACK_RGB
 
@@ -641,12 +641,10 @@
     mov       dlp, dlpstrt          ; Reset the dataList
 
     movq      mm1, [vertex+X]       ; y | x of vertex C
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write y | x of vertex C
+
     add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
 
-    WRITE_MM1_FIFO_ALIGNED -8       ; PCI write y | x of vertex C
-    nop                             ; filler
-
-
 %IF GLIDE_PACK_RGB
 %IF GLIDE_PACK_ALPHA
     ;; assumes color and alpha values < 256.0
@@ -693,25 +691,27 @@
           
 .__paramLoop3a:
     movd      mm2, [eax+vertex]     ; get next parameter
-    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
 
     mov       eax, [dlp]            ; offset = *(dlp + 1)
-    add       dlp, 8                ; dlp += 2
+    add       dlp, 4                ; dlp++
 
-    test      eax, eax              ; at end of offset list (offset == 0) ?
     punpckldq mm1, mm2              ; current param | previous param
-  
-    WRITE_MM1_FIFO_ALIGNED -8       ; PCI write current param | previous param
-    jz        .__paramLoopDoneWBzero3; exit, "write buffer" empty
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write current param | previous param
 
+    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
+    test      eax, eax              ; at end of offset list (offset == 0) ?
+
+    jz        .__paramLoopDoneWBzero3; exit, "write buffer" empty
     movd      mm1, [eax+vertex]     ; get next parameter
-    mov       eax, [dlp-4]          ; offset = *(dlp + 1)
+
+    mov       eax, [dlp]            ; offset = *(dlp + 1)
+    add       dlp, 4                ; dlp++
 
     test      eax, eax              ; at end of offset list (offset == 0) ?
-    jnz       .__paramLoop3a         ; nope, copy next parameter
+    jnz       .__paramLoop3a        ; nope, copy next parameter
 
     nop                             ; filler
-    jmp       .__paramLoopDoneWBone3 ; merge back into common stream
+    jmp       .__paramLoopDoneWBone3; merge back into common stream
 
 %ELSE ; ! GLIDE_PACK_RGB
 
@@ -727,22 +727,23 @@
     movd      mm1, [eax+vertex]     ; get next parameter
     mov       eax, [dlp]            ; offset = *(dlp + 1)
 
+    add       dlp, 4                ; dlp++
     test      eax, eax              ; at end of offset list (offset == 0) ?
-    jz        .__paramLoopDoneWBone3 ; exit, write buffer contains one DWORD
 
+    jz        .__paramLoopDoneWBone3; exit, write buffer contains one DWORD
     movd      mm2, [eax+vertex]     ; get next parameter
-    add       dlp, 8                ; dlp += 2
 
-    mov       eax, [dlp-4]          ; offset = *(dlp + 1)
-    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
+    mov       eax, [dlp]            ; offset = *(dlp + 1)
+    add       dlp, 4                ; dlp++
 
     punpckldq mm1, mm2              ; current param | previous param
+
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write current param | previous param
+    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
+
     test      eax, eax              ; at end of offset list (offset == 0) ?
+    jnz       .__paramLoop3a        ; nope, copy next parameter
 
-    WRITE_MM1_FIFO_ALIGNED -8       ; PCI write current param | previous param
-    jnz       .__paramLoop3a         ; nope, copy next parameter
-
-    mov       esp, esp              ; filler
     jmp       .__paramLoopDoneWBzero3; write buffer empty
 
 %ENDIF ; GLIDE_PACK_RGB
@@ -758,11 +759,10 @@
     movd      mm2, [vertex+X]       ; 0 | x if vertex C
     punpckldq mm1, mm2              ; x | old param
 
-    WRITE_MM1_FIFO_ALIGNED 0        ; PCI write: x | old param
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write: x | old param
     add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
 
     movd      mm1, [vertex+Y]       ; 0 | y of vertex C
-    mov       esp, esp              ; filler
 
 %IF GLIDE_PACK_RGB
 %IF GLIDE_PACK_ALPHA
@@ -801,35 +801,37 @@
 %ENDIF ; !GLIDE_PACK_ALPHA
 
     punpckldq mm1, mm4              ; RGB(A) | y
-    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
-
     mov       eax, [dlp]            ; get first offset from the data list
+
     add       dlp, 4                ; dlp++
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write y | RGB(A)
 
-    WRITE_MM1_FIFO_ALIGNED -8       ; PCI write y | RGB(A)
-    nop                             ; filler
-
+    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
     test      eax, eax              ; end of list ?
+
     jz        .__paramLoopDoneWBzero3; yes, "write buffer" is empty
+    nop                             ; filler
 
 .__paramLoop3b:
     movd      mm1, [eax+vertex]     ; get next parameter
     mov       eax, [dlp]            ; offset = *(dlp + 1)
 
     test      eax, eax              ; at end of offset list (offset == 0) ?
-    jz        .__paramLoopDoneWBone3 ; exit, write buffer contains one DWORD
+    jz        .__paramLoopDoneWBone3; exit, write buffer contains one DWORD
 
     movd      mm2, [eax+vertex]     ; get next parameter
-    add       dlp, 8                ; dlp++
+    add       dlp, 4                ; dlp++
 
-    mov       eax, [dlp-4]          ; offset = *(dlp + 1)
-    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
+    mov       eax, [dlp]            ; offset = *(dlp + 1)
+    add       dlp, 4                ; dlp++
 
     punpckldq mm1, mm2              ; current param | previous param
     test      eax, eax              ; at end of offset list (offset == 0) ?
 
-    WRITE_MM1_FIFO_ALIGNED -8       ; PCI write current param | previous param
-    jnz       .__paramLoop3b         ; nope, copy next parameter
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write current param | previous param
+    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
+
+    jnz       .__paramLoop3b        ; nope, copy next parameter
 
     nop                             ; filler
     jmp       .__paramLoopDoneWBzero3; write buffer empty
@@ -840,26 +842,28 @@
     add       dlp, 4                ; dlp++
 
     test      eax, eax              ; end of list ?
-    jz        .__paramLoopDoneWBone3 ; yes, "write buffer" has y data
+    jz        .__paramLoopDoneWBone3; yes, "write buffer" has y data
 
 .__paramLoop3b:
     movd      mm2, [eax+vertex]     ; get next parameter
-    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
-
     mov       eax, [dlp]            ; offset = *(dlp + 1)
-    add       dlp, 8                ; dlp += 2
 
     punpckldq mm1, mm2              ; current param | previous param
-    cmp       eax, 0                ; at end of offset list (offset == 0) ?
-  
-    WRITE_MM1_FIFO_ALIGNED -8       ; PCI write current param | previous param
+    add       dlp, 4                ; dlp++
+
+    WRITE_MM1_FIFO_ALIGNED          ; PCI write current param | previous param
+    add       fifo, 8               ; fifoPtr += 2*sizeof(FxU32)
+
+    test      eax, eax              ; at end of offset list (offset == 0) ?
     jz        .__paramLoopDoneWBzero3; exit, "write buffer" empty
 
     movd      mm1, [eax+vertex]     ; get next parameter
-    mov       eax, [dlp-4]          ; offset = *(dlp + 1)
+    mov       eax, [dlp]            ; offset = *(dlp + 1)
+
+    add       dlp, 4                ; dlp++
 
     test      eax, eax              ; at end of offset list (offset == 0) ?
-    jnz       .__paramLoop3b         ; nope, copy next parameter
+    jnz       .__paramLoop3b        ; nope, copy next parameter
 %ENDIF
 
 .__paramLoopDoneWBone3:
@@ -873,7 +877,7 @@
 
     ;; Update gc->fifoPtr and gc->fifoRoom
 
-    mov       ecx, [_GlideRoot + trisDrawn]    ; _GlideRoot.stats.trisDrawn
+    mov       ecx, [_GlideRoot + trisDrawn]    	; _GlideRoot.stats.trisDrawn
     mov       eax, fifo                         ; new fifo pointer
     
     mov       ebx, [gc + fifoPtr]               ; old fifo pointer
@@ -885,7 +889,7 @@
     mov       ebp, [_GlideRoot + trisProcessed]; _GlideRoot.stats.trisProcessed
     sub       eax, ebx                          ; new fifo ptr - old fifo ptr = additional fifo space used
 
-    mov       [_GlideRoot + trisDrawn], ecx    ;
+    mov       [_GlideRoot + trisDrawn], ecx    	;
     sub       edx, eax                          ; new fifo space available
 
     mov       eax, 1h                           ; return value = triangle drawn
@@ -893,35 +897,35 @@
 
     ;; Restore trashed registers    
 
-    inc       ebp                   ; _GlideRoot.stats.trisProcessed++
-    pop       esi                   ; restore caller's register variable
+    inc       ebp				; _GlideRoot.stats.trisProcessed++
+    pop       esi				; restore caller's register variable
 
-    mov       [_GlideRoot + trisProcessed], ebp ;
-    pop       ebx                   ; restore caller's register variable
+    mov       [_GlideRoot + trisProcessed], ebp	;
+    pop       ebx				; restore caller's register variable
 
-    pop       ebp                   ; restore frame pointer
-    pop       edi                   ; restore caller's register variable
+    pop       ebp				; restore frame pointer
+    pop       edi				; restore caller's register variable
 
-    femms                           ; no more AMD3D code, clear FPU/MMX regs
+    femms					; no more AMD3D code, clear FPU/MMX regs
 
-    ret                             ; return to caller
+    ret						; return to caller
 
 
 %IF GLIDE_CULLING
 .__cullFail:
-    mov       ebp, [_GlideRoot + trisProcessed]; triangles processed so far
-    xor       eax, eax              ; return value = triangle not drawn
+    mov       ebp, [_GlideRoot + trisProcessed]	; triangles processed so far
+    xor       eax, eax				; return value = triangle not drawn
 
-    femms                           ; no more AMD3D code, clear FPU/MMX regs
+    femms					; no more AMD3D code, clear FPU/MMX regs
 
     ;; Restore trashed registers    
-    inc       ebp                   ; _GlideRoot.stats.trisProcessed++;    
+    inc       ebp				; _GlideRoot.stats.trisProcessed++;    
     pop       esi
 
     mov       [_GlideRoot + trisProcessed], ebp
     pop       ebx
     
-    pop       ebp                   ; restore frame pointer
+    pop       ebp				; restore frame pointer
     pop       edi
 
     ret
@@ -968,9 +972,6 @@
     mov     cull, [gc + cull_mode]    
     mov     fc, [esp + _vc$]
 
-    test    cull, cull
-    jz      .nocull
-
     shl     cull, 31                    ; culltest << 31    
         
 ;Area_Computation:    
@@ -1011,7 +1012,6 @@
     
     jge	    .__triDone
 
-.nocull:
 %ENDIF ; GLIDE_CULLING    
 
 	    align 4
@@ -1062,39 +1062,39 @@
     nop						; Avoid p5 agi w/ load of vertex ptr
     nop
     
-    mov	    eax, dword [vertex]		; X
+    mov	    eax, [vertex + X]		        ; X
     lea	    dlp, [gc + tsuDataList]		; Reset the dataList
 
     GR_FIFO_WRITE fifo, -8, eax			; PCI write X
-    mov	    eax, dword [vertex + 4]		; Y 
+    mov	    eax, [vertex + Y]			; Y 
 
     xor	    packCol, packCol			; Clear packed color
     GR_FIFO_WRITE fifo, -4, eax			; PCI write Y
 
 %IF GLIDE_PACK_RGB
-    fld	    dword [vertex + b]		; B
+    fld	    dword [vertex + b]		        ; B
     fadd    dword [_GlideRoot + fBiasLo]	; BC GC
 
-    fld	    dword [vertex + g]		; G B
+    fld	    dword [vertex + g]		        ; G B
     fadd    dword [_GlideRoot + fBiasHi]	; GC B
     
-    fld	    dword [vertex + r]		; R GC BC
+    fld	    dword [vertex + r]		        ; R GC BC
     fadd    dword [_GlideRoot + fBiasHi]	; RC GC BC
 
-    fxch    st2				; BC GC RC
+    fxch    st2				        ; BC GC RC
     fstp    dword [bias0]			; GC RC
 
     fstp    dword [bias1]			; RC
-    mov	    packCol, dword [bias0]		; B + bias
+    mov	    packCol, [bias0]			; B + bias
 
     fstp    dword [bias0]
-    mov	    eax, dword [bias1]		; G + bias
-    
+    mov	    eax, [bias1]		        ; G + bias
+
 %IF GLIDE_PACK_ALPHA
     fld	    dword [vertex + a]
     fadd    dword [_GlideRoot + fBiasHi]
 
-    and	    packCol, 00FFh			; B color component
+    and	    packCol, 000000FFh			; B color component
     and	    eax, 0000FF00h			; G component << 8
 
     add	    dlp, 8				; Packed RGB + A dataList entry
@@ -1104,13 +1104,13 @@
     nop
 
     fstp    dword [bias1]
-    mov	    eax, dword [bias0]		; R + bias
+    mov	    eax, [bias0]		        ; R + bias
     
-    mov	    esi, dword [bias1]		; A + bias
+    mov	    esi, [bias1]		        ; A + bias
     and	    eax, 0000FF00h			; R component << 8
     
-    and	    esi, 0FFFFFF00h			; A component << 8
-    shl	    eax, 8				; R << 16
+    and	    esi, 0000FF00h			; A component << 8
+    shl	    eax, 8				; R << 8
     
     or	    packCol, eax			; 00RRGGBB
     shl	    esi, 16				; A << 16
@@ -1118,16 +1118,16 @@
     or	    packCol, esi			; AARRGGBB
     nop
 %ELSE ; !GLIDE_PACK_ALPHA    
-    and	    packCol, 00FFh			; B color component
+    and	    packCol, 000000FFh			; B color component
     and	    eax, 0000FF00h			; G component << 8
 
     add	    dlp, 4				; Next dataList item    
     or	    packCol, eax
     
-    mov	    eax, dword [bias0]		; R + bias
+    mov	    eax, [bias0]		        ; R + bias
     and	    eax, 0000FF00h			; R component << 8
 
-    shl	    eax, 8				; R << 16
+    shl	    eax, 8				; R << 8
     or	    packCol, eax			; 00RRGGBB
 %ENDIF ; !GLIDE_PACK_ALPHA
 
@@ -1136,7 +1136,7 @@
 %ENDIF ; GLIDE_PACK_RGB
 
 .__doParams:
-    mov	    eax, dword [dlp]		; Get first offset from the data list
+    mov	    eax, [dlp]				; Get first offset from the data list
     add	    dlp, 4				; dlp++
     
     cmp	    eax, 0				; Are we done?
@@ -1148,10 +1148,10 @@
     nop
         
 .__paramLoop:
-    mov     tempVal, dword [eax + vertex]	; Get the parameter from teh vertex
+    mov     tempVal, [eax + vertex]		; Get the parameter from teh vertex
     add     fifo, 4				; fifoPtr += sizeof(FxU32)
 
-    mov     eax, dword [dlp]		; offset = *(dlp + 1)
+    mov     eax, [dlp]		        	; offset = *(dlp + 1)
     add     dlp, 4				; dlp++
     
     cmp     eax, 0				; Are we done?
@@ -1175,7 +1175,7 @@
     mov	    [gc + fifoPtr], fifo
     sub	    eax, ebx
 
-    mov	    ebx, [_GlideRoot + trisDrawn]		; _GlideRoot.stats.trisDrawn++;    
+    mov	    ebx, [_GlideRoot + trisDrawn]	; _GlideRoot.stats.trisDrawn++;    
     sub	    [gc + fifoRoom], eax
 
     add	    ebx, 1