summaryrefslogtreecommitdiff
path: root/src/gallium/drivers/cell/ppu/cell_vertex_fetch.c
diff options
context:
space:
mode:
authorBrian Paul <brian.paul@tungstengraphics.com>2008-10-08 20:44:32 -0600
committerBrian Paul <brian.paul@tungstengraphics.com>2008-10-08 20:44:32 -0600
commitd48a92e88040470f93e2186f8eb23e4797a09860 (patch)
tree7fbfd01dc04b42e42b4457eae507693ef8dc5f3b /src/gallium/drivers/cell/ppu/cell_vertex_fetch.c
parenta4e477433f485a39b5de448d0a9cb6f4bf9bb90f (diff)
cell: implement function calls from shader code. fslight demo runs now.
Used for SIN, COS, EXP2, LOG2, POW instructions. TEX next. Fixed some bugs in MIN, MAX, DP3, DP4, DPH instructions. In rtasm code: Special-case spe_lqd(), spe_stqd() functions so they take byte offsets but low-order 4 bits are shifted out. This makes things consistant with SPU assembly language conventions. Added spe_get_registers_used() function.
Diffstat (limited to 'src/gallium/drivers/cell/ppu/cell_vertex_fetch.c')
-rw-r--r--src/gallium/drivers/cell/ppu/cell_vertex_fetch.c30
1 files changed, 15 insertions, 15 deletions
diff --git a/src/gallium/drivers/cell/ppu/cell_vertex_fetch.c b/src/gallium/drivers/cell/ppu/cell_vertex_fetch.c
index 566df7f59e..18969005b0 100644
--- a/src/gallium/drivers/cell/ppu/cell_vertex_fetch.c
+++ b/src/gallium/drivers/cell/ppu/cell_vertex_fetch.c
@@ -73,8 +73,8 @@ emit_matrix_transpose(struct spe_function *p,
int col3;
- spe_lqd(p, shuf_hi, shuf_ptr, 3);
- spe_lqd(p, shuf_lo, shuf_ptr, 4);
+ spe_lqd(p, shuf_hi, shuf_ptr, 3*16);
+ spe_lqd(p, shuf_lo, shuf_ptr, 4*16);
spe_shufb(p, t1, row0, row2, shuf_hi);
spe_shufb(p, t2, row0, row2, shuf_lo);
@@ -122,13 +122,13 @@ emit_matrix_transpose(struct spe_function *p,
*/
switch (count) {
case 4:
- spe_stqd(p, col3, dest_ptr, 3);
+ spe_stqd(p, col3, dest_ptr, 3 * 16);
case 3:
- spe_stqd(p, col2, dest_ptr, 2);
+ spe_stqd(p, col2, dest_ptr, 2 * 16);
case 2:
- spe_stqd(p, col1, dest_ptr, 1);
+ spe_stqd(p, col1, dest_ptr, 1 * 16);
case 1:
- spe_stqd(p, col0, dest_ptr, 0);
+ spe_stqd(p, col0, dest_ptr, 0 * 16);
}
@@ -166,17 +166,17 @@ emit_fetch(struct spe_function *p,
float scale_signed = 0.0;
float scale_unsigned = 0.0;
- spe_lqd(p, v0, in_ptr, 0 + offset[0]);
- spe_lqd(p, v1, in_ptr, 1 + offset[0]);
- spe_lqd(p, v2, in_ptr, 2 + offset[0]);
- spe_lqd(p, v3, in_ptr, 3 + offset[0]);
+ spe_lqd(p, v0, in_ptr, (0 + offset[0]) * 16);
+ spe_lqd(p, v1, in_ptr, (1 + offset[0]) * 16);
+ spe_lqd(p, v2, in_ptr, (2 + offset[0]) * 16);
+ spe_lqd(p, v3, in_ptr, (3 + offset[0]) * 16);
offset[0] += 4;
switch (bytes) {
case 1:
scale_signed = 1.0f / 127.0f;
scale_unsigned = 1.0f / 255.0f;
- spe_lqd(p, tmp, shuf_ptr, 1);
+ spe_lqd(p, tmp, shuf_ptr, 1 * 16);
spe_shufb(p, v0, v0, v0, tmp);
spe_shufb(p, v1, v1, v1, tmp);
spe_shufb(p, v2, v2, v2, tmp);
@@ -185,7 +185,7 @@ emit_fetch(struct spe_function *p,
case 2:
scale_signed = 1.0f / 32767.0f;
scale_unsigned = 1.0f / 65535.0f;
- spe_lqd(p, tmp, shuf_ptr, 2);
+ spe_lqd(p, tmp, shuf_ptr, 2 * 16);
spe_shufb(p, v0, v0, v0, tmp);
spe_shufb(p, v1, v1, v1, tmp);
spe_shufb(p, v2, v2, v2, tmp);
@@ -241,11 +241,11 @@ emit_fetch(struct spe_function *p,
switch (count) {
case 1:
- spe_stqd(p, float_zero, out_ptr, 1);
+ spe_stqd(p, float_zero, out_ptr, 1 * 16);
case 2:
- spe_stqd(p, float_zero, out_ptr, 2);
+ spe_stqd(p, float_zero, out_ptr, 2 * 16);
case 3:
- spe_stqd(p, float_one, out_ptr, 3);
+ spe_stqd(p, float_one, out_ptr, 3 * 16);
}
if (float_zero != -1) {