#include "brw_eu.h"
#include "brw_state.h"
+#define FILE_DEBUG_FLAG DEBUG_BLORP
/**
* Helper function for handling mirror image blits.
}
void
-brw_blorp_blit_miptrees(struct intel_context *intel,
+brw_blorp_blit_miptrees(struct brw_context *brw,
struct intel_mipmap_tree *src_mt,
unsigned src_level, unsigned src_layer,
struct intel_mipmap_tree *dst_mt,
float src_x1, float src_y1,
float dst_x0, float dst_y0,
float dst_x1, float dst_y1,
- bool mirror_x, bool mirror_y)
+ GLenum filter, bool mirror_x, bool mirror_y)
{
- intel_miptree_slice_resolve_depth(intel, src_mt, src_level, src_layer);
- intel_miptree_slice_resolve_depth(intel, dst_mt, dst_level, dst_layer);
-
- brw_blorp_blit_params params(brw_context(&intel->ctx),
+ /* Get ready to blit. This includes depth resolving the src and dst
+ * buffers if necessary. Note: it's not necessary to do a color resolve on
+ * the destination buffer because we use the standard render path to render
+ * to destination color buffers, and the standard render path is
+ * fast-color-aware.
+ */
+ intel_miptree_resolve_color(brw, src_mt);
+ intel_miptree_slice_resolve_depth(brw, src_mt, src_level, src_layer);
+ intel_miptree_slice_resolve_depth(brw, dst_mt, dst_level, dst_layer);
+
+ DBG("%s from %s mt %p %d %d (%f,%f) (%f,%f)"
+ "to %s mt %p %d %d (%f,%f) (%f,%f) (flip %d,%d)\n",
+ __FUNCTION__,
+ _mesa_get_format_name(src_mt->format), src_mt,
+ src_level, src_layer, src_x0, src_y0, src_x1, src_y1,
+ _mesa_get_format_name(dst_mt->format), dst_mt,
+ dst_level, dst_layer, dst_x0, dst_y0, dst_x1, dst_y1,
+ mirror_x, mirror_y);
+
+ brw_blorp_blit_params params(brw,
src_mt, src_level, src_layer,
dst_mt, dst_level, dst_layer,
src_x0, src_y0,
src_x1, src_y1,
dst_x0, dst_y0,
dst_x1, dst_y1,
- mirror_x, mirror_y);
- brw_blorp_exec(intel, ¶ms);
+ filter, mirror_x, mirror_y);
+ brw_blorp_exec(brw, ¶ms);
intel_miptree_slice_set_needs_hiz_resolve(dst_mt, dst_level, dst_layer);
}
static void
-do_blorp_blit(struct intel_context *intel, GLbitfield buffer_bit,
+do_blorp_blit(struct brw_context *brw, GLbitfield buffer_bit,
struct intel_renderbuffer *src_irb,
struct intel_renderbuffer *dst_irb,
GLfloat srcX0, GLfloat srcY0, GLfloat srcX1, GLfloat srcY1,
GLfloat dstX0, GLfloat dstY0, GLfloat dstX1, GLfloat dstY1,
- bool mirror_x, bool mirror_y)
+ GLenum filter, bool mirror_x, bool mirror_y)
{
/* Find source/dst miptrees */
struct intel_mipmap_tree *src_mt = find_miptree(buffer_bit, src_irb);
struct intel_mipmap_tree *dst_mt = find_miptree(buffer_bit, dst_irb);
/* Do the blit */
- brw_blorp_blit_miptrees(intel,
+ brw_blorp_blit_miptrees(brw,
src_mt, src_irb->mt_level, src_irb->mt_layer,
dst_mt, dst_irb->mt_level, dst_irb->mt_layer,
srcX0, srcY0, srcX1, srcY1,
dstX0, dstY0, dstX1, dstY1,
- mirror_x, mirror_y);
+ filter, mirror_x, mirror_y);
intel_renderbuffer_set_needs_downsample(dst_irb);
}
}
static bool
-try_blorp_blit(struct intel_context *intel,
+try_blorp_blit(struct brw_context *brw,
GLfloat srcX0, GLfloat srcY0, GLfloat srcX1, GLfloat srcY1,
GLfloat dstX0, GLfloat dstY0, GLfloat dstX1, GLfloat dstY1,
GLenum filter, GLbitfield buffer_bit)
{
- struct gl_context *ctx = &intel->ctx;
+ struct gl_context *ctx = &brw->ctx;
/* Sync up the state of window system buffers. We need to do this before
* we go looking for the buffers.
*/
- intel_prepare_render(intel);
+ intel_prepare_render(brw);
const struct gl_framebuffer *read_fb = ctx->ReadBuffer;
const struct gl_framebuffer *draw_fb = ctx->DrawBuffer;
fixup_mirroring(mirror_y, srcY0, srcY1);
fixup_mirroring(mirror_y, dstY0, dstY1);
- /* Linear filtering is not yet implemented in blorp engine. So, fallback
- * to other blit paths.
- */
- if ((srcX1 - srcX0 != dstX1 - dstX0 ||
- srcY1 - srcY0 != dstY1 - dstY0) &&
- filter == GL_LINEAR)
- return false;
-
/* If the destination rectangle needs to be clipped or scissored, do so.
*/
if (!(clip_or_scissor(mirror_x, srcX0, srcX1, dstX0, dstX1,
for (unsigned i = 0; i < ctx->DrawBuffer->_NumColorDrawBuffers; ++i) {
dst_irb = intel_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[i]);
if (dst_irb)
- do_blorp_blit(intel, buffer_bit, src_irb, dst_irb, srcX0, srcY0,
+ do_blorp_blit(brw, buffer_bit, src_irb, dst_irb, srcX0, srcY0,
srcX1, srcY1, dstX0, dstY0, dstX1, dstY1,
- mirror_x, mirror_y);
+ filter, mirror_x, mirror_y);
}
break;
case GL_DEPTH_BUFFER_BIT:
intel_renderbuffer(draw_fb->Attachment[BUFFER_DEPTH].Renderbuffer);
if (!formats_match(buffer_bit, src_irb, dst_irb))
return false;
- do_blorp_blit(intel, buffer_bit, src_irb, dst_irb, srcX0, srcY0,
+ do_blorp_blit(brw, buffer_bit, src_irb, dst_irb, srcX0, srcY0,
srcX1, srcY1, dstX0, dstY0, dstX1, dstY1,
- mirror_x, mirror_y);
+ filter, mirror_x, mirror_y);
break;
case GL_STENCIL_BUFFER_BIT:
src_irb =
intel_renderbuffer(draw_fb->Attachment[BUFFER_STENCIL].Renderbuffer);
if (!formats_match(buffer_bit, src_irb, dst_irb))
return false;
- do_blorp_blit(intel, buffer_bit, src_irb, dst_irb, srcX0, srcY0,
+ do_blorp_blit(brw, buffer_bit, src_irb, dst_irb, srcX0, srcY0,
srcX1, srcY1, dstX0, dstY0, dstX1, dstY1,
- mirror_x, mirror_y);
+ filter, mirror_x, mirror_y);
break;
default:
assert(false);
}
bool
-brw_blorp_copytexsubimage(struct intel_context *intel,
+brw_blorp_copytexsubimage(struct brw_context *brw,
struct gl_renderbuffer *src_rb,
struct gl_texture_image *dst_image,
+ int slice,
int srcX0, int srcY0,
int dstX0, int dstY0,
int width, int height)
{
- struct gl_context *ctx = &intel->ctx;
+ struct gl_context *ctx = &brw->ctx;
struct intel_renderbuffer *src_irb = intel_renderbuffer(src_rb);
struct intel_texture_image *intel_image = intel_texture_image(dst_image);
/* Sync up the state of window system buffers. We need to do this before
* we go looking at the src renderbuffer's miptree.
*/
- intel_prepare_render(intel);
+ intel_prepare_render(brw);
struct intel_mipmap_tree *src_mt = src_irb->mt;
struct intel_mipmap_tree *dst_mt = intel_image->mt;
/* BLORP is not supported before Gen6. */
- if (intel->gen < 6)
+ if (brw->gen < 6)
return false;
if (!color_formats_match(src_mt->format, dst_mt->format)) {
mirror_y = true;
}
- brw_blorp_blit_miptrees(intel,
+ brw_blorp_blit_miptrees(brw,
src_mt, src_irb->mt_level, src_irb->mt_layer,
- dst_mt, dst_image->Level, dst_image->Face,
+ dst_mt, dst_image->Level, dst_image->Face + slice,
srcX0, srcY0, srcX1, srcY1,
dstX0, dstY0, dstX1, dstY1,
- false, mirror_y);
+ GL_NEAREST, false, mirror_y);
/* If we're copying to a packed depth stencil texture and the source
* framebuffer has separate stencil, we need to also copy the stencil data
dst_mt = dst_mt->stencil_mt;
if (src_mt != dst_mt) {
- brw_blorp_blit_miptrees(intel,
+ brw_blorp_blit_miptrees(brw,
src_mt, src_irb->mt_level, src_irb->mt_layer,
- dst_mt, dst_image->Level, dst_image->Face,
+ dst_mt, dst_image->Level,
+ dst_image->Face + slice,
srcX0, srcY0, srcX1, srcY1,
dstX0, dstY0, dstX1, dstY1,
- false, mirror_y);
+ GL_NEAREST, false, mirror_y);
}
}
GLbitfield
-brw_blorp_framebuffer(struct intel_context *intel,
+brw_blorp_framebuffer(struct brw_context *brw,
GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
GLbitfield mask, GLenum filter)
{
/* BLORP is not supported before Gen6. */
- if (intel->gen < 6)
+ if (brw->gen < 6)
return mask;
static GLbitfield buffer_bits[] = {
for (unsigned int i = 0; i < ARRAY_SIZE(buffer_bits); ++i) {
if ((mask & buffer_bits[i]) &&
- try_blorp_blit(intel,
+ try_blorp_blit(brw,
srcX0, srcY0, srcX1, srcY1,
dstX0, dstY0, dstX1, dstY1,
filter, buffer_bits[i])) {
void decode_msaa(unsigned num_samples, intel_msaa_layout layout);
void kill_if_outside_dst_rect();
void translate_dst_to_src();
+ void clamp_tex_coords(struct brw_reg regX, struct brw_reg regY,
+ struct brw_reg clampX0, struct brw_reg clampY0,
+ struct brw_reg clampX1, struct brw_reg clampY1);
void single_to_blend();
- void manual_blend(unsigned num_samples);
+ void manual_blend_average(unsigned num_samples);
+ void manual_blend_bilinear(unsigned num_samples);
void sample(struct brw_reg dst);
void texel_fetch(struct brw_reg dst);
void mcs_fetch();
struct brw_reg dst_x1;
struct brw_reg dst_y0;
struct brw_reg dst_y1;
+ /* Top right coordinates of the rectangular grid used for scaled blitting */
+ struct brw_reg rect_grid_x1;
+ struct brw_reg rect_grid_y1;
struct {
struct brw_reg multiplier;
struct brw_reg offset;
*/
struct brw_reg y_coords[2];
+ /* X, Y coordinates of the pixel from which we need to fetch the specific
+ * sample. These are used for multisample scaled blitting.
+ */
+ struct brw_reg x_sample_coords;
+ struct brw_reg y_sample_coords;
+
+ /* Fractional parts of the x and y coordinates, used as bilinear interpolation coefficients */
+ struct brw_reg x_frac;
+ struct brw_reg y_frac;
+
/* Which element of x_coords and y_coords is currently in use.
*/
int xy_coord_index;
* that we want to texture from. Exception: if we are blending, then S is
* irrelevant, because we are going to fetch all samples.
*/
- if (key->blend) {
- if (brw->intel.gen == 6) {
+ if (key->blend && !key->blit_scaled) {
+ if (brw->gen == 6) {
/* Gen6 hardware an automatically blend using the SAMPLE message */
single_to_blend();
sample(texture_data[0]);
} else {
/* Gen7+ hardware doesn't automaticaly blend. */
- manual_blend(key->src_samples);
+ manual_blend_average(key->src_samples);
}
+ } else if(key->blend && key->blit_scaled) {
+ manual_blend_bilinear(key->src_samples);
} else {
/* We aren't blending, which means we just want to fetch a single sample
* from the source surface. The address that we want to fetch from is
* the same as the configuration of the texture, then we need to adjust
* the coordinates to compensate for the difference.
*/
- if (tex_tiled_w != key->src_tiled_w ||
- key->tex_samples != key->src_samples ||
- key->tex_layout != key->src_layout) {
+ if ((tex_tiled_w != key->src_tiled_w ||
+ key->tex_samples != key->src_samples ||
+ key->tex_layout != key->src_layout) &&
+ !key->bilinear_filter) {
encode_msaa(key->src_samples, key->src_layout);
/* Now (X, Y, S) = detile(src_tiling, offset) */
translate_tiling(key->src_tiled_w, tex_tiled_w);
decode_msaa(key->tex_samples, key->tex_layout);
}
- /* Now (X, Y, S) = decode_msaa(tex_samples, detile(tex_tiling, offset)).
- *
- * In other words: X, Y, and S now contain values which, when passed to
- * the texturing unit, will cause data to be read from the correct
- * memory location. So we can fetch the texel now.
- */
- if (key->tex_layout == INTEL_MSAA_LAYOUT_CMS)
- mcs_fetch();
- texel_fetch(texture_data[0]);
+ if (key->bilinear_filter) {
+ sample(texture_data[0]);
+ }
+ else {
+ /* Now (X, Y, S) = decode_msaa(tex_samples, detile(tex_tiling, offset)).
+ *
+ * In other words: X, Y, and S now contain values which, when passed to
+ * the texturing unit, will cause data to be read from the correct
+ * memory location. So we can fetch the texel now.
+ */
+ if (key->tex_layout == INTEL_MSAA_LAYOUT_CMS)
+ mcs_fetch();
+ texel_fetch(texture_data[0]);
+ }
}
/* Finally, write the fetched (or blended) value to the render target and
#define CONST_LOC(name) offsetof(brw_blorp_wm_push_constants, name)
#define ALLOC_REG(name) \
this->name = \
- brw_vec1_reg(BRW_GENERAL_REGISTER_FILE, base_reg, CONST_LOC(name) / 4)
+ brw_vec1_reg(BRW_GENERAL_REGISTER_FILE, \
+ base_reg + CONST_LOC(name) / 32, \
+ (CONST_LOC(name) % 32) / 4)
ALLOC_REG(dst_x0);
ALLOC_REG(dst_x1);
ALLOC_REG(dst_y0);
ALLOC_REG(dst_y1);
+ ALLOC_REG(rect_grid_x1);
+ ALLOC_REG(rect_grid_y1);
ALLOC_REG(x_transform.multiplier);
ALLOC_REG(x_transform.offset);
ALLOC_REG(y_transform.multiplier);
= retype(brw_vec8_grf(reg, 0), BRW_REGISTER_TYPE_UD);
reg += 2;
}
+
+ if (key->blit_scaled && key->blend) {
+ this->x_sample_coords = brw_vec8_grf(reg, 0);
+ reg += 2;
+ this->y_sample_coords = brw_vec8_grf(reg, 0);
+ reg += 2;
+ this->x_frac = brw_vec8_grf(reg, 0);
+ reg += 2;
+ this->y_frac = brw_vec8_grf(reg, 0);
+ reg += 2;
+ }
+
this->xy_coord_index = 0;
this->sample_index
= retype(brw_vec8_grf(reg, 0), BRW_REGISTER_TYPE_UD);
brw_pop_insn_state(&func);
}
+#define X_f retype(X, BRW_REGISTER_TYPE_F)
+#define Y_f retype(Y, BRW_REGISTER_TYPE_F)
+#define Xp_f retype(Xp, BRW_REGISTER_TYPE_F)
+#define Yp_f retype(Yp, BRW_REGISTER_TYPE_F)
/**
* Emit code to translate from destination (X, Y) coordinates to source (X, Y)
* coordinates.
void
brw_blorp_blit_program::translate_dst_to_src()
{
- struct brw_reg X_f = retype(X, BRW_REGISTER_TYPE_F);
- struct brw_reg Y_f = retype(Y, BRW_REGISTER_TYPE_F);
- struct brw_reg Xp_f = retype(Xp, BRW_REGISTER_TYPE_F);
- struct brw_reg Yp_f = retype(Yp, BRW_REGISTER_TYPE_F);
-
brw_set_compression_control(&func, BRW_COMPRESSION_COMPRESSED);
/* Move the UD coordinates to float registers. */
brw_MOV(&func, Xp_f, X);
brw_MUL(&func, Y_f, Yp_f, y_transform.multiplier);
brw_ADD(&func, X_f, X_f, x_transform.offset);
brw_ADD(&func, Y_f, Y_f, y_transform.offset);
- /* Round the float coordinates down to nearest integer by moving to
- * UD registers.
- */
- brw_MOV(&func, Xp, X_f);
- brw_MOV(&func, Yp, Y_f);
- SWAP_XY_AND_XPYP();
+ if (key->blit_scaled && key->blend) {
+ /* Translate coordinates to lay out the samples in a rectangular grid
+ * roughly corresponding to sample locations.
+ */
+ brw_MUL(&func, X_f, X_f, brw_imm_f(key->x_scale));
+ brw_MUL(&func, Y_f, Y_f, brw_imm_f(key->y_scale));
+ /* Adjust coordinates so that integers represent pixel centers rather
+ * than pixel edges.
+ */
+ brw_ADD(&func, X_f, X_f, brw_imm_f(-0.5));
+ brw_ADD(&func, Y_f, Y_f, brw_imm_f(-0.5));
+
+ /* Clamp the X, Y texture coordinates to properly handle the sampling of
+ * texels on texture edges.
+ */
+ clamp_tex_coords(X_f, Y_f,
+ brw_imm_f(0.0), brw_imm_f(0.0),
+ rect_grid_x1, rect_grid_y1);
+
+ /* Store the fractional parts to be used as bilinear interpolation
+ * coefficients.
+ */
+ brw_FRC(&func, x_frac, X_f);
+ brw_FRC(&func, y_frac, Y_f);
+
+ /* Round the float coordinates down to nearest integer */
+ brw_RNDD(&func, Xp_f, X_f);
+ brw_RNDD(&func, Yp_f, Y_f);
+ brw_MUL(&func, X_f, Xp_f, brw_imm_f(1 / key->x_scale));
+ brw_MUL(&func, Y_f, Yp_f, brw_imm_f(1 / key->y_scale));
+ SWAP_XY_AND_XPYP();
+ } else if (!key->bilinear_filter) {
+ /* Round the float coordinates down to nearest integer by moving to
+ * UD registers.
+ */
+ brw_MOV(&func, Xp, X_f);
+ brw_MOV(&func, Yp, Y_f);
+ SWAP_XY_AND_XPYP();
+ }
brw_set_compression_control(&func, BRW_COMPRESSION_NONE);
}
+void
+brw_blorp_blit_program::clamp_tex_coords(struct brw_reg regX,
+ struct brw_reg regY,
+ struct brw_reg clampX0,
+ struct brw_reg clampY0,
+ struct brw_reg clampX1,
+ struct brw_reg clampY1)
+{
+ brw_CMP(&func, vec16(brw_null_reg()), BRW_CONDITIONAL_L, regX, clampX0);
+ brw_MOV(&func, regX, clampX0);
+ brw_set_predicate_control(&func, BRW_PREDICATE_NONE);
+
+ brw_CMP(&func, vec16(brw_null_reg()), BRW_CONDITIONAL_G, regX, clampX1);
+ brw_MOV(&func, regX, clampX1);
+ brw_set_predicate_control(&func, BRW_PREDICATE_NONE);
+
+ brw_CMP(&func, vec16(brw_null_reg()), BRW_CONDITIONAL_L, regY, clampY0);
+ brw_MOV(&func, regY, clampY0);
+ brw_set_predicate_control(&func, BRW_PREDICATE_NONE);
+
+ brw_CMP(&func, vec16(brw_null_reg()), BRW_CONDITIONAL_G, regY, clampY1);
+ brw_MOV(&func, regY, clampY1);
+ brw_set_predicate_control(&func, BRW_PREDICATE_NONE);
+}
+#undef X_f
+#undef Y_f
+#undef Xp_f
+#undef Yp_f
+
/**
* Emit code to transform the X and Y coordinates as needed for blending
* together the different samples in an MSAA texture.
void
-brw_blorp_blit_program::manual_blend(unsigned num_samples)
+brw_blorp_blit_program::manual_blend_average(unsigned num_samples)
{
if (key->tex_layout == INTEL_MSAA_LAYOUT_CMS)
mcs_fetch();
brw_ENDIF(&func);
}
+void
+brw_blorp_blit_program::manual_blend_bilinear(unsigned num_samples)
+{
+ /* We do this computation by performing the following operations:
+ *
+ * In case of 4x, 8x MSAA:
+ * - Compute the pixel coordinates and sample numbers (a, b, c, d)
+ * which are later used for interpolation
+ * - linearly interpolate samples a and b in X
+ * - linearly interpolate samples c and d in X
+ * - linearly interpolate the results of last two operations in Y
+ *
+ * result = lrp(lrp(a + b) + lrp(c + d))
+ */
+ struct brw_reg Xp_f = retype(Xp, BRW_REGISTER_TYPE_F);
+ struct brw_reg Yp_f = retype(Yp, BRW_REGISTER_TYPE_F);
+ struct brw_reg t1_f = retype(t1, BRW_REGISTER_TYPE_F);
+ struct brw_reg t2_f = retype(t2, BRW_REGISTER_TYPE_F);
+
+ for (unsigned i = 0; i < 4; ++i) {
+ assert(i < ARRAY_SIZE(texture_data));
+ s_is_zero = false;
+
+ /* Compute pixel coordinates */
+ brw_ADD(&func, vec16(x_sample_coords), Xp_f,
+ brw_imm_f((float)(i & 0x1) * (1.0 / key->x_scale)));
+ brw_ADD(&func, vec16(y_sample_coords), Yp_f,
+ brw_imm_f((float)((i >> 1) & 0x1) * (1.0 / key->y_scale)));
+ brw_MOV(&func, vec16(X), x_sample_coords);
+ brw_MOV(&func, vec16(Y), y_sample_coords);
+
+ /* The MCS value we fetch has to match up with the pixel that we're
+ * sampling from. Since we sample from different pixels in each
+ * iteration of this "for" loop, the call to mcs_fetch() should be
+ * here inside the loop after computing the pixel coordinates.
+ */
+ if (key->tex_layout == INTEL_MSAA_LAYOUT_CMS)
+ mcs_fetch();
+
+ /* Compute sample index and map the sample index to a sample number.
+ * Sample index layout shows the numbering of slots in a rectangular
+ * grid of samples with in a pixel. Sample number layout shows the
+ * rectangular grid of samples roughly corresponding to the real sample
+ * locations with in a pixel.
+ * In case of 4x MSAA, layout of sample indices matches the layout of
+ * sample numbers:
+ * ---------
+ * | 0 | 1 |
+ * ---------
+ * | 2 | 3 |
+ * ---------
+ *
+ * In case of 8x MSAA the two layouts don't match.
+ * sample index layout : --------- sample number layout : ---------
+ * | 0 | 1 | | 5 | 2 |
+ * --------- ---------
+ * | 2 | 3 | | 4 | 6 |
+ * --------- ---------
+ * | 4 | 5 | | 0 | 3 |
+ * --------- ---------
+ * | 6 | 7 | | 7 | 1 |
+ * --------- ---------
+ */
+ brw_FRC(&func, vec16(t1_f), x_sample_coords);
+ brw_FRC(&func, vec16(t2_f), y_sample_coords);
+ brw_MUL(&func, vec16(t1_f), t1_f, brw_imm_f(key->x_scale));
+ brw_MUL(&func, vec16(t2_f), t2_f, brw_imm_f(key->x_scale * key->y_scale));
+ brw_ADD(&func, vec16(t1_f), t1_f, t2_f);
+ brw_MOV(&func, vec16(S), t1_f);
+
+ if (num_samples == 8) {
+ /* Map the sample index to a sample number */
+ brw_CMP(&func, vec16(brw_null_reg()), BRW_CONDITIONAL_L,
+ S, brw_imm_d(4));
+ brw_IF(&func, BRW_EXECUTE_16);
+ {
+ brw_MOV(&func, vec16(t2), brw_imm_d(5));
+ brw_CMP(&func, vec16(brw_null_reg()), BRW_CONDITIONAL_EQ,
+ S, brw_imm_d(1));
+ brw_MOV(&func, vec16(t2), brw_imm_d(2));
+ brw_set_predicate_control(&func, BRW_PREDICATE_NONE);
+ brw_CMP(&func, vec16(brw_null_reg()), BRW_CONDITIONAL_EQ,
+ S, brw_imm_d(2));
+ brw_MOV(&func, vec16(t2), brw_imm_d(4));
+ brw_set_predicate_control(&func, BRW_PREDICATE_NONE);
+ brw_CMP(&func, vec16(brw_null_reg()), BRW_CONDITIONAL_EQ,
+ S, brw_imm_d(3));
+ brw_MOV(&func, vec16(t2), brw_imm_d(6));
+ brw_set_predicate_control(&func, BRW_PREDICATE_NONE);
+ }
+ brw_ELSE(&func);
+ {
+ brw_MOV(&func, vec16(t2), brw_imm_d(0));
+ brw_CMP(&func, vec16(brw_null_reg()), BRW_CONDITIONAL_EQ,
+ S, brw_imm_d(5));
+ brw_MOV(&func, vec16(t2), brw_imm_d(3));
+ brw_set_predicate_control(&func, BRW_PREDICATE_NONE);
+ brw_CMP(&func, vec16(brw_null_reg()), BRW_CONDITIONAL_EQ,
+ S, brw_imm_d(6));
+ brw_MOV(&func, vec16(t2), brw_imm_d(7));
+ brw_set_predicate_control(&func, BRW_PREDICATE_NONE);
+ brw_CMP(&func, vec16(brw_null_reg()), BRW_CONDITIONAL_EQ,
+ S, brw_imm_d(7));
+ brw_MOV(&func, vec16(t2), brw_imm_d(1));
+ brw_set_predicate_control(&func, BRW_PREDICATE_NONE);
+ }
+ brw_ENDIF(&func);
+ brw_MOV(&func, vec16(S), t2);
+ }
+ texel_fetch(texture_data[i]);
+ }
+
+#define SAMPLE(x, y) offset(texture_data[x], y)
+ brw_set_access_mode(&func, BRW_ALIGN_16);
+ for (int index = 3; index > 0; ) {
+ /* Since we're doing SIMD16, 4 color channels fits in to 8 registers.
+ * Counter value of 8 in 'for' loop below is used to interpolate all
+ * the color components.
+ */
+ for (int k = 0; k < 8; ++k)
+ brw_LRP(&func,
+ vec8(SAMPLE(index - 1, k)),
+ offset(x_frac, k & 1),
+ SAMPLE(index, k),
+ SAMPLE(index - 1, k));
+ index -= 2;
+ }
+ for (int k = 0; k < 8; ++k)
+ brw_LRP(&func,
+ vec8(SAMPLE(0, k)),
+ offset(y_frac, k & 1),
+ vec8(SAMPLE(2, k)),
+ vec8(SAMPLE(0, k)));
+ brw_set_access_mode(&func, BRW_ALIGN_1);
+#undef SAMPLE
+}
+
/**
* Emit code to look up a value in the texture using the SAMPLE message (which
* does blending of MSAA surfaces).
SAMPLER_MESSAGE_ARG_V_FLOAT
};
- texture_lookup(dst, GEN5_SAMPLER_MESSAGE_SAMPLE, args, ARRAY_SIZE(args));
+ texture_lookup(dst, GEN5_SAMPLER_MESSAGE_SAMPLE, args,
+ ARRAY_SIZE(args));
}
/**
SAMPLER_MESSAGE_ARG_V_INT
};
- switch (brw->intel.gen) {
+ switch (brw->gen) {
case 6:
texture_lookup(dst, GEN5_SAMPLER_MESSAGE_SAMPLE_LD, gen6_args,
s_is_zero ? 2 : 5);
for (int arg = 0; arg < num_args; ++arg) {
switch (args[arg]) {
case SAMPLER_MESSAGE_ARG_U_FLOAT:
- brw_MOV(&func, retype(mrf, BRW_REGISTER_TYPE_F), X);
+ if (key->bilinear_filter)
+ brw_MOV(&func, retype(mrf, BRW_REGISTER_TYPE_F),
+ retype(X, BRW_REGISTER_TYPE_F));
+ else
+ brw_MOV(&func, retype(mrf, BRW_REGISTER_TYPE_F), X);
break;
case SAMPLER_MESSAGE_ARG_V_FLOAT:
- brw_MOV(&func, retype(mrf, BRW_REGISTER_TYPE_F), Y);
+ if (key->bilinear_filter)
+ brw_MOV(&func, retype(mrf, BRW_REGISTER_TYPE_F),
+ retype(Y, BRW_REGISTER_TYPE_F));
+ else
+ brw_MOV(&func, retype(mrf, BRW_REGISTER_TYPE_F), Y);
break;
case SAMPLER_MESSAGE_ARG_U_INT:
brw_MOV(&func, mrf, X);
}
/* Prior to Gen7, all MSAA surfaces use IMS layout. */
- if (brw->intel.gen == 6) {
+ if (brw->gen == 6) {
assert(true_layout == INTEL_MSAA_LAYOUT_IMS);
}
GLfloat src_x1, GLfloat src_y1,
GLfloat dst_x0, GLfloat dst_y0,
GLfloat dst_x1, GLfloat dst_y1,
+ GLenum filter,
bool mirror_x, bool mirror_y)
{
+ struct gl_context *ctx = &brw->ctx;
+ const struct gl_framebuffer *read_fb = ctx->ReadBuffer;
+
src.set(brw, src_mt, src_level, src_layer);
dst.set(brw, dst_mt, dst_level, dst_layer);
- src.brw_surfaceformat = dst.brw_surfaceformat;
+ /* Even though we do multisample resolves at the time of the blit, OpenGL
+ * specification defines them as if they happen at the time of rendering,
+ * which means that the type of averaging we do during the resolve should
+ * only depend on the source format; the destination format should be
+ * ignored. But, specification doesn't seem to be strict about it.
+ *
+ * It has been observed that mulitisample resolves produce slightly better
+ * looking images when averaging is done using destination format. NVIDIA's
+ * proprietary OpenGL driver also follow this approach. So, we choose to
+ * follow it in our driver.
+ *
+ * Following if-else block takes care of this exception made for
+ * multisampled resolves.
+ */
+ if (src.num_samples > 1)
+ src.brw_surfaceformat = dst.brw_surfaceformat;
+ else
+ dst.brw_surfaceformat = src.brw_surfaceformat;
use_wm_prog = true;
memset(&wm_prog_key, 0, sizeof(wm_prog_key));
break;
}
- if (brw->intel.gen > 6) {
+ if (brw->gen > 6) {
/* Gen7's rendering hardware only supports the IMS layout for depth and
* stencil render targets. Blorp always maps its destination surface as
* a color render target (even if it's actually a depth or stencil
wm_prog_key.persample_msaa_dispatch = true;
}
+ /* Scaled blitting or not. */
+ wm_prog_key.blit_scaled =
+ ((dst_x1 - dst_x0) == (src_x1 - src_x0) &&
+ (dst_y1 - dst_y0) == (src_y1 - src_y0)) ? false : true;
+
+ /* Scaling factors used for bilinear filtering in multisample scaled
+ * blits.
+ */
+ wm_prog_key.x_scale = 2.0;
+ wm_prog_key.y_scale = src_mt->num_samples / 2.0;
+
+ if (filter == GL_LINEAR)
+ wm_prog_key.bilinear_filter = true;
+
/* The render path must be configured to use the same number of samples as
* the destination buffer.
*/
y0 = wm_push_consts.dst_y0 = dst_y0;
x1 = wm_push_consts.dst_x1 = dst_x1;
y1 = wm_push_consts.dst_y1 = dst_y1;
+ wm_push_consts.rect_grid_x1 = read_fb->Width * wm_prog_key.x_scale - 1.0;
+ wm_push_consts.rect_grid_y1 = read_fb->Height * wm_prog_key.y_scale - 1.0;
+
wm_push_consts.x_transform.setup(src_x0, src_x1, dst_x0, dst_x1, mirror_x);
wm_push_consts.y_transform.setup(src_y0, src_y1, dst_y0, dst_y1, mirror_y);
brw_blorp_blit_params::get_wm_prog(struct brw_context *brw,
brw_blorp_prog_data **prog_data) const
{
- uint32_t prog_offset;
+ uint32_t prog_offset = 0;
if (!brw_search_cache(&brw->cache, BRW_BLORP_BLIT_PROG,
&this->wm_prog_key, sizeof(this->wm_prog_key),
&prog_offset, prog_data)) {