From 007d8a6d04a807f5c65c8545398ee5639f2eee31 Mon Sep 17 00:00:00 2001 From: Jason Ekstrand Date: Fri, 19 Aug 2016 03:53:33 -0700 Subject: [PATCH] i965: Move get_fast_clear_rect to blorp_clear.c This has been the only caller since we deleted the meta fast clear code. Signed-off-by: Jason Ekstrand Reviewed-by: Topi Pohjolainen --- src/mesa/drivers/dri/i965/blorp_clear.c | 127 +++++++++++++++++++++- src/mesa/drivers/dri/i965/brw_meta_util.c | 122 --------------------- src/mesa/drivers/dri/i965/brw_meta_util.h | 6 - 3 files changed, 124 insertions(+), 131 deletions(-) diff --git a/src/mesa/drivers/dri/i965/blorp_clear.c b/src/mesa/drivers/dri/i965/blorp_clear.c index 60e40e9771e..0b10926065b 100644 --- a/src/mesa/drivers/dri/i965/blorp_clear.c +++ b/src/mesa/drivers/dri/i965/blorp_clear.c @@ -24,7 +24,6 @@ #include "util/ralloc.h" #include "blorp_priv.h" -#include "brw_meta_util.h" #include "brw_defines.h" #include "nir_builder.h" @@ -85,6 +84,128 @@ brw_blorp_params_get_clear_kernel(struct blorp_context *blorp, ralloc_free(mem_ctx); } +/* The x0, y0, x1, and y1 parameters must already be populated with the render + * area of the framebuffer to be cleared. + */ +static void +get_fast_clear_rect(const struct isl_device *dev, + const struct isl_surf *aux_surf, + unsigned *x0, unsigned *y0, + unsigned *x1, unsigned *y1) +{ + unsigned int x_align, y_align; + unsigned int x_scaledown, y_scaledown; + + /* Only single sampled surfaces need to (and actually can) be resolved. */ + if (aux_surf->usage == ISL_SURF_USAGE_CCS_BIT) { + /* From the Ivy Bridge PRM, Vol2 Part1 11.7 "MCS Buffer for Render + * Target(s)", beneath the "Fast Color Clear" bullet (p327): + * + * Clear pass must have a clear rectangle that must follow + * alignment rules in terms of pixels and lines as shown in the + * table below. Further, the clear-rectangle height and width + * must be multiple of the following dimensions. If the height + * and width of the render target being cleared do not meet these + * requirements, an MCS buffer can be created such that it + * follows the requirement and covers the RT. + * + * The alignment size in the table that follows is related to the + * alignment size that is baked into the CCS surface format but with X + * alignment multiplied by 16 and Y alignment multiplied by 32. + */ + x_align = isl_format_get_layout(aux_surf->format)->bw; + y_align = isl_format_get_layout(aux_surf->format)->bh; + + x_align *= 16; + + /* SKL+ line alignment requirement for Y-tiled are half those of the prior + * generations. + */ + if (dev->info->gen >= 9) + y_align *= 16; + else + y_align *= 32; + + /* From the Ivy Bridge PRM, Vol2 Part1 11.7 "MCS Buffer for Render + * Target(s)", beneath the "Fast Color Clear" bullet (p327): + * + * In order to optimize the performance MCS buffer (when bound to + * 1X RT) clear similarly to MCS buffer clear for MSRT case, + * clear rect is required to be scaled by the following factors + * in the horizontal and vertical directions: + * + * The X and Y scale down factors in the table that follows are each + * equal to half the alignment value computed above. + */ + x_scaledown = x_align / 2; + y_scaledown = y_align / 2; + + /* From BSpec: 3D-Media-GPGPU Engine > 3D Pipeline > Pixel > Pixel + * Backend > MCS Buffer for Render Target(s) [DevIVB+] > Table "Color + * Clear of Non-MultiSampled Render Target Restrictions": + * + * Clear rectangle must be aligned to two times the number of + * pixels in the table shown below due to 16x16 hashing across the + * slice. + */ + x_align *= 2; + y_align *= 2; + } else { + assert(aux_surf->usage == ISL_SURF_USAGE_MCS_BIT); + + /* From the Ivy Bridge PRM, Vol2 Part1 11.7 "MCS Buffer for Render + * Target(s)", beneath the "MSAA Compression" bullet (p326): + * + * Clear pass for this case requires that scaled down primitive + * is sent down with upper left co-ordinate to coincide with + * actual rectangle being cleared. For MSAA, clear rectangle’s + * height and width need to as show in the following table in + * terms of (width,height) of the RT. + * + * MSAA Width of Clear Rect Height of Clear Rect + * 2X Ceil(1/8*width) Ceil(1/2*height) + * 4X Ceil(1/8*width) Ceil(1/2*height) + * 8X Ceil(1/2*width) Ceil(1/2*height) + * 16X width Ceil(1/2*height) + * + * The text "with upper left co-ordinate to coincide with actual + * rectangle being cleared" is a little confusing--it seems to imply + * that to clear a rectangle from (x,y) to (x+w,y+h), one needs to + * feed the pipeline using the rectangle (x,y) to + * (x+Ceil(w/N),y+Ceil(h/2)), where N is either 2 or 8 depending on + * the number of samples. Experiments indicate that this is not + * quite correct; actually, what the hardware appears to do is to + * align whatever rectangle is sent down the pipeline to the nearest + * multiple of 2x2 blocks, and then scale it up by a factor of N + * horizontally and 2 vertically. So the resulting alignment is 4 + * vertically and either 4 or 16 horizontally, and the scaledown + * factor is 2 vertically and either 2 or 8 horizontally. + */ + switch (aux_surf->format) { + case ISL_FORMAT_MCS_2X: + case ISL_FORMAT_MCS_4X: + x_scaledown = 8; + break; + case ISL_FORMAT_MCS_8X: + x_scaledown = 2; + break; + case ISL_FORMAT_MCS_16X: + x_scaledown = 1; + break; + default: + unreachable("Unexpected MCS format for fast clear"); + } + y_scaledown = 2; + x_align = x_scaledown * 2; + y_align = y_scaledown * 2; + } + + *x0 = ROUND_DOWN_TO(*x0, x_align) / x_scaledown; + *y0 = ROUND_DOWN_TO(*y0, y_align) / y_scaledown; + *x1 = ALIGN(*x1, x_align) / x_scaledown; + *y1 = ALIGN(*y1, y_align) / y_scaledown; +} + void blorp_fast_clear(struct blorp_batch *batch, const struct brw_blorp_surf *surf, @@ -102,8 +223,8 @@ blorp_fast_clear(struct blorp_batch *batch, memset(¶ms.wm_inputs, 0xff, 4*sizeof(float)); params.fast_clear_op = BLORP_FAST_CLEAR_OP_CLEAR; - brw_get_fast_clear_rect(batch->blorp->isl_dev, surf->aux_surf, - ¶ms.x0, ¶ms.y0, ¶ms.x1, ¶ms.y1); + get_fast_clear_rect(batch->blorp->isl_dev, surf->aux_surf, + ¶ms.x0, ¶ms.y0, ¶ms.x1, ¶ms.y1); brw_blorp_params_get_clear_kernel(batch->blorp, ¶ms, true); diff --git a/src/mesa/drivers/dri/i965/brw_meta_util.c b/src/mesa/drivers/dri/i965/brw_meta_util.c index 52b7be4b693..499b6eacc0c 100644 --- a/src/mesa/drivers/dri/i965/brw_meta_util.c +++ b/src/mesa/drivers/dri/i965/brw_meta_util.c @@ -441,125 +441,3 @@ brw_meta_set_fast_clear_color(struct brw_context *brw, return updated; } - -/* The x0, y0, x1, and y1 parameters must already be populated with the render - * area of the framebuffer to be cleared. - */ -void -brw_get_fast_clear_rect(const struct isl_device *dev, - const struct isl_surf *aux_surf, - unsigned *x0, unsigned *y0, - unsigned *x1, unsigned *y1) -{ - unsigned int x_align, y_align; - unsigned int x_scaledown, y_scaledown; - - /* Only single sampled surfaces need to (and actually can) be resolved. */ - if (aux_surf->usage == ISL_SURF_USAGE_CCS_BIT) { - /* From the Ivy Bridge PRM, Vol2 Part1 11.7 "MCS Buffer for Render - * Target(s)", beneath the "Fast Color Clear" bullet (p327): - * - * Clear pass must have a clear rectangle that must follow - * alignment rules in terms of pixels and lines as shown in the - * table below. Further, the clear-rectangle height and width - * must be multiple of the following dimensions. If the height - * and width of the render target being cleared do not meet these - * requirements, an MCS buffer can be created such that it - * follows the requirement and covers the RT. - * - * The alignment size in the table that follows is related to the - * alignment size that is baked into the CCS surface format but with X - * alignment multiplied by 16 and Y alignment multiplied by 32. - */ - x_align = isl_format_get_layout(aux_surf->format)->bw; - y_align = isl_format_get_layout(aux_surf->format)->bh; - - x_align *= 16; - - /* SKL+ line alignment requirement for Y-tiled are half those of the prior - * generations. - */ - if (dev->info->gen >= 9) - y_align *= 16; - else - y_align *= 32; - - /* From the Ivy Bridge PRM, Vol2 Part1 11.7 "MCS Buffer for Render - * Target(s)", beneath the "Fast Color Clear" bullet (p327): - * - * In order to optimize the performance MCS buffer (when bound to - * 1X RT) clear similarly to MCS buffer clear for MSRT case, - * clear rect is required to be scaled by the following factors - * in the horizontal and vertical directions: - * - * The X and Y scale down factors in the table that follows are each - * equal to half the alignment value computed above. - */ - x_scaledown = x_align / 2; - y_scaledown = y_align / 2; - - /* From BSpec: 3D-Media-GPGPU Engine > 3D Pipeline > Pixel > Pixel - * Backend > MCS Buffer for Render Target(s) [DevIVB+] > Table "Color - * Clear of Non-MultiSampled Render Target Restrictions": - * - * Clear rectangle must be aligned to two times the number of - * pixels in the table shown below due to 16x16 hashing across the - * slice. - */ - x_align *= 2; - y_align *= 2; - } else { - assert(aux_surf->usage == ISL_SURF_USAGE_MCS_BIT); - - /* From the Ivy Bridge PRM, Vol2 Part1 11.7 "MCS Buffer for Render - * Target(s)", beneath the "MSAA Compression" bullet (p326): - * - * Clear pass for this case requires that scaled down primitive - * is sent down with upper left co-ordinate to coincide with - * actual rectangle being cleared. For MSAA, clear rectangle’s - * height and width need to as show in the following table in - * terms of (width,height) of the RT. - * - * MSAA Width of Clear Rect Height of Clear Rect - * 2X Ceil(1/8*width) Ceil(1/2*height) - * 4X Ceil(1/8*width) Ceil(1/2*height) - * 8X Ceil(1/2*width) Ceil(1/2*height) - * 16X width Ceil(1/2*height) - * - * The text "with upper left co-ordinate to coincide with actual - * rectangle being cleared" is a little confusing--it seems to imply - * that to clear a rectangle from (x,y) to (x+w,y+h), one needs to - * feed the pipeline using the rectangle (x,y) to - * (x+Ceil(w/N),y+Ceil(h/2)), where N is either 2 or 8 depending on - * the number of samples. Experiments indicate that this is not - * quite correct; actually, what the hardware appears to do is to - * align whatever rectangle is sent down the pipeline to the nearest - * multiple of 2x2 blocks, and then scale it up by a factor of N - * horizontally and 2 vertically. So the resulting alignment is 4 - * vertically and either 4 or 16 horizontally, and the scaledown - * factor is 2 vertically and either 2 or 8 horizontally. - */ - switch (aux_surf->format) { - case ISL_FORMAT_MCS_2X: - case ISL_FORMAT_MCS_4X: - x_scaledown = 8; - break; - case ISL_FORMAT_MCS_8X: - x_scaledown = 2; - break; - case ISL_FORMAT_MCS_16X: - x_scaledown = 1; - break; - default: - unreachable("Unexpected MCS format for fast clear"); - } - y_scaledown = 2; - x_align = x_scaledown * 2; - y_align = y_scaledown * 2; - } - - *x0 = ROUND_DOWN_TO(*x0, x_align) / x_scaledown; - *y0 = ROUND_DOWN_TO(*y0, y_align) / y_scaledown; - *x1 = ALIGN(*x1, x_align) / x_scaledown; - *y1 = ALIGN(*y1, y_align) / y_scaledown; -} diff --git a/src/mesa/drivers/dri/i965/brw_meta_util.h b/src/mesa/drivers/dri/i965/brw_meta_util.h index d5172373331..b9c4eac9c6c 100644 --- a/src/mesa/drivers/dri/i965/brw_meta_util.h +++ b/src/mesa/drivers/dri/i965/brw_meta_util.h @@ -42,12 +42,6 @@ brw_meta_mirror_clip_and_scissor(const struct gl_context *ctx, GLfloat *dstX1, GLfloat *dstY1, bool *mirror_x, bool *mirror_y); -void -brw_get_fast_clear_rect(const struct isl_device *dev, - const struct isl_surf *aux_surf, - unsigned *x0, unsigned *y0, - unsigned *x1, unsigned *y1); - bool brw_meta_set_fast_clear_color(struct brw_context *brw, struct intel_mipmap_tree *mt, -- 2.30.2