Lines Matching refs:depth
13 /* Low-resolution Z buffer is very similar to a depth prepass that helps
15 * be subsequently discarded by the depth test afterwards.
20 * In the binning pass it is possible to store the depth value of each
21 * vertex into internal low resolution depth buffer and quickly test
27 * - Writing depth while:
28 * - Changing direction of depth test (e.g. from OP_GREATER to OP_LESS);
30 * - Clearing depth with vkCmdClearAttachments;
31 * - (pre-a650) Not clearing depth attachment with LOAD_OP_CLEAR;
39 * unknown direction, the direction is set first time when depth write occurs
53 * - Parameters of the last used depth view.
62 * because depth buffer may have several layers and mip levels, on the
65 * LRZ direction between renderpasses is disabled when underlying depth
66 * buffer is changed, the following commands could change depth image:
76 * '0' means block has original depth clear value, and '1' means that the
81 * - To 0.0 if depth comparison is GREATER;
82 * - To 1.0 if depth comparison is LESS;
85 * fast-clear if depth clear value is not 0.0 or 1.0 because it may be worse
86 * for perf if some primitives are expected to fail depth test against the
87 * actual depth clear value.
99 * However, it seems that depth comparisons with LRZ values have some "slack"
100 * and nothing special should be done for such depth clear values.
122 * - Cache for direction byte + depth view params.
169 /* Disable direction by writing invalid depth view. */
205 * depth attachment, there are any secondaries, and GPU tracking is
251 /* We may not have the depth attachment when executing in a secondary
253 * the normal case and enable fast clear even if the depth image doesn't
288 (clear.depthStencil.depth == 0.f ||
289 clear.depthStencil.depth == 1.f);
315 "different depth attachments in a single renderpass");
382 * to fail the comparison of depth views.
412 * TODO: we could avoid this if we don't store depth and don't
470 /* Make sure depth view comparison will fail. */
512 /* Clear LRZ, used for out of renderpass depth clears. */
524 /* We cannot predict which depth subresource would be used later on,
525 * so we just pick the first one with depth cleared and clear the LRZ.
539 bool fast_clear = image->lrz_fc_size && (pDepthStencil->depth == 0.f ||
540 pDepthStencil->depth == 1.f);
594 * to depth test, in this case we potentially need to disable early
608 * conceptually stencil test and write happens before depth-test.
653 /* If depth test is disabled we shouldn't touch LRZ.
654 * Same if there is no depth attachment.
687 * depth test or shader could disable LRZ until depth buffer is cleared.
712 /* OP_ALWAYS and OP_NOT_EQUAL could have depth value of any direction,
713 * so if there is a depth write - LRZ must be disabled.
752 * LRZ buffer encodes a min/max depth value per block, but if
811 /* Without depth write it's enough to make sure that depth test