AlexIndustrial/mesa
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

radv: Use flush postamble on GFX7 with different flags.

Browse Source 
Flush caches at the end of each submission on GFX7.
This significantly improves stability on Hawaii
when running the CTS on multiple threads.

Keep previous behaviour on GFX6.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33563>
This commit is contained in:
Timur Kristóf
2025-02-22 09:07:18 +01:00
committed by Marge Bot
parent 7f6e28db26
commit 3f3a5d8068
1 changed files with 14 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
src/amd/vulkan/radv_queue.c
View File

@@ -1338,11 +1338,9 @@ radv_update_preambles(struct radv_queue_state *queue, struct radv_device *device
return radv_update_preamble_cs(queue, device, &needs);
}
/* Creates a postamble CS that executes cache flush commands
/**
* Creates a postamble CS that executes cache flush commands
* that we can use at the end of each submission.
*
* GFX6: The kernel flushes L2 before shaders are finished.
* Therefore we need to wait for idle at the end of each submission.
*/
static VkResult
radv_create_flush_postamble(struct radv_queue *queue)
@@ -1359,10 +1357,18 @@ radv_create_flush_postamble(struct radv_queue *queue)
radeon_check_space(ws, cs, 256);
const enum amd_gfx_level gfx_level = pdev->info.gfx_level;
enum radv_cmd_flush_bits flush_bits = RADV_CMD_FLAG_CS_PARTIAL_FLUSH | RADV_CMD_FLAG_WB_L2;
enum radv_cmd_flush_bits flush_bits = 0;
if (ip == AMD_IP_GFX)
flush_bits |= RADV_CMD_FLAG_PS_PARTIAL_FLUSH;
if (gfx_level == GFX6) {
/* GFX6: The kernel flushes L2 before shaders are finished. */
flush_bits = RADV_CMD_FLAG_CS_PARTIAL_FLUSH | RADV_CMD_FLAG_WB_L2;
if (ip == AMD_IP_GFX)
flush_bits |= RADV_CMD_FLAG_PS_PARTIAL_FLUSH;
} else {
/* Improves stability on Hawaii. */
flush_bits =
RADV_CMD_FLAG_INV_ICACHE | RADV_CMD_FLAG_INV_SCACHE | RADV_CMD_FLAG_INV_VCACHE | RADV_CMD_FLAG_INV_L2;
}
enum rgp_flush_bits sqtt_flush_bits = 0;
radv_cs_emit_cache_flush(ws, cs, gfx_level, NULL, 0, queue->state.qf, flush_bits, &sqtt_flush_bits, 0);
@@ -1969,7 +1975,7 @@ radv_queue_init(struct radv_device *device, struct radv_queue *queue, int idx,
goto fail;
}
if (pdev->info.gfx_level == GFX6 &&
if (pdev->info.gfx_level <= GFX7 &&
(queue->state.qf == RADV_QUEUE_GENERAL || queue->state.qf == RADV_QUEUE_COMPUTE)) {
result = radv_create_flush_postamble(queue);
if (result != VK_SUCCESS)