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

panfrost: Separate core ID range from core count

Browse Source 
To query the core count, the hardware has a SHADERS_PRESENT register containing
a mask of shader cores connected. The core count equals the number of 1-bits,
regardless of placement. This value is useful for public consumption (like
in clinfo).

However, internally we are interested in the range of core IDs.
We usually query core count to determine how many cores to allocate various
per-core buffers for (performance counters, occlusion queries, and the stack).
In each case, the hardware writes at the index of its core ID, so we have to
allocate enough for entire range of core IDs. If the core mask is
discontiguous, this necessarily overallocates.

Rename the existing core_count to core_id_range, better reflecting its
definition and purpose, and repurpose core_count for the actual core count.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17265>
This commit is contained in:
Alyssa Rosenzweig
2022-06-24 17:43:09 -04:00
committed by Marge Bot
parent 5aa740bc8e
commit 6f3eea5ddb
11 changed files with 33 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/gallium/drivers/panfrost/pan_cmdstream.c
+4 -4
View File
@@ -1585,7 +1585,7 @@ panfrost_emit_shared_memory(struct panfrost_batch *batch,
panfrost_batch_get_scratchpad(batch,
ss->info.tls_size,
dev->thread_tls_alloc,
dev->core_count);
dev->core_id_range);
info.tls.ptr = bo->ptr.gpu;
}
@@ -1593,7 +1593,7 @@ panfrost_emit_shared_memory(struct panfrost_batch *batch,
unsigned size =
pan_wls_adjust_size(info.wls.size) *
pan_wls_instances(&info.wls.dim) *
dev->core_count;
dev->core_id_range;
struct panfrost_bo *bo =
panfrost_batch_get_shared_memory(batch, size, 1);
@@ -2732,7 +2732,7 @@ emit_tls(struct panfrost_batch *batch)
panfrost_batch_get_scratchpad(batch,
batch->stack_size,
dev->thread_tls_alloc,
dev->core_count):
dev->core_id_range):
NULL;
struct pan_tls_info tls = {
.tls = {
@@ -2754,7 +2754,7 @@ emit_fbd(struct panfrost_batch *batch, const struct pan_fb_info *fb)
panfrost_batch_get_scratchpad(batch,
batch->stack_size,
dev->thread_tls_alloc,
dev->core_count):
dev->core_id_range):
NULL;
struct pan_tls_info tls = {
.tls = {
src/gallium/drivers/panfrost/pan_context.c
+2 -2
View File
@@ -869,7 +869,7 @@ panfrost_begin_query(struct pipe_context *pipe, struct pipe_query *q)
case PIPE_QUERY_OCCLUSION_COUNTER:
case PIPE_QUERY_OCCLUSION_PREDICATE:
case PIPE_QUERY_OCCLUSION_PREDICATE_CONSERVATIVE: {
unsigned size = sizeof(uint64_t) * dev->core_count;
unsigned size = sizeof(uint64_t) * dev->core_id_range;
/* Allocate a resource for the query results to be stored */
if (!query->rsrc) {
@@ -953,7 +953,7 @@ panfrost_get_query_result(struct pipe_context *pipe,
if (query->type == PIPE_QUERY_OCCLUSION_COUNTER) {
uint64_t passed = 0;
for (int i = 0; i < dev->core_count; ++i)
for (int i = 0; i < dev->core_id_range; ++i)
passed += result[i];
if (dev->arch <= 5 && !query->msaa)
src/gallium/drivers/panfrost/pan_job.c
+2 -2
View File
@@ -388,11 +388,11 @@ struct panfrost_bo *
panfrost_batch_get_scratchpad(struct panfrost_batch *batch,
unsigned size_per_thread,
unsigned thread_tls_alloc,
unsigned core_count)
unsigned core_id_range)
{
unsigned size = panfrost_get_total_stack_size(size_per_thread,
thread_tls_alloc,
core_count);
core_id_range);
if (batch->scratchpad) {
assert(batch->scratchpad->size >= size);
src/gallium/drivers/panfrost/pan_job.h
+1 -1
View File
@@ -251,7 +251,7 @@ void
panfrost_batch_adjust_stack_size(struct panfrost_batch *batch);
struct panfrost_bo *
panfrost_batch_get_scratchpad(struct panfrost_batch *batch, unsigned size, unsigned thread_tls_alloc, unsigned core_count);
panfrost_batch_get_scratchpad(struct panfrost_batch *batch, unsigned size, unsigned thread_tls_alloc, unsigned core_id_range);
struct panfrost_bo *
panfrost_batch_get_shared_memory(struct panfrost_batch *batch, unsigned size, unsigned workgroup_count);