Reducing Bucket index calculation to O(1).
This algorithm calculates the index using matrix method. Assuming
PAGE_SIZE is 4096, matrix arrangement is as below:
1*4096 2*4096 3*4096 4*4096
5*4096 6*4096 7*4096 8*4096
10*4096 12*4096 14*4096 16*4096
20*4096 24*4096 28*4096 32*4096
... ... ... ...
... ... ... ...
... ... ... max_cache_size
From this matrix its clearly seen that every row follows the below way:
... ... ... n
n+(1/4)n n+(1/2)n n+(3/4)n 2n
Row is calculated as log2(size/PAGE_SIZE) Column is calculated as
converting the difference between the elements to fit into power size of
two and indexing it.
Final Index is (row*4)+(col-1)
Tested with Intel Mesa CI.
Improves performance of 3DMark on BXT by 0.705966% +/- 0.229767% (n=20)
v4: Review comments on style and code comments implemented (Ian).
v3: Review comments implemented (Ian).
v2: Review comments implemented (Jason).
Signed-off-by: Aravindan Muthukumar <aravindan.muthukumar@intel.com>
Signed-off-by: Kedar Karanje <kedar.j.karanje@intel.com>
Reviewed-by: Yogesh Marathe <yogesh.marathe@intel.com>
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
#define memclear(s) memset(&s, 0, sizeof(s))
#define memclear(s) memset(&s, 0, sizeof(s))
+#define PAGE_SIZE 4096
+
#define FILE_DEBUG_FLAG DEBUG_BUFMGR
static inline int
#define FILE_DEBUG_FLAG DEBUG_BUFMGR
static inline int
return ALIGN(pitch, tile_width);
}
return ALIGN(pitch, tile_width);
}
+/**
+ * This function finds the correct bucket fit for the input size.
+ * The function works with O(1) complexity when the requested size
+ * was queried instead of iterating the size through all the buckets.
+ */
static struct bo_cache_bucket *
bucket_for_size(struct brw_bufmgr *bufmgr, uint64_t size)
{
static struct bo_cache_bucket *
bucket_for_size(struct brw_bufmgr *bufmgr, uint64_t size)
{
+ /* Calculating the pages and rounding up to the page size. */
+ const unsigned pages = (size + PAGE_SIZE - 1) / PAGE_SIZE;
+
+ /* Row Bucket sizes clz((x-1) | 3) Row Column
+ * in pages stride size
+ * 0: 1 2 3 4 -> 30 30 30 30 4 1
+ * 1: 5 6 7 8 -> 29 29 29 29 4 1
+ * 2: 10 12 14 16 -> 28 28 28 28 8 2
+ * 3: 20 24 28 32 -> 27 27 27 27 16 4
+ */
+ const unsigned row = 30 - __builtin_clz((pages - 1) | 3);
+ const unsigned row_max_pages = 4 << row;
- for (i = 0; i < bufmgr->num_buckets; i++) {
- struct bo_cache_bucket *bucket = &bufmgr->cache_bucket[i];
- if (bucket->size >= size) {
- return bucket;
- }
- }
+ /* The '& ~2' is the special case for row 1. In row 1, max pages /
+ * 2 is 2, but the previous row maximum is zero (because there is
+ * no previous row). All row maximum sizes are power of 2, so that
+ * is the only case where that bit will be set.
+ */
+ const unsigned prev_row_max_pages = (row_max_pages / 2) & ~2;
+ int col_size_log2 = row - 1;
+ col_size_log2 += (col_size_log2 < 0);
+ const unsigned col = (pages - prev_row_max_pages +
+ ((1 << col_size_log2) - 1)) >> col_size_log2;
+
+ /* Calculating the index based on the row and column. */
+ const unsigned index = (row * 4) + (col - 1);
+
+ return (index < bufmgr->num_buckets) ?
+ &bufmgr->cache_bucket[index] : NULL;
list_inithead(&bufmgr->cache_bucket[i].head);
bufmgr->cache_bucket[i].size = size;
bufmgr->num_buckets++;
list_inithead(&bufmgr->cache_bucket[i].head);
bufmgr->cache_bucket[i].size = size;
bufmgr->num_buckets++;
+
+ assert(bucket_for_size(bufmgr, size) == &bufmgr->cache_bucket[i]);
+ assert(bucket_for_size(bufmgr, size - 2048) == &bufmgr->cache_bucket[i]);
+ assert(bucket_for_size(bufmgr, size + 1) != &bufmgr->cache_bucket[i]);