diff options
author | Tom Lane <tgl@sss.pgh.pa.us> | 2021-07-25 14:02:27 -0400 |
---|---|---|
committer | Tom Lane <tgl@sss.pgh.pa.us> | 2021-07-25 14:02:27 -0400 |
commit | 28d936031a86d94806c6604480ff3f3f169b371c (patch) | |
tree | 5484c0b625433483dbff26b9da48881c8e3426c0 /src/backend/executor/nodeAgg.c | |
parent | d9d8aa9bb9aa2a850f53facd2c91e454129af57b (diff) | |
download | postgresql-28d936031a86d94806c6604480ff3f3f169b371c.tar.gz postgresql-28d936031a86d94806c6604480ff3f3f169b371c.zip |
Get rid of artificial restriction on hash table sizes on Windows.
The point of introducing the hash_mem_multiplier GUC was to let users
reproduce the old behavior of hash aggregation, i.e. that it could use
more than work_mem at need. However, the implementation failed to get
the job done on Win64, where work_mem is clamped to 2GB to protect
various places that calculate memory sizes using "long int". As
written, the same clamp was applied to hash_mem. This resulted in
severe performance regressions for queries requiring a bit more than
2GB for hash aggregation, as they now spill to disk and there's no
way to stop that.
Getting rid of the work_mem restriction seems like a good idea, but
it's a big job and could not conceivably be back-patched. However,
there's only a fairly small number of places that are concerned with
the hash_mem value, and it turns out to be possible to remove the
restriction there without too much code churn or any ABI breaks.
So, let's do that for now to fix the regression, and leave the
larger task for another day.
This patch does introduce a bit more infrastructure that should help
with the larger task, namely pg_bitutils.h support for working with
size_t values.
Per gripe from Laurent Hasson. Back-patch to v13 where the
behavior change came in.
Discussion: https://postgr.es/m/997817.1627074924@sss.pgh.pa.us
Discussion: https://postgr.es/m/MN2PR15MB25601E80A9B6D1BA6F592B1985E39@MN2PR15MB2560.namprd15.prod.outlook.com
Diffstat (limited to 'src/backend/executor/nodeAgg.c')
-rw-r--r-- | src/backend/executor/nodeAgg.c | 42 |
1 files changed, 23 insertions, 19 deletions
diff --git a/src/backend/executor/nodeAgg.c b/src/backend/executor/nodeAgg.c index 914b02ceee4..39bea204d16 100644 --- a/src/backend/executor/nodeAgg.c +++ b/src/backend/executor/nodeAgg.c @@ -1802,15 +1802,15 @@ hash_agg_set_limits(double hashentrysize, double input_groups, int used_bits, { int npartitions; Size partition_mem; - int hash_mem = get_hash_mem(); + Size hash_mem_limit = get_hash_memory_limit(); /* if not expected to spill, use all of hash_mem */ - if (input_groups * hashentrysize < hash_mem * 1024L) + if (input_groups * hashentrysize <= hash_mem_limit) { if (num_partitions != NULL) *num_partitions = 0; - *mem_limit = hash_mem * 1024L; - *ngroups_limit = *mem_limit / hashentrysize; + *mem_limit = hash_mem_limit; + *ngroups_limit = hash_mem_limit / hashentrysize; return; } @@ -1835,10 +1835,10 @@ hash_agg_set_limits(double hashentrysize, double input_groups, int used_bits, * minimum number of partitions, so we aren't going to dramatically exceed * work mem anyway. */ - if (hash_mem * 1024L > 4 * partition_mem) - *mem_limit = hash_mem * 1024L - partition_mem; + if (hash_mem_limit > 4 * partition_mem) + *mem_limit = hash_mem_limit - partition_mem; else - *mem_limit = hash_mem * 1024L * 0.75; + *mem_limit = hash_mem_limit * 0.75; if (*mem_limit > hashentrysize) *ngroups_limit = *mem_limit / hashentrysize; @@ -1992,32 +1992,36 @@ static int hash_choose_num_partitions(double input_groups, double hashentrysize, int used_bits, int *log2_npartitions) { - Size mem_wanted; - int partition_limit; + Size hash_mem_limit = get_hash_memory_limit(); + double partition_limit; + double mem_wanted; + double dpartitions; int npartitions; int partition_bits; - int hash_mem = get_hash_mem(); /* * Avoid creating so many partitions that the memory requirements of the * open partition files are greater than 1/4 of hash_mem. */ partition_limit = - (hash_mem * 1024L * 0.25 - HASHAGG_READ_BUFFER_SIZE) / + (hash_mem_limit * 0.25 - HASHAGG_READ_BUFFER_SIZE) / HASHAGG_WRITE_BUFFER_SIZE; mem_wanted = HASHAGG_PARTITION_FACTOR * input_groups * hashentrysize; /* make enough partitions so that each one is likely to fit in memory */ - npartitions = 1 + (mem_wanted / (hash_mem * 1024L)); + dpartitions = 1 + (mem_wanted / hash_mem_limit); + + if (dpartitions > partition_limit) + dpartitions = partition_limit; - if (npartitions > partition_limit) - npartitions = partition_limit; + if (dpartitions < HASHAGG_MIN_PARTITIONS) + dpartitions = HASHAGG_MIN_PARTITIONS; + if (dpartitions > HASHAGG_MAX_PARTITIONS) + dpartitions = HASHAGG_MAX_PARTITIONS; - if (npartitions < HASHAGG_MIN_PARTITIONS) - npartitions = HASHAGG_MIN_PARTITIONS; - if (npartitions > HASHAGG_MAX_PARTITIONS) - npartitions = HASHAGG_MAX_PARTITIONS; + /* HASHAGG_MAX_PARTITIONS limit makes this safe */ + npartitions = (int) dpartitions; /* ceil(log2(npartitions)) */ partition_bits = my_log2(npartitions); @@ -2030,7 +2034,7 @@ hash_choose_num_partitions(double input_groups, double hashentrysize, *log2_npartitions = partition_bits; /* number of partitions will be a power of two */ - npartitions = 1L << partition_bits; + npartitions = 1 << partition_bits; return npartitions; } |