Thread: Postgres process count GCC vs Clang is Different on autovaccum=on
Software/Hardware used:
Command used to count process: ps -eaf | grep postgres
Case2:AutoVaccum=off
===================
PostgresV14.v
OS: RHELv8.4
Benchmark:HammerDB v4.3
Hardware used: Apple/AMD Ryzen.
RAM size: 256 GB
SSD/HDD: 1TB
CPU(s): 256(0-255)
Thread(s) per core: 2
Core(s) per socket: 64
Socket(s): 2
NUMA node(s): 8
OS: RHELv8.4
Benchmark:HammerDB v4.3
Hardware used: Apple/AMD Ryzen.
RAM size: 256 GB
SSD/HDD: 1TB
CPU(s): 256(0-255)
Thread(s) per core: 2
Core(s) per socket: 64
Socket(s): 2
NUMA node(s): 8
Command used to count process: ps -eaf | grep postgres
Case1: AutoVaccum=on
vu | GCC | Clang |
32 | 43 | 42 |
64 | 76 | 74 |
192 | 203 | 202 |
250 | 262 | 262 |
Case2:AutoVaccum=off
vu | GCC | Clang |
32 | 40 | 40 |
64 | 72 | 72 |
192 | 200 | 200 |
250 | 261 | 263 |
In Case1 why is the process different in Clang vs GCC.
In postgresql process dependent wrt compiler GCC/Clang?
Is any recommendation or suggestion to check on this in Postgresv14
In postgresql process dependent wrt compiler GCC/Clang?
Is any recommendation or suggestion to check on this in Postgresv14
On 11/24/21 13:05, hpc researcher_mspk wrote: > Software/Hardware used: > =================== > PostgresV14.v > OS: RHELv8.4 > Benchmark:HammerDB v4.3 > Hardware used: Apple/AMD Ryzen. > RAM size: 256 GB > SSD/HDD: 1TB > CPU(s): 256(0-255) > Thread(s) per core: 2 > Core(s) per socket: 64 > Socket(s): 2 > NUMA node(s): 8 > > Command used to count process: ps -eaf | grep postgres > > Case1: AutoVaccum=on > vu GCC Clang > 32 43 42 > 64 76 74 > 192 203 202 > 250 262 262 > > > Case2:AutoVaccum=off > vu GCC Clang > 32 40 40 > 64 72 72 > 192 200 200 > 250 261 263 > > > In Case1 why is the process different in Clang vs GCC. > In postgresql process dependent wrt compiler GCC/Clang? No, it's not. The most likely explanation is that you're seeing different number of autovacuum workers. Those are dynamic, i.e. may appear/disappear. Or maybe there are more connections to the DB. regards -- Tomas Vondra EnterpriseDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company