Re: [HACKERS] Should we cacheline align PGXACT?

Okay. As suggested by Alexander, I have changed the order of reading and doing initdb for each pgbench run. With these changes, I got following results at 300 scale factor with 8GB of shared buffer.

pgbench settings:
pgbench -i -s 300 postgres
pgbench -M prepared -c $thread -j $thread -T $time_for_reading  postgres

where, time_for_reading = 30mins

non default GUC param
shared_buffers=8GB
max_connections=300

pg_xlog is located in SSD.

CLIENT COUNT	TPS (HEAD)	TPS (PATCH)	% IMPROVEMENT
4	2803	2843	1.427042455
8	5315	5225	-1.69332079
32	19755	19669	-0.4353328271
64	28679	27980	-2.437323477
128	28887	28008	-3.042891266
156	27465	26728	-2.683415256
180	27425	26697	-2.654512306
196	28826	27396	-4.960799278
256	29787	28107	-5.640044315

The test machine details are as follows,

Architecture:          x86_64
CPU op-mode(s):        32-bit, 64-bit
Byte Order:            Little Endian
CPU(s):                128
On-line CPU(s) list:   0-127
Thread(s) per core:    2
Core(s) per socket:    8
Socket(s):             8
NUMA node(s):          8
Vendor ID:             GenuineIntel
CPU family:            6
Model:                 47
Model name:            Intel(R) Xeon(R) CPU E7- 8830  @ 2.13GHz

Ashutosh Sharma