Hitnrun
Member
- Joined
- Mar 1, 2008
- Messages
- 427
I found this while looking for allegro information:
http://osdir.com/ml/handhelds.maemo.devel/...3/msg00168.html
It's basically memset and memcpy ARM-assembler optimized. Anyone tried it, does it make any difference?
I run it on my gp2x, I got this:
CODE
--- running correctness tests ---
all the correctness tests passed
--- running performance tests (memory bandwidth benchmark) ---:
memset() memory bandwidth: 223.10MB/s
memset8() memory bandwidth: 262.14MB/s
memcpy() memory bandwidth (perfectly aligned): 49.00MB/s
memcpy16() memory bandwidth (perfectly aligned): 51.15MB/s
memcpy() memory bandwidth (16-bit aligned): 40.25MB/s
memcpy16() memory bandwidth (16-bit aligned): 46.60MB/s
--- testing performance for random blocks (size 0-15 bytes) ---
memset time: 0.600
memset8 time: 0.610
--- testing performance for random blocks (size 0-511 bytes) ---
memset time: 1.740
memset8 time: 1.550
Can it give any additional performance?
http://osdir.com/ml/handhelds.maemo.devel/...3/msg00168.html
It's basically memset and memcpy ARM-assembler optimized. Anyone tried it, does it make any difference?
I run it on my gp2x, I got this:
CODE
--- running correctness tests ---
all the correctness tests passed
--- running performance tests (memory bandwidth benchmark) ---:
memset() memory bandwidth: 223.10MB/s
memset8() memory bandwidth: 262.14MB/s
memcpy() memory bandwidth (perfectly aligned): 49.00MB/s
memcpy16() memory bandwidth (perfectly aligned): 51.15MB/s
memcpy() memory bandwidth (16-bit aligned): 40.25MB/s
memcpy16() memory bandwidth (16-bit aligned): 46.60MB/s
--- testing performance for random blocks (size 0-15 bytes) ---
memset time: 0.600
memset8 time: 0.610
--- testing performance for random blocks (size 0-511 bytes) ---
memset time: 1.740
memset8 time: 1.550
Can it give any additional performance?