Przemek:
Harbour
ST: total application time: 25.10
MT: total application time: 27.99
xHarbour:
ST: total application time: 38.23
MT: total application time: 51.95
Harbour MT overhead to ST mode is: 11.51%
xHarbour MT overhead to ST mode is: 35.89%
xHarbour ST mode is 52.31% slower then Harbour
xHarbour MT mode is 86.60% slower then Harbour
In your results:
Harbour MT overhead to ST mode is: 10.27%
xHarbour MT overhead to ST mode is: 67.34%
xHarbour ST mode is 51.75% slower then Harbour
xHarbour MT mode is 130.26% slower then Harbour
As I said the only one noticeable difference is bigger MT overhead
in xHarbour for OS2 then in xHarbour for Linux. The rest is in practice
proportionally the same.
The overhead ~10-15 percent seems to be minimal cost we have to pay for
MT mode in Harbour and cannot be significantly reduced.
Well detailed and interesting results.
At last we can live with 10-15 % penalty for benefits of MT :-)
I'll commit both inline asm functions to SF repository in a while
and I would like to ask BCC and MinGW users to make speed tests for
MT mode for Harbour compiled with and without -DHB_NO_TLS macro.
It will be the answer if it's worth to keep them or not.
As I saw in different posts performance with -DHB_NO_TLS is better in
general with exceptions in particular, so I suppose it should be kept
I made my pending tests for comparison of two generations processors
AMD Athlon 2200+ 2.0 GHz/1MbL2, 1 Gb RAM (2003)
========================================
Each OS in own partition, with AirBoot 1.05 as boot manager
Windows, BCC 5.5.1
Harbour 1.1.0dev (Rev. 9508), Windows XP 5.1.2600 Service Pack 2
-------------
ST
total application time: 51.28
total real time: 51.58
MT
total application time: 62.28
total real time: 62.42
OS/2: eComStation 1.2 MR
Harbour 1.1.0dev (Rev. 9508), OS/2 4.50
-------------
ST
total application time: 46.42
total real time: 46.42
MT
total application time: 52.99
total real time: 53.02
OS/2: IBM OS/2 MCP
Harbour 1.1.0dev (Rev. 9508), OS/2 4.50
-------------
ST
total application time: 45.59
total real time: 45.60
MT
total application time: 53.01
total real time: 53.01
Linux Mandriva 2008
Harbour 1.1.0dev (Rev. 9507), Linux 2.6.22.18-desktop-1mdv i686
-------------
ST
total application time: 64.25
total real time: 67.25
MT
total application time: 75.71
total real time: 78.90
Comments:
- Very different results between OS in same hardware
- OS/2 is best in both versions :-)
AMD Athlon - OS/2 show a good combination for performance
- Surprisely Linux is too slow ( is desktop )
- Overhead for MT is uniform
Intel Core2Duo 2.0 GHz/4MbL2, 2 Gb RAM (Santa Rosa, July 2007)
======================================
Each OS in own partition, with AirBoot 1.05 as boot manager
Windows, BCC 5.5.1
Harbour 1.1.0dev (Rev. 9508), Windows Vista 6.0.6000
-------------
ST
total application time: 34.32
total real time: 35.19
MT
total application time: 45.97
total real time: 46.53
Harbour 1.1.0dev (Rev. 9508), Windows XP 5.1.2600 Service Pack 2
-------------
ST
total application time: 39.91
total real time: 40.78
MT
total application time: 52.20
total real time: 52.95
Harbour 1.1.0dev (Rev. 9508), Windows XP 5.1.2600 Service Pack 2
Windows under VirtualBox 2.0.2 on Linux Mandriva 2008.1
-------------
ST
total application time: 24.12
total real time: 34.39
MT
total application time: 35.39
total real time: 45.77
Linux Mandriva 2008.1
Harbour 1.1.0dev (Rev. 9507), Linux 2.6.24.4-laptop-1mnb i686
-------------
ST
total application time: 42.09
total real time: 42.09
MT
total application time: 50.37
total real time: 50.37
Comments:
- Very different results between OS with same hardware
- Surprisely Linux is slow
- Overhead for MT is uniform but greater
- Strange results for VirtualBox with 10 sec difference between
application/real times
Even so it was faster than both real Windows Vista/XP
- Disappointment for Core2Duo
I expected much better results, even results are better for same OS
Logic is: if PIV 3.3/1MbL2 has better performance than AMD Athlon
2200+/256KbL2 and Core2Duo 2.0/4MbL2 has better performance than PIV
3.3, then Core2Duo must have a better performance than AMD Athlon 2200+,
BUT to much :-)
I little justificacion: is Core2Duo mobile
David Macias
_______________________________________________
Harbour mailing list
Harbour@harbour-project.org
http://lists.harbour-project.org/mailman/listinfo/harbour
