Optimized version of Rebol ASPI to DyBASE
[1/4] from: knizhnik::garret::ru at: 19-Dec-2003 13:49
New optimized version of Rebol API to DyBASE is available at
http://www.garret.ru/~knizhnik/dybase.html
I was managed to increase speed of testindex.r example 4 times by optimizing work with
hash
tables. Unfortunately, the problem is not completely solved and for
large number of cached objects, performance is still bad because of
inefficient hash implementation. I will continue investigations.
--
Best regards,
Konstantin mailto:[knizhnik--garret--ru]
[2/4] from: robert:muench:robertmuench at: 20-Dec-2003 15:47
On Fri, 19 Dec 2003 13:49:40 +0300, Konstantin Knizhnik
<[knizhnik--garret--ru]> wrote:
> New optimized version of Rebol API to DyBASE is available at
> http://www.garret.ru/~knizhnik/dybase.html
> I was managed to increase speed of testindex.r example 4 times by
> optimizing work with hash tables.
Hi, can you please update your speed comparision so we can see what effect
the optimizations have? You are the only one having access to the
reference platform. Thanks. Robert
[3/4] from: knizhnik:garret:ru at: 20-Dec-2003 21:28
Hello Robert,
Saturday, December 20, 2003, 5:47:51 PM, you wrote:
RMM> On Fri, 19 Dec 2003 13:49:40 +0300, Konstantin Knizhnik
RMM> <[knizhnik--garret--ru]> wrote:
>> New optimized version of Rebol API to DyBASE is available at
>> http://www.garret.ru/~knizhnik/dybase.html
>> I was managed to increase speed of testindex.r example 4 times by
>> optimizing work with hash tables.
RMM> Hi, can you please update your speed comparision so we can see what effect
RMM> the optimizations have? You are the only one having access to the
RMM> reference platform. Thanks. Robert
I have already done it.
Results are here. Unfortunately my "reference platform" was changed
slightly changed - after hard drive crash, I replaced disk and OS
(WinXP instead of Win2k). Also version of compiler were changed (from
VC 6.0 to VS.NET 2003). I didn't rerun all tests for all products, but
results for example for python results are worse than measured by me
before. Also DyBASE itself was significantly changed from the time
when these measurements was made. So I am going to update all results,
but it will take some time.
--
Best regards,
Konstantin mailto:[knizhnik--garret--ru]
[4/4] from: knizhnik:garret:ru at: 20-Dec-2003 21:39
Hello Konstantin,
Saturday, December 20, 2003, 9:28:25 PM, you wrote:
KK> Hello Robert,
KK> Saturday, December 20, 2003, 5:47:51 PM, you wrote:
RMM>> On Fri, 19 Dec 2003 13:49:40 +0300, Konstantin Knizhnik
RMM>> <[knizhnik--garret--ru]> wrote:
>>> New optimized version of Rebol API to DyBASE is available at
>>> http://www.garret.ru/~knizhnik/dybase.html
>>> I was managed to increase speed of testindex.r example 4 times by
>>> optimizing work with hash tables.
RMM>> Hi, can you please update your speed comparision so we can see what effect
RMM>> the optimizations have? You are the only one having access to the
RMM>> reference platform. Thanks. Robert
KK> I have already done it.
KK> Results are here.
Sorry, forget to attach the reference.
http://www.garret.ru/~knizhnik/compare.html
Comparison table is also included in DyBASE distribution.
KK> --
KK> Best regards,
KK> Konstantin mailto:[knizhnik--garret--ru]
--
Best regards,
Konstantin mailto:[knizhnik--garret--ru]