I suppose you run it too shortly.Run test in 2 or 3 hours. Look also at
document count i have used
Regards,Karol Bieniaszewski
nullFirebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
2019. 01. 04. 13:38 keltezéssel, Gabor Boros írta:
2019. 01. 03. 16:43 keltezéssel, liviuslivius írta:
17% bigger and the difference grow with time
After change 2.5's firebird.conf (DefaultDbCachePages = 262144,
TempCacheLimit = 2147483648 and comment out CpuAffinityMask) I got the
below
2019. 01. 03. 16:43 keltezéssel, liviuslivius írta:
17% bigger and the difference grow with time
After change 2.5's firebird.conf (DefaultDbCachePages = 262144,
TempCacheLimit = 2147483648 and comment out CpuAffinityMask) I got the
below numbers. I want to know why 3.0 faster (than 2.5) for
2019. 01. 03. 16:20 keltezéssel, Leyne, Sean írta:
Gabor,
Thanks for the numbers.
It would have been more appropriate, IMO, to have simply pasted the numbers as
text in the body of your message, rather than as image attachments.
I am confused by the meaningfulness of the numbers, given that
17% bigger and the difference grow with time
Regards,Karol Bieniaszewski
nullFirebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
, which I would have fully expected to
yield different performance.
Sean
> -Original Message-
> From: Gabor Boros
> Sent: Thursday, January 3, 2019 7:27 AM
> To: firebird-devel@lists.sourceforge.net
> Subject: Re: [Firebird-devel] Some OLTP numbers
>
> 2018. 11.
2018. 11. 28. 15:44 keltezéssel, liviuslivius írta:
i have sent my settings few weeks ago, maybe you have it in the spam folder?
I tried your configuration (with one modification: test_time = 10) on
Windows 10 and the size of databases:
2.5.9.27125 SC - 2.77GB
3.0.5.33085 SS - 3.25GB
> > I do not know what was your settings.
>
>
> Same as before.
i have sent my settings few weeks ago, maybe you have it in the spam folder?
>
> > But for me, it is more interesting why db size for Fb2.5 is e.g 6GB but
> > for FB3 it is 15GB after test.
>
>
> Because 3.0 is faster for you
2018. 11. 26. 6:10 keltezéssel, liviuslivius írta:
I do not know what was your settings.
Same as before.
But for me, it is more interesting why db size for Fb2.5 is e.g 6GB but
for FB3 it is 15GB after test.
Because 3.0 is faster for you than the 2.5?
Gabor
Firebird-Devel mailing
I do not know what was your settings.But for me, it is more interesting why db
size for Fb2.5 is e.g 6GB but for FB3 it is 15GB after test.Regards,Karol
Bieniaszewski
nullFirebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
2018. 10. 22. 15:37 keltezéssel, Gabor Boros írta:
4.0 slower than 3.0 and 3.0 slower than 2.5?
More numbers, more fun. Every case executed three times and the chart
show the average of AVG_TIMES_PER_MINUTE results.
Any comment?
Gabor
Firebird-Devel mailing list, web interface at
02.11.2018 9:23, Karol Bieniaszewski wrote:
Hi,
Result for Firebird 2.5 SuperClassic is amazing.
And i do not know why FB3 here is not so fast? FB2.5 is the fastest. It do 2x
more job then FB4.
Interesting is also db size after.
In FB 2.5 SuperServer it is ~5GB
In FB 2.5 SuperClassic which
2018. 10. 25. 22:08 keltezéssel, Vlad Khorsun írta:
You set DefaultDbCachePages to 65536 and not changed
FileSystemCacheThreshold.
Thus you completely disabled file system cache.
FileSystemCacheThreshold value increased and the 2 hours cycle
re-executed. 2.5 still the fastest. Why does
2018. 10. 25. 22:08 keltezéssel, Vlad Khorsun írta:
You set DefaultDbCachePages to 65536 and not changed
FileSystemCacheThreshold.
Thus you completely disabled file system cache.
Sorry. I just followed the OLTP's readme_quick_start.txt which not wrote
anything about
25.10.2018 19:05, Gabor Boros wrote:
2018. 10. 25. 17:45 keltezéssel, Vlad Khorsun írta:
25.10.2018 18:37, Gabor Boros wrote:
2018. 10. 25. 16:16 keltezéssel, Vlad Khorsun írta:
I have the 2 hours numbers and attached the refreshed picture. 2.5 is still the fastest. Is it the truth or I made a
I have the 2 hours numbers and attached the refreshed picture. 2.5 is still the fastest. Is it the truth or I made a mistake
somewhere?
It is kinda expected. How many worker isqls you had?
Five.
Five ?! It is not serious, try 50 at least
I do not want to blow up my test machine.
2018. 10. 25. 17:52 keltezéssel, liviuslivius írta:
I am also interested
Regards,
Karol Bieniaszewski
I sent it to your address.
Gabor
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
2018. 10. 25. 17:45 keltezéssel, Vlad Khorsun írta:
25.10.2018 18:37, Gabor Boros пишет:
2018. 10. 25. 16:16 keltezéssel, Vlad Khorsun írta:
I have the 2 hours numbers and attached the refreshed picture. 2.5
is still the fastest. Is it the truth or I made a mistake somewhere?
Could you
2018. 10. 25. 17:44 keltezéssel, Vlad Khorsun írta:
25.10.2018 18:35, Gabor Boros wrote:
2018. 10. 25. 16:06 keltezéssel, Dimitry Sibiryakov írta:
25.10.2018 15:05, Gabor Boros wrote:
I have the 2 hours numbers and attached the refreshed picture. 2.5
is still the fastest. Is it the truth or I
I am also interestedRegards,Karol Bieniaszewski
nullFirebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
25.10.2018 18:37, Gabor Boros пишет:
2018. 10. 25. 16:16 keltezéssel, Vlad Khorsun írta:
I have the 2 hours numbers and attached the refreshed picture. 2.5 is still the fastest. Is it the truth or I made a mistake
somewhere?
Could you provide html reports generated by the test runs ?
25.10.2018 18:35, Gabor Boros wrote:
2018. 10. 25. 16:06 keltezéssel, Dimitry Sibiryakov írta:
25.10.2018 15:05, Gabor Boros wrote:
I have the 2 hours numbers and attached the refreshed picture. 2.5 is still the fastest. Is it the truth or I made a mistake
somewhere?
It is kinda expected.
2018. 10. 25. 16:16 keltezéssel, Vlad Khorsun írta:
I have the 2 hours numbers and attached the refreshed picture. 2.5 is
still the fastest. Is it the truth or I made a mistake somewhere?
Could you provide html reports generated by the test runs ?
Yes. Here or in private?
Gabor
2018. 10. 25. 16:06 keltezéssel, Dimitry Sibiryakov írta:
25.10.2018 15:05, Gabor Boros wrote:
I have the 2 hours numbers and attached the refreshed picture. 2.5 is
still the fastest. Is it the truth or I made a mistake somewhere?
It is kinda expected. How many worker isqls you had?
I have the 2 hours numbers and attached the refreshed picture. 2.5 is still the
fastest. Is it the truth or I made a mistake somewhere?
Could you provide html reports generated by the test runs ?
Regards,
Vlad
Firebird-Devel mailing list, web interface at
25.10.2018 15:05, Gabor Boros wrote:
I have the 2 hours numbers and attached the refreshed picture. 2.5 is still the fastest.
Is it the truth or I made a mistake somewhere?
It is kinda expected. How many worker isqls you had?
--
WBR, SD.
Firebird-Devel mailing list, web interface at
2018. 10. 23. 10:23 keltezéssel, Gabor Boros írta:
2018. 10. 22. 17:31 keltezéssel, Dimitry Sibiryakov írta:
22.10.2018 17:28, Gabor Boros wrote:
How long distance is enough for the correct results?
Usually it is run for 2-3 hours at least.
The attached numbers come from 1 hour. I will
2018. 10. 22. 17:31 keltezéssel, Dimitry Sibiryakov írta:
22.10.2018 17:28, Gabor Boros wrote:
How long distance is enough for the correct results?
Usually it is run for 2-3 hours at least.
The attached numbers come from 1 hour. I will try with 2 and 3 hours.
Gabor
Firebird-Devel
2018. 10. 22. 19:23 keltezéssel, Vlad Khorsun írta:
As for 2.5 - are you sure you test SS ?
Yes. Installed from FirebirdSS-2.5.9.27119-0.amd64.tar.gz
Gabor
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
22.10.2018 16:37, Gabor Boros wrote:
Hi All,
After executed an OLTP test with multiple Firebird SS versions got the attached numbers. The test executed three times per Firebird
version. Same OLTP configuration, same server (Debian 9.5 64bit) and same client (Windows 10 64bit). The result
22.10.2018 17:28, Gabor Boros wrote:
How long distance is enough for the correct results?
Usually it is run for 2-3 hours at least.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
2018. 10. 22. 15:41 keltezéssel, Dimitry Sibiryakov írta:
22.10.2018 15:37, Gabor Boros wrote:
Why the big difference between numbers which result from the same
version?
OLTP emul is based on random. It cannot be used as a reliable
benchmark on short distances.
How long distance is
2018. 10. 22. 16:21 keltezéssel, liviuslivius írta:
can you bring some light what that numbers mean?
Main results of an OLTP test.
https://www.ibphoenix.com/files/conf2014/Firebird-OLTP-Zotov-2014.pdf
Gabor
Firebird-Devel mailing list, web interface at
Hi,can you bring some light what that numbers mean?I do not understand your
table.I can imagine something like this:Column1: total number of requests e.g.
1000kColumn2: total timeColumn3: requests per second.and maybe alsoColumn4:
total fetchesColumn5: total readsRegards,Karol Bieniaszewski
22.10.2018 15:37, Gabor Boros wrote:
Why the big difference between numbers which result from the same version?
OLTP emul is based on random. It cannot be used as a reliable benchmark on
short distances.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
Hi All,
After executed an OLTP test with multiple Firebird SS versions got the
attached numbers. The test executed three times per Firebird version.
Same OLTP configuration, same server (Debian 9.5 64bit) and same client
(Windows 10 64bit). The result database deleted and server rebooted
36 matches
Mail list logo
