Am 15.12.2015 um 13:49 schrieb Dimitris Chloupis <kilon.alios@gmail.com>:

30% faster , not bad at all

What about the sudden peak at 24 of August , what caused that ?

There are differences on each run. This depends on what is running at that time on the machine. If I remember correctly this particular peak was part of a bigger refactoring. You can see the curve going slightly down the subsequent runs. These were gains of that refactoring. And it is part of "making it right" not "making it fast" :) Norbert

On Tue, Dec 15, 2015 at 2:32 PM Norbert Hartl <norbert@hartl.name <mailto:norbert@hartl.name>> wrote: Here is a picture of our performance build for TCAP. We just switched to a spur image. I think you can see when we did it.

You can see it here: https://ci.hartl.name/job/TCAP-pharo5-nightly/plot/ <https://ci.hartl.name/job/TCAP-pharo5-nightly/plot/>

Norbert

<Bildschirmfoto 2015-12-15 um 13.28.28.png><Bildschirmfoto 2015-12-15 um 13.28.28.png>

Am 15.12.2015 um 14:09 schrieb Sven Van Caekenberghe <sven@stfx.eu>:

Impressive !

If only we knew what TCAP and the benchmark were ;-)

https://en.wikipedia.org/wiki/Transaction_Capabilities_Application_Part <https://en.wikipedia.org/wiki/Transaction_Capabilities_Application_Part> It is the first link when you search for TCAP in google :P The benchmark is decoding a couple of binary messages with the TCAP stack. The code can be found in the TCAP-Benchmark package in the TCAP Project on smalltalkhub here: http://smalltalkhub.com/#!/~osmocom/TCAP <http://smalltalkhub.com/#!/~osmocom/TCAP> Norbert

...

On 15 Dec 2015, at 13:31, Norbert Hartl <norbert@hartl.name> wrote:

Here is a picture of our performance build for TCAP. We just switched to a spur image. I think you can see when we did it.

You can see it here: https://ci.hartl.name/job/TCAP-pharo5-nightly/plot/

Norbert

<Bildschirmfoto 2015-12-15 um 13.28.28.png>