Re: [MSEide-MSEgui-talk] Benchmarks.

[Fred van Stappen]

Re-hello Sieghard.

> The correct comarison always is between corresponding items, i.e. both
> the plain versions (fpc 3.2.2 vs. fpc 3.3.1) or - if available -
> equivalent variants (fpc 3.2.2 llvm vs. fpc 3.3.1 llvm), which do not exist 
> for fpc.

Yes, I agree with you but, for the game, comparing fpc 3.3.1 vs. fpc 3.3.1 llvm 
seems to show that llvm does not do much better.

Anyway, very interesting to see the huge progress done by fpc team in float 
optimization (and the same for integer would be welcome too ;-) ).

Fre;D

___
mseide-msegui-talk mailing list
mseide-msegui-talk@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mseide-msegui-talk

Re: [MSEide-MSEgui-talk] Benchmarks.

[Fred van Stappen]

Hello Sieghard.

>Yes, it has improved a whole lot compared to fpc 3,2,2, ESPECIALLY on
>floating point operations.

In my case, because using extendly float calculation in uos sound library, it 
is a big plus.

> With integer operations, it has slowed down
> appreciably, by about 30% (i.e. it takes nearly 50% LONGER).

And this is a few sad ( and difficult to undertand, maybe change all integer 
calculation into float and then round it, should be faster? )

Very interesting benchmark.


Fre;D
___
mseide-msegui-talk mailing list
mseide-msegui-talk@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mseide-msegui-talk

Re: [MSEide-MSEgui-talk] Benchmarks.

[Sieghard via mseide-msegui-talk]

Hallo Fred,

vous ecrit au Sun, 27 Nov 2022 23:04:08 +:

> fpc 3.3.1 does better than fpc-llvm 3.3.1...

No, that is not the case. But llvm-based fpc 3.3.1 is only marginally
faster than plain fpc 3.3.1 on integer calculations and not much faster
(perhaps around 20%) on floating point (doubles) operations.

> So fpc 3.3.1 is very good

Yes, it has improved a whole lot compared to fpc 3,2,2, ESPECIALLY on
floating point operations. With integer operations, it has slowed down
appreciably, by about 30% (i.e. it takes nearly 50% LONGER).

> or fpc-llvm 3.3.1 dont use yet all the power of llvm
> or llvm is a joke.

I'd not say hat, llvm is yet another C/C++ compiler for many systems,
and it probabely has its place. And, llvm certainly has a place as the
tool to build compilers for other languages that it is.
There may be established "competitors" it DOES improve a lot against,
e.g. gcc is not usually called notoriously speedy.
But fpc, i.e. fpc 3.3.1, obviously is not one of them. And what these
results seem to display is that it is probabaly a bad idea to compare
a variant of an advanced version of an application with an older plain
version of it. E.g., fpc 3.2.2 and fpc 3.3.1 llvm. They DO NOT COMPARE.
The correct comarison always is between corresponding items, i.e. both
the plain versions (fpc 3.2.2 vs. fpc 3.3.1) or - if available -
equivalent variants (fpc 3.2.2 llvm vs. fpc 3.3.1 llvm), which do not
exist for fpc.

So, this seems to suggest that the llvm version of fpc is not worth
switching to it, for now at least, for fpc 3.3.1. This could change in
the future, but it WILL require a thorough examination of the results.
Much more thorough even than I did here, if it should be definitive.

-- 
(Weitergabe von Adressdaten, Telefonnummern u.ä. ohne Zustimmung
nicht gestattet, ebenso Zusendung von Werbung oder ähnlichem)
---
Mit freundlichen Grüßen, S. Schicktanz
---



___
mseide-msegui-talk mailing list
mseide-msegui-talk@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mseide-msegui-talk