Bringing Up Advancements In Compiler Testing

[Michael]

yet again i ask for a full x264 supplied "checkasm --bench" , as that would be a very good option to add to the Phoronix standard tests too for both old and new platforms

Patches / new test profiles are always welcome too.

[popper]

Patches / new test profiles are always welcome too.

great, lets hope some PHP dev writes some one day, and you get around to writing a simple working and tested howto write test and finaly integrate a working PHP test profile for your data collecting product where people can actually find it so people can help you if they are so inclined to for free.

[Michael]

great, lets hope some PHP dev writes some one day, and you get around to writing a simple working and tested howto write test and finaly integrate a working PHP test profile for your data collecting product where people can actually find it so people can help you if they are so inclined to for free.

Huh? There's already many independent people creating their own test profiles and test suites.... There is some documentation as well but looking at other existing test profiles is often the easiest and simplest way. Test profiles don't even touch PHP at all but is just a set of bash scripts and XML files.

[enrico.tagliavini]

Very nice feature, be sure to highlight the difference between the same compiler but with different version (e.g. gcc-4.4 vs gcc-4.6)

Thank you very much

[Michael]

Very nice feature, be sure to highlight the difference between the same compiler but with different version (e.g. gcc-4.4 vs gcc-4.6)

That's long been reflected from the system information table.

[enrico.tagliavini]

That's long been reflected from the system information table.

Eheh true, but I always forgot about it .

[ulenrich]

1. Why pushing a symlink of an ought to be compiler, which is kept after a crash. Why not simply a bind mount which vanishes after a reboot?

Other topic, optical easen of results:

2. Why not, if results are better the other direction, this also optical, as follows:

normal results:
|xxxx
|xx

other way round also optical other direction - more obvious:
| xxxx|
| xxxxxx|

[Michael]

1. Why pushing a symlink of an ought to be compiler, which is kept after a crash. Why not simply a bind mount which vanishes after a reboot?

This way is more compatible and theoretically will even work under Windows (I just haven't done my Windows qualification tests yet).

2. Why not, if results are better the other direction, this also optical, as follows:

normal results:
|xxxx
|xx

other way round also optical other direction - more obvious:
| xxxx|
| xxxxxx|

That's the first time I've heard that request, nor heard any complaints of the bar graph layout from enterprise users, so it hasn't been a problem.

[ulenrich]

This way is more compatible and theoretically will even work under Windows (I just haven't done my Windows qualification tests yet).

This anti-crash misconfig feature of a bind mount just for linux would be implementable totally compatible by using a simple
if ENVIRONMENT statement. Though very unprofessional not to do it ...

That's the first time I've heard that request, nor heard any complaints of the bar graph layout from enterprise users, so it hasn't been a problem.

Yes, professionals which do often benchmarking, are used to look up the direction. And me also learned this. But I remember when I looked up your website for results the first time. I wondered why so many results where contradiciting my expectations.
This "optical" feature is for easy and fast look up results mainly for beginners and first time visitors of your benchmarking website!

[Michael]

This anti-crash misconfig feature of a bind mount just for linux would be implementable totally compatible by using a simple
if ENVIRONMENT statement. Though very unprofessional not to do it ...

There's no threat of a mis-config in this scenario for how it's implemented.