[Bug 37028] Amnesia/HPL2 Demo: Strange graphical bugs on r600g

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=37028

--- Comment #15 from Maggioni Marcello <hayarms at gmail.com> 2011-06-07 11:46:29 PDT ---
Yes, you are perfectly right, I looked better to the application and you are
right, but considering that you don't set the "random seed" value with srand
the random sequence is always the same, so the first execution of the of the
"while" cycle should always have the same 68698 samples drawn. So at least in
the first cycle my test app should report

Correct: 50
Non correct: 0

right?


This is an example of a sequence of values reported in the first cycle:

Param is: 68698
Param is: 2147483647
Param is: 2147483647
Param is: 2147483647
Param is: 2147483647
Param is: 68698
Param is: 2147483647
Param is: 68698
Param is: 68698
Param is: 2147483647
Param is: 68698
Param is: 68698
Param is: 68698
Param is: 2147483647
Param is: 2147483647
Param is: 2147483647
Param is: 2147483647
Param is: 68698
Param is: 68698
Param is: 68698
Param is: 68698

The patch is applied, of this I'm sure.

I tried reducing the value of the "size" variable (as you suggested) to 16
(instead of the value that usually is set by num_results that is 32), and the
result is :

Param is: 34594
Param is: 2147483647
Param is: 34594
Param is: 34594
Param is: 34594
Param is: 2147483647
Param is: 34594
Param is: 34594
Param is: 34594
Param is: 34594
Param is: 2147483647
Param is: 34594
Param is: 34594
Param is: 2147483647
Param is: 34594

The correct values are almost half of the ones with "size = query->num_results"
and the non correct ones are the same ...

It seems like the memory gets corrupted or not correctly initialized sometimes.

Any idea?

-- 
Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.