Please file a bug AND if you find a simple shader that has the problem
please attach it (or in the worst case attach what you have that can
repro the failure)
BTW have you tried other HW? Are you checking for SW fallback?
What happens if you force SW?
- Nick
On Mar 3, 2008, at 2:41 PM, Jocelyn Houle wrote:
(forgot to reply to all, so allow me a quick follow up)
It sounds like I might be hitting some compiler limits somewhere,
since simplifying my shader makes exp work again.
I still don't have a nice reproducer, but I thought I should
publicly state that the exp function does indeed work in most
circumstances, and I only seem to be hitting a corner case.
Unless people start replying to this thread, I'll likely only
continue inside my bug report (#5777193).
On 2008-03-03, at 16:41 , Nick Burns wrote:
Please list ...
1 - CPU (PPC? x86? / MacBook? Mac Pro?)
2 - GFX (ATI x1xxx? NV GeForce7xxx)
3 - OS (Tiger? Leopard? What build?)
4 - Shader that has the problem + directions to see problem (either
GLUT or GLSLEditorSample)
- Nick
On Mar 3, 2008, at 9:46 AM, Jocelyn Houle wrote:
It seems like the 'exp' function is not quite up to spec in
fragment programs. Unless I'm mistaken, it should yield e^x
(while exp2() should yield 2^x).
Of course, I can fake it by calling pow(2.71828182846, x), but
somehow, it doesn't feel right! ;-)
Hum... Now, I haven't tested the vertex shader version, but this
sounds like... how should I put it... a suboptimal situation...?
Did anyone else file a bug on that one?
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Mac-opengl mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Mac-opengl mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden
