time.time() strangeness

[Nitro]

Ok, my final solution is to add the D3DCREATE_FPU_P RESERVE flag. It didn't
harm performance in a noticeable way at all. I was under the impression
SSE would be affected by this, too. Additionally I was under the
impression that float precision would suffice for time.time(). Obviously I
was blatantly wrong :-)
Thanks to Gabriel, Ross and Roel for commenting on this and sharing their
insights!

They should really make the fpu preserve flag the default. It just causes
very sneaky bugs.

-Matthias

[Roel Schroeven]

Re: time.time() strangeness

Nitro schreef:

Ok, my final solution is to add the D3DCREATE_FPU_P RESERVE flag. It didn't
harm performance in a noticeable way at all. I was under the impression
SSE would be affected by this, too. Additionally I was under the
impression that float precision would suffice for time.time(). Obviously I
was blatantly wrong :-)

Float precision might even be enough but DirectX (without the flag)
switches to even lower precision than that; only 22 bits IIRC.

They should really make the fpu preserve flag the default. It just causes
very sneaky bugs.

Absolutely.


Cheers,
Roel

[Ross Ridge]

Re: time.time() strangeness

Nitro <nitro@dr-code.orgwrote:

>They should really make the fpu preserve flag the default. It just causes
>very sneaky bugs.

They did in Direct3D 10, which doesn't change the flags. It's too late
to change the behaviour Direct3D 9 which was created a time where changing
FPU precision could have an effect on low end configurations.

Ross Ridge

--
l/ // Ross Ridge -- The Great HTMU
[oo][oo] rridge@csclub.u waterloo.ca
-()-/()/ http://www.csclub.uwaterloo.ca/~rridge/
db //