Zoals ik wel verwacht had, heeft Tnet geen gehoor gegeven aan m'n nieuwsubmit over het dramatisch inleveren van fillrate als anisotropische filtering wordt gebruikt bij de GF4, de performancehit is zelfs zo hoog dat de GF3Ti500 op dat moment een grotere fillrate heeft!
Ik ondekte zelf al direkt dat de kaart zeer snel was, maar zodra ik aniso op 2x zette, de performance nauwelijks boven die van de GF3(Ti500) ligt.
Enkele sites maakte hiervan een opmerking, maarja verder werd er geen gehoor aangegeven, dat gebeurd alleen als ATI iets fout doet ;)vooral nVidia's asslicker 1e klass, Anand Shimpi, rept er geen woord over terwijl hij donders goed op de hoogte is.
Vanochtend las ik op nvnewsnet weer eens dat iemand wakker wilde worden en het issue duidelijk aan de kaak stelde.
Hier praten we over:
a problem! No this isn't space related, I promise you. I've decided to post this on the front page, in part, to have the powers-that-be take notice of this problem.
If you read my Leadtek Ti4400 preview last week, you may have taken notice of some graphs which I tried to illustrate why (at least partially) the GeForce4 would tank in performance when anisotropic filtering was enabled.
I was able to show that both OpenGL and Direct3D exhibited significant fill-rate reduction, to the point where I was left scratching my head. I basically wanted to prove, beyond a shadow of a doubt, that NVIDIA has a nasty bug on their hands that needs to be addressed and/or acknowledged.
This is what I did. I simply removed the GeForce4 Ti 4400 from my system, and slapped in the GeForce3. I re-ran the same tests, so let's take a look. BTW, the GeForce3 is clocked to Ti 500 levels.
First up is Serious Sam2.


The bottom line is that the GeForce3 is kicking out ~50% more fill-rate than the GeForce4 when anisotropic filtering is enabled.
MikeC: This has been an unresolved issue from day 1 with the GeForce4. Inquiring minds want to know...
De bottomline er is dat er mogelijk wat mis met de drivers of de hardware, ik hoop het eerste, zodat we nog enige verbetering kunnen verwachten, maar bovenstaande grafieken en m'n eigen ervaring zeggen dat er iet niet lekker ziet.
IK speel nou zelf bv. met een lod van -1 en zet aniso lekker uit, want je verliest gewoon 50% van je renderingspeed...
Ik ondekte zelf al direkt dat de kaart zeer snel was, maar zodra ik aniso op 2x zette, de performance nauwelijks boven die van de GF3(Ti500) ligt.
Enkele sites maakte hiervan een opmerking, maarja verder werd er geen gehoor aangegeven, dat gebeurd alleen als ATI iets fout doet ;)vooral nVidia's asslicker 1e klass, Anand Shimpi, rept er geen woord over terwijl hij donders goed op de hoogte is.
Vanochtend las ik op nvnewsnet weer eens dat iemand wakker wilde worden en het issue duidelijk aan de kaak stelde.
Hier praten we over:
a problem! No this isn't space related, I promise you. I've decided to post this on the front page, in part, to have the powers-that-be take notice of this problem.
If you read my Leadtek Ti4400 preview last week, you may have taken notice of some graphs which I tried to illustrate why (at least partially) the GeForce4 would tank in performance when anisotropic filtering was enabled.
I was able to show that both OpenGL and Direct3D exhibited significant fill-rate reduction, to the point where I was left scratching my head. I basically wanted to prove, beyond a shadow of a doubt, that NVIDIA has a nasty bug on their hands that needs to be addressed and/or acknowledged.
This is what I did. I simply removed the GeForce4 Ti 4400 from my system, and slapped in the GeForce3. I re-ran the same tests, so let's take a look. BTW, the GeForce3 is clocked to Ti 500 levels.
First up is Serious Sam2.


The bottom line is that the GeForce3 is kicking out ~50% more fill-rate than the GeForce4 when anisotropic filtering is enabled.
MikeC: This has been an unresolved issue from day 1 with the GeForce4. Inquiring minds want to know...
De bottomline er is dat er mogelijk wat mis met de drivers of de hardware, ik hoop het eerste, zodat we nog enige verbetering kunnen verwachten, maar bovenstaande grafieken en m'n eigen ervaring zeggen dat er iet niet lekker ziet.
IK speel nou zelf bv. met een lod van -1 en zet aniso lekker uit, want je verliest gewoon 50% van je renderingspeed...
Niet op voorraad.
