Posts: 820
Threads: 2
Joined: Aug 2011
Location: just a box
08-31-2011, 06:31 AM
(This post was last modified: 08-31-2011, 06:59 AM by xstyla.)
(08-31-2011, 05:55 AM)taltamir Wrote: I am guessing most people don't get this. How about including actual figures?
As in, show up in the drop down menu as "6x Native (3072x2688)"
it doesn't really work to show the numbers without knowing what the actual native is.
and for the example if I did the math correct it's only 512MB at 4x FSAA.
Posts: 90
Threads: 4
Joined: Aug 2009
Reputation:
0
(08-31-2011, 06:31 AM)xstyla Wrote: it doesn't really work to show the numbers without knowing what the actual native is.
and for the example if I did the math correct it's only 512MB at 4x MSAA.
So? vRAM isn't the biggest issue with such insane resolutions.
Also, did you account for double/triple buffering? (does PCSX2 use double/triple buffering?) And you vRAM to store game assets beyond just the finished image.
I do not have a superman complex; for I am God, not superman!
Rig: Q9400, 4GB DDR2, eVGA GTX260 SC, gigabyte EP35-DS3R. X25-M 80GB G2.
Posts: 3.559
Threads: 21
Joined: Jul 2010
Reputation:
61
Location: Australia
08-31-2011, 06:54 AM
(This post was last modified: 08-31-2011, 06:55 AM by Squall Leonhart.)
Both Vram and Vram bandwidth are the limiting factors of high internal resolutions.
4xMSAA can not be compared to internally rendering the image at 4-5-6x the original.
Posts: 820
Threads: 2
Joined: Aug 2011
Location: just a box
08-31-2011, 07:06 AM
(This post was last modified: 08-31-2011, 07:08 AM by xstyla.)
had to correct. sorry squall. I meant that old FSAA upscaling technique. that's the 512MB for a single frame.
as for the new solution it's only 32MB for a frame 64MB for double 96MB for triple buffering and the computational FXAA algorithm. I don't even know if this is comparable with MSAA. but for that FXAA sampling bandwidth and shader speed is really important.
the rest of the vram is free to use for whatever the plugin caches.
Posts: 3.018
Threads: 88
Joined: Jul 2010
Location: FREE SYRIA
(08-31-2011, 07:06 AM)xstyla Wrote: had to correct. sorry squall. I meant that old FSAA upscaling technique. that's the 512MB for a single frame.
as for the new solution it's only 32MB for a frame 64MB for double 96MB for triple buffering and the computational FXAA algorithm. I don't even know if this is comparable with MSAA. but for that FXAA sampling bandwidth and shader speed is really important.
the rest of the vram is free to use for whatever the plugin caches.
OMG ~512MB for one frame !!!
was there a GPU at the time that capable of running it
Posts: 90
Threads: 4
Joined: Aug 2009
Reputation:
0
(08-31-2011, 12:49 PM)zpinto Wrote: I think the problem is solved. I was using Boot CDVD (FULL) and i tried Fast a couple of times and the see through bug was fixed.
So im guessing it was because i was using FULL boot.
Thanks for the help
wait, there are problems in full boot that are avoidable by using fast? I thought fast would carry risks but never imagined full boot would too.
I do not have a superman complex; for I am God, not superman!
Rig: Q9400, 4GB DDR2, eVGA GTX260 SC, gigabyte EP35-DS3R. X25-M 80GB G2.
Posts: 820
Threads: 2
Joined: Aug 2011
Location: just a box
(08-31-2011, 10:12 AM)abdo123 Wrote: OMG ~512MB for one frame !!!
was there a GPU at the time that capable of running it
I think when FSAA was used for pc games the resolutions were a lil lower so it would have been less than 128MB for already double buffered 4xFSAA. I dunno if a card would have been able to do more than that.