From: Paul Jakma (Paul.Jakma at domain Digital.com)
Date: Thu 20 May 1999 - 15:25:25 IST
> ----------
> From: Colm Buckley[SMTP:colm at domain tuatha.org]
> Sent: 20 May 1999 15:11
> To: Paul Jakma
> Subject: Re: [ILUG] Re: Fonts...
>
>
> > I think the latency of RAM access is more of an issue to keeping the
> > graphics engine fed than anything else. Which is why modern cards are
> > so heavily pipelined, eg the G200. Then the graphics engine outputs to
> > the DAC, so wouldn't it be something like this:
>
> [... snippage ...]
>
> I'm not talking about game FPS, I'm talking about the maximum refresh
> rate which the card can support at a given resolution. Clearly, this
> depends on the slowest component of the overall data-moving
> system... even if the RAMDAC can handle 300MHz, the RAM might not be
> able to.
>
my argument is that refresh rate is a constant totally dependant on the
capabilities of the RAMDAC, and totally decoupled from RAM. And that the
speed of the RAM is only of influence on keeping the graphics engine
saturated.
whether your ram is fast or slow is only relevant to the graphics engine. If
the ram is too slow, then the graphics engine stalls - but the RAMDAC will
still output an image at X refresh rate, it'll just be the same image as on
the previous scan.
Of course I agree that a card has to be balanced, there's no point having a
350MHz RAMDAC hooked up to a TGUI8900 with 2MB of DRAM, but i don't see how
you can connect the specifics of refresh rates to ram latencies.
> Colm
>
-Paul.
This archive was generated by hypermail 2.1.6 : Thu 06 Feb 2003 - 13:04:13 GMT