Re: [gentoo-amd64] Re: Re: Re: More ATi driver madness

[Bob Sanders]

Duncan, mused, then expounded:
> Bob Sanders <rsanders@×××.com> posted 20060814151338.GA102938@×××.com,
> excerpted below, on  Mon, 14 Aug 2006 08:13:38 -0700:
>
> > Gfx, especially 3D, is about memory bandwidth.  Move the memory out
> > of direct contact with the gpu chip - say via a socket, and it's
> > necessary to drop the frequency that the memory interface runs at.
> > It's basic electronics - add more capacitaince and inductance, and
> > the speed of the interface goes down.
>
> Two factors in counter-point.  We're talking FB-DIMM tech time this comes
> out, and this will be using upgrades to the same memory controller on the
> CPU (only in this case GPU) AMD is already using to maintain its lead in
> multi-socket performance against Intel (even while Intel's Core-Duo has
> caught up at the low single-socket end).
>

Problems with this approach - it takes 4 FB-DIMMs to a special northbridge
for Intel to acieve 21.5 GB/s bandwidth to the cpu in a Woodcrest based server.
And that's a fair chunk of change for the northbridge and dimms.  Plus the
increased latency.  Btw - the current DDR-2 controller in the AMD Socket AM-2s
can talk to FB-DIMMs.  It's not allowed to yet, probably not tweaked up right
and AMD like to take things slowly so that risk is minimized - one or two
things at a time.   But the current memory controller should work 
with FB-DIMMs.

While 21.5 GB/s may seem fast, that's the peak speed, not the sustained
bandwidth.  Effective bandwidth is going to be less.  And it's a lot
cheaper to build a dedicated PCIe card with memory than to even purchase
one stick of FB-DIMM memory.  Even with just 2 DIMMS, DDR-2 or FB-DIMMS,
the memory bandwidth peaks around 10.5 GB/s with a sustained bandwidth
of around 8 GB/s.  Still less than the 12.8 GB/s an Nvidia 600GS card
attains while retailing for $109.
 
> > Remember, Intel doesn't make real 3D Gfx chips.  Unlike, Nvidia and
> > Ati, Intel does most of it's 3D processing in software.  Thus opening up
> > the chips specs and driver has little impact on any IP outside of Intel as
> > it doesn't expose any IP that might belong to Micrsoft or SGI.
>
> Does <> will do.  With aeroglass driving dx10 requirements, the low end of
> the gfx market is about to get very legacy, very fast.  Their new chips
> do more in hardware because they must, due to the performance
> requirements.  That's the reason Intel's latest free driver release made
> the headlines in the FLOSS world -- it's the first time we've had full
> sources to something that capable.
>


Based on all I've read, what I'm putting forth is strictly my take on the
situation.  No source to go to, just speculation.  What Intel is doing
is stopping Gfx development any laying off most of it's Gfx Engineering
staff.  They are outsourcing most of the IP development by licensing
cores from Imagination Technologies - the PowerVR SGX, that has been widely
reported, but not explicitly confirmed (unless I missed it).  As the SDK
for the PowerVR is already publically available, all Intel did was go
and say - all this old crap we made and are abandoning, well - here are
the specs for it.  And since the new cores already have an open api and
sdk, we'll just "say" we're opening the development to the Open Source
Community, because, well, it already is.  Great Marketing move.

Of course I could be wrong on this, but time will tell.

Bob
-  
-- 
gentoo-amd64@g.o mailing list