Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

In the early days the address and data buses were pretty
straighforward, but I haven't kept up with all that, and all the
frontside and backside buses and stuff.

But it seems to me that unless they've done a workaround, the
biggest slowdown from integrated video is not the shared ram per se,
but rather the sharing of the address and data buses with the
integrated video controller. In other words, with a separate card,
all the refreshing of display output would take place within the
card until the main processor changes it, but an integrated video
controller is going to be constantly refreshing from the system ram,
during which time the CPU can't be using those buses too.

Or, do I have it wrong?

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

Peabody <waybackKILLSPAM44@yahoo.com> wrote:
> But it seems to me that unless they've done a workaround,
> the biggest slowdown from integrated video is not the shared
> ram per se, but rather the sharing of the address and data
> buses with the integrated video controller. In other words,
> with a separate card, all the refreshing of display output
> would take place within the card until the main processor
> changes it, but an integrated video controller is going to
> be constantly refreshing from the system ram, during which
> time the CPU can't be using those buses too.

I believe this is essentially correct. It is bus contention,
not actually RAM size. Integrated graphics aren't a burden
when there is separate vidRAM on the mobo (GPU?). Then it
just is less upgradeable.

-- Robert

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On 7/22/2005 14:27, Robert Redelmeier wrote:
> [...] Integrated graphics aren't a burden
> when there is separate vidRAM on the mobo (GPU?). Then it
> just is less upgradeable.

Are there non-server motherboards with Intel chipset that have dedicated a
dedicated framebuffer? I seem to remember something that you could put in
an agp slot that was essentially a "local ram" upgrade for integrated video.

~Jason

--

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Mon, 25 Jul 2005 14:16:56 -0400, Jason Gurtz <ask@NOmeSPAM.where> wrote:

>On 7/22/2005 14:27, Robert Redelmeier wrote:
>> [...] Integrated graphics aren't a burden
>> when there is separate vidRAM on the mobo (GPU?). Then it
>> just is less upgradeable.
>
>Are there non-server motherboards with Intel chipset that have dedicated a
>dedicated framebuffer? I seem to remember something that you could put in
>an agp slot that was essentially a "local ram" upgrade for integrated video.

There was a thing called an AIMM (AGP Inline Memory Module) which was
supposed to plug into an AGP slot. I never saw one and don't recall seeing
a mbrd which claimed to support this though... assuming some special
support was required??

--
Rgds, George Macdonald

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

George Macdonald wrote:

> There was a thing called an AIMM (AGP Inline Memory Module) which was
> supposed to plug into an AGP slot. I never saw one and don't recall seeing
> a mbrd which claimed to support this though... assuming some special
> support was required??

The 810e chip set supported a 4 MB video cache on the motherboard that
allowed faster access. I had one of those boards some time back. I
couldn't really notice any difference in performance with the cache
activated, and the on-board video was pretty bad: fuzzy 2D and pathetic 3D.

The 815e chip set supported a video cache with a memory module in the
AGP slot, as you indicated. I have an Intel 815EEA2 board that supports
this feature. The sales of these memory modules were almost
non-existant, but I did buy one made by Kingston some time back on
clearance for less than $10. Mine is a 4 MB module labeled "GPA card"
which I think stood for "Graphics Performance Adapter" or something like
that. Again, it seems to make no difference in video performance. The 2D
is ok (less fuzzy than the 810 but still not great) and the 3D is still
pathetic (but marginally better than the 810). I still use the system as
a file server on my network, with an old video card in the AGP slot.
--
Gary L.
Reply to the newsgroup only

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

On Mon, 25 Jul 2005 16:29:51 -0400, George Macdonald
<fammacd=!SPAM^nothanks@tellurian.com> wrote:

>On Mon, 25 Jul 2005 14:16:56 -0400, Jason Gurtz <ask@NOmeSPAM.where> wrote:
>
>>On 7/22/2005 14:27, Robert Redelmeier wrote:
>>> [...] Integrated graphics aren't a burden
>>> when there is separate vidRAM on the mobo (GPU?). Then it
>>> just is less upgradeable.
>>
>>Are there non-server motherboards with Intel chipset that have dedicated a
>>dedicated framebuffer? I seem to remember something that you could put in
>>an agp slot that was essentially a "local ram" upgrade for integrated video.
>
>There was a thing called an AIMM (AGP Inline Memory Module) which was
>supposed to plug into an AGP slot. I never saw one and don't recall seeing
>a mbrd which claimed to support this though... assuming some special
>support was required??

I don't think there was much too special that was required. I saw a
few of them in old Compaq Deskpro machines. i810 chipset as I recall,
4MB AIMM card. I don't know that it did much of anything for
performance, certainly it wasn't anything you would see without
benchmarking.

I don't think any of the newer chipsets support this spec anymore. I
believe it died after the i815, though I wouldn't swear to the exact
date. I've never seen another Intel chipset with local frame buffer
memory, though I have seen some from SiS and ATi. I expect that
nVidia will have a similar solution in the not-too-distant future as
well.

-------------
Tony Hill
hilla <underscore> 20 <at> yahoo <dot> ca