RDRam

  • Thread starter Thread starter Diane
  • Start date Start date
On Sun, 12 Dec 2004 17:51:45 -0500, Tony Hill


Not sure about that - here, the i820 never did fall into affordability
before the scandal blew up.

Chris, you must admit that you're not exactly in the center of the PC
universe. Perhaps you mean before *you* heard the ex(im)plosion?
The scandal was hot and on the front-burner well before Intel shipped
chipset one. DRDRAM was the wrong idea at the wrong time. Trying to
corner the memory market (and others) using extortion didn't help the
minimal technical case any at all.
I stayed with the venerable i440BX in
those days, even if it required a slotket card; then when 815 and 815e
came out, I jumped over to that as native S370.

The 440BX was an excellent chipset (the last of the Intel goodies). I had
several systems with 'em. The last one (ASUS with integrated SCSI-UW) is
cluttering my office, waiting to be thrown into the dumpster (minus
keyboard ;-).
I haven't seen or sold a single i820, tho I built one i850 for a big
multimedia dev system.


It wasn't long after i845 that the i845G came out, which was DDR-based,
and I switched to that. I think I built about 3 pre-DDR SDRAM-based P4
systems; I don't remember them being slow.

But they made up for it by being expensive. ;-)
 
Timmy keeps comming back to pump every time RMBS gets a little good news.

No wonder if we haven't seen him for a while, then ;-)


--------------- ----- ---- --- -- - - -
Tech Support: The guys who follow the
'Parade of New Products' with a shovel.
 
Perhaps I should have said "price-conscious and buzzword happy"
consume grade systems. Dell sold a few of those as their "high-end"
systems because they had the latest and greatest chipset from Intel.

Ah, Dell. I remember a Dell that had only DIMM slots and was based on
a chipset that doesn't support SDRAM. Had to use EDO DIMMs, and guess
how available and cost-effective *those* were,
That's what a lot of people did, though many others started using VIA
chipsets instead as they offered better performance and a MUCH better
price than anything Intel had.

I've never been happy to trust VIA. If I was, I'd likely be AMD too.
And now the i845GV, which is basically same as the G but without AGP.

Run away screaming! I don't do crippleware such as no-ATX or mATX.
Well, when PCI Express catches on, I'll drop the AGP requirement :-)


-------------------- ----- ---- --- -- - - - -
Treat Yourself - You're Worth It! #29:
Next bath night, try using the "soft"
side of the sponge/scourer.
 
Chris, you must admit that you're not exactly in the center of the PC
universe. Perhaps you mean before *you* heard the ex(im)plosion?

Sure - hence the word "here". my point was that i820 was still costly
here when the scandal broke, so no-one had been drawn to it on a
low-cost basis. I had an eye on it in cases where a high-end user
wanted to go SDRAM now, RDRAM later.

But AFAIK even that wasn't an option, as mobos didn't have both slots.
And the lower RAM capacity of i820 made it IMO unfit for use in that
hi-end sector; when I needed that, I went i850 instead.
The scandal was hot and on the front-burner well before Intel shipped
chipset one.

I remember i820 not being able to do as many banks of RDRAM as
expected - thus RAM drop, from 1G to 512M I think it was - and that
the MTH made SDRAM slower than i440BX.

But while that had me asking "why are we doing this again?" it was
later that MTX-vs.-SDRAM unreliability blew up into product recall.
At *that* time, i820 was still priced at a premium over i440BX, so it
pretty much died on the vine (as it should have).
DRDRAM was the wrong idea at the wrong time.

I'm not sure if it had a "right" time. It was old, and there were no
mass-market takers until Intel's hubris, which reminded me of IBM's
"let's re-invent the PC as PS/2 and they will have to follow" fiasco.

Way before Intel and system RDRAM, I remember a Cirrus Logic "Laguna"
card that used RDRAM. It was a weak performer, as CL generally were
by then, and was a 2M SVGA priced at near-1M levels. Because SVGA
manages its own RAM, oddball RAM types were common - even so, RDRAM
was minor, along with early DDR and Tseng's MDRAM.

RAMbus are an Exhibt A with what sucks about US business, in that what
ever success there might have been, would have been delivered from the
legal rather than R&D budget. And manufacturing? Hell no, we leave
that to the peasants, let them choke on the pollution while we milk
our "bright idea" patent-right teats for the next few years.

There's a lot that doesn't suck about US biz; that's found elsewhere
:-)
The 440BX was an excellent chipset (the last of the Intel goodies).

I wasn't adverse to 815e, 845G or 865G - but having switched to Intel
rather than generic motherboards, I have QA concerns (as posted
elsewhere as another thread).

-- Risk Management is the clue that asks:
"Why do I keep open buckets of petrol next to all the
ashtrays in the lounge, when I don't even have a car?"
 
No wonder if we haven't seen him for a while, then ;-)

It's my understanding that over on the Yahoo stock boards they manufacture
the good news.. which is i suspect one of the reasons that the RMBS charts
have interesting saw-tooth and square wave patterns.

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??
 
Ah, Dell. I remember a Dell that had only DIMM slots and was based on
a chipset that doesn't support SDRAM. Had to use EDO DIMMs, and guess
how available and cost-effective *those* were,



I've never been happy to trust VIA. If I was, I'd likely be AMD too.


Run away screaming! I don't do crippleware such as no-ATX or mATX.
Well, when PCI Express catches on, I'll drop the AGP requirement :-)

Mid-range AGP cards are getting hard to find in the U.S. Nvidia recently
announced that the 6600GTs would "support" AGP but judging by the extra
$50. or so cost over PCI-Express I suspect that the boot is on the other
foot as regards extra bridge circuitry now. Whereas a few months ago, the
bridge was necessary to get PCI-Express..... There are couple of AGP
6600GT cards being sold but I haven't seen any 6600s(non-GT) yet.

I'm a *little* surprised that PCI-Express has caught on as fast as it has
but the fast fade-away of mid-range AGP is even more surprising. BTW, in
case you hadn't noticed, the favorite e-tailer for DIYers here is
www.newegg.com .... and no I don't think it's an EggHead revival.:-)

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??
 
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Way before Intel and system RDRAM, I remember a Cirrus Logic "Laguna"
card that used RDRAM. It was a weak performer, as CL generally were
by then, and was a 2M SVGA priced at near-1M levels. Because SVGA
manages its own RAM, oddball RAM types were common - even so, RDRAM
was minor, along with early DDR and Tseng's MDRAM.

While we're on the subject of weird memory on video cards, can anyone
identify the type that was used on this ancient video card?

http://alfter.us/graphics/vidcard.jpg

It's a 2MB ISA SVGA card based on the Trident TVGA8900D. The eight metal
boxes on the left are the card's memory. They're about a quarter-inch tall
and have 23 pins each in a 5x5 grid pattern (with two pins missing,
presumably for keying). Is this some sort of standard memory technology in
an oddball package, or is it something truly weird?

_/_
/ v \ Scott Alfter (remove the obvious to send mail)
(IIGS( http://alfter.us/ Top-posting!
\_^_/ rm -rf /bin/laden >What's the most annoying thing on Usenet?

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (Linux)

iD8DBQFBvzSNVgTKos01OwkRAvEjAJ90p9NxOGi3Cosv3vCB7OPheYESugCgsEfd
GUL1GyWHPuL/kPeQ0FvMcbY=
=8xYR
-----END PGP SIGNATURE-----
 
: -----BEGIN PGP SIGNED MESSAGE-----
: Hash: SHA1
:
: In article <[email protected]>,
:: Way before Intel and system RDRAM, I remember a Cirrus Logic
:: "Laguna" card that used RDRAM. It was a weak performer, as CL
:: generally were by then, and was a 2M SVGA priced at near-1M
:: levels. Because SVGA manages its own RAM, oddball RAM types were
:: common - even so, RDRAM was minor, along with early DDR and
:: Tseng's MDRAM.
:
: While we're on the subject of weird memory on video cards, can
: anyone identify the type that was used on this ancient video card?
:
: http://alfter.us/graphics/vidcard.jpg
:
: It's a 2MB ISA SVGA card based on the Trident TVGA8900D. The eight
: metal boxes on the left are the card's memory. They're about a
: quarter-inch tall and have 23 pins each in a 5x5 grid pattern (with
: two pins missing, presumably for keying). Is this some sort of
: standard memory technology in an oddball package, or is it
: something truly weird?

Can't say, but what I'd like to know is who is that strange man behind the
curtain...er...graphics card? Too funny! :-)

j.
 
I'm a *little* surprised that PCI-Express has caught on as fast as it has
but the fast fade-away of mid-range AGP is even more surprising.

I think it's DirectX 9 that's driving SVGA (or rather, 3d) -
especially now that new PCs ship with XP SP2 and DX9c.

The new pixel shading stuff looks drop-dead beautiful, but bottom-end
cards can't do it at all. Even chipsets we aspire to here (e.g.
FX5700) that can do, do it slowly. It takes boss ones to do it well.

So I think there may have been quite a bit of new development with
this in mind - just in time to start off on tomorrow's bus.

The other thing, is that PCI Express seems such a U-turn on the AGP
concept ("get the SVGA off the general bus, give it it's own fat
pipe") that I wonder if something hasn't gone wrong with AGP.


---------- ----- ---- --- -- - - - -
On the 'net, *everyone* can hear you scream
 
While we're on the subject of weird memory on video cards, can anyone
identify the type that was used on this ancient video card?

Very small bee hives?
It's a 2MB ISA SVGA card based on the Trident TVGA8900D. The eight metal
boxes on the left are the card's memory. They're about a quarter-inch tall
and have 23 pins each in a 5x5 grid pattern (with two pins missing,
presumably for keying). Is this some sort of standard memory technology in
an oddball package, or is it something truly weird?

I have to tell you, "truly wierd" comes to mind. Was this a Windows
accelerator chipset? I wasn't aware that Trident got into that in the
ISA era, I thought the only mainstream players were Cirrus Logic (the
first I heard of) and S3 (who rocked at the time).


---------- ----- ---- --- -- - - - -
On the 'net, *everyone* can hear you scream
 
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1



While we're on the subject of weird memory on video cards, can anyone
identify the type that was used on this ancient video card?

http://alfter.us/graphics/vidcard.jpg

It's a 2MB ISA SVGA card based on the Trident TVGA8900D. The eight metal
boxes on the left are the card's memory. They're about a quarter-inch tall
and have 23 pins each in a 5x5 grid pattern (with two pins missing,
presumably for keying). Is this some sort of standard memory technology in
an oddball package, or is it something truly weird?

Those cans look suspiciously like IBM 1/2" DRAM modules, last used
*moons* ago. If it really is a 2MB card, the modules would be 2Mb,
which seems high for that package.

Are there any IBMish part numbers[*] on the memory or the card?

[*] IBM component/FRU part numbers are of the form NNAXXXX
Where: N= numeric
A= Alpha (usually, but *really* old stuff may be numeric)
X= Alpha-numeric (for components usually all numeric)
 
I think it's DirectX 9 that's driving SVGA (or rather, 3d) -
especially now that new PCs ship with XP SP2 and DX9c.

The new pixel shading stuff looks drop-dead beautiful, but bottom-end
cards can't do it at all. Even chipsets we aspire to here (e.g.
FX5700) that can do, do it slowly. It takes boss ones to do it well.

So I think there may have been quite a bit of new development with
this in mind - just in time to start off on tomorrow's bus.

The other thing, is that PCI Express seems such a U-turn on the AGP
concept ("get the SVGA off the general bus, give it it's own fat
pipe") that I wonder if something hasn't gone wrong with AGP.

The paradigm is virtually the same, it's just the transport that's been
changed: most desktop systems will have their PCIe graphics card slot on a
dedicated PCI Express hose - preferably 16b wide...

/daytripper
 
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Those cans look suspiciously like IBM 1/2" DRAM modules, last used
*moons* ago. If it really is a 2MB card, the modules would be 2Mb,
which seems high for that package.

I think it's a 2MB card; it could be just 1MB. It's not currently installed
in anything, and hooking up a junk-box 486 or Pentium to check is more than
I want to do right now.
Are there any IBMish part numbers[*] on the memory or the card?

Nope...the memory "cans" have no identifying markings on them at all. The
three chips (graphics processor, RAMDAC, and firmware ROM) are all Trident
parts, and the rest is a bunch of passive components. It appears to be a
generic "made-in-China" video card that could've been sold by nearly
anybody.

_/_
/ v \ Scott Alfter (remove the obvious to send mail)
(IIGS( http://alfter.us/ Top-posting!
\_^_/ rm -rf /bin/laden >What's the most annoying thing on Usenet?

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (Linux)

iD8DBQFBv3pwVgTKos01OwkRAqbSAJ9/kKR9L9+zjDxcdmq7E/B7Bgbx2QCeIvh0
W5R5Acf7LM/tzn/FGFgaAlo=
=DqSw
-----END PGP SIGNATURE-----
 
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I have to tell you, "truly wierd" comes to mind. Was this a Windows
accelerator chipset? I wasn't aware that Trident got into that in the
ISA era, I thought the only mainstream players were Cirrus Logic (the
first I heard of) and S3 (who rocked at the time).

The machine it was pulled from (a few years ago) was running Win95. An old
copy of the Linux Hardware Compatibility HOWTO indicates that it's just a
framebuffer with no acceleration. There seems to be even less documentation
of the card running under Windows than under Linux.

I actually found a driver package for the card with DOS, Win3.x, Win9x, and
NT 3.5 drivers. The readme for the Win9x driver (beta, dated 27 Jul 95,
which was a few weeks before Win95 was released) refers to it as a
non-accelerated driver, so I'm guessing that it's just a framebuffer.

(Followup to an earlier reply: I thought it was a 2MB card, but the limited
info I've run across suggests that the chip only supports 1MB.)

_/_
/ v \ Scott Alfter (remove the obvious to send mail)
(IIGS( http://alfter.us/ Top-posting!
\_^_/ rm -rf /bin/laden >What's the most annoying thing on Usenet?

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (Linux)

iD8DBQFBv34gVgTKos01OwkRAi3FAKDZJoGTT/WKM0Q5hR3DtJGQMNfZ7gCgzeXp
FPtvNDZ2cPmS9mc7TB8Sio0=
=jhOJ
-----END PGP SIGNATURE-----
 
I think it's DirectX 9 that's driving SVGA (or rather, 3d) -
especially now that new PCs ship with XP SP2 and DX9c.

The new pixel shading stuff looks drop-dead beautiful, but bottom-end
cards can't do it at all. Even chipsets we aspire to here (e.g.
FX5700) that can do, do it slowly. It takes boss ones to do it well.

I'm no video expert but it appears that, with the FX5700 in particular,
Nvidia dropped the ball on pixel shading and allowed ATI to nose ahead...
something few would have predicted. In pixel shading, the ATI 9600 series
supposedly did much better.
So I think there may have been quite a bit of new development with
this in mind - just in time to start off on tomorrow's bus.

The other thing, is that PCI Express seems such a U-turn on the AGP
concept ("get the SVGA off the general bus, give it it's own fat
pipe") that I wonder if something hasn't gone wrong with AGP.

AIUI, in that PCI-Express is not a bus at all - more several point to point
interconnects - I suppose it is a swing away. The idea is that video will
have a fat pipe (16-lane) within PCI-Ex specs; most regular devices will
have a 1-lane version and if necessary, down the road some high bandwidth
devices may get more lanes as required.

For video, it *is* also a break, I suppose, in that it's not a mezzanine
bus but that term is getting to have less meaning in modern systems.
PCI-Express is less accessible as a standard to mere mortals, so I'm not
sure what design parameters there are available for the implementers but it
would also appear that having a bi-directional high bandwidth inter-connect
for video is a bit of a waste.

At any rate it's here and going at full steam ahead - Intel finally got its
way on NGIO.:-) As for AGP, the DIME seems to have been a red herring and
and with the strobe clocks running at quadruple the common clock for 8x, it
was pretty much tapped out on future directions.

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??
 
No wonder if we haven't seen him for a while, then ;-)

Timmy? He was shilling here as recently as two weeks ago (December 4)
....at least that's the date according to the server I use.
 
On Mon, 13 Dec 2004 16:31:01 -0500, Tony Hill


Ah, Dell. I remember a Dell that had only DIMM slots and was based on
a chipset that doesn't support SDRAM. Had to use EDO DIMMs, and guess
how available and cost-effective *those* were,

Those were probably more common than RDRAM chips, you can even still
find them today if need be. A fair number of the 430HX based boards
used them.
I've never been happy to trust VIA. If I was, I'd likely be AMD too.

I'm no big VIA fan (though IMO nVidia makes some really top-notch
chipsets, and for the time being at least they are AMD-only), but
whether you were a fan or not, people definitely bought their chipsets
but the truckload at that time.
Run away screaming! I don't do crippleware such as no-ATX or mATX.
Well, when PCI Express catches on, I'll drop the AGP requirement :-)

I wouldn't touch 'em either, even if AGP was a non-issue (as it is for
the vast majority of business systems). Still, they hit a price point
that few other chipsets can touch, cheaper even than most SiS based
boards.
 
AIUI, in that PCI-Express is not a bus at all - more several point to point
interconnects - I suppose it is a swing away.

I can intuit what you mean, but can't quite grasp the difference
between several interconnects and a bus. I presume it means that
traffic on the same wires (I assume they are the same wires?) is
mediated in a different way or at a different level?
The idea is that video will have a fat pipe (16-lane) within PCI-Ex
specs; most regular devices will have a 1-lane version and if
necessary, down the road some high bandwidth
devices may get more lanes as required.

Now I get it! It's that other devices are now crowding PCI into
obselescence, e.g. Giga-LAN, S-ATA etc. so instead of AGP + PCI slots,
we need at least 3 x fast slots. PCI-Ex sounds better designed to
handle this gracefully, i.e. allocate width as needed without having
to shatter old standards and set new ones (as AGP ?x now does)

Ultimately, it shakes out to:
- the highest CPU clock the CPU('s cache) can handle
- the highest RAM clock the current RAM standard can handle
- a high standard for bits that have to be in the case (PCI Ex?)
- a standard for bits that have to be outside the case (USB?)
- a standard for bits that are wire-less

The trend will be to either toss stuff out of the case (so that dumb
retail can sell them safely) or build it into the mobo, and ultimately
processor core, as Moore's Law allows. Perhaps at some stage we won't
have the "has to be inside the case for speed" layer at all.
For video, it *is* also a break, I suppose, in that it's not a mezzanine
bus but that term is getting to have less meaning in modern systems.

In the original VL-Bus vs. PCI sense, I doubt if we will ever see a
"local bus" again, given how RAM out-paces other cards and devices.

In what sense is PCI Ex not a mezzanine bus?
PCI-Express is less accessible as a standard to mere mortals, so I'm not
sure what design parameters there are available for the implementers but it
would also appear that having a bi-directional high bandwidth inter-connect
for video is a bit of a waste.

Unless they foresee the GPU as generating system input in some way?
At any rate it's here and going at full steam ahead - Intel finally got its
way on NGIO.:-) As for AGP, the DIME seems to have been a red herring and
and with the strobe clocks running at quadruple the common clock for 8x, it
was pretty much tapped out on future directions.

Yep. I only understand the last bit about AGP clock; also thinking
that whenever they up the data rate, they have to drop voltage to stop
the wires frying, and I'm wondering at what point VR will be too
granular to maintain voltage consistency.

So does PCI Ex solve this by adding more physical wires? That's
interesting if so, given the original "parallel for data speed"
approach that swung to the "serial to avoid cross-talk and reduce pin
count" phase we are currently enjoying with S-ATA and USB.


Thinking back on it (especially the initial rocky and costly rollout),
PCI's been a pretty good bus. It gained traction here in around 1995,
so it's served us for 10 years - ?as long as ISA-16.

And yet it seems that VIA still couldn't get the hang of it, as
recently as a few years back (the UIDE corruption scandal).


---------- ----- ---- --- -- - - - -
On the 'net, *everyone* can hear you scream
 
Those were probably more common than RDRAM chips, you can even still
find them today if need be. A fair number of the 430HX based boards
used them.

EDO 32-bit SIMMs, yes. EDO DIMMs is another matter... in just about
all cases, mobos stayed SIMMs for EDO, DIMMs for SDRAM, and where both
were supported, both types of slots (just don't use both at once).

The cruel and unusual thing about these Dells were that they were
i830HX (lovely 64M+ chipset, shame it pre-dates SDRAM) but had only
DIMM slots, when there was no reason not to have only SIMM slots.

At least I think they were i430HX; they may have been pre-SDRAM Slot
One, i.e. the old PPro-generation i440FX. That, too, would have no
reason to have DIMM slots, as there's no SDRAM support.
whether you were a fan or not, people definitely bought their chipsets
but the truckload at that time.

Sure; I'm on the fringe by drawing the lines that I do. If I didn't
have those requirements, I'd have been even less likely to i820.
I wouldn't touch 'em either, even if AGP was a non-issue (as it is for
the vast majority of business systems). Still, they hit a price point
that few other chipsets can touch

True; at one time, I contemplated using i810 in such cases (this was
before Intel had done the 815, i.e. set the precident for SVGA+AGP)
but as it turned out, I didn't have anyone who needed that niche.
Even the office folks went i440BX and SVGA card (often i740).


---------- ----- ---- --- -- - - - -
On the 'net, *everyone* can hear you scream
 
On Tue, 14 Dec 2004 23:58:25 GMT,
(Followup to an earlier reply: I thought it was a 2MB card, but the limited
info I've run across suggests that the chip only supports 1MB.)

That makes sense, as non-accelerated cards only need SVGA RAM to hold
the display image, and nothing else - bearing in mind that the
RAM-hungry SVGA graphic resolution don't do page swapping.

So; 1600 x 1200 x 3 / 1024 / 1024 = > 4M, but more realistically for
mid/low-range unaccel. cards, 1024 x 768 x 3 / 1024 / 1024 = 2.25M.

That's an interesting figure, as it's exactly the amount of RAM
Tseng's ET6000 MDRAM had on the card (but that was
Windows-accelerated, tho prolly not at 24-bit 1024x768)

So my impression was wrong; there was a need for 2M on unaccelerated
SVGA cards, if they had a TrueColor RAMDAC.

Those old Trident SlowVGA prolly pre-date TrueColor; it wasn't long
after Tseng's ET3000 offered that, than the switch to
Windows-accelerated and various local bus cards was in swing.


---------- ----- ---- --- -- - - - -
On the 'net, *everyone* can hear you scream
 
Back
Top