P 256 meg, 4-pin molex connector

[Jutai]

one of the factors holding the P's clock back is the voltage restrictions put on the AGP specs. and from the looks of things i'de say all the other video- card companies are having the same problem... Until they can make the move to .13 micron it takes to much power to get the clock rates as high as they can go. Having said that looking at pictures of the 9700 i noticed the nice little 4-pin molex on the card and i have to tip my hat to the ATI boys for it. it's genius, they can make the card fit the AGP specs. and deliver the extra power neccissary to get those nice high clock rates. So why doesn't M use that innovation on their 256-meg version of the P to increase the clock rate, maybe they could even reach the original target clock rate of 300mhz for the core. In the end more RAM alone will not justify the price increase. In the end what we all want to see is not necissarily that the P is Killing the competition, but just delivering the numbers we know it's capable of. We all know the technology is outstanding, it's just tough to see the small details holding it back. Finally i want to explain i am NOT saying the P is a dissapointment, or that it doesn't perform outstanding, so don't flame me. I just want to see Matrox really push the P as far as it will go and blow us all away with the technology they designed.

P.S. thought this might be a subject for the soap box, but because the P-256 meg is an announced product and not vaporware it would fit better here.

[az]

I think it still belongs in the new forum, the crystal ball

AZ

[Pace]

I don't know if it is announced really, just that Matrox have it in a planning stage.

Plus speculation on an extra power connector is...speculation

Go read some threads in the Crystal Ball, I think you'll see some answers

P.

[Kruzin]

Moved to the Crystal Ball speculation forum...

[Greebe]

Pardon me but to consider something as simple as a very inexpense molex connector as the niftiest cureall idea then we're already in so deep that a diving suit wouldn't suffice. LOL

[CaineTanathos]

I thought original clockrate was 260 mhz
and p-256 isn't anounced yet , so it's vaporware

[az]

Greebe, and yet you have to admit it's a simple, but effective and kinda elegant way to ensure power supply and AGP compliance - if you have a device that needs those amounts of power (and it looks as if all newer hardware tends to use more power, so maybe the AGP spec was just a little short-sighted - but who could've seen such an explosion in graphics cards complexity, power consumption, etc back then? )

AZ

[Greebe]

For someone who admits they don't care enough to delve into specs to find out for themself how this stuff works that's a perdy big bite you just tore off. First if you would even take the time to understand why it's not there in the first place (and no ATi wasn't the first todo this but 3dfx of years past) then you might has a case for your cause, but you don't.

Needless to say boys if you want to think this is just the niftiest wizbang then knock yourselves out, but in all honesty it's a joke from the word go to consider it as such.

[az]

Greebe - why should I take time and try to understand specs I can't understand because a) my technical english is by far not sufficient, and b) I have no degree in electronics engineering (not even close)?

I know 3dfx was the first to do this - and this was really not the point of my post. I'm a bit angry you attack me in this way without even really reading what I said. Sure this wouldn't be a nice feature if your card wouldn't draw too much power (like the parhelia). But IF your card draws that much power, it's a good way to ensure it gets enough without running the AGP out of spec and risking instability, damage to the motherboard, etc. Sure it would be better designing a card that doesn't use that high amounts of energy, because all energy that goes in has to get out somewhere, and moch of it gets out as heat - and we all know modern PCs are hot enough already.

It's just that I think the industry as a whole is moving towards ever faster, more feature-packed graphics cards, and that comes at a price, and a part of that price is higher power consumtion (yeah, I know, that can be countered by smaller processes, power management features, etc. - but recent history shows that those countermeasures haven't kept pace with the ever-growing power hunger of graphics cards).

Phew...

Besides, I stated in that thread that I really wanted to learn more, but you have to have a certain knowledge to understand technical papers, and I just thought the MURC might be a good place to start, with technology-savvy people here, and all.

AZ

[Wombat]

az, there are some things that can be explained in forums, and some that can't. Most of the questions you ask lately pertain to information that can't. With respect to the RAM issue earlier: A couple months of a good circuit design class, and a transistor class would get you <I>started</I> on the skills you would need to understand what's going on in there. There's a reason circuit engineering has a back-breaking 4-year curriculum to get started: It's a hell of a lot of information and acquired skill, and a learning curve that goes pretty much straight up.

The molex connector is neither elegant nor effective, and certainly not simple. If you're worried about risking stability, this would NOT be a solution around that.

[az]

OK

But I can't know this before somebody tells me that it is so - if I knew, I wouldn't ask

AZ

[rylan]

As an Electrical Engineer, I have to say that the molex power connector is a joke. Its a quick and very dirty way to get extra power to the board instead of using a more efficient regulator design and optimising the core to use less power.

While sucking juice off the system power supply may seem like a nice idea, its usually has the highest electrical noise in the system, which means unless you filter the hell out of it and use more inneficient linear regulators, you're going to have crappy stability and noise bleeding to the output stage on the board.

[az]

Thank you for the explanation, Rylan

I actually understood it

AZ

[Umfriend]

Another smart-ass around the block:

So if I understand correctly, you r saying that the current feed from the PS is "dirty", doesn't behave well enough to meet the Gfx-card requirements. This same dirty feed is fed to the MoBo, which does have the required filters b4 it feeds the card (and CPU etc)?

If so, would it then be an idea to give the card a 2nd connector that would fit in a PCI slot just to get some extra power from there? Would that at least be "easier" from a stability point of view then the molex option? Just theoretically speaking.

Umf

[Wombat]

It might be better than the molex, but mechanically it's nasty, and could not be called AGP compliant. Still not really worth thinking about. If you've gotta do it, do what 3Dfx did: external power supply.