a new version of the parhelia

[Supernaut]

I read that the banding artifcats do not occur when using dvi. So tell me what u guys think of this...

1.) Parhelia with only dvi out, i wonder how much cheaper this would be

2.) parhelia with dual dvi out only with no faa

Do u think the absence of a DAC will make the card alot cheaper?

[K6-III]

Crystal Ball....

[DGhost]

won't make it cheaper as the RAMDACs are integrated into the core, and there is no way they would release a version of the chip without them. the only real way to lowers costs on manufacturing of the P is, from the little i know about board design and the P's design to 1) remove DVI capability (lowers costs by not requiring the external TMDS) 2) reduce amount of memory (use cheaper memory modules) 3) reduce the size of the memory bus (less complex board design with less traces to route) 4) lower clocked core could possibly allow cheaper power regulation circuitry 5) a redesign of the core to make it smaller and easier to produce... the only problem with that is that it would take engineering efforts on the part of Matrox that i don't think they have...

anyways, the banding issue has supposedly been fixed in the AGP 8x compatable core revision...

my idea for a good "budget" P in a lower cost environment would be to make it so that it only can output to two monitors (it still does have the best dualhead, but get rid of triplehead and surround gaming, instead leave FAA and the rest of the features in place, including the two display + feature modes), bundle two dvi-hd15 adaptors, move the tv-output onto the board itself isntead of through one of the dongles, and have two versions, one with 64mb of ram, the other with 128mb... you could market it as the Parhelia DH or some, sell it to the Multimon market, or possibly even the desktop market... this assumes that you drop the price down to like $200 for the 64mb and $300 for the 128mb or something... you could probably sell it to business easier, especially those that do not require the use of three monitors...

[JaG]

Building it cheaper...

Implied by Parhelia's actual board design it may be possible, without a redesign, to create a 'dual head only' model, DGhost.

Here you wouldn't need the second TMDS by Chrontel (chip in the ring). Instead of the second DVI-I connector, where the triple head adapter would have been installed to, on common Parhelias, you may also put a standard HD15 for only one analog CRT or TFT.

Additionally you can save the money for all adapters, because a svideo connector suits well besides this standard HD15. Together with the standard bulk clock and only 64 megs, it should be possible to build a cheaper modell on this way. But who knows...

[DGhost]

the Parhelia honestly looks like it was designed with that in mind (dropping the second DVI connector in favor of an HD15 connector) - it has the spots on the board... however, i never said thats what i wanted nor did i ever imply that it was possible without a redesign... i noticed that a long time ago...

but, i think it would be a terrible waste for them to drop the second DVI port... i would honestly rather see them leave it on because the card could benefit more from having that capability than it could from the cost savings from not having it...

an S-Video connector on the board would be very nice... hehe...

[JaG]

I agree again DGhost, the second DVI is one of these certain things which make Parhelia unique in market at the moment.

But as an additional, cheaper member in Parhelia family or for oems it might be a way...

[Kurt]

Re: a new version of the parhelia

Originally posted by Supernaut
I read that the banding artifcats do not occur when using dvi. So tell me what u guys think of this...

1.) Parhelia with only dvi out, i wonder how much cheaper this would be

2.) parhelia with dual dvi out only with no faa

Do u think the absence of a DAC will make the card alot cheaper?

Drop both DVI in favor of HD15, forget about the TV-OUT, use 64MB SDRAM instead of DDR (if it's possible?).

Leave the other features in, as long as they don't depend on external chips, and lower the clock speed to improve yields and thus reduce costs.

This is a fairly marketable product: a replacement for the G550 with improved 2D quality. I guess they could reasonably sell it around 149$US.

You could have a variation at 199$US with Dual DVI, Triplehead and TV-OUT enabled.

They wouldn't be for gaming, you'd have the actual P128 or P-II for that. Consider them as the Radeon VE of Matrox

All that without changing the core

[Supernaut]

the reason I brought this topic up is because I was wondering if all of those defective parhelias can be reused. If they just have dvi out, no more banding problems right?

[knirfie]

My plan:
1. take all the parhelia (1) chips and start using them for the replacement of the G550 (around $150) and a new mms series: 6 monitors with one card; both of these cards dont really need 3d as they are buisiness oriented thus you can disable some if the broken features.
2. release the Parhelia 1.1 revision with 128 mb and 256 mb, mainstream and high-end cards.
3. get the parhelia 2 out as fast as possible for the gaming market.

[Chrono_Wanderer]

Does anybody think that parhelia is actually in works?

[Dogbert]

Bah, I say drop the price to $200 and you'll see them disappearing faster than _

[Wombat]

How much do you guys think it costs to make a card?

[ZokesPro]

Originally posted by Wombat
How much do you guys think it costs to make a card?

It's probably more expensive than we think.

[thop]

doesnt matrox get them for free by the lovely card fairie?

[William]

Originally posted by Wombat
How much do you guys think it costs to make a card?

Let's see, last time we had some boards (4 layer, no components on them) made (5) we paid about £600 for the lot. I'm guessing that if we'd had 5 boards of Parhelia complexity done, it'd have cost about £1000 for a similar number. That's 200 quid before we start talking about anything that gets soldered on (of course, if we'd wanted 10 boards, we wouldn't have paid much more than maybe £700 -ballpark; we wanted 5).

I know nothing about chip manufacture, so I won't speculate.