Next Version Of Parhelia

[Android604]

I am with Kurt on the P64... sorry, but I work in 2D with gigabyte size TIFF files 10 hours a day and as much as I like using the G450 and G550 (which are in our computers currently)... my P128 at home kicks them around the block all day long.

If M releases a P64 which is much cheaper than a P128 it will be in my next workstation and everyone after that (I hope the price is around G550 price please).

Andrew

[runderwo]

>> Issues to correct on a new Parhelia rev:

How about Bump mapping?? I can't believe they actually left some features behind that the G400 had.

[MagerValp]

Stupid, stupid, stupid

If there is a a single feature that I require in my next video card, it's F(S)AA. That's what I miss the most in my G400Max.

Compared to the announced Radeon 9500, this is a joke.

[GNEP]

OK - my 2p's worth:

One of Matrox's big markets is the trading rooms and financial workstations where you need lots of monitors.

I happen to work in one as a financial analyst-type. It would seem that the P-64 might well be aimed right here - to be quite honest this makes a lot of sense. 3 monitors is quite a common configuaration for us. And to do this on 1 AGP card cheaply would be great. And so all the 3-d stuff is pretty much irrelevant.

But there is one big problem - another common thing in this environment is the sole use of digital flat panels - otherwise the office space needed gets ridiculous. As does the peripheral-vision flicker. And the P can't drive 3 DVI's. Which is surely a bit of a must have for something new in this market segment. Because to be quite honest, the drivers are the only other differentiator - 2d performance is really no issue at all.

All we need is a cheap AGP card that can do 2d well on 3 DVI panels at 1024x768 or 1280x1024. The Parhelia, whilst nice, is probably (a) far too noisy and (b) offers far too many "redundant" 3d features.

Why do you think that the MMS cards are still based around the G200 core?

So a nice simple low-power card which is cheap (although cost is not so much of an issue here), and with drivers that are 100% rock-solid with no conflicts is what we want. Even with a G450 at the moment I have to turn off some of the DH features as some custom trading apps don't like them.

Oh, and screen-rotation would be nice as well - running 3x1024 wide by 1280 high would be a real real sweet spot for 2d.

This P-64 just seems to miss the home/enthusiast bit be reducing 3d performance still further (FAA) whilst being waaay too expensive.

And for trading floors, there is a lot in there that is not needed, and the 3rd DVI is not present.

The two stools seem to be sliding apart...

Oh, and by the way, I have actually been very impressed with my P at home so far. I really like it as a card.

gnep

gnep

[Kurt]

why would you insist on 3 DVI when the analog outputs can drive higher resolution screens???

I guess if I were an analyst I'd go for 3x 1600*1200 (or higher) rather than 3x 1280*1024 (or 1024*1280 - just get screens than rotate 90°)...

[Wombat]

why would you insist on 3 DVI when the analog outputs can drive higher resolution screens

That's easy. They're using LCDs to save space.

[Umfriend]

But if it is targetted at 2D, Business and LCD, why bother going P at all? Won't G450MMS do? 4 screens, and I can;t imagine that with DVI, 2D quality between G450MMS and P. would differ...

But then, what do I know....

Umf

[VigilAnt]

Originally posted by runderwo
>> Issues to correct on a new Parhelia rev:

How about Bump mapping?? I can't believe they actually left some features behind that the G400 had.

Parhelia has more bump mapping features than the G400.

VigilAnt

[K6-III]

Originally posted by VigilAnt

Parhelia has more bump mapping features than the G400.

VigilAnt

Pixel shaders...

[Kurt]

Originally posted by Wombat
That's easy. They're using LCDs to save space.

I was refering to flat panels with a regular VGA input

[WayneHu]

It seems Matrox marketing does not do their homework quite well? In my opinion, it is quite diffcult to persuade the average users pay $250 to buy a Parhelia-64 without any special feature except to 3 inputs.

I don't deny that Matrox's Parhelia-64 decision could gain some market share because there is still no similar single solution within the same price range to get 2 digital outputs or 3 analog output simultaneously. However, since they have been able to pay $250 for this solution, it is impossible that the customers do not have enought money to pay extra $70~100 to get current's more featured and powerful Parhelia-128.

In addition, why don't Matrox provide more general-purposed solutions within the accepatble price range for regular users?

If I were Matrox, I would try to maintain three product lines below as the graphics adapter line:

1. G550 DH 32/64MB + improved memory controller for 3D core, 1 DVI / Analog output , 1 Analog-output / TV-out. $50~90.
(for average users)

2. Parhelia-64/128 DH, 2 DVI / Analog outputs, 1 TV-out from any display's mirrioring or overlay space. $150~250
(for regular gamer/semi-professional users)

3. Parhelia-128/256 TH, 3 DVI / Analog outputs, 1~2 TV-out from any display's mirroring or overlay space. $350~500
(for professional users)

Of course, reasonable 3D performance for the given price tag although most people do not expect Matrox can beat nVidia in 3D performance.

[diewlei]

this is coming down to a "wishing well" thread. c'mon throw in a penny or two. lol...

[KngtRider]

Parhelia-512 GPU family is their new building block, G chip

need

*entry level card (cut down 64)
*mainstream (parh-128)
*professional (quadro/firegl style)

This is what the company see and what the customers (OEM/System Builders need)

If you lot had bothered to read the readmes with the new 'stuff' (drivers/plugins) you will see 256MB mentioned.

Also the DCC plugins need 128MB parh minimum

What I am hoping to too see and this will probably become true is that the 256MB ver will be the updated one, with DX9/PS 2.1(read several times on the russian sites that the update will be dx9/ps 2.1)

But knowning the direction the company ones this could be unlikely

Why you all suprised at the price? Shops here sell the 450 marvel for $750 AU and the G450/550 for 300/350 approx, Parh about 800/1000/1200

R9000PRO is <$200 AU here... go figure

Do they care if marvel is discontinued? heck they woudlnt even believe it if a customer told them. Anothe place listed G550 (DB15+DVI onboard) under 2d workstation cards next to a 3dlabs card...

If they wanna sell Parhelia they gotta half the price, but they could make up volume. Darkcrow bench showe parh faster than r8500 in UT

Hey I dont like the situation but what can you do, only rant and winge to them, a startup run by consumers/fans/users wouldnt work

hrrm i think need something to end this post, how about Long Live Rendition ?

[Yxalitis]

Make's sense to me

Ok, enough is enough!

There are several good reasons why Matrox would choose to pursue a lower spec'ed 64 MB version.

1. This is for the high-end OEM market, where organisations are looking for new hardware to run multi monitor setups. They don't need high 3D speed, or FAA, so why include them?

2. By the looks of overclocking results, the Parhelia core is pretty close to the edge, with some retail units shipping under the 220 MHz mark. Given that many chips are only just running at 220, there must be a bundle that didn't make the grade, perhaps not even 200 MHz for oem stock. Wy not use those chips? Would you rather Matrox tossed them out??!!

3. Revenue, A company needs to keep the money flowing, getting some deals signed with the major vendors is the main way to do this, and having a new product to entice them is mandatory. The G550 MMS is an old chipset, and it only makes sense to retire it, and replace it with new technology.

4. Matrox don't need to make a 64 MB card with everything else the same, the cost savings would not be sufficiently beneficial to do so. I guess they figure tht 64MB is too low for future games. I think everbody thinks that a 64 MB version wiould cost hepas less, bit it wouldn't make a sufficient diference to warrent the purchase. I guess $50 less.

Any way, that's what i think. At least when the Parhela 2 comes out, I will still get good money for my old card. Matrox cards hold their value well, precisely becasue Matrox don't release newer chipsets every 6 months, with substantial price cuts on their old line.

[rylan]

Cutting the memory in half would probably save them $15 max in components, since memory chips are sooo incredibly cheap now. Only other thing they could do is drop the speed like you mentioned and use some of the chips that couldn't run at 220, but at say 200 instead. But seriously, the price has gotta come down a lot more for it to be a reasonable solution for OEMs when compared to the current version of the card. From what I see the biggest thing going for the card right now is that it can drive two digital 1600x1200 flat panels.