G550 Test in CHIP Online

[EvilDonnyboy]

Holy crap. M could of at least put in some faster DDR memory. I mean, they put in the faster core with 2X the pipelines, only to feed it with (effectively) the same speed ram as a G400/G450.

I don't see any point for Matrox to waste their money adding those extra pipelines. Maybe they might have needed the higher clock speed to do have a TnL fast enough to do headcasting. But the extra pipelines are a waste of money, when hooked up to the extra slow memory bus.

[Hati]

Aces's hardware has a review of G550 too.
Look at :

http://www.aceshardware.com/Spades/read.php?article_id=45000188

[K6-III]

According to MSI's website the 826 is also an MX400.

[harkpabst_meliantrop]

Originally posted by K6-III
According to MSI's website the 826 is also an MX400.

Actually, they make two cards, both referred to as StarForce 826. The 826/32 is an MX-200 board, the 826/64 is an MX-400 board. The results (absolute as well as relative) give no other explanation than that the board was in fact the 826/32.

Apart from that: I got my Hercules 3D Prophet 4500 on saturday and obviously this was the right decision. Period.

--E<:|

[harkpabst_meliantrop]

Originally posted by Ali
Oh, and 166/166 core clock. Shouldnt be too hard to overclock a little.

Contrary to what Chip says. Ace's Hardware reports the core to be running at 125 MHz still. Usualy I would put more faith in Ace's than in Chip, but who knows.

On one hand this could mean that a noticeably beefed up MAX version of the G550 might be able to even beat an MX400. But it could also mean, that the chip simply won't run at a much higher speed, even when fed a higher voltage and cooled actively (memory speed should mainly be an issue of DDR-RAM prices).

And of course there's still the lack of T&L (may it be usefull or not) and texture compression and the question what limits the triangle setup of this card. When I remember right, even the G400 MAX was falling back to the same level as the G400 when very high numbers of polygones came into play (and thanks to nVidia beeing rather good at high polygon counts we're going to see more games like Max Paine in the future). From the specs, I don't see any advantages for the G550 over the older G4x0 cards in this area.

--E<:|

[dZeus]

Ace's hardware probably got it wrong, as well as the AGP texturing method (DMA vs. DiME).

Anyway, I wasn't saying that the G550 would make a good performer in games, I was merely pointing out that you can't say things like 'it's as slow as my g400max' when you compare it to the scores of a test you don't know anything about, except the result.

I don't understand the point about Texture Compression that Ace makes at all. Would it help Matrox sell more G550s one bit by adding support for TC? Adding TC wouldn't transform the card into a card good for gaming. And it also wouldn't make the card more attractive for their targeted buyers, which aren't gamers. Does TC help much in apps like CAD/CAM en modelling apps?

[Nuno]

Why to carry on on this already beaten subject? Of course it hurts seeing Matrox newest card getting spanked in 3d benchmarks by preety much everything on the market, but Matrox clearly stated it´s NOT a 3d gaming card. It´s a 2D oriented card with too many features that you can+´t find on the competition. If you play games, this isn´t the card for you.
$125 isn´t a steal either.

Of course, IMHO Matrox still could have released a higher clocked card with 128 bit DDR memory, ask $25 more abd they would have the better of 2 worlds. They didn´t choose that way. So live with it.

[harkpabst_meliantrop]

Originally posted by dZeus
Ace's hardware probably got it wrong, as well as the AGP texturing method (DMA vs. DiME).

One doesn't have to do much with the other, does it?

... I was merely pointing out that you can't say things like 'it's as slow as my g400max' when you compare it to the scores of a test you don't know anything about, except the result.

As the basic specs of the testing system are clearly laid out, not only the results, I don't see your point here. It's at least very save to say that this card doesn't provide any progress for G400MAX owners.

I don't understand the point about Texture Compression that Ace makes at all. Would it help Matrox sell more G550s one bit by adding support for TC? Adding TC wouldn't transform the card into a card good for gaming. And it also wouldn't make the card more attractive for their targeted buyers, which aren't gamers. Does TC help much in apps like CAD/CAM en modelling apps?

Yes, I know, Matrox always stated that the G550 was not a gamers card. And of course all of the beta testers and super murcers would have strongly agreed (if only they could) and have never really hoped for anything better or even implied the possibility.

Come on, get real. Does adding dual-pixel pipelines make the card more attractive for their targetet buyers? Does - ahem - Head casting?

I agree to Nuno that this thread is obsolete in a way (as the product turned out just as excepted). But this is a very legitimate place to vent frustration about what M might have been doing over the last couple of years. If Matrox decided to make no more gamers card they (and all the people in this forum) will have to live with people expressing their disappointment (or hopes) about it.

--E<:|

[dZeus]

[i]Originally posted by harkpabst_meliantrop

One doesn't have to do much with the other, does it?

Yes it does. Ace tend to not get their facts straight.


As the basic specs of the testing system are clearly laid out, not only the results, I don't see your point here. It's at least very save to say that this card doesn't provide any progress for G400MAX owners.

I don't see any testing specs there other than the test system specs. I can make quake 3 run in 1024x768x32 at anything between let's say 10fps and 60fps... it's all in the settings used. It's not like it's a static number that you can use to compare with your own results.


Yes, I know, Matrox always stated that the G550 was not a gamers card. And of course all of the beta testers and super murcers would have strongly agreed (if only they could) and have never really hoped for anything better or even implied the possibility.

hoped for better? It's a bit cheap to suddenly start venting your frustrations about something you could have known beforehand. Matrox never said it was going to be a gaming card. The minute Matrox announces a gaming card that is as at least half as bad as the G550 (or let's say 1/8 as bad), you can start all the whining you want

Come on, get real. Does adding dual-pixel pipelines make the card more attractive for their targetet buyers? Does - ahem - Head casting?

uhm, yes. What else would make people chose the G550 over the G450 (I agree that headcasting isn't going to help there). The reason to buy a G550 over a G450 is higher performance and lower costs. Nothing else.

I agree to Nuno that this thread is obsolete in a way (as the product turned out just as excepted). But this is a very legitimate place to vent frustration about what M might have been doing over the last couple of years. If Matrox decided to make no more gamers card they (and all the people in this forum) will have to live with people expressing their disappointment (or hopes) about it.

now have a bit more patience....

[harkpabst_meliantrop]

Originally posted by dZeus
What else would make people chose the G550 over the G450 ...

Exactly. That's the question.

The reason to buy a G550 over a G450 is higher performance and lower costs. Nothing else.

Forget about the former. The higher texel fill rate won't improve performance in anything the target market requires. This leaves a lower price as the results of 2 years of R&D.

I don't want to argue. While I am disappointed, im not by no means impatient at all. I went with a different brand. However - may it be good or bad news for you - that won't keep me from coming back to the murc and it's forums...

[dZeus]

what makes you think that the G550 was the result of 2 years of R&D? It has every sign of being an evolutionary step of the G450 (which is very similar to the G400 chip). You have no idea about what Matrox has on their sleeves as their next 3d architecture

[harkpabst_meliantrop]

Originally posted by dZeus
You have no idea about what Matrox has on their sleeves as their next 3d architecture

Very true. Unfortunately true. I had hoped to finally have an idea about it by now, hehehe.

And I'm always open to learn that they did more than creating vertex models of whatever parts of their bodies after the G400 was released.

--E<:|

[snn47]

1. Chip tries to favor everyone, they lack the necessary details to allow you to compare their reviews with others (even their own in the same paper) Details praised at one product are totally ignored at others. I buy CHIP only for their program-collection on their "complementary CD('s).
From my experience read them but trust them only when it is confirmed by a second reliable source!

2. Unless there are some Memory tricks disabled I don't think the GPU and RAM is capable of much higher clock-rates/improvement, without major cooling of RAM and GPU.


Personally I am looking for information on the Dual-Head-Output for Video use (S)VCD & DVD, e.g. what signals it generates (Macrovision?), how the signal form looks like, if in PAL/NTSC is it standard-TV-Norm or something else. Does it generate the same problems as the asian DVD-player with ESS chips?

Any information/link/comment in this filed or should I just get the G450?

[Alec]

Crap. I thought I would have a significatively faster 3D card from Matrox... I don´t want to lose one bit of the 2D quality I have with a G400 and a Trinitron monitor. I´m a designer, but a gamer, too... I´ll never buy Nvidia while they have those crappy RAMDACS... maybe a Kyro II? I only have a 466 Celeron, and I heard the Kyro II is very CPU-dependant. What do you reckon I should do for a graphics card upgrade? Maybe a Radeon?

[DosFreak]

The specs of the machine it was tested on do not matter. Stick a G400 in a 1g P3. Run Q3 at 1024X768 will all settings to max and you'll get less than 30fps framerates.