Parhelia, poor TV output?

[Flying dutchman]

I just read a Parhelia review in the German CT magazine. They report that (in CURRENT games) it's 1/3 slower than the current Nvidia cards and also slower than a Radeon 8500.

What I found most disturbing is the remark that the composite TV output has a bandwidth of only 2.4 Mhz (less than half that of a Marvel 450).

[KuroNeko]

I just read that article too. What I found utterly baffling and disturbing was that the _VGA_ outputs were of the same lousy output quality as Nvidia's cards.

Above 1280x1024, fuzzy as hell. And then they brag about their special filter. Yeah right.

So why in hell would I still buy a Matrox? The excellent 2D (aka VGA output) quality was the only thing that kept me with Matrox G450 - I don't play many games anyway, I just need razorsharp desktops.

There's another test in the same C't BTW, about ATI cards. Or rather, the European market 3rd party cards, cause ATI no longer sells their cards here but has licenced Gigabyte and other brands to make them.

Result? Same misery as with Nvidia. All these cheapo chinese boards cut corners on signal quality.

Looks like Matrox decided to follow a negative trend

Luckily, this issue is getting more and more attention, so hopefully these guys will get a clue quick!

Neko

[thop]

the review was discussed before in this thread: http://forums.murc.ws/showthread.php?s=&threadid=34196

[KuroNeko]

read it. Deviates quickly into discussion about HDTV etc.

Lots of people saying quality is "gorgious!" just like my co-workers who are avid Nvidia fans say about their GF4s.

I trust C't much more that the Matrox groupies.

Not ONE of those supporters dared mention at which resolution they worked and if they did at the very least a side-by-side comparison with a G450 or an ATI or other card. Wouldn't be surprised if some were actually running DVI instead of VGA...

I expect we will get a Parhelia at work eventually (we have just about every bit of hardware represented in our test park) and will run tests at 1600x1200 and higher, but I definitely won't get a 500+ Euro card if it can't even handle those high-res images right.

Anyone know if P10 cards are out yet?

Neko

[Greebe]

Kuro, I'd be very careful when making such comments about us beta testers. If you think we're biased "groupies" then you are completely off base bud!

Errrrrrrr

Abuse us and we'll shut up completely and leave you in the dark. Now how refreshing would that be?!

[Dr Mordrid]

Not only that but groupies wouldn't put Matrox through as much digital hell as we do, would they Greebe?

That said IMHO those articles have done poor research and testing.

I've checked the output on a vectorscope & waveform monitor and they look great.

Putting high quality signals out to a Sony WEGA makes 'em look even better.

Now...there may be a difference between NTSC (here) and PAL (there), but others will have to test the PAL output...and I don't mean some magazine that doesn't seem to know how to go about it.

Dr. Mordrid

[Greebe]

I can pick apart c't's review to pieces...

for example: The G450.. there were 4 different speed versions released. Which did they test... the cream or the crop or the slowest, noisey version? They don't say... I think the latter. This in itself would be considered biased.

Same deal with the G550, did they test retail or OEM...

Now you get the bigger picture. So I'd drop that bull and smell the roses.

[Greebe]

Yes Doc we sure do

Sure they are great engineers, but they are not the be all end all of engineers!

Heck I'm of German decent and have spent plenty of time calibrating such equipment they're using to gain these results and know all it takes is one DOH! to mess up all the results. That comes from PMEL (Precision Measurement and Evaluation Laboratories) training among many other things.

[thop]

i don't want to start that discussion all over again but i think c't is maybe the most unbiased magazine there is. and they are professionals (not regular guys who happen to like hardware, but mostly engineers, programmers and things of that sort). they know their job.

but then the guys at matrox are probably better so that leaves me a bit puzzled. i guess i will have to judge it with my own eyes.

Heck I'm of German decent

does that mean your ancestors originated from germany? you speak some german?

[KuroNeko]

Greebe, your comments DO make you sound like a groupie. Sorry. Just my opinion as a professional test engineer.

You obviously haven't read any C't review, yet you claim to be able to pick them apart.

C't is the highest respected independant computer magazine in Germany and a large part of Europe _because_ of their thorough and unbiased testing. In fact, on multiple occasions I have caught them in the act of trashing a product of a company they had several page-size ads for in previous and current issues. Not to mention that C't has always been a defender of Matrox, clarifying to the public that all is not games in PC-land.

I certainly trust _their_ professional test results more than your obviously biased opinion.

Doc, much as I always appreciate your insightful comments, you are missing the ball on this one.

Fact: as beta testers your are most certainly Matrox fans.

Fact: either of you have still not mentioned the setup and resolutions you tested on.

Fact: you are running beta hardware. Handpicked and possibly handmodified or even hand-assembled cards. To claim that the C't results are worthless and accusing a reputed magazine and test engineers of false test results on a _retail_ card reflects poorly on the value of your comments and opinions.

I will reserve my final judgement until I have run my own tests, but I definitely will not buy this card blind after reading the C't article.

Finally, the discussion is about both VGA and TV output quality - or lack thereof. Don't mix them up. A TV output can be very easily roughly tested without expensive equipment - just run a test image through the output to the Svideo input of a TV and see where the resolution bars are no longer clearly separated. At 2.4 MHz limitation, as measured by C't with more accurate means, half of what it should be, such poor quality will be easy to spot even at home. Once more owners in Europe come forward, we'll see how PAL fares.

Neko

[Dr Mordrid]

Fact: as beta testers your are most certainly Matrox fans.

Your presumption is totally bogus. I've given them hell before and will again, and they know that to be a fact. They asked me to do it because they know they'll hear the truth, not because I'm a buttboy. Any of the other betas should be able to confirm this and I can confirm that the others do likewise.

Fact: either of you have still not mentioned the setup and resolutions you tested on.

See below

Fact: you are running beta hardware. Handpicked and possibly handmodified or even hand-assembled cards. To claim that the C't results are worthless and accusing a reputed magazine and test engineers of false test results on a _retail_ card reflects poorly on the value of your comments and opinions.

Actually they have better hardware than we do. Most of us have OEM cards running with slower clocks, but other wise they are the same.

Now...moving on....

I did read the article.

Anyone who knows anything about analog video would know that when you test it on this side of the pond it's NTSC standard interlaced video at 29.970fps & 1v p-p. For DV and MPEG-2, the most commonly used formats for editing and export these days, this means a resolution of 720x480.

Also: PROPERLY testing "TV" output (really NTSC as discribed above) does indeed require expensive gear....as in the vectorscope and waveform monitors mentioned above.

Eyeballing a set of colorbars tells you NOTHING compared when compared to what they can do as eyeballing depends on a coinsistantly inaccurate organic sensor mechanism; the human eye. The eye can easily miss saturation and contrast differences as high as 20-30%. 'scopes don't lie near that much.

Do your freaking homework.

Dr. Mordrid

[Greebe]

LOL, while I sound like a "groupie" , you come off as a biased hypocrate unable to evalutate anything greater than your ordinary snack machine.

I have also seen poor quality vid out/2D on G450's and G550's but then the above reasons I gave where also true. bottom of the barrel OEM cards only purchased because the buyer was penny wise and pound foolish.

So go on believing what you must. I'll stick to what I do best. Oh and by the way your vending machine just gave me back 50 cents more than what I put in it!

[KuroNeko]

Originally posted by Dr Mordrid
Fact: either of you have still not mentioned the setup and resolutions you tested on.

See below

Fact: you are running beta hardware. Handpicked and possibly handmodified or even hand-assembled cards. To claim that the C't results are worthless and accusing a reputed magazine and test engineers of false test results on a _retail_ card reflects poorly on the value of your comments and opinions.

Actually they have better hardware than we do. Most of us have OEM cards running with slower clocks, but other wise they are the same.

Now...moving on....

I did read the article.

Anyone who knows anything about analog video would know that when you test it on this side of the pond it's NTSC standard interlaced video at 29.970fps & 1v p-p. For DV and MPEG-2, the most commonly used formats for editing and export these days, this means a resolution of 720x480.

Also: PROPERLY testing "TV" output (really NTSC as discribed above) does indeed require expensive gear....as in the vectorscope and waveform monitors mentioned above.

Eyeballing a set of colorbars tells you NOTHING compared when compared to what they can do as eyeballing depends on a coinsistantly inaccurate organic sensor mechanism; the human eye. The eye can easily miss saturation and contrast differences as high as 20-30%. 'scopes don't lie near that much.

Do your freaking homework.

Dr. Mordrid

I will not even dignify the insults by Greebe with a reply. Anyone without bias reading this thread can easily weigh the worth of his comments.

As for my "homework"... any moron knows that a test image is not just color bars but incorporates a frequency testpattern as well. At least it does here in PAL land. Quite sure it also does in NTSC country.

If you'd have taken the time to read my post, you would also have noticed I did mention a _rough_ test, with minimal equipment, so anyone could verify the _professionally obtained_ test results from C't at home.

And you would have bloody well also realised that the resolution I referred to was for _VGA_ and not TV out. I even specifically warned readers not to get discussions of the 2 mixed up.

Since my posts aren't read anyway, I'll keep my voice from now on. No need to further upset the holy leaders. What needed be said has been said, for those that are willing to listen.

Neko

[Dr Mordrid]

Please note the header on this thread: TV OUTPUT. That doesn't mean VGA.....it means analog video.

Yes, we have the SMPTE images here as well colorbars, but again they are no substitute for HARDWARE.

Now..if you want to have a pissing contest with my 'scopes over who can read video quality better that's your business, but you're likely to get electrocuted

Dr. Mordrid

[xortam]

KuroNeko ... watch the AVS HTPC forums for feedback on the Parhelia as used in a Home Theatre application. There are some very critical and well-versed users there but they are mostly evaluating hi-res outputs (VGA, DVI, component etc.) to hi-def FPTVs or RPTVs. Most serious videophiles use these higher grade systems so its hard to get much good critique on NTSC/PAL systems. No one on the AVS forums has posted their Parhelia experience yet (still awaiting delivery).