The Parhelia's Fragment AA IS an amazing technological leap, but it has quite a few downsides. It can't always be applied, misses some jaggies and results in eratic frame rates. Where FxAA can't be used the 4xAA mode is laughable, you are talking GF4MX perf, not what you expect from a $400, or even $150 gfx card. Where it comes to Aniso, nVidia are king. The quality is superb and the perf hit, while larger than (double) Rad8500, is still not huge. Rad8500 has very fast Aniso but the quality is inferior. As the Rad9700 hasn't been finalised yet it is difficult to comment, but I would expect it to be inbetween ATI & nVidia. Well all of the links relating to the Rad9700 that I found either didn't mention 10bit RGB (what Matrox call Gigacolor) or said that Rad9700 can do it, not 1 link said it couldn't. As I said the Rad9700 hasn't truly been finalised yet, mostly clock speeds and a few driver tweaks, as such we will know NOTHING for certain until late August if not September. It would seem that DX9 calls for 128bit colour and floating point precision, this would be far superior to 10bit RGB, and would work for games too, if you don't mind perf being quartered! If you like 2D apps which can benefit from 10bit RGB (not very many at all) then it may be worth having, but paying $400 for only GF3 3D perf is quite a price to pay. I would suggest you wait for solid Rad9700 info, price drops and actual releases, if not Rad9500, in the meantime try a Rad8500 (built by ATI) or GF4TI4200 (pref Leadtek as '2D' IQ varies), you may find these cards suit your needs and are a world apart from the GF3 you have, you may decide saving $250-300 is worth it if you only lose 10bit RGB. In 6-12 months prices, availability and loads more info should be out, and that would be a mutch better time to buy a Parhelia, Rad9500/9700 or the new nVidia card.....But now I find myself rethinking....I need a powerhouse jack of all trades card to support my needs because along with being a avid PC gamer I also do a lot of video editing through Premiere as well as 3D animation using Lightwave and After Effects.Some of things that I need is rock solid stability,GOOD multi monitor support,fast OpenGL rendering,Crystal Clear 2D performance,and support that is geared towards Pro and consumer level applications. LOL, to seriously say the Parhelia can compete with GF4 is a laugh. I can pick some benchmarks for you showing the 4200 beating the 4600 when both are at defaults etc etc, it certainly isn't representative of the whole picture. Check this out. Please excuse my shorthand but I believe most people will still understand it. You may think that the 4600 is an unfair comparison but rem that is where the P512's price point places it. The Radeon8500 is about a third the cost and beats the P512 almost every time too! So if I wanted to quote a few extreme examples (see the above links for the sources): Easy to pick a few odd benchmarks to say something, isn't it. Hey TheGoose I mean no disrespect, but I do find it very annoying when a couple of select benchmark results are quoted. I am more aware than most the power of statistics and how they can be used (and unused) to make a point. I still stand 100% behind the advice I gave earlier, IMHO the best thing to do would be to WAIT, if you're peed off with your GF3 then get a GF4TI4200 (pref Leadtek for assured great IQ) or Rad8500 (or LE but definitely a true ATI either way), if they aren't enough of an improvement then you've lost nothing (but could have saved yourself $250+) as in a few month's time prices of Parhelia512 and Rad9700 will have fallen, plus there should be plenty of concrete info and reviews AND budget versions of both cards (Parhelia256 & Rad9500?). If 3D perf doesn't really bother you then I would rec a Matrox card (but prob not the hugely priced Parhelia), but since you must want superior 3D for your $400 the only real choice is the Rad9700, not only hugely faster and superior to the GF3 in all areas, but IQ is as good as guearanteed to be far superior to the GF3 too. If you didn't work in high resolutions so much then GF4TI would be the wisest choice, Rad9700 if you can afford it. Since you have a GF3, and apparantly one of the worser manu'd ones, then you have good solid 3D perf, it is really the IQ that sucks and as such you must want an improvement in both IQ & 3D if you're spending over $100, let alone $400! Of course the Rad9700 also sports 10bit RGB, essentially all Gigacolor is. But it is also far better technically with full FP precision throughout and support for 128bit colour, which is far more likely to become a widespread standard as part of DX9 AND being applicable to games as well as applications. ATI & nVidia have both caught up with Matrox, gone are the days when Matrox were best for multi-monitor support and IQ, ATI for multimedia (VIVO/TV etc) and nVidia for pure 3D perf. With a market as active as the gfx arena is, it makes sense for ALL companies to close in on their adversaries' strengths and enhance their own weaknesses, as such we have already witnessed nVidia improving IQ, standards and dual monitor support, ATI catching up with AA and pure 3D speed (catching up?), and finally Matrox catching up (just about) with 3D perf. If ATI and nVidia weren't closing in on Matrox's specialities, then Matrox would be unlikely to bother enhancing the IQ, or providing 3 monitor support ... Matrox wants to play with the big boys in the mass market, I hope they succeed, but it will take a lot more than an overpriced card with inperceptibly better IQ, 3 monitor support (precious few have room or money for that), technically brilliant but a little flawed AA and last year's 3D perf. Sure a few loony Matrox fanboys and IQ enthusiasts will buy one at $400, heck that's why they set that price, but it will never become mainstream unless they halve the price, and even then it would still be tough. Whatever did they do with that 20GB/s bandwidth, esp considering the 4600's 10GB/s? I simply can't understand the logic behind someone recommending the Parhelia512 at $400 to EddNog, who is looking for better IQ and surely an improvement over his GF3 3D, something the P512 really doesn't have. Sure the IQ is great, and since he works at 'ridiculous resolutions' it makes snese to have the best IQ possible. But even the Matrox PDF file inadvertantly shows that both Rad8500 and GF4 have caught right up in IQ, just as everywhere seems to be saying. I believe the difference between Rad's and GF4's IQ and the P512 will be very slight and as such NOT worth the huge price tag the P512 carries. EddNog will notice a huge improvement for a simple $100 Rad8500 or $150 GF4TI4200 (better IQ & 3D), and in a few months if still unsatisfied should find both cards easy to sell on at a very similar price as he paid, and then decide to purchase a Parhelia, Rad9x00 or nVidia newby at significantly lower prices and with far more reviews, user experiences and good solid information to go on. Just as a note, I say '2D' image quality and put the '2D' in ' ' because we both know this stands true for both 2D and 3D (beware of WinXP setting the 3D refresh to 60Hz though), it is just far more noticable in 2D AND too many people get confused and think that 'image quality' refers to Anisotropic Filtering, TriLinear Filtering, Mipmap levels, AA , even supported refresh rates and simple RAMDAC speed. So I find '2D' IQ is the best way to put it. LOL Bingo13, I handle a lot more PC hw than I personally own. And I agree to a small extent that you can't beat personal experience. But even then, handling 1 or 2 samples of a product doesn't add up to all that much, esp with vast array of options and configurations out there, almost every setup is different and so even using a particular component doesn't ensure you can say very much with any degree of certainty. I don't need to drive a BMW to know it will be nicer than a mini, I don't need to have 8 children to know they'll take more time and care than 2, I don't need to use a 500MB HD to realise its limitations, I don't need to handle every single component out there in order to have formed an educated opinion and give advice. In any case there is no denying that the P512 falls very short on gaming perf, esp for somebody already very used to GF3 3D perf, this is something that needs more than clock increases in order to even reach GF$TI4200 levels let alone $400 Rad9700 levels. What the heck have Matrox done with double the memory bandwidth that the P512 has? In any case, where have you been Bingo13, you are pretty much just echoing what has been said already. The P512 most likely does have the best '2D' IQ, 3 monitor support and moderate GF3 level 3D perf. The P512 has its market, but I don't think it is worth EddNog shelling out $400 only to improve his '2D' IQ. IMHO he would be far better off getting a Rad8500 ($100) or GF4TI4200 ($150) and seeing if these fit his needs, if not then it will at least make the few months of wait for prices to fall and concrete info to come out, if not budget versions of Rad9700 and P512, now what's wrong with that? There is no doubt that both Rad8500 and GF4TI cards will have significantly improved '2D', the Rad8500 won't break the bank and the GF4TI4200 won't either, and will increase 3D perf at the same time. In my educated opinion based upon many individuals' experiences and independent reviews from multiple reliable sources I have given it how I see it, without bias or malice. So what exectly are you getting at, that I should not express an opinion without having tried every flavour of component out there? Unlike some, I don't simply rec a product based upon my own purchases in order to justify to myself I have not wasted my money. If EddNog wasn't too concerned about 3D perf, needed the absolute top in '2D' IQ and didn't mind shelling out $400, then the P512 would certainly be the top candidate, but I'm sure EddNog expects an increase in his 3D as well as '2D' IQ even when spending $200 let alone $400. Pretty much every user I've spoken to, even avid Matrox fans, admit that Rad8500 and GF4 have caught up with Matrox's pre-P512 '2D' IQ, surely it makes sense to give the Rad8500 or GF4TI4200 a spin for the relatively small cost and see how they fit his requirements, even if they don't he will likely find he can sell them on in a few months at pretty much the same price and still get a better deal on the P512 or Rad9700, that is if P256, Rad9500 or new nVidia cards aren't out by then! SSXeon5, regarding point number 1; I got the GF4 IQ being easily on par, and usually exceeding the very capable and respectable Rad8500 from a whole wealth of GF4 reviews, admittedly most review sites do fixate on benchmarks and such, but if you look hard you find most sites talk about IQ as well. Many users, and I'm not just talking nVidia fanboys here, have also mentioned the increased IQ of the GF4 cards, that's after upgrading from GF2, GF3, Matrox and Radeon cards mostly. But I think the most telling and impartial account of the GF4 being better at IQ than Rad8500 (and actually very close to Parhelia) comes inadvertantly from Matrox themselves: As I said, you have to scour GF4 reviews to find IQ references, also Parhelia reviews are also worth checking too. As I simply don't have the time to go through each and every site here's the main review sites I use: I use others but these are the ones I use most, I won't hold your hand nor do the ground work for you, if you look through the above for GF4 and Parhelia reviews you will find that, where mentioned, the GF4 IQ is a world apart from the GF3 and are right up there with ATI. As to your other point, YES! Rad9700 does stick 4600 in the anus as you so eliquently put it, who ever said it didn't? However, bear in mind that no independent has conducted a real review of a final product yet, the clocks aren't even finalised yet! However, the 4200 at $150 o/c's to either 4400 or 4600 perf and as such will give a nice boost over the GF3 EddNog currently uses in ALL departments, but especially IQ. Sure the Rad9700 whips the Parhelia, but if you listened to EddNog you'd realise it is IQ he's more concerned about, he still wants 3D, but is unsure of the actual diffs in IQ between GF4, Rad and Parhelia. Since $400 is a lot to spend if a $150 card may certainly do the job, added to which he should find a 4200 easy to sell on in a few months AND by then a lot of hard info will be out, as well as possible budget versions of Parhelia and Rad9700 (even if not the prices will have dropped). If it was my $400 (and I had it to spend) I would not hesitate to go Rad9700, that thing rocks, but everybody's requirements and preferences are different now aren't they SSXeon5.
Facebook
Twitter