[Anonymous]

the FX 5600 Ultra seems to be just about the same or slower than a GeForce 4 Ti 4200 8x and Radeon 9500 PRO.

The problem is.. for at least 3 months now the 9500pro has been neck and neck with the Ti 4600 in base FPS..

Why, oh why do websites post endless, endless endlesss misrepresentative information???

[Anonymous]

Quite significant that this is the first NVIDIA product update in the last two years or so which hasn't been coupled with rafts of OEM win announcements - in fact, has there been any yet? On the otherhand, for the first time in the same period ATI have already announced a bunch for their new line.

[cellarboy]

Originally Posted by ISV

I see a lack of DX9 functionality tests missing from those comments, using 3Dmark 2001 for DX9 class hardware is very strange indeed.

Seems to be a trend on review sites anymore.

Doesn't NVidia's apparant 'suggestion' that 3Dmark2003 not be used to test their hardware somewhat hinder this proceadure?

[RussSchultz]

Originally Posted by Anonymous

Quite significant that this is the first NVIDIA product update in the last two years or so which hasn't been coupled with rafts of OEM win announcements - in fact, has there been any yet? On the otherhand, for the first time in the same period ATI have already announced a bunch for their new line.

http://www.beyond3d.com/forum/viewtopic.php?t=4725, the paragraph right above "About NVIDIA":

Available in April 2003, the GeForce FX family of cinematic graphics processors is being offered by a record number of add-in-card partners, including Abit Computer, Albatron, AOpen, ASUSTeK, Chaintech, Creative, Gainward, Leadtek Research, MSI, Pine XFX, PNY Technologies, Sparkle Technology and Terratec. For additional details, list of partners, log on to http://www.nvidia.co.uk.

[Anonymous]

Originally Posted by Ostsol

I think the problem is that if the video card isn't fast enough to play games that use DirectX 9 features. . . what's the point?

Well then what was the point of criticizing GF4MX for not having DX8 features? GF4MX delivered significant improvements over the GF2MX and was a credible product for the <$100 market. Do you really think that a game that utilized lots of DX8 pixel shaders and significant amounts of geometry would have performed well on a value card with very low bandwidth and fillrate?

You can't have your cake and eat it to. To criticize NVidia for not introducing a $80 DX8.1 capable part, but then to turn around and criticize them for delivering an $80 DX9 capable part is the height of hypocrisy.

The fact is, on low end cards, even some heavy geometry and multitextured DX7 games would be unplayable. It is not the pixel shading capable of the card which makes the game too slow, it is the level of detail used. Halo could be called a DX8 title, and runs "ok" at 640x480 on a GF3/GF4-4200 equivalent. This game would crush any entry level GPU, so a GF4MX w/DX8 wouldn't be able to run it either.


NVidia was lambasted for "flooding the market with cheap DX7" cards in a time when DX8 was out. They were criticized for this because people thought it would make developers target only DX7. Now Nvidia is about to flood the market with cheap DX9 cards. Regardless of whether or not people will have to "turn down the details", or run in 640x480 with no AA, developers will be able to target DX9 instead.


My guess is, if the NV34 was not DX9 pixel shader capable, the same people criticizing them now, would be doubly critical of Nvidia sabotaging our DX9 nirvana future by shipping non-DX9 entry level cards.

If people would be objective for just one moment, they would see that one of the two biggest graphics vendors in the world shipping a entry level DX9 card is a good thing.

[Anonymous]

Originally Posted by RussSchultz

http://www.beyond3d.com/forum/viewtopic.php?t=4725, the paragraph right above "About NVIDIA":

Seemed like the standard AIB list there, not OEM announcments, such as:

http://www.beyond3d.com/forum/viewtopic.php?t=4723

[Gateway, Fujitsu Siemens and NEC are all announcing systems powered by the new RADEON boards. Gateway has chosen the RADEON 9800 PRO as the graphics powerhouse in their new 700XL system, which will be available soon. Fujitsu Siemens has selected the RADEON 9200, RADEON 9600 and RADEON 9800 products for its Scaleo PCs and NEC has chosen RADEON 9200 products for its Packard Bell range and the RADEON 9800 PRO for the NEC iSelect. All the major gaming-focused system integrators have also chosen the new RADEON products. ABS Computers, Alienware Corp., CyberPower, Falcon Northwest, iBuyPower, MDG, and Voodoo PC will include RADEON 9800, RADEON 9600 and RADEON 9200 products in their current and upcoming systems.

[Anonymous]

NVidia was lambasted for "flooding the market with cheap DX7" cards in a time when DX8 was out

Since when do IHVs need to be praised for doing the right thing !
It is ludicrous for a modern value card not to support DX 8/9 today, then again if the card is not able to render scenes with any speed then to a developer, even though the card can render the scene in reality it is not useable. To a developer then, it might as well not be there.

[Ostsol]

Originally Posted by Anonymous

Well then what was the point of criticizing GF4MX for not having DX8 features? GF4MX delivered significant improvements over the GF2MX and was a credible product for the <$100 market. Do you really think that a game that utilized lots of DX8 pixel shaders and significant amounts of geometry would have performed well on a value card with very low bandwidth and fillrate?

The point was the name. The difference between the Geforce 4 MX and ti series wasn't just speed. It was capability. It was two generations behind (the last previous generation from the Geforce 4 being the Geforce 3). Yet it retained the "4". Why? Why didn't NVidia just call it a Geforce 2?

Of course you can argue the same thing against ATI with the Radeon 9000 (which was slower than the 8500) and I'd agree. A heck of alot of people would agree, actually, and they do.

My guess is, if the NV34 was not DX9 pixel shader capable, the same people criticizing them now, would be doubly critical of Nvidia sabotaging our DX9 nirvana future by shipping non-DX9 entry level cards.

We certainly would if it was still called a GeforceFX, but only supported DirectX 8 and PS1.3. Cinematic graphics indeed. . . Supporting DX9, however. . . well, that's certainly a step forward.

[Joe DeFuria]

Originally Posted by Anonymous

GF4MX delivered significant improvements over the GF2MX and was a credible product for the <$100 market.

...except they actually tried to push it in the $179 bracket. But then, the Geforce4MX 460 is an easy card to forget...

NVidia was lambasted for "flooding the market with cheap DX7" cards in a time when DX8 was out.

That was the war cry, yes. But more accurately, they tried to replace the Geforce3 Ti 200 (DX8 card) with a GeForce4 MX in the same price bracket. It was seen as a step backwards.

It would be as if ATI tried to replace the Radeon 9500, with the Radeon 9200 at the same price.

My guess is, if the NV34 was not DX9 pixel shader capable, the same people criticizing them now, would be doubly critical of Nvidia sabotaging our DX9 nirvana future by shipping non-DX9 entry level cards.

Actually, I doubt it....considering no other vendor is shipping DX9 cards at that price bracket.

If people would be objective for just one moment, they would see that one of the two biggest graphics vendors in the world shipping a entry level DX9 card is a good thing.

I don't know if it's a good thing or not at this time...awaiting benchmarks. It mostly depends on how well it stacks up, performance wise, to the Radeon 9200. If performance is generally about on par, then its definitely a good thing. If performance is lackluster (with the assumption being that performance could be better if they dropped DX9 support), then it's not so good.

[Anonymous]

Well, using the same reasoning as the past posters, I would criticize ATI for preparing to flood the entry level market with non-DX9 capable cards, thus influencing developers to continue delivering DX7 titles just because the card has ok DX7 performance. Basically, ATI would be better off not shipping the 9200, because it ain't gonna run Doom3 fast, and all it will do, if it achieves significant installed base, is stall the delivery of advanced DX9 titles.


You may as well argue that there shouldn't even be an entry level market if your argument is that these cards don't have the power to run modern games. Either you are going to hold back developer progress because you don't support DX9, or you're going to hold it back because you lack CPU, fillrate, and bandwidth. Either way, the entry level market ties the hands of developers.

Basically, you're saying, NVidia sux for not delivering a value DX8 card (even though any game specifically designed for DX8 wouldn't run well anyway), and you are now criticizing them for delivering a value DX9 card, which does have modern features, so you find a different (and contrary) reason for criticizing them.

The fact of the matter is, it is not DX9 which determines the performance required for a card. Doom3, UT2003, C&C Generals, etc are not DX9 titles, but would bring any DX7 value card to its knees. Any game that makes heavy use of geometry, textures, or multi-pass techniques is going to crush entry level cards *regardless* of the DX version they support because entry level cards lack fillrate and bandwidth.

The arguments being made against the NV34 basically apply whether it has DX9 or not, DX8 or not.

I would suggest that there will be a sweet spot where DX9 effects can be used on these entry level cards, and they can run with acceptable performance in low resolutions. Turning off stencil shadow volumes, for example, would boost performance significantly, but not affect many DX9 fancy shader effects.


Having one standard API supported across all platforms is a boon for developers, even if the game will run slow (it will run slow, regardless of DX9 or not), or drop out features, or resolution, on some cards. I'd rather write 5 different PS2.0 shaders (one for each detail level) then write 5 different shaders, each a different version in a different programming language.

I think all entry level chipsets should be DX9 minimum, and I don't think you can criticize the NV34 on the basis that it will be slow on some titles, since that will be the case with or without DX9. The important thing is to have all these cards speak the same language.

[duncan36]

If you were actually listening you might understand what people were saying.

ATi's 9000 release was in no meaningful way comparabe to the GF4MX launch because it was almost as fast and at least in retail was priced less. No two ways about it the launch of the GF4MX was deceitful and disgusting. They purposefully confused consumers with their deceptive branding and many, many people bought GF4MX cards for $140+ at retail expecting to get at least a 'better than GF3' card, and they were basically robbed.

The 5200 launch is very different, people rightfully are saying 'at that clockspeed theres no big reason to get excited about DX9', you can't blame people for being thoughtful instead of being fanboys. The limitation on these low end cards is clockspeed and bandwidth not DX9 features. Its very possible that the low end 9200 which is DX8 could be the more desireable card despite having DX8.

[Anonymous]

Sorry, but there are two types of people who criticized the GF4MX. Those who criticized it because of deceptive labeling (the hardware was fine with them, as long as it was named something else), and those who criticized it for lack of DX8.

I am speaking about the latter arguments. I do not find ATI's numbering scheme any more logical. Secondly, I said nothing about the 9000, I am talking about the 9200. In any case, the numbering schemes are just as deceptive. It is impossible to determine, from product number alone, what API or chip core is supported. As a consumer, if I don't read the fine print, I might think that the 9000, 9200 have some relation to the 9500 and 9700.

The fact of the matter is, DX9 is orthogonal to performance in most games. Game performance will scale relative to the geometry, texture, and fillrate load, not whether someone is using DX8 or DX9 shaders, or the old register combiners. Most DX9 shaders, even on the NV30 or R300, will be relatively short and few. It is certainly possible to use DX9 shaders on the NV34, it is also possible to use no DX8 or DX9 shaders at all, and suffer tremendous performance problems. Any game that uses global lighting (stencil shadow volumes, shadow buffers, etc) is going to be dogged on any entry level hardware.

A $79 card that has DX9 certainly is something to be interested in, anyway you slice it. I'd rather have the market flooded with $79 cards with DX9 than have it flooded with $79 DX8 cards. Neither card is going to run top tier games at performance, but the preponderance of API support at the bottom does influence developers. Now that DOT3 has filtered all the way down to the bottom, we are just starting to see games designed that assume DOT3 in the pipeline.


I can't think of Q3, UT, or modern game engine today that runs well on entry level hardware. Last non Q3/UT games I bought were Warcraft3, BattleField 1942, and C&C Generals, and none of these games run acceptably on any entry level HW.


To criticize NVidia for delivering a full-featured entry level card because it won't run the most complex games at highest detail levels is not sound, and hypocrisy at best.

You may as well argue that all entry level cards should be DX5-class due to performance reasons. I look forward to the same reasoning being applied to the 9200 when it is shown that it won't run Doom3 acceptably. Did they waste transistors on features for not reason at all?


This board has gone way way down hill. I can understand the negatively with regards to the NV30, but the bias against entry level products is ridiculous, especially since 99% of the people posting would never ever buy an entry level card anyway.

[Pete]

Warcraft 3 runs acceptably at 8x6 on my nForce-220's integrated GF2MX (XP 1700+). BF1942, however, runs like carp (you 'eard me!).

[Reverend]

Dunno 'bout you guys but Guest here (why is it so hard to register, I'm wondering...) makes some very pertinent points.

[LeStoffer]

Originally Posted by Anonymous

This board has gone way way down hill.

Well, the main thing is that it's really just wonderful that both the two big companies are doing such a great effort to bring DX9 to the mainstream (and even at the value segment with the NV34!).

But that doesn't mean that people should just act like PR-brainwashed happy-going fools. At least not yet : We still need at lot of info on what kind of compromise that nVidia has made to the Cine FX (especially on NV34) just as we have to wait to Monday to get an idea on the performance.

BTW: It would be a nice gesture of you to log in when you're attacking the board for going 'way, way down hill' IMO.

[Pete]

Originally Posted by Reverend

Dunno 'bout you guys but Guest here (why is it so hard to register, I'm wondering...) makes some very pertinent points.

I disagree, but I'm too tired to elaborate. If no one else does by tomorrow, and I find my way back here, I'll try to make a fool of myself for everyone's enjoyment.

Let me just say I disliked the GF4MX both for its deceptive name and its repackaged, lacklustre hardware (compare it to a similarly-priced, "entry level" Radeon 9000). People who buy entry-level hardware don't expect to play the latest games well--they expect to play older games well. Thus the logic of releasing a top-end card with X features first, followed by mid-end cards also X-featured, and finally X-featured budget cards when the next top-end is released, by then XYZ-featured. I fail to see the logic in releasing a DX9 card if it won't be fast enough to play DX9 games (not to mention forcing you to consider the CPU with your vertex shaders, as now vertex shading impinges on your AI and other things). There's no point in driving developers to standardise on DX9 if the majority of "DX9" cards won't play at acceptable speeds.

Realistically, DX-compliance is not entirely orthogonal to performance, as it (along with price) generally clues you in to how a card will perform. We're worried that NV34's "DX9" compatability may be nothing more than another marketing ploy--sure, it may support DX9, but it will be so slow as to make the feature useless? We're cringing and sighing and groaning in anticipation of another deceptively-marketed card. Don't think we won't be happy if nVidia manages to deliver a budget card that delivers more than the competition. But don't expect us not to be once bitten, twice shy.

The ideal would be if IHV's can deliver a vertical offering of cards with the same features, but hardware-optimized for certain resolutions. The entry level would run well at 6x4 or 8x6, the mid at 10x7, the high at 12x10, each with just enough shaders and pipelines for their target res. But it's apparently difficult for IHV's to develop chips that way, so we're left with many iterations of many capabilities. Perhaps you (Guest) are saying the same, and perhaps nV will deliver with their NV3x line-up--but I doubt it. Perhaps ATi thinks DX9 hardware won't be effective at an entry-level cost, whereas nV does. Perhaps nV doesn't, but they're counting on the marketing appeal of DX9 compliance to attract/sucker more customers. Or perhaps ATi just doesn't have the engineers to develop three chips based on the same architecture at once, and nV's market, money, and talent advantage will inevitably prevail. I await the outcome with pleasure, as the competition will benefit us all, muddied though it is.

So much for a short post. I'll be back later to see just how badly my arguments fall in on themselves.

[Anonymous]

Article Info Summary:
After launching the GeForce FX5800 for the enthusiast segment, NVIDIA now brings the new DirectX 9 technology to the mainstream and entry level segments. Here's a first look at the features in the new GeForce FX 5600 and 5200 series.


However, we can already give you some of our first impressions about the performance of the new cards. In the standard tests without FSAA and anitotropic filtering, the FX 5600 Ultra seems to be just about the same or slower than a GeForce 4 Ti 4200 8x and Radeon 9500 PRO. This might be due to the reduced pixel pipelines (4x1 alias 2x2 (NV31) as opposed to 4x2 in NV25). With 4xFSAA, it appears to reach nearly double the performance of the 4200, beating a 4800 as well, but it loses out to the Radeon 9500 PRO. It's a similar picture with the anisotropic filtering. In the pixel shader tests from 3D Mark 2001, it beats the 4200/2800, but loses in the vertex shader tests. In both tests, it clearly loses to the Radeon 9500 PRO.

The FX 5200 Ultra is quite a bit slower than the 5600 Ultra in the standard tests. In anti-aliasing, it's just a tad behing the Ti 4200. It's the same with anisotropic filtering - at least in Balanced mode. In Performance mode, it even manages to beat the 4800.

We'll have more for you on this in a few days, including extensive benchmark tables. But this much can be said for now: the FX 5600 and 5200 cards won't have an easy time against the new ATI products.

[Anonymous]

Originally Posted by fat freddie

Article Info Summary:
After launching the GeForce FX5800 for the enthusiast segment, NVIDIA now brings the new DirectX 9 technology to the mainstream and entry level segments. Here's a first look at the features in the new GeForce FX 5600 and 5200 series.


However, we can already give you some of our first impressions about the performance of the new cards. In the standard tests without FSAA and anitotropic filtering, the FX 5600 Ultra seems to be just about the same or slower than a GeForce 4 Ti 4200 8x and Radeon 9500 PRO. This might be due to the reduced pixel pipelines (4x1 alias 2x2 (NV31) as opposed to 4x2 in NV25). With 4xFSAA, it appears to reach nearly double the performance of the 4200, beating a 4800 as well, but it loses out to the Radeon 9500 PRO. It's a similar picture with the anisotropic filtering. In the pixel shader tests from 3D Mark 2001, it beats the 4200/2800, but loses in the vertex shader tests. In both tests, it clearly loses to the Radeon 9500 PRO.

The FX 5200 Ultra is quite a bit slower than the 5600 Ultra in the standard tests. In anti-aliasing, it's just a tad behing the Ti 4200. It's the same with anisotropic filtering - at least in Balanced mode. In Performance mode, it even manages to beat the 4800.

We'll have more for you on this in a few days, including extensive benchmark tables. But this much can be said for now: the FX 5600 and 5200 cards won't have an easy time against the new ATI products.

This summary is from Tomshardware Preview.

So IMHO the FX5200 Ultra is an nice card but IMHO a littly bit too expensive at US $149. The card has nearly(?) Ti4200 performance (without AA+AF, but DAU's don't use it anyway ) at an slightly lower price and on top basic DX9 support. So when the street price goes down to < US$129 this will be an really good card for casual gamers, cause most games will run ok on the card.

The FX5600 on the other hand is imho not such an good card, cause the performance is only at an Ti4200 level without AF+AA and an lower level than the R9500Pro with AF+AA. So the card is not bad, but imho no "winner" cause the R9600Pro should be faster ( according to DaveB the R9600Pro is faster then the R9500Pro most of the time due to the higher bandwidth).

[Anonymous]

Anyone noticed NO review site got the FX 5200 Regular to review?
"Oh, sure, it's the world's slower card. Sure, no one would buy it if they really knew what it was.
But, you know, it's just that we didn't have sufficent extra cards to ship to all of you... That's because our employees have been stealing them for a while now, because they're so darn fast!"


Uttar

[Dave Baumann]

Anyone noticed NO review site got the FX 5200 Regular to review?

I've just been told its still under NDA, which is why its removed from this post.

[DeanoC]

I'm all for Dx9 features across the boards, it makes my life easier

Effectively supporting Dx9 and Dx8 generation means completely seperate paths through the engine, just as Dx8 vs Dx7 had seperate paths.

Fewer paths allows us to concentrate on making it run faster/prettier across all boards. Its a lot easier to drop some lights, only do 1 shadow, reduce to poly count, reduce the screen res then to have a completely seperate path with different render-states, shaders etc. It means only writing a skinner once, shadow system once, special fx once.

In reality will still have to support the Dx8 (and Dx7 and Dx6... ) so the paths will be there anyway. If they run the Dx8 paths faster, we will just chip ID them and ignore there Dx9 status, which kind of defeats the point.

[Dave Baumann]

This board has gone way way down hill.

It doesn't actually add to you case when a regular user such as yourself effectively contributes to the 'board going downhill' when he doesn't want to represent his own comments by not logging in.

Still, just becuase there is a certain element of noise in the news forum does not mean that the entire thing has gone 'way downhill' -- the news forums are purposefully made so non registrants can leave their opions on whats here and our reviews, and that may sometime be a little flamatory if you don' like it then stick the registration only forums.

However, IMO, the registartion forums haven't gone downhil, they have just got more active. And, contrary to my fears of the general level being 'dumbed down' I'd actually say we are garnering a more diverse level of support, that doesn't really appear to be driving developer types away - on the contrary perhaps.

[Anonymous]

Well I kind of agree with Guest that its a good thing that nV is pushing a DX9. I want to wait until we see reviews and numbers before I am 100% concerned. With nV marketing of late I am not goint to trust what they say at face value.

And I was a little bit miffed that the R9200 is only a DX8.1 card as I was hoping for DX9 cards from "top to bottom".

I suppose the only thing I disagree with Guest is that the R9200 is much closer to DX9 card then the GF4MX was to being a DX8 card.....

[Ostsol]

It had better be a DirectX 9 card. If it's a DX 8 card, but still retains the FX suffix, then it's no better than the GF4MX.

[MuFu]

Originally Posted by DaveBaumann

However, IMO, the registartion forums haven't gone downhil, they have just got more active. And, contrary to my fears of the general level being 'dumbed down' I'd actually say we are garnering a more diverse level of support, that doesn't really appear to be driving developer types away - on the contrary perhaps.

Always wanted to hear your opinion on that, Dave. Sounds encouraging.

Originally Posted by DeanoC

I'm all for Dx9 features across the boards, it makes my life easier

See that's also great to hear - I couldn't really back up my argument last week that DX9-compliance across the board would be an unquestionably good thing (I have the programming skills of a cashewnut), despite the performance of mainstream products when using advanced shaders etc being awful.

The GFMX series has been holding the industry back for way too long now, IMHO.

MuFu.