Commentary On AMD’s Next-Gen GPU

H

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
You know, I’m just going to be honest with you…I like the way this guy thinks. I don’t like him simply because he agrees with our assessment of the HD 2900 XT, it is the fact that he believes that real world game testing above synthetic benchmarks and is not afraid to say so.

My two cents: When it comes to games specifically, avoid sites that use synthetic benchmarks or more importantly, appear to base their final opinions on said results. The ideal way to benchmark a video card is simply to run it through a gauntlet of real-world game tests.
Click to expand...
 
Frankly... I'm underwhelmed.

If Crysis already gives the 8800GTX a run for its money in beta, and the R600 is not the savior it was supposed to be.... I don't know how anyone will be able to really enjoy the wonder of the new level of graphics in Crysis.

I'm sad. :rolleyes: :eek: :(
 
It doesn't take a genius to see the value of real world gaming over premade demo's at this stage of video evolution. Glad someone outside of the [H]ardOCP staff took notice of it.

I sort of remember back when EverQuest 2 shipped, with levels of detail so high that even the top of the line machine of the day couldn't run it all.

Heck, even today a top machine can barely run it at max settings.

Same for Vanguard - Saga of Heroes. They admitted they put the graphics bar too high for the hardware that was available at launch.

As much of a beast as the 8800 GTX's are, it's still possible to overwhelm them if you try hard enough especially the higher and higher the resolution goes, and Crysis is one title that will do it.

I'm not sure what to call the 2900, it sure isnt a challenger to the 8800 series at all, and I'm thinking it won't be on the list of cards recommended for Crysis either.

Granted I'm sure you'll be able to drop the visuals enough to make it run, but it won't look half as good either.

Crysis is a next gen title for next gen hardware, the old and weak need not apply, or the new and weak.
 
It's a shame R600 hasn't turned out better. I've always liked ATI's drivers just because they play nice with more software and hardware than Nvidia's. There's a lot to be said for an easily implemented and maintained hardware/software package. Nvidia holds the crown for the best GPU out there but, it hasn't made their drivers any easier to live with.
 
xX_Jack_Carver_Xx said:
Frankly... I'm underwhelmed.

If Crysis already gives the 8800GTX a run for its money in beta, and the R600 is not the savior it was supposed to be.... I don't know how anyone will be able to really enjoy the wonder of the new level of graphics in Crysis.

I'm sad. :rolleyes: :eek: :(
Click to expand...

Don't understand why you say that. It's been noted by several Crytek staff members, that Crysis is supposed to run well on a 7800 GTX (which was also the card used in earlier demos of the game), with most of the eye candy turned on (except for DX10 features of course).
Now, you also've seen a 8800 card in action, in some of the Crysis DX10 footage we've seen around. And it seems pretty smooth to me, with all the bells and whistles on.
 
Well... I just completed my Newegg order for a GTS....

I was planning on buying one last week... but figured I'd wait and see what was coming out with the NDA's lifting today... and lo and behold.... I should have bought last week ;-)

I hope ATi get's their stuff together for the next round (or has some really competitive midrange pricing with this generation). As it is, this card is absolutely not worth it.

Friedmud
 
Magnus said:
It's a shame R600 hasn't turned out better. I've always liked ATI's drivers just because they play nice with more software and hardware than Nvidia's. There's a lot to be said for an easily implemented and maintained hardware/software package. Nvidia holds the crown for the best GPU out there but, it hasn't made their drivers any easier to live with.
Click to expand...

Only way I can agree with you on this one is in regards to the Omega drivers...otherwise I've found nvidia drivers much easier to swallow than ATI's...plus the actually have linux driver support.
 
Rather hilarious of the list of apologists. Good that someone pointed out that a lot of websites dont have the guts to actually admit(clearly) a product that is an inferior choice. Possibly avoiding to piss of sponsors or just fanboyism. Hope [H] keeps telling like it is.
 
Steve said:
...it is the fact that he believes that real world game testing above synthetic benchmarks and is not afraid to say so.
Click to expand...

To be quite honest it's so long ago now that any serious review sites and publications abandoned synthetic 3D gaming benchmarks in favour of in-game tests that I'm surprised anybody still uses synthetics, or considers it a matter worth discussing.

Sure, you find quite a few sites and publications which still include 3DMark results amongst the graphs and other paraphenalia, but if you look more closely hardly any of those really use that data to inform their conclusions. It mostly just gets a mention and little else. Understandable enough, I suppose. There are still a lot of readers who can't make 'meaningful' head or tail from anything other than the 3DMark scores. The in-game results are jibberish to those readers insofar as being comparitive indications. Nothing wrong with including a token reference for the benefits of readers who only really want to see the synthetic scores. Those readers need to have something to read.



But, on the other hand, we also can't assume that a comparitive review is 'more meaningful' purely because in-game benchmarks have been used. Comparitive reviews using in-game benchmarks can still fall short of being 'meaningful' if the test design assumptions fall short. The conclusions drawn won't necessarily be widely applicable. The [H]ard 2900XT review, for example, makes some assumptions. Whilst I enjoyed reading it, and consider it a 'better' article than most, it still left me with some questions.

"Why?" you might well ask, and the answer is quite simple. The tests there have used 'maxxed out' in-game detail settings and they've been predominately restricted to screen resolutions of 1600x1200 and 1920x1200. There is no comparitive figures given for earlier generation cards running at those settings - the only comparison given is the immediate generation competitor. The stated assumption is that this reflects the "highest playable settings", and the inference is that because it's newer it must necessarily be better.

There are inherent problems here. We're looking at benchmarks at the very edge of accaptable playability, so do those really reflect what we do with the hardware and what we expect from it? I see quite a bit amongst those results which would have me thinking "No thanks! Crank the settings back a bit!" Some of it is simply not demands I'd be reasonably expecting of a single-card rig. And with detail settings cranked back a bit there's no indication given as to whether or not the cards perform better or worse than earlier generation cards when the other reason for seeking a better card is introduced - higher screen resolutions. Let's face it, perhaps the biggest motivation to get a better card is the need to run higher screen resolutions. People having 30" monitors which run at 2560x1600 have the most reason for looking at more capable graphics cards.

Inevitably, though, conclusions drawn in such articles are generalised and all-embracing, and [H]ard's article is no exception. " As it stands right now the Radeon HD 2900 XT, in our opinion, is a flop." we are told. But, whilst I've seen evidence which suggests the card falls short for those people who want to 'maxx out' every available graphics detail setting, I've seen nothing to warrant any conclusion about its capabilities when cranking up the screen resolution rather than the details levels. I've seen nothing to suggest it is any better or worse for those people who want the best of all worlds, and who rightly accept that dual-card rigs are required to maximise performance with both detail settings and screen resolution cranked up beyond what the majority of people use.

The article and its conclusions are fine in comparison to many others, but they still fall short of being fully explanatory. And those shortcomings don't get acknowledgement.
 
From Matt Peckham's article....
I'm especially disappointed in virtually all of these sites for lazily failing to take Windows XP into consideration. A lot of folks are holding off on Vista because of the OS's serious stability and performance issues, and the first noteworthy DX10 games are still some six months off. I'm likewise weary of reading wasted pages on DX10's future with self-evident filler like "the real story will be DX10 performance," something everyone's been regurgitating since Nvidia's 8800-series arrived last year. We get it! Thank you! In any event, these really should have been XP vs. Vista benchmarks of current DX9 games. Instead we have numbers that tell gamers still running XP (i.e. most of us) next to nothing.
Click to expand...
Hip-fricking-hooray for someone finally saying that. Jeez, I'm so sick of Vista and DX10 blah blah blah. I know a lot of computer nerds where I am and none will touch Vista for another year or two. Some even go so far as to say they'll (Microsoft) pry WinXP Pro from their cold dead hands. While I think that's a little severe, Vista is a dog right now in my opinion (others have different opinions, I get it).

More on topic, is the comment about real world benchmarks vs synthetic. I agree that you can NOT depend only on synthetic. Infact is Synthetic benchies disappeared I wouldn't care. They are mostly worthless except for e-peen rights on forums and such.

Anyhow, good article, glad Steve linked it. :D

P
 
Phosphoros said:
Hip-fricking-hooray for someone finally saying that. Jeez, I'm so sick of Vista and DX10 blah blah blah. I know a lot of computer nerds where I am and none will touch Vista for another year or two. Some even go so far as to say they'll (Microsoft) pry WinXP Pro from their cold dead hands. While I think that's a little severe, Vista is a dog right now in my opinion (others have different opinions, I get it).
Click to expand...
Have to agree with the "Hip-fricking-hooray" myself, albeit for not quite the same reason. It's a point I noticed myself whilst reading through the linked article. I might be vocal in my endorsement of Vista, but that doesn't alter the fact that the vast majority of people use Windows XP, and especially so as a gaming platform.
 
I thought ignoring synthetic benchmarks has been the standard for a few years now...?

I certainly haven't paid them any attention. Why should anyone else?
 
You must log in or register to reply here.
Back
Top