AMD: R7xx Speculation

Sorry...


And this ? -> LINK

 

He probably does see that, given he wrote it!

 

I am aware of that which is why I would rather see real world gaming tests than flybys, walkthrus, ingame benches or cutscenes used for bench marking purposes as all 4 can be optimized for and give results that dont relate in reality to real world performance. Oblivion has an in game fps counter, Crysis, use devmode and there is absolutely nothing wrong with using FRAPS to get FPS readings.

 

It wouldn't be the first time FRAPs was detected and optimisations turned on in drivers.

 

I have never heard of that one before.. I'm sure its possible. But I have never heard of it...

Chris

 

What? You want a real life experience account?

Good. I installed the 8.5s, and everything was considerably smoother, from Crysis DX10 to WIC DX10 and even Hell-gate got some improvements on non-AA mode (smooth enough for all-round playing)


Not that you'd buy any of it though, given I'm an ATI card owner (as if that tag is worthy. Wait. It better be. )

 

If there is any increase in-game performance from a quoted driver that increases performance then it shows that the optimization is not specific to the benchmark.

Driver optimizations look for particularly slow paths and find ways to improve them. However, rarely is the game going to use exactly the same rendering properties across the entire thing - what may hit a slow path exessively in one level (or even one part of a level) may not on at all elsewhere. So, the level of improvement a driver is bringing can have massive variations dependent on where the user is testing. If one review looks at a particular level and showed an improvement less than the quoted level of improvement that doesn't mean to say that other levels, or areas tested, won't show as great, if not even greater, numbers than those quoted.

 

<COUGH> CRYSIS <COUGH> 20% boost only showed up in the bench, not in the game.

 

If your in game experience is showing the improvement, then good, but I'm not gonna lay down and take AMDs word for it given their past "perofrmance boosting" driver releases. NV aint safe with me either, show me in real gaming, not canned benches.

 

You've tested every rendering scenario in the game?

 

Good point.

Also, notice how XMAN26 makes broad blanket statements, and doesn't specify any details to his claims.

What driver set, what site, what settings, what benchmark tools, etc?

 

geee
I still remember the time when b3d forum was the place where bench-specific bashing started years ago... when someone found how NV30 was getting so high results in 3dmark.
Back then, most of participating reviewers agreed on doing more real-game tests, trying to fool drivers what game is being executed...
Now, if someone says that he don't believe any driver improvements until this being proven by 3rd party, he's beeing bashed and trolled.
A progress, indeed.
And to post something on-topic.
I bet AMD won't have magic drivers for x2 cards, they'll stick to AFR and will need to lower prices in order to compete.
And architecture wise the improvements will be as big as they were when going from 2xxx to 3xxx...

 

Even though B3D today seems to have more the tone of a technology "fan site" rather than "fact site", I have a feeling that the reaction is against the delivery and/or the delivery boy, rather than the message per se.

I have no reason to doubt that benchmark driven optimizations are still commonplace today, and that some benchmark specific optimizations occur for the small subset of reviewer darlings. The market mechanisms that drove that practise is still firmly in place.

LOL!
And now who is trolling?

 

Why trolling? That (what chavvdarrr said) seems quite like a logical assumption.

 

Really?
So thinking R600 - RV670 improvements is equal to RV670 - RV770 improvements is logical?
Shrink vs architectural tweaks?

Edit- Now if you are talking about the first part, that I would have to agree on, somewhat.

 

over 12% at the average for single RV670:
http://www.computerbase.de/news/treiber/grafikkarten/ati/2008/mai/ati_catalyst_85_dx10/

 

RV670 wasn't just a mere die shrink though. It implemented DX10.1 along with other things that did hamper R600s performance and as a result it outperformed R600 across many benchmarks/games.

 

Only that this time the progress effectively exists, and his statements were made without having an inkling of direct insight, they were made just because.

What is a real-game test?Why has this become some magical thing?We all do real-game tests, it's not some bloody breakthrough. We probably all have some spots we like to test with new cards/new drivers.

The thing that is bothersome is the blanket assumption that all reviewers are basically idiots who don't check beyond an inbuilt test (if one exists). This is untrue, anyone that is even remotely serious about this goes beyond the inbuilt test, but if what the inbuilt test correlates properly with what you're getting in-game, then you sure as hell use that. Due to ease of use and comparability.

It's the same story with timedemos/walkthroughs/whatever- you check if what you're getting with an "automated" test correlates with what you're getting in game(especially in those tight spots mentioned above), and then you use it.

 

The biggest problem, IMO, with testing in-game is the fact that it's nowadays hard to replicate the exact run in many games due the fact that AI etc might work differently on different runs resulting unbalanced testing conditions