R520 info thread #(9!)^9

FEAR, with all the eye candy on, is performing a soft-shadowing algorithm that D3 can't match.

Jawed

 

True, but it will be more interesting once we see dynamic branching used in games before drawing any conclusions about how important this performance difference is. Synthetic benchmarks are one thing, games are another.

One game in particular that I think will be good for testing dynamic branching performance will be the first UE3 engined game. I expect the G70 card to be fairly competitive in indoor areas, with the R520 pulling ahead for outdoor areas.

 

And why the difference between outdoor and indoor areas if I may ask? I expect the R520 pulling ahead for the simple fact that at first sight it seems to excell at new games that take shaders to the extreme. If Dave's article will prove that, then R520 I find more interesting. For the simple fact that it will be better for newer games, and has performance enough to run current games at great speed.

 

I'm suspecting that in outdoor areas, the shadow edges will be dominated by objects that are far away, whereas indoors the shadow edges will be dominated by objects that are nearby.

Nearby objects (typically) cast big shadows with wide edges, so that the savings in rendering a smaller border around the edges becomes less important.

 

Ummm, certainly aspects of FEAR look considerably better / more detailed than Doom 3. Others (like the lighting engine) do not, but it would be brave to make any conclusions at this stage.

 

Sorry if I wasn't clear, but I'm almost positive $200 was the estimated street price, whereas the actual MSRP was $219 or $229. I know the difference b/w retail and street and all that, but I'm pretty sure it happened with a few cards at the $200 level that the price the IHVs put out were their estimates of street prices, or "retail" prices after a rebate.

Yes, I'm being nit-picky. I know that, given sufficient supply, cards will drop below MSRP as stores attempt to net higher overall profits by shifting from higher profit per sale to higher overall sales, and thus overall profits (especially with competition from other stores). I was just pointing out another case of having to read between the IHV's lines.

 

I don't buy that at all.

Well, there's more to it than that. For one, a significant portion of the cost of the chips is R&D, and once that's recouped the chips can be sold at a lower cost. As products acquire more design wins and production ramps up, economies of scale can help to further lower costs and therefore prices. Competition from other retailers can also help to reduces prices for consumers.

 

I'm pretty sure about this, Chal, as I asked a reviewer about it at launch. Can't remember which card or reviewer, tho. A quick check of TR's comments for several reviews turned up nothing, and I'm not up to searching AT or B3D. It may well have been the 9600P I'm thinking of, per Wavey's "circa $199" price.

Edit: In any case, Orton seems adamant that we're seeing actual MSRPs, not street prices, according to geo's recollection of today's CC:

I'm going to take SRP to mean street price, as I have no idea how to deacronyze it without ending up with MSRP without the M.

 

Hey, somebody does read them. Kewl.

I took that to deacronyze as "Street Retail Price".

 

Ok, but be careful. MSRP is not mean price, it's Manufacturer's Suggested Retail Price.

I still don't get the original quote, stating it's not MSRP, but SRP. What's that supposed to mean? (MSRP and SRP are interchangeable terms) Some new meaning or SRP?

Doh...I should read more and try to comprehend what I am reading while I am at it. Ok, so this is Geo's little acronym when he wears Orton's hat. I think I get it now.

I think someone as debonaire as Geo channeling Orton should use four letter acronyms like CARP or CMRP: Current Actual Retail Price or Current Mean Retail Price.

Of course, answering the question "How much is Jim-bob asking for it?" works for me.

 

Well, considering nobody but the manufacturer is going to be suggesting prices, I'd say SRP just comes out of the "M" being redundant. I don't think you can really discern between MSRP and any estimated street price. Any difference in street prices from MSRP would be indicative of the manufacturer not properly forecasting supply and/or demand.

 

Well, darn it, I can go back and listen, but my memory is I actually heard "SRP" and "street" used interchangeably in that part of the conversation.

Darn it, be right back. . .

Edit: Yes, he actually used the acronym "SRP" and did not break it down. I never heard him use "street" even. Tho he did use "North American Retail Pricing" at one point as analgous to MSRP. He also used more words then I conveyed (of course --it was a summary except where quote marks were used) to explain it, but clearly whatever the actual literal breakdown of "SRP" as an acronym, there is no doubt he meant what we all here would call "street price". There. So not me being debonair, but thanks for the sideways compliment. I think.

 

Oh, no, no, no. Am I going to have to literally transcribe that section? He clearly meant that the street price for both X1800 and X1600 lines will be less than the MSRPs that they announced, but that because there is a greater absolute delta between the street price of the enthusiast line (X1800) and its MSRP than the absolute delta between X1600's MSRP and its street price, that the apparent $200 gap in their pricing configuration isn't really by the time folks are buying these cards on the street.

That was the whole context of the question --why is there a $200 gap? Answer: There won't be at street prices.

The conversation is at about 1:55:30 on the webcast for anyone who wants to listen for themselves.

 

Heh, I tried giving you rep props, but I was chided to spread the love first. So a "good effort" will have to suffice for now.

OK, makes sense. I hate acronyms.

Hee-hee! </Homer giggle> I didn't mean to recreate the acronym. I meant to say, "I'm going to interpret SRP as street price," not, "I'm going to interpret SRP as MSRP." Now I'm dizzy.

Is this always true, or can a manufacturer (surreptitiously/blatantly) boost the MSRP in order to charge more per item?

I hope Orton is wrong for the X1600XT's sake. That sucker don't look too good in most current benchmarks against even a 6600GT, considering its price.

 

Well, now that I'm done defending the accuracy of my deathless report . . . it does seem a little bit of a facile answer even on its own terms. I mean, how much difference could it make? Reduce the gap to $180? Maybe $150? It's still a pretty big hole as these things go. I look at it, and I think of X800XL pricing falling pretty rapidly, and something more 550/1200ish stuffed between XL and XT. I mean, that's a whopping big bandwidth advantage that XT has over XL for a "one step" in the enthusiast line.

 

I thought Unreal 3.0 was developed to make use of SM 3.0 hardware ... and was used for marketing the NV40. Does'nt that mean Nvidia has an advantage ?

 

Unreal 3.0 is one of the major engines for the Xbox-360 too. So wouldn't that mean ATI has an equal advantage?

 

One thing that may give nVidia some advantage is that the shader library setup with UE3 will make it easier to optimize for a specific architecture, and since nVidia's architecture requires more optimization for optimal performance, UE3 may well do relatively better.

But UE3 is going to make use of dynamic branching, so that's a bonus for ATI's hardware.

 

X1800XT overclock

Now, if the XT is clocking so well, I hope the XL won't be so XL as it's name suggests. (X800XL being not that great OC-er)

 

Errr, they got a 10% core overclock and a 19% memory overclock, and somehow got more than a 25% performance boost in Doom3? Something is not right with Driverheaven's benchmarks.