Last edited by a moderator:
NVIDIA GF100 & Friends speculation
- Thread starter Arty
- Start date
Heh I love it. A couple of months ago, tessellation was the big thing for some people (here and in other forums). Now that leaks seem to suggest that Fermi's very good at it, the same people call it a gimmick. It never fails
I'm really curious to see the alleged decrease in TMUs vs GT200, because the leaked performance numbers suggest much higher performance vs GT200. The in-depth architecture analysis will be very interesting.
I for one am happy that GF100 will shine in tesselated workloads!
This can only be good for PC gaming and general adoption of this technology. If for some reason nVidia would make half-baked tessellation then that would really slowed adoption because of all of these TWIMTBP games coming.
I think AMD solution is adequate for first DX11 card - a lot faster than XBOX can push which means sufficient for any console port plus some extra (aka Dirt2). GF100 will simply be faster in heavily tessellated scenarios. In most cases (AvP2?) this will manifest as even more polys on close objects than RV870 can push without decreasing FPS thanks to adaptive algorithms, but who needs so many sub-pixel triangles anyway?
I have a feeling we will only see tessellation used in more rational ways when next-gen consoles start to appear or shortly before that.
Bring them on ASAP Microsoft and Sony!!!
What I was meaning was saying the letters out loud, so 4x SSAA = Four times super sampling anti aliasing! I wondered if the industry was getting a little rediculous with the abreviations of technical terms, so much so that people know that AA means less jaggies and AF means better textures but have no idea what AA or AF actually means.
As far as your electricity bill is concerned, OUCH! You pay 2* more than me. I take it that its going to increase over the coming years due to Copenhagen etc? Im just wondering what impact increasing power costs will have on the high end market. In addition to this the trends towards more mobilie laptop and desktop replacement units may be accelerated due to increasing power costs.
So what does this mean towards Fermi? Is having a large die and being unable to deliver on time for the OEMs liking going to dictate the future of GPUs right up the chain to the enthusiast parts? Thats the question im thinking about, because when does Nvidia even get a shot at getting some design wins with their Fermi architecture for the mobile space? Next year? Then what? Are they facing down potential AMD 32nm refresh parts with even better performance charactaristics for the mobile space?
It depends on game/engine on how it loads the hardware. Some have much higher perfomance other less and then the question if CPU was limiting.
There are already games with DX11 support and many more to come (both exclusive and multiplatforms). Early adoption and better and clever use as time passes. Such thing as tesselation will make for a even more drastic difference than already by highlighting the difference between 'lego brick characters' and CGI alike ones.
Also DX11 being something far far better in comparision to DX9->DX10 in ease of use, freedom etc.
Last edited by a moderator:
DavidGraham
Veteran
Nope ,Mr.Rys made his statement because there was no revealed info about hardware tessellation , also he couldn't identify it's presence in the die shot ,and hes said he could be wrong , I don't think any of those excuses apply to Charlie though , for obvious reasons !
neliz is referring to the fact that there's a thread for Charlie "loving"/"bashing" and this isn't it, though it is funny to see that only those "bashing" take heat, while the ones "loving" don't
Anyway, that bjorn3d post got me curious. The NDA lifts for all launch products ? I mean, will we see both Fermi based GeForces (380 and 360) or just one of them ?
I'm asking because the previous leaked info/videos, suggested that the GTX 360 was the only one being shown and since the launch is still around March, NVIDIA may still want to keep the performance of the full Fermi chip in the 380, under NDA.
PSU-failure
Newcomer
It has still to be proven otherwise, though.
The main thing which could improve setup rate and tessellation is exactly that, but at the same time it requires a sensible ALU usage, so it would not be that good in the end.
It couldn't be the FF rasterizer has moved to the hot clock as it would in fact be half the said hot clock as only the ALUs run at full clock, which is already less impressive.
By no mean I'm assuming it's a bad idea to improve tessellation performance, but it remains to be seen what this performance is, and why.
PSU-failure
Newcomer
Quite the reason why I'd really prefer GF100 to fail.Millions of g92's were rebranded in order to bring you this 'radical' new card.
NV becomes worse and worse with each day they have more than 50% market shares, a little humiliation would only give them a reason to change their minds about their customers/partners.
AIBs are not cash-cows, neither are customers, and devs don't have to choose which manufacturer's products to target, that's exactly why APIs exist, and that's exactly what killed OpenGL... and guess who already included numerous "proprietary extensions" back then?
FrameBuffer
Banned
I honestly don't understand how anyone could want the GF100 to fail. Are these the same people who wanted the R600 to fail ? If Fermi (GF100/300) does well it's better for everyone, more competition, forcing better prices for the consumer and force ATI to fight back.
"and guess who already included numerous "proprietary extensions" back then?"
Answer: ATi and nVidia as well as S3 and Matrox (I'm not sure about Intel).
LOVE the bunny suit! You did forget to mention how he's been right about Fermi so far in spite of so many people trying to bash him along the way.
Are you the same guy who started this site? If so, please update it...I'm dying to see what happens next.
Good grief, so he got the tape out(which some still think he was only partially right on, depends on what you call tapeout) and when it will launch.
His other predictions concerning have been poor performance, not much better than GTX285, completely dismal to name a few.
I find your honesty refreshing.
I'd say OpenGL is still on life support, and Direct3D's rapid progression is a far more important reason, but implying that the extensions were a bad thing is the most serious flaw in your reasoning. Using extensions to give access to new hardware capabilities has probably helped to keep the API relevant, if anything.
Charlie writing a piece of NV's tessellation/setup design? I don't believe he has the necessary knowledge to write such a paper, so this amounts implies someone at AMD feeding him talking points, like a healthcare lobbyist writing a speech for a Senator.
Frankly, I'm losing respect for some of the forum regulars and some of their sudden turnarounds on tessellation. It was a great thing to have back when it was assumed that AMD had the only true DX11 HW implementation, and everyone thought NVidia was going to do it all on the ALUs. Now many people who were cheering tessellation just a few months ago are talking about how it's not really important or needed.
How about we leave out personal feelings about tessellation from the standpoint of whether you like it, or think it matters, and just evaluate, from a technical standpoint, what they've done?
Frankly, I'm losing respect for some of the forum regulars and some of their sudden turnarounds on tessellation. It was a great thing to have back when it was assumed that AMD had the only true DX11 HW implementation, and everyone thought NVidia was going to do it all on the ALUs. Now many people who were cheering tessellation just a few months ago are talking about how it's not really important or needed.
How about we leave out personal feelings about tessellation from the standpoint of whether you like it, or think it matters, and just evaluate, from a technical standpoint, what they've done?
No they aren't. I'm very well aware what Rys said in the past and how exactly he meant it. And with all due respect Rys was probably damn close to the real tesselation implementation of Fermi in complete contrast to others that were completely wrong.
In hindsight all that matters is what shoots out at the other end. And if GF100 should turn out as efficient with adaptive tesselation as NV wants everyone to believe then Rys for sure hasn't a shred to apologize about.
Perhaps you should consider all metrics before coming to such conclusions. i.e The performance differences being talked about, how do they measure up in terms of die size difference/no of transistors, board powers etc., etc...
What consumer cares about die size or # trannies? People care about performance and cost. When I buy a Plasma TV, I could care less how many rejects came off the assembly line, or what the expenses were that went into making my Honda Civic, all I care about is it's performance record and cost. Anything else is spin.
The average consumer doesn't even care about power consumption, otherwise , they'd stop using halogen/incandescent lights, buy more efficient cars, refrigerators, etc.
jimmyjames123
Regular
Oh puhleeze
Never, ever, ever put Rys on the same level as Chuck. Sometimes it's not what one says but the way one says it that is important PSU-failure
Newcomer
Pure performance competition is one thing, questionable practices clearly is something else.
You can't say you act to change things while at the same time it's clearly the opposite.
BTW, price will never climb such high as I or most of us won't be able to buy even without competition, any manufacturer needs to improve its products over time to maintain sales. They even need to maintain their prices in certain ranges so that they can sell, and with just 1 manufacturer you're almost guaranteed to have the best performance... which is exactly what NV is expecting for some months now.
Last edited by a moderator:
Speaking of Rys, I wanna see how close he came with the guess-work in his TR article from a few months ago. http://www.techreport.com/articles.x/17815
jimmyjames123
Regular
Millions of g92's were rebranded in order to bring you this 'radical' new card.
Not sure what point you are trying to make. Consumers shouldn't be excited about a brand new product that gives much higher performance per dollar, performance per watt, and features per dollar vs the prior gen architecture......due to lower end products being rebranded in the past? Umm, okay. I suppose no nation should do business with any other nation due to all the atrocities committed by these nations in the past?
The reality is that sometimes products do have to be essentially "rebranded" for various reasons: need for limited R&D investment, need for quicker time to market vs developing brand new cards for that segment, need for new naming scheme to maintain clarity with respect to performance heirarchy vs newer higher performance products from the same company. The only way to twist this into something truly evil is if a company rebranded a product and raised prices on the rebranded product. If prices stay the same or go down for the rebranded product, then it can be justifiable for the reasons given above.Good to know.
I hope the mods/admins keep the Charlie haters (dont see any lovers) in some other thread. This thread was spun off as speculation doesnt mean its a free for all to go offtopic.
Similar threads
- Deleted member 2197
- Architecture and Products
