If the pipeline doesn't have logic in to sort out overlapping read/writes, then it is faulty. Essentially, each pixel has to be 'checked out' and 'checked in' - it can't be read if it's checked out. Therefore, overlapping rendering generates pipeline bubbles, not incorrect rendering.DemoCoder said:
So what do you think about S3's DeltaChrome DX9 chip
- Thread starter Brent
- Start date
Nappe1 said:
You appear to be working under the assumption that 3Dlabs stopped developing parts after P9/10. Given that Creative have stated they bought 3Dlabs for consumer and they have sent out mixed signals on the immediate viablility of P10 for that market I'd think it silly to assume that no other development was going on.
Someone said on page 2 (I can't remember who right now 
) the DC chip might be faster than R300 if running at 300MHz and the deferred rendering stuff was working as advertised.
Well, will it really be faster? For pure pixel-filling, maybe. Then again, if it isn't TRULY a deferred renderer, but merely an immediate mode renderer who cheats by doing a Z-only first pass, it's not neccessarily going to be a beater. It might be the chip lacks most, if not all kinds of advanced occlusion elimination technology and thus HAS to use this software approach to remain competitive with the big-boys.
Also, its shaders might have low IPC, or its memory controller could be inferior to R300s. It might not have the bandwidth to support eight pipes + additional PS/VS demands at 300MHz if it's only 128-bit and regular DDR memory.
*G*
) the DC chip might be faster than R300 if running at 300MHz and the deferred rendering stuff was working as advertised.Well, will it really be faster? For pure pixel-filling, maybe. Then again, if it isn't TRULY a deferred renderer, but merely an immediate mode renderer who cheats by doing a Z-only first pass, it's not neccessarily going to be a beater. It might be the chip lacks most, if not all kinds of advanced occlusion elimination technology and thus HAS to use this software approach to remain competitive with the big-boys.
Also, its shaders might have low IPC, or its memory controller could be inferior to R300s. It might not have the bandwidth to support eight pipes + additional PS/VS demands at 300MHz if it's only 128-bit and regular DDR memory.
*G*
Forgive me as I have not gotten through all of the documentation on this new part...but did any one hear of a hint on when we might see these show up and/or reviews? I would gues late 1Q/ early 2Q.
I also am glad to see more players out there as that will help to drive down the cost and give us more main stream features. Its a good thing
I also am glad to see more players out there as that will help to drive down the cost and give us more main stream features. Its a good thing
Deflection
Newcomer
DaveBaumann said:
I agree with Dave. If anyone is equipped to compete with NVIDIA and ATI it would be 3DLabs. Anymore these three seem to be the ones at the front next-gen tech development as evidenced by their work in the OpenGL meetings. If a company is not pushing their future tech into DX and OpenGL it's unlikely to assume they have their sights on the high-end. We've seen before that without the leading edge tech of high-end parts, companies are ill equipped to compete for the middle too.
Look...
Here, the 300MHz Clockspeed, use of 300MHz DDR (1) RAM and a target performance of no less than 20000 3DMarks is revealed.
This compared to 4500 3DMarks for a AlphaChrome (S3 Mobile Chip).
They also have an interesting roadmap here..
(please notice that they use a "logaritmic performance scale")
The extremetech readers, however, don't look too impressed..
With regards
Kjetil
Here, the 300MHz Clockspeed, use of 300MHz DDR (1) RAM and a target performance of no less than 20000 3DMarks is revealed.
This compared to 4500 3DMarks for a AlphaChrome (S3 Mobile Chip).
They also have an interesting roadmap here..
(please notice that they use a "logaritmic performance scale")
The extremetech readers, however, don't look too impressed..
With regards
Kjetil
Nebuchadnezzar
Legend
0.13µ and still 300 MHz ??? 
Maybe there won't be any Desktop version of it cause of all that low-power bungus marketing... EDIT: ...and of "Mobile integration" (I didn't see that)
Maybe there won't be any Desktop version of it cause of all that low-power bungus marketing... EDIT: ...and of "Mobile integration" (I didn't see that
)DaveBaumann said:
sorry, but your conlusion is wrong. I just said that there's competition on their own market segment too. and I don't see why nVidia wouldn't do Quadro version from their GF FX as well as ATI does FireGL variant of R300 and both of them are definately trying to make on top of professional market. Still this does not mean that 3DLabs would be in trouble, but more like that they should not underestimate the importance keeping their position on that market. if they lose it when chasing gamers, they are on in the air.
and another point, I never assume anything if there's also knowledge available. we all know that few latest drivers have been having strings indicating future products.
:drumroll: I am just trying to be realistic on my comments this time. S3 is the only one (from SiS, PowerVR, S3, 3DLabs.) that has introduced / released something. All the others are still on the rumours only. some of them are pretty strong and others less strong. Only time shows which these will materialize. SiS propably comes with their octa pipe chip and 3DLabs will definately introduce something new at least to preserve their situation on their own market segment. The thing that I doubt is, their possibilities to design chip so flexible it would be competitive on both segments. Making two differnt versions from one chip is possibility, but then we come the problem so often said on case of Bitboys: available resources. (and afaik, BB solved this with multi-chip scalability, which could be also 3DLabs way to do it.) the direction of development sounds good to 3DLabs because DX9 is step forward on direction that they are taking their chip development, but is the next generation chip already that chip which combines rendering and gaming worlds, is the question mark. IMO, time is not ready yet for that. (but it's just my IMO.)
and Dave, you must remember that info that is under NDA is called rumour to general public. (until it's confirmed by the company themselves. and when they do it, it's not anymore rumour and not anymore under NDA.) and I respect you fully on 3DLabs news, knowing how good relations you have there, but my guess still goes that you need one more generation before you can combine very flexible programmable rendering targetted chip and gaming optimized chip efficiently and cost wisely. (Still nothing can stop 3DLabs trying that. And you know how I love positive suprises.
)one more thing I want to remember to everyone:
No one of us knows how many of us knows information that others don't know and is not allowed to talked publically. Some of you might know some of that what someone else knows here. Some here still knows what I knew when I knew. (Do they believe my words anymore? hard to say... most of questions are still open though Dave Barron's BB article did most of the job.) Only rare ones have lived the good times and bad times on this flow of news that I have been following last 1.5 years. And of course there's stuff that no one has heard from me. I used to know lot more than I know right now. (practically nothing, though I keep my eyes open and mind sharp.) Logical reason for this, is that companies I had contacts with are not doing much with this market anymore.
You're still not quite getting it. 3Dlabs are responsible for 3Dlabs market segment, Creative are responsible for the consumer market. Creative did not buy 3Dlabs for the Workstation market.
Nebuchadnezzar said:0.13µ and still 300 MHz ???
Maybe there won't be any Desktop version of it cause of all that low-power bungus marketing
I'll quote this from the article.
""The initial emphasis is in mobile," said Young Kwon, a senior product marketing manager for S3. "That's where we can win, where we can claim we're the best in the world."
Still, Kwon said, the DeltaChrome will also be sold as a desktop part, the company's first foray back into the desktop space since Via bought up the graphics assets of S3 Inc. However, Via's S3 division will market its chips to low-cost PC builders, preferring to compete in the $20-per-chip space rather than in the higher-margin, higher-performance segments controlled by ATI and Nvidia."
Which sounds pretty sensible, IMHO.
Should be a strong contender in the mobile space, and at $20 it might find its way into quite a few machines. Portables, after all, is growing towards a third of the market in units shipped. In terms of value, the mobile market share might be higher still.
If you fed the chip a higher voltage you could probably push clock speed up on a stationary card, (although processes can favour either low power or high speed, and just giving a low power part higher voltage will not make it perform quite as well as a part that was designed for, and produced with a high speed process to begin with.)
Hey, if SIS can find willing manufacturers of Xabre cards, this part should have no problems finding its way to market either, regardless of performance - after all it is buzzword complete. And as long as the performance isn't too shabby, it will definitely put some price pressure on the market leaders.
Entropy
DaveBaumann said:
well, heh
I am getting it but obiously I can't find the right words...look, what I am talking about is the chips. those are the _the thing_ making card alive. And If I have got this straight, 3DLabs is responsible for chip design, right? so, creative needs a chip to put in cards that they are manufacturing (??) and marketing, right? as what comes to design process, design chips is the hard part, right? I am not saying that nowadays multilayer PCB's would be easy to design, but easier than chips and Creative could be said as a veteran on AIB manufacturer/designer so they propably fit on that task better than any other possible company.
hmmh... maybe you didn't read my post thru before answering. (I feel like saying these things again.) and yes, my post wasn't very well aligned. Sorry about that.
Your own post had that answer. You point to NVIDIA and ATI's encrouchment on the worstation market, but they do that with desktop designs (albiet with minor tweaks in NV's case, to stop soft Quadros occuring again) - they are proof that Workstation and high end Consumer can exist on the same parts. 3Dlabs have already proved their designs have en eye on consumer with P9 having deferred rendering - absolutely no use in the CAD market with wireframes!
The other thing I'm saying is that you shouldn't necessarily assume that the levels of resource that 3Dlabs had available before the creative purchase will stay static. Even if they don't have an influx of Engineers, Creative aren't exactly green in this area themselves.
K.I.L.I.R:
The morrowind thing was mostly just rumors from their forums that were floating around. Now that I think about it, I believe there were a couple of people here saying that the assumption was flawed in another thread, so perhaps it's bogus. I still have to wonder what the heck they are doing to get such abysmal performance figures.
Nite_Hawk
The morrowind thing was mostly just rumors from their forums that were floating around. Now that I think about it, I believe there were a couple of people here saying that the assumption was flawed in another thread, so perhaps it's bogus. I still have to wonder what the heck they are doing to get such abysmal performance figures.
Nite_Hawk
Joe DeFuria
Legend
What I'm mainly disappointed about right now is timing. According to Extreme tech, the first chips (mobile) are expected to be "begin sampling" in Q2, with desktop versions "to follow."
I get the uneasy feeling with how these things tend to encounter a snag or two, that the the first desktop parts won't be seen until 2004....
I get the uneasy feeling with how these things tend to encounter a snag or two, that the the first desktop parts won't be seen until 2004....