R520 launch delayed?
- Thread starter kemosabe
- Start date
Tahir said:
Preassembled SLI rigs?
Sales of SLI chipsets mean nothing. I own one, the NForce4 was the only game in town for AMD PCI-Express, and it is still arguably the best chipset for AMD cpus. Does that in any way, shape or form imply that the customers of the nForce4 are actually running SLI?
No.
And while you insist that SLI makes a major difference in the marketplace, my anecdotal evidence (local retailers) say that it just doesn't happen. The price/performance simply isn't there. And even in this place when the call has gone out, the number of raised hands have been very easy to count....
We have no data that show us if SLI has any significant marketshare whatsoever. None. Again, until we have independant statistics, (and as far as I can see, Valve is the only party who could supply it,) it is all marketspeak and wishful thinking either way.
Lack of data is lack of data. We just don't know. But not only do I not know of anyone who owns an SLI rig, I don't even know a retailer who has sold even a single one. Until I do, I will continue to think of it as a feature in the marketing war of the IHVs.
egore said:
Dave Orton isn't a Shaman. He doesn't wake up one morning, stir the chicken bones, and say "Today is a good day to release a card." Or, conversely, a black cat crosses his path and he decides they better wait another three months. They had lead up, partners in line, roadmaps out, winks to various groups, several folks independently reporting they were pointing at Computex. Probably planned synergy with the Valve "ATI levels", etc. The decision to delay would not have made lightly --that there were factors that made it attractive to do so, obviously.
For one thing, it's another little knick on their reputation for their ability to execute --a reputation that over the last year or so is starting to look a little like the fender of a '57 DeSoto out behind the barn. Which is not to say it wasn't the right choice --getting it out the door on half-baked drivers would have been much more disastrous to their hard-earned reputation in that area. R8500 never did recover its reputation from the early months of driver woes (oy, do I know on that one), and a lot of good work in the Catalyst program to improve their reputation in that area could have easily been flushed in just a couple months of serious driver issues with R520. In fact, I don't have too much trouble imagining Terry M. throwing a red-faced arm-waving hissy fit with much stronger language making that same point internally.
To my knowledge, no, it doesn't. Yet. (Nor does it seem that Futuremark do so either.)Natoma said:
I hope they will implement a way of detecting SLI though. Valves' is a much, much better database over what kind of equipment is actually in use by online gamers, whereas FutureMarks database contains a lot more specific performance data. They complement each other, but to get a reasonable estimate of the market penetration of SLI, we would need to compare against Valves body of data. FutureMarks ORB is useless for that. Unless of course we assume that everyone who has an SLI rig has tried running 3DMark on it. Which may actually not be such an unreasonable assumption.
caboosemoose
Regular
The driver issue is interesting isn't it? Probably someone has already mentioned this, but ATI has essentially been using the same architecture since R300 / 9700 so maybe they are have a few teething issues with their first properly new chip for a while. And as someone rightly said, even if the chip was great, a couple of months of dodgy drivers at launch would be enough to do huge damage to ATI reputation. It's a tough business, I wouldn't want to be in their shoes!
caboosemoose said:The driver issue is interesting isn't it? Probably someone has already mentioned this, but ATI has essentially been using the same architecture since R300 / 9700 so maybe they are have a few teething issues with their first properly new chip for a while. And as someone rightly said, even if the chip was great, a couple of months of dodgy drivers at launch would be enough to do huge damage to ATI reputation. It's a tough business, I wouldn't want to be in their shoes!
Yeah, as much excellent work as CATALYST has done. . .well, "Welcome to the Big Leagues son, they throw the slider here." Be very interesting to see how much transparency and responsiveness they can maintain this time.
Personally after seeing so many launches and hype wars, I think nearly all spectualtion nowadays is pre-emptive and largely wasted. A thread with 52 pages on what R520 might be - waste, a thread on will a product be delayed - no hard data to comment on so why bother?
Wait and see - all will be revealed.
Wait and see - all will be revealed.
Maybe it is not he R520 itself that is causing the delay. If it proves to be a great performer then it has the potential to really move the rest of the lineup down in the market. So until the other market segments are covered (Rv520?) ATI might be weary of segmenting its product range to much.
Tahir said:http://www.theinq.com/?article=21489
I know it's by Fuad, but 350'000 compared to 1'000'000 single slot solutions is indeed a "shedload" by my definition.
350.000 chipsets != 350.000 SLI setups, but whatever. I don't know exactly and you neither. All I know is that I know absolutely nobody with a SLI setup among people I know here. Which is a shedload of people
Does anyone here @B3D have an SLI setup at all?
Unknown Soldier
Veteran
geo said:
Why would they have driver issues with the R520? I mean it's on the same architecture as the R3xx/R42x as far as I understand.
The only reason where they might have a driver issue is the AMR, and if it is AMR that's keeping ATI waiting.. then I say "so what, RELEASE THE DAMN CARD ALREADY!!!" Nvidia had issues with SLI originally .. so I don't see why people would jump down ATI's throat (heavily) if AMR doesn't work 100%, as long as it works.
US
No, its a development from the R300 architecture, it its not the same - at the most basic level there is a large difference between the shader capabilities which will require new shader compiler optimiser, and that doesn't even consider the architectural differences.
Personally, so far, I would liken R420->R520 as the change from TNT->GeForce 256 or GF256->GF4 or GF4->GFFX.
