Bigus Dickus
Regular
What good is having a [ code ] [ /code ] tag if it doesn't use a fixed width font????
Here are my thoughts on the matter:
I've heard several people make statements about John Carmack and his comments about the NV30, R300, etc. JC has known a long time what was coming in the next generation of cards... especially for Nvidia because he is on their advisory board and he knows which direction they are headed long before any of us. He wouldn't "say" anything different because maybe he didn't "know" what the R300 or NV30 was going to bring to the table.
Secondly, I don't think Nvidia was "caught off guard" at all by ATI... I'm sure they've known for a long time what ATI had coming... they both had similar goals in mind (OpenGL 2.0/DX 9)... the big difference is that ATI decided to go for a .15 product launch and Nvidia waiting for a .13 launch. In fact, I'm sure they somehow get insider information about future products from their competitors somehow...
Thirdly, the NV30 has been on the drawing board for an incredibly long time. In fact, I remember reading prior to the NV10 or original Geforce launch a site publishing some thoughts from an Nvidia employee on the architectures of the future. They stated that the NV10 was going to be a big change, but the bigger changes would come in the NV20 and NV30 while saying that the latter was going to "rock" and this was back in 1999... While I'm sure they may have just had rough outlines for the NV30 architecture back then, it was being developed way back then believe it or not!
Fourthly, I think the biggest change for Nvidia in the past few years that I can forsee in any way stalling development the most was the acquisition of 3dfx... Especially in the light of the fact that over 100 new engineers were incorporated into Nvidia. Also, the NV30 was to be the first chip to employee 3dfx technology. Now, while the NV30 may have been much further along in development, after Nvidia acquired 3dfx and their engineers, ip, etc., they may have taken a step back and said, "look, we've got some breathing room here... what could we do to change this chip that would make it better without delaying it too much?"
Fifthly, as far as Nvidia not always keeping to their "6 month product cycle"... I believe many of these "delays" or extensions of older architectures were more of a marketing decision than anything... In fact, I believe the NV25 was actually being tested way back in December (or possibly earlier?... I remember reading that at Anandtech I believe) but it's introduction would only erode Geforce 3 Ti sales which were already beating the 8500 at that time. While it may be true some of these new architectures may have been a few months delayed, I believe Nvidia chose to extend product life cycles as a financial decision and introduce newer parts in a later cycle rather than releasing new architectures right away.
Finally, the NV30 isn't late... yet. People keep saying that they were beat to the market by ATI, etc. Nvidia has always stated that the NV30 would be a fall part and it's still summer folks! Nvidia (if memory serves me correctly) have never announced a high end graphics part in June/July. The official (at least paper launches) have always been in late August to early September with cards being released for public consumption a month or so later. And... if they are delayed a month or so, who cares? Geforce 1 anyone??? That card was delayed slightly and the DDR parts even longer!!!! Nvidia still made the holiday season and I see them continuing to do that for this fall as well.
I've heard several people make statements about John Carmack and his comments about the NV30, R300, etc. JC has known a long time what was coming in the next generation of cards... especially for Nvidia because he is on their advisory board and he knows which direction they are headed long before any of us. He wouldn't "say" anything different because maybe he didn't "know" what the R300 or NV30 was going to bring to the table.
Secondly, I don't think Nvidia was "caught off guard" at all by ATI... I'm sure they've known for a long time what ATI had coming... they both had similar goals in mind (OpenGL 2.0/DX 9)... the big difference is that ATI decided to go for a .15 product launch and Nvidia waiting for a .13 launch. In fact, I'm sure they somehow get insider information about future products from their competitors somehow...
Thirdly, the NV30 has been on the drawing board for an incredibly long time. In fact, I remember reading prior to the NV10 or original Geforce launch a site publishing some thoughts from an Nvidia employee on the architectures of the future. They stated that the NV10 was going to be a big change, but the bigger changes would come in the NV20 and NV30 while saying that the latter was going to "rock" and this was back in 1999... While I'm sure they may have just had rough outlines for the NV30 architecture back then, it was being developed way back then believe it or not!
Fourthly, I think the biggest change for Nvidia in the past few years that I can forsee in any way stalling development the most was the acquisition of 3dfx... Especially in the light of the fact that over 100 new engineers were incorporated into Nvidia. Also, the NV30 was to be the first chip to employee 3dfx technology. Now, while the NV30 may have been much further along in development, after Nvidia acquired 3dfx and their engineers, ip, etc., they may have taken a step back and said, "look, we've got some breathing room here... what could we do to change this chip that would make it better without delaying it too much?"
Fifthly, as far as Nvidia not always keeping to their "6 month product cycle"... I believe many of these "delays" or extensions of older architectures were more of a marketing decision than anything... In fact, I believe the NV25 was actually being tested way back in December (or possibly earlier?... I remember reading that at Anandtech I believe) but it's introduction would only erode Geforce 3 Ti sales which were already beating the 8500 at that time. While it may be true some of these new architectures may have been a few months delayed, I believe Nvidia chose to extend product life cycles as a financial decision and introduce newer parts in a later cycle rather than releasing new architectures right away.
Finally, the NV30 isn't late... yet. People keep saying that they were beat to the market by ATI, etc. Nvidia has always stated that the NV30 would be a fall part and it's still summer folks! Nvidia (if memory serves me correctly) have never announced a high end graphics part in June/July. The official (at least paper launches) have always been in late August to early September with cards being released for public consumption a month or so later. And... if they are delayed a month or so, who cares? Geforce 1 anyone??? That card was delayed slightly and the DDR parts even longer!!!! Nvidia still made the holiday season and I see them continuing to do that for this fall as well.
bdmosky said:
Well all indications now point to the NV30 being delayed a little more than a "month". There are now debates occurring about when fall actually ends, which doesn't look good for Nvidia (granted none of the people debating the point really have any inside information on Nvidia's workings). Technically if Nvidia gets the chip done before December 20th, they'll make their "fall" deadline. But usually when people speak of fall they are thinking of September/October.
Also there is one major difference from the GF1/GF2 situation: At the time Nvidia didn't have any serious competition. They might have been a little delayed, but 3DFX was EXTREMELY delayed (or their product just sucked), anyway that's just water under the bridge now. I think Nvidia is probably feeling the heat a lot more now than with GF1/GF2.
well i say if it comes out on sept 6th you should all chip in and buy me one , oh and also send me some beer since i will be 21 . As a matter of fact you can all come down to rutgars and get drunk with me ... or the following weekend you can all come to my house and go swiming ... wouldn't that be better than all the fighting we are doing ?
. As a matter of fact you can all come down to rutgars and get drunk with me ... or the following weekend you can all come to my house and go swiming ... wouldn't that be better than all the fighting we are doing ?Nagorak, my biggest point was basically to debunk many of the invalid arguments people have been throwing around... Not to say "Nvidia will be on time this fall for sure..." I was only making hypotheses based on their track record which I believe speaks more volume than any rumor of the hardware delays, etc., that people seem to be having a ball throwing around lately. An on time launch? I believe it possible. A successful launch? I say most definitely. Whether it's delayed or not, I believe the latter should never be questioned about Nvidia.
hughJ said:
Thanks. I didn't dare quote this link myself http://www.reactorcritical.com/chips/summary-full.shtml if that is what you do, but the info seems plausible enough. And it pretty much implies that nVidia prefer to introduce major updates on new processes. Value parts are something else of course. Unfortunately for our crystal balls it seems nVidia will spread over both TSMC and UMC, otherwise you might possibly make pretty accurate predictions on nVidia architectural revisions by looking at TSMCs process schedules.
Entropy
Thanks. Nvidia got where they are by making successful product launches and it'll take more than one "misstep" to screw things up. The message I was trying to convey was that even if it is delayed a bit, it will be successful launch, mainly because Nvidia knows how to market their products. If you think differently, fine, but next time please don't sling mud and choose to support your opinion instead. I really don't think that remark will attest to your ability to comment fairly either.
Evildeus said:
Well...the issue isn't that the taping out itself is going to take them 2-3 months, it's that even if they taped out tomorrow, it's still going to take 2-3 months for them to go into production. It's also safe to say that production is fairly far off since they haven't made any noise yet, and you know they'd have fouled up the R9700 launch if they could have.
bdmosky said:
What?
Its not slinging mud to say someone has a bias if they say soemthing rediculous like "A successful launch .... should never be questioned about Nvidia."
You just stated that you are unwilling to even consider the idea that nVidia will not have a succesful product launch.
You conveyed your message just fine, thanks.
BTW: I dont have an opinion yet. Unlike you, i prefer to have data before i make up my mind.
You refuse to consider an nVidia failure. Meaning you have an enormous blind spot. Its exactly this kind of blatant f@nboyism that causes problems here. Take your bias and go, please.
Well, i see. But i don't think that's important for the futur of the R300. What's important is its price, its availabality (talking of mid-september or later nowNagorak said:
), and when the NV30 will be available it's comparaison to the NV30. Till then i think that Ati have a small period of joy . Now sayng 3 months it could still be october (yeah 2 months and 25 days actualy )Evildeus said:
I will take the official word from the CEO all day over the wispers from reaktorcritical. If the NV30 would have taped out already then the CEO would have given this information, cause his "has not taped out yet" had made a lot of noise, bad press and reduced the value of the company (see the other thread)
Typedef Enum
Regular
There were other claims that the "tapeout" took place in the June timeframe as well...RC has been a pretty good site for getting non Inquirer-ish info (ahem...Det. 5's that were to be released about...oh...was it like 7 weeks ago or so?).
Just think back to all the ridiculous bickering that took place over John Carmack's words...I mean, there were all of these different interpretations of "well, but what he REALLY meant was..." or "No...By NOT saying this, he clearly is saying..."
Personally, I've been of the opinion that nVidia did, indeed, attempt to manufacture the thing...call it a sample...call it what you will...and they've been finalizing it since. To some, this means they haven't completed the "tapeout." To others, it means something else. The fact that nVidia's CEO didn't go into this level of detail doesn't really say much, one way or the other. You could make an argument either way (damned if you do...damned if you don't).
Just think back to all the ridiculous bickering that took place over John Carmack's words...I mean, there were all of these different interpretations of "well, but what he REALLY meant was..." or "No...By NOT saying this, he clearly is saying..."
Personally, I've been of the opinion that nVidia did, indeed, attempt to manufacture the thing...call it a sample...call it what you will...and they've been finalizing it since. To some, this means they haven't completed the "tapeout." To others, it means something else. The fact that nVidia's CEO didn't go into this level of detail doesn't really say much, one way or the other. You could make an argument either way (damned if you do...damned if you don't).
Don't know if this was already posted, but it seems NV30 cards are at least 6months away :
http://www.tomshardware.com/business/02q3/020802/siggraph-03.html
If tomshardware says it it must be true, cause they will do everything to let Nvidia shine bright
http://www.tomshardware.com/business/02q3/020802/siggraph-03.html
If tomshardware says it it must be true, cause they will do everything to let Nvidia shine bright
Typedef Enum said:
If you read the news from reactorcritical then they have an odd definition of tape-out. It seems that parts of the NV30 had an tape-out, but not the complete chip. So now when they tried to merge the parts of the chip in the final design it did not work and they had to alter the chipdesign to make it work. So it seems even reactorcritical says that the NV30 (the complete chip) had not taped out yet. IMHO this means we are at an preA01 step. Hopefully with this subparts being tested and optimised before they can make it with only one revision and get A02-chips and boards out in Jan/Feb 2003.
