Jen Hsun Huang on Fermi's Failure, in the Development Stages

[CDdude55]

Retail chips are NOT called Fermi, to me. Fermi was the chip's name while in development, but that chip, that is is retail cards, is now known as GF100, right?

Fermi has come and gone, to me, so to look back at what happened during that time is some of the best "news" there is...that's real journalism, and I love it!

The architecture is Fermi, i refer to the cards themselves and the retail GPU's themselves as GTX 4xx. But that's just me.

Great video though.

 

[cadaveca]

The architecture is Fermi, i refer to the cards themselves and the retail GPU's themselves as GTX 4xx. But that's just me.

Yeah. To me it's the same as calling the 6-series cards from AMD, "Northern Islands". Or whatever they are. Southern Island, etc...

What even more interesting is that Huang seems very confident that the next gen won't have these same issues...it almost seems liek they've taped out, and maybe are not that far behind AMD in releasing new chips...maybe they've been relatively quiet about the new gen so they can crush AMD's next chips.


I mean, I consider the 58xx-series reference retail cards a failure, 100%. It's not like I hype AMD, and bash nV...I bash AMD far more.

 

[wolf]

The architecture is Fermi

Seems a lot of people don't realise or choose to ignore this while enjoying paying out GTX400 series cards.

statements like "Fermi is hot and power hungry", however vague, does somewhat apply to GF100, but hardly GF104/106.

 

[crazyeyesreaper]

it still does apply to the 460 bro a 460 max overclocked consumes more power then a 5870 and is still slower and at stock it uses equal or greater power then a faster 5850, but i digress the differences are not enough to use the blanket statement hot and power hungry. What really aggrevates me is seeing ppl say get a 460 get a 460 when they already have a 285 its not really a real upgrade its pretty much a side grade with DX11 well unless you buy 2

 

[cadaveca]

Yeah, but that's not important here, and is off topic. That's what got CDdude going in the first place.

Retail products have no place in this thread. This is about the Fermi dev process. I didn't say Fermi was a failure...Fermi failed(and the teams behind it as well), in a specific scenario, early in development. THAT's what this thread is about.

Please stop talking about retail products here, they are unimportant in the context of this discussion. Watch the video, and comment on that, please.

 

[crazyeyesreaper]

in which case nvidia fumbled and lost the game so to speak but just like everything else in sports or in technology theres always another round to come to make up for past mistakes.

 

[cadaveca]

in which case nvidia fumbled and lost the game so to speak but just like everything else in sports or in technology theres always another round to come to make up for past mistakes.

Yeah, exactly, and that's why I put "In the Past Tense" in the title, but seems my choice of wording didn't get the point across...it only makes sense if you watch the video on the link I provided. Of course, I watched the video before posting the thread...seems many commenting did not, by the early posts here.

 

[crazyeyesreaper]

oh i watched and i also agree its good to see some actual honesty from the not so jolly green giant

i honestly think come 28nm Nvidia will have the edge but for now im glad AMD is on top they basically needed ati gpu sales to hold the line so to speak in order to keep development going and it seems its finally starting to pay off. As an enthusiast i want proper competition not half baked scandal. so i suspect sandybridge vs bulldozer and 28nm geforce vs radeon will be very interesting both to see how the companies have learned from past mistakes and to see whats new that gets brought to the table. Im also interested in seeing how gf100 arch matures and evolves

 

[mdsx1950]

Its just the GF100 that sucks. Rest are good.

 

[cadaveca]

oh i watched and i also agree its good to see some actual honesty from the not so jolly green giant

i honestly think come 28nm Nvidia will have the edge but for now im glad AMD is on top they basically needed ati gpu sales to hold the line so to speak in order to keep development going and it seems its finally starting to pay off. As an enthusiast i want proper competition not half baked scandal. so i suspect sandybridge vs bulldozer and 28nm geforce vs radeon will be very interesting both to see how the companies have learned from past mistakes and to see whats new that gets brought to the table. Im also interested in seeing how gf100 arch matures and evolves

Yeah, and I think I need a title change. MY own fault, there, I guess. Let's see if I can't get it changed to "Jen Hsun Huang on Fermi's Failure, in the Development Stages. "

 

[wolf]

I think its fantastic he actually went ahead and talked about what really went wrong and caused the delays. It's a little bit late now to gloss over the delays in development, but it's food for thought and gives the market some answers as to why it was so late. kinda helps dissolve the anger and bad feelings that surrounded the delayed product.

Maybe a better title would be something simple like "Jen Hsun Huang sheds light on the Fermi delay", having failure in the title, while correct for the scenario, just begs for arguments IMO.

 

[cadaveca]

I think its fantastic he actually went ahead and talked about what really went wrong and caused the delays. It's a little bit late now to gloss over the delays in development, but it's food for thought and gives the market some answers as to why it was so late. kinda helps dissolve the anger and bad feelings that aurrounded the delayed product.

Maybe a better title would be something simple like "Jen Hsun Huang sheds light on the Fermi delay", having failure in the title, while correct for the scenario, just begs for arguments IMO.

Yeah, my sentiments exactly. In the end, it makes alot of sense, to me, for nV to be talknig about it at this point. They've moved on to bigger and better, and part of how you do that is by correcting past mistakes.


I PM'ed a mod to try to get the title changed.

 

[Sasqui]

Yeah, my sentiments exactly. In the end, it makes alot of sense, to me, for nV to be talknig about it at this point. They've moved on to bigger and better, and part of how you do that is by correcting past mistakes.


I PM'ed a mod to try to get the title changed.

Interesting read, really two points come out:

1. They wanted to do something that was technologically risky
2. "Two teams and no boss" - that speaks volumes.

...I think you can change the title by editing your first post and then "Go to advanced"

 

[cadaveca]

Interesting read, really two points come out:

1. They wanted to do something that was technologically risky
2. "Two teams and no boss" - that speaks volumes.

...I think you can change the title by editing your first post and then "Go to advanced"

Yeah. It's a very eye-opening "interview". I really like this version of Mr. nV.


Thread titles can only be changed within a limited time after your first post it....it's fixed now though.