NVIDIA Fermi: Architecture discussion

It's unfortunate, Charlie, that you would think that you could drop your [removed colour] where real people are sitting.

I *personally* worked with the Jaguar and you are talking out of your [removed colour].

It had a few problems but you are massively overplaying them.

You are *way* out of your depth here.

I suggest you shut [removed colour] up before you dig yourself into a hole you cannot get out of.

OK, shall we talk about the inability for the serial port to do 8-N-1 like it was promised, that was the one that bit me pretty hard. I could go dig my dev docs out of the basement and look for the others, but it has been 10+ years since I last looked at them, so I don't remember the list.

If you worked on making the jag fine, it was a cool piece of hardware, and I like it a lot. Bug free it was not. Neither is any console I have seen the dev kit for.

The problem I am referring to was almost fatal to my project. FWIW, I wrote a mostly full blown TCP/IP stack a somewhat HTML2.0 compliant browser, and a few other utilities for the Jag. I use mostly and somewhat because Atari folded before I could complete it.

They made a breakout box for my project with serial, parallel, mouse and keyboard ports, and only when I was in the middle of doing it did they say that the serial ports were unfixably fuxored, something that they promised would be patched/worked around before I was done. Trust me when I say that hosed my world. A brief look through newsgroup searches can't find references to it, but they are there if you look.

What did you do on the platform, or are you talking out of your (likely to be removed by admins)?

-Charlie
 
http://www.xbitlabs.com/news/video/display/20091105234423_Nvidia_to_Ramp_Up_Fermi_Graphics_Cards_Only_in_2010.html

Nvidia Corp.’s chief executive officer said during quarterly conference call with financial analysts that the company would only ramp up production of its next-generation flagship Fermi-G300 graphics processing unit in the company’s first quarter of fiscal year 2011, which means January – April timeframe.

“Next year it is going to be an interesting first quarter because, in fact, we will need more wafers than ever in Q1. The reason for that is because – and I mean more 40nm wafers than ever in Q1 –we are […] fully ramping Fermi for three different product lines: GeForce, Quadro and Tesla,” said Jen-Hsun Huang, chief executive officer of Nvidia, in the conference call with financial analysts.

Nvidia’s first quarter of fiscal year 2011 begins on the 26th of January and ends on the 26th of April, 2010. Therefore, the claim by Mr. Huang means that Nvidia will only be able to start commercial shipments its next-generation graphics cards based on Fermi architecture in calendar 2010, not in calendar 2009, as the company promised originally.

Besides the next-generation high-end graphics processor, which is usually called Fermi (also known as NV60, G300, GT300), Nvidia will also ramp up manufacturing of its next-generation core-logic for Intel Core 2 processors and other products.
article continues...
 
OK, shall we talk about the inability for the serial port to do 8-N-1 like it was promised, that was the one that bit me pretty hard. I could go dig my dev docs out of the basement and look for the others, but it has been 10+ years since I last looked at them, so I don't remember the list.

Since when does a broken serial port render an entire console buggy as hell? Sure it's there and some peripherals use it, but it's far from the core functionality of a console. That's like declaring an entire GPU buggy as hell because an integrated I2C fan controller is busted.
 
http://www.xbitlabs.com/news/video/display/20091105234423_Nvidia_to_Ramp_Up_Fermi_Graphics_Cards_Only_in_2010.html

"Quote:
Nvidia Corp.’s chief executive officer said during quarterly conference call with financial analysts that the company would only ramp up production of its next-generation flagship Fermi-G300 graphics processing unit in the company’s first quarter of fiscal year 2011, which means January – April timeframe.

“Next year it is going to be an interesting first quarter because, in fact, we will need more wafers than ever in Q1. The reason for that is because – and I mean more 40nm wafers than ever in Q1 –we are […] fully ramping Fermi for three different product lines: GeForce, Quadro and Tesla,” said Jen-Hsun Huang, chief executive officer of Nvidia, in the conference call with financial analysts.

Nvidia’s first quarter of fiscal year 2011 begins on the 26th of January and ends on the 26th of April, 2010. Therefore, the claim by Mr. Huang means that Nvidia will only be able to start commercial shipments its next-generation graphics cards based on Fermi architecture in calendar 2010, not in calendar 2009, as the company promised originally.

Besides the next-generation high-end graphics processor, which is usually called Fermi (also known as NV60, G300, GT300), Nvidia will also ramp up manufacturing of its next-generation core-logic for Intel Core 2 processors and other products.
"

article continues...

hmm so theres NOTHING New in the shops Werth buying over the christmas season then! thats Odd.

it seems not many non gaming people at least are going to be buying any newest AMD/ATI this year, as their OpenCL codebase just doesnt work on their gfx, and they still dont intent releasing any UVD ASIC documentation for 3rd party devs to finally use in simple video decode/Encode apps etc...

so whats new on the table for intel likely getting your cash this cristmas season!
 
Last edited by a moderator:
Since when does a broken serial port render an entire console buggy as hell? Sure it's there and some peripherals use it, but it's far from the core functionality of a console. That's like declaring an entire GPU buggy as hell because an integrated I2C fan controller is busted.

It doesn't, there were a lot of others, that is just the one I remember because it bit me.

-Charlie
 
it seems not many non gaming people at least are going to be buying any newest AMD/ATI this year, as their OpenCL codebase just doesnt work on their gfx
If you want the best performance per dollars or Watt you will probably try to get it to run just to be able to run the numbers. Of course GPGPU is generally more about generating papers than about generating something useful :p
and they still dont intent releasing any UVD ASIC documentation for 3rd party devs to finally use in simple video decode/Encode apps etc...
Which is very disappointing for the couple hundred people in release groups who would use it ... but really not a big issue in the grand scheme of things.
 

Now you pray that they'll get something back they can actually start production on come January, add 2-3 months and we'll see Fermi fight a RV870 refresh.
What Fuad fails to mention is that the chip in the GTX295 was ready for more than one Quarter.
 
hmm so theres NOTHING new in the shops worth buying over the christmas season then! thats Odd.

Well that depends. If you're trying to get your game on there's AMD's stuff if you can find it in stock. If you're all about OpenCL and computing then I'm not sure why Christmas season is relevant. There are a lot of cheap cards out there now that you could pick up to get started on the compute side of things, you don't need Fermi for that.
 
Back
Top