NVIDIA Fermi: Architecture discussion

[MrBelmontvedere]

Wow -- what kind of resize filter did you use for that?

I would also like to know this please :smile:

 

[Groo The Wanderer]

From a customer POV, it doesn't make a difference: you couldn't buy it either.

These violent reactions against paper launches or product pre-announcements are really strange and seem to be unique to the GPU world.

Actually, I think NVidia stoked those flames, or at least lit them when they promised hard launches a few years ago. Since ATI was floundering badly at the time, it seemed like a smart thing to do, unless you don't see far enough in the future to realize that this industry is cyclical.

They made it an issue, so now it is.

As for the buying either, I think it depends on how the launch is presented. If it is a technology demo, or something else, that is fine. If a CEO with an ego problem led by PR drones that don't have a clue gets up on stage saying "This puppy is Fermi", when it clearly isn't, and they haven't even taped out the final spin.... that is different. When a company is pushing their tame journalists to promote heavily that the launch is imminent, that adds another dimension of slime.

If you hold something up and say, "this isn't ready, and it will be by Day X", and it is, or is quite early, hey, no problem.

-Charlie

 

[Malo]

I really think Nvidia's actions in other aspects of their business lately have left a bad taste in a lot of people's mouths and so some are reacting in more negative ways to anything that Nvidia is releasing in regards to Fermi.

 

[Groo The Wanderer]

There is a huge difference between Intel showing Larrabee and talking about it months or even years in advance, vs Nvidia talking about Fermi. Larrabee is not Intel's core business. And if Larrabee is never released, or never makes any noise in the marketplace, at least in the mid term Intel will hardly be affected.

Fermi is Nvidia's core business and vital to their future. So of course there is going to be great discussion when it turns into a fiasco. Let's be honest here, we don't have a launch date commitment from Nvidia because they simply don't know when it will happen. "Fully ramping Q1-2010" etc. doesn't exactly clear up the situation.

To the best of my knowledge, Intel has never said that Larrabee would be productized, has never promised a release date, has never stated target markets, and has never mentioned much. Publicly.

They have given a public demo, have talked about the ISA, programming model for the family etc etc. Again, no dates for hardware.

All the dates have been leaks or information from mostly Fudo and I, or at least the one that are in the ballpark, there are others that have MIMD/512b memory levels of believability too, but I discount them.

-Charlie

 

[Groo The Wanderer]

You're talking about presentation tactics, which is not my point. But if you insist: Nvidia shows a mock-up enclosure and Intel shows a wafer full of Larrabee's. Both were not showing a working prototype, but both were presenting concrete details of the chip architecture. Practical difference for everybody involved: nihil (* exept that one will supposedly launch much earlier than the other.) Yet nobody complained about Intel paper launching Larrabee...

If Nvidia had decided to show a wafer full of dies instead, do you really believe people would not have complained about a paper launch?

Anyway, I don't care about presentation tactics. The main question is: why is it considered evil for GPU vendors to disclose technical details of their next architecture when it's not for anybody else?

No, Intel showed full Larrabee cards, and had the demo running live, you could trace the cables to the monitor. I did.

There were also some that were shown in private in much more details.

-Charlie

 

[Arty]

Maybe. I don't know. I do not know the reason for everything Nvidia does. Unfortunately everyone seems to think I do. If something I say does come from Nvidia. I am pretty clear on outlining that. TBH I am surprised at the reactions here. Most forums ((with the exception hardocp/B3D that I have read. Have mostly been positive. Most people are just happy to see a picture of the card.

Chris, again with the exception of the nvfansites I don't see the reaction as anything but positive and it hasn't created a stir either.

Whatever Nvidia thought as a reason to release the picture, wasn't sound at all. They surely must have processed that after the fake card fiasco they were subject to more scrutiny. And what do they do? Release a low quality photo that doesn't answer much but raises more questions.

 

[trinibwoy]

In which case, it's mission accomplished if their goal was just to keep people talking about Fermi.

 

[MrBelmontvedere]

perhaps they were just throwing gamers a bone by releasing something relating to a new generation of Geforce. if you read forums related to PC gaming, you hear much grumbling that Nvidia has spent so much time talking about the GPGPU applications of Fermi but had been relatively silent about any new Geforce.

 

[SiliconAbyss]

Wow -- what kind of resize filter did you use for that?

Alien Skin Blowup (PS plugin) plus some hand tweaking. Great plugin if you have a large number of images that need to be upsized.

In which case, it's mission accomplished if their goal was just to keep people talking about Fermi.

I highly doubt people are going to forget about Fermi and stop talking about it.

 

[Kombatant]

In which case, it's mission accomplished if their goal was just to keep people talking about Fermi.

Wow, someone actually knows how PR works

 

[digitalwanderer]

I got the impression from them that they were surprised this is even an issue.

Thank you, thank you.

digisan bows and looks way to pleased with himself

 

[DuckThor Evil]

digisan

Heh why do you call yourself digisan? Japanese people never use san when they talk about themselves. or is it something else

Sorry for the off topic.

 

[Mize]

He's not referring to himself - that's the narrator in his head.

 

[digitalwanderer]

Heh why do you call yourself digisan?

Just a shorter version I use for IRC and such, a little friendlier on the fingers. I also go by "digiwand" at sites that only allow 8 character names.

He's not referring to himself - that's the narrator in his head.

My soundtrack is even better, you should hear it!

 

[Arty]

In which case, it's mission accomplished if their goal was just to keep people talking about Fermi.

Okay, negative press is still press I guess. (No matter how small it is)

 

[Ailuros]

Wow, someone actually knows how PR works

I have severe doubts you know though; you're far too objective to have a fair "understanding" of PR. You may pay me later the agreed sum to make you look good

 

[Razor1]

Wow, someone actually knows how PR works

Depends on how you want or need the PR, bad adverstising at this point when the product is close, is not good. But its a select few on each side, so its nothing that wasn't discussed an analyized before hand. Good vs. Bad PR, depends on what the target is and if that target is benefited by good or bad PR. If it was a rapper, bad PR is great (ok being silly, not everyone, but you get the idea)

 

[silent_guy]

Actually, I think NVidia stoked those flames, or at least lit them when they promised hard launches a few years ago.

But this was a reaction towards the negative press they got after the paper launch NV30. And I believe at some point, after the hard launch of one of their products (don't remember which one), they stated that they wanted to back away from hard launches because it's so hard to coordinate from a logistics pov.

As for the buying either, I think it depends on how the launch is presented. If it is a technology demo, or something else, that is fine.

That's exactly how I understood the Fermi launch in September. The theatricals that come with it are good for entertainment. They don't bother me.

 

[Ailuros]

Depends on how you want or need the PR, bad adverstising at this point when the product is close, is not good. But its a select few on each side, so its nothing that wasn't discussed an analyized before hand. Good vs. Bad PR, depends on what the target is and if that target is benefited by good or bad PR. If it was a rapper, bad PR is great (ok being silly, not everyone, but you get the idea)

The problem in terms of quality starts exactly when you adjust to stereotypes. It's far easier to raise quality when you de-couple from the latter and that's irrelevant to the field of operation.

 

[seahawk]

Is it still PR or is it FUD?

FUD is aimed to damge the competition and that is what they are trying to do. Keep people from buying the ATI cards and have them waiting an unknown time for unrealsed, untested and unfinished NV products.