Tesselation review by xbitlabs

[Scali]

Maybe it is because in my opinion people aren't supposed to get away with all this slander towards me and nVidia.
I merely stated the truth, and was attacked post after post. People saying I'm crazy, don't know what I'm talking about etc. And basically saying that the nVidia engineers are ignorant about things like double vias and are incompetent at designing a chip.
I think it is important to make a stand. Let truth prevail. Death to FUD.
Maybe it is because I'm one of the few people with a sense of integrity on these forums.
Most people apparently don't even see the WHY of my actions.

 

[dug777]

Maybe it is because in my opinion people aren't supposed to get away with all this slander towards me and nVidia.
I merely stated the truth, and was attacked post after post. People saying I'm crazy, don't know what I'm talking about etc. And basically saying that the nVidia engineers are ignorant about things like double vias and are incompetent at designing a chip.
I think it is important to make a stand. Let truth prevail. Death to FUD.
Maybe it is because I'm one of the few people with a sense of integrity on these forums.
Most people apparently don't even see the WHY of my actions.

Mate, don't let this place get to you, or to give you delusions of grandeur.

It's a cranky, tetchy hardware forum full of people who have considerable attachment to their video cards and video card vendors, just like car people in the Garage.

Adjust your expectations accordingly, it isn't going to change and it certainly hasn't over the time I've been a member. Whatever people fondly remember as some kind of video forum 'golden age' is like those long hot summers of your childhood that don't seem to exist anymore

 

[Scali]

It's a cranky, tetchy hardware forum full of people who have considerable attachment to their video cards and video card vendors, just like car people in the Garage.

Haha, I think you have a point there: the AnandTetch forums

 

[dug777]

Haha, I think you have a point there: the AnandTetch forums

Very appropriate

 

[Scali]

At any rate, for me the golden age was accelerated Amiga AGA, and 486DX2-66 PCs with localbus. Raise your hands if you had a Cirrus Logic 5426/8

 

[Keysplayr]

Commodore Vic20, Commodore64. First real PC was 486SX-25 with a math-coprocessor socket which I used (400.00 for the coprocessor) and made it a 486DX-50. Had a lot of fun, and learned most of my PC knowledge from that system. What a chore getting the CD-ROM (when they first came out) to work. LOL. Fun times.

 

[dug777]

At any rate, for me the golden age was accelerated Amiga AGA, and 486DX2-66 PCs with localbus. Raise your hands if you had a Cirrus Logic 5426/8

I can appreciate that generation, but I came to computers pretty late and my first '3D' accelerator was a Voodoo 2 when GF4 cards were new and top dog (CounterStrike with Glide was awesome!). First proper 'standalone' card was a GF4 440MX, and then I got a 9800 Pro...man I loved that card so much, FarCry at 1024x768 with 2xAA/16xAF and everything else cranked to the max looked so good

 

[Scali]

I started on a ZX81, not exactly the 'golden age' of video
C64 was great, but pretty much strictly 2D.
I learnt most of my 3D computer graphics engineering skills by implementing software renderers on Amiga and 386/486. They were the first to be able to handle reasonably complex 3D scenes in realtime.
Still feels a bit strange that the current generation can only get anything 3D on screen via a prefab API such as OpenGL or D3D. To them it's a black box system.

 

[Idontcare]

Heard back from NV. They utilize a mixture of single and double vias in all Fermi GPU's.

Using double-vias in circuit layout and design is as old as the technology of using metal interconnect for wiring together the circuits itself.

Naturally there is a balance to be struck between current density, lifetime requirements, yields, and everything else that goes into designing an IC when deciding where to use double-vias and where to just go with single-vias.

This may also strike some people as a surprise, except those folks who work in the industry, but it doesn't stop with double-vias either. You can have triple, quadruple, etc vias as well. I've seen layouts that had thousands of vias connecting two metal wires.

But leave it to AMD to figure out a way to make it into a marketing story, and boy did they ever.

Tapping into the sympathies of the fanbase is a tactic that is also probably as old as the concept of marketing itself, nothing new here either.

I must admit I have a new-found curiosity for AMD's marketing dept, I'm really interested to see what common industry practice they next intend to spin as innovation for their next parts. So far they've got "harvesting" and "double-via" on their innovation list...who knows, next time it could be "ZOMG we figured out how to use transistors and stuff to make electrical circuits! we are teh w00t!".

 

[Scali]

You have to want to believe.

 

[dug777]

Using double-vias in circuit layout and design is as old as the technology of using metal interconnect for wiring together the circuits itself.

Naturally there is a balance to be struck between current density, lifetime requirements, yields, and everything else that goes into designing an IC when deciding where to use double-vias and where to just go with single-vias.

This may also strike some people as a surprise, except those folks who work in the industry, but it doesn't stop with double-vias either. You can have triple, quadruple, etc vias as well. I've seen layouts that had thousands of vias connecting two metal wires.

But leave it to AMD to figure out a way to make it into a marketing story, and boy did they ever.

Tapping into the sympathies of the fanbase is a tactic that is also probably as old as the concept of marketing itself, nothing new here either.

I must admit I have a new-found curiosity for AMD's marketing dept, I'm really interested to see what common industry practice they next intend to spin as innovation for their next parts. So far they've got "harvesting" and "double-via" on their innovation list...who knows, next time it could be "ZOMG we figured out how to use transistors and stuff to make electrical circuits! we are teh w00t!".

Makes a change from their usual marketing fail in the face of nVidia's generally slick marketing.

Next they need to master vendor-specific features in games, and rebranding cards over at least three generations. That's doctorate material

Not that it even comes close to intel's dark marketing arts, bless them and I do love them dearly (hugs his Q6600)

 

[jvroig]

I must admit I have a new-found curiosity for AMD's marketing dept, I'm really interested to see what common industry practice they next intend to spin as innovation for their next parts. So far they've got "harvesting" and "double-via" on their innovation list...who knows, next time it could be "ZOMG we figured out how to use transistors and stuff to make electrical circuits! we are teh w00t!".

Just good old marketing. In fact, I'm sure they teach this in marketing textbooks. I think the case study about this still remains to be about the beer company that hired a new marketing agency, and they toured the representative of the agency around their facility, and explained to him in passing how the whole facility runs and how beer is made from grains into the delicious drink suitable for all ages and walks of life (except for ages below 18, I'm sure).

The tour was just a courtesy, but the marketing agent was amazed with all the hi-tech gear and said "This place is amazing! You should tell your customers about this!"

Naturally, the beer people said "What? But all our competitors have the same exact equipment and do practically the exact same things to produce their own beers."

To which the agent replied, "Yeah, but they haven't gotten the story out yet, this is amazing stuff and you'd be the first to tell it!"

And the rest is history. The marketing lesson was: customers loved stories, and you win them over despite having more-or-less similar/equal products as competitors (either through amazement or fascination) and the next people (your competitors) who try to do the same marketing move won't have as much impact.

At least, that's how I remember it back in school years ago, long before my master's or college degree. I was forced to take one entrepreneurship subject, they put it in all the course curricula, from multimedia arts to my own course, computer science. Not that I bitched about it back then, getting some formal schooling in business is never a bad idea, even if all I was interested in was comsci.

So it's not surprising AMD's marketing dept made that move. It's practically a textbook example of one effective marketing/PR technique.

 

[Madcatatlas]

You see, there is something that you're missing. It is an ability to let things go, or let them end. The argument is long over, yet you persist, and insist, on being the winner.
You won't, or can't, end it. Why you can not, will not (whichever) recognize this remains a mystery. It's also the source of all your ills on these forums.
Here is a test: End it by not posting about it any further. Not even in response to my post here. It's really easy and it's a first step.

I.. cant get over the notion that Keys is trying to subvert you to the dark side Scali...

whatever you do. dont belive him..pls

 

[Scali]

So it's not surprising AMD's marketing dept made that move. It's practically a textbook example of one effective marketing/PR technique.

I think the irony of it all is that the people who fell for AMD's marketing trick are pretty much the same people who are constantly on Keys' case for being an 'nVidia viral marketer' and whatnot.

PS: Keys, can I talk to you in private, off the record?

 

[Madcatatlas]

Let me add the following: being paranoid or overly sensitive does not make your day better or your points more valid.

Its just forum talk. If we dont have small conflicts, we dont have much to talk about. I dont know about you, but id hate it if everyone agreed with me..

 

[Keysplayr]

I.. cant get over the notion that Keys is trying to subvert you to the dark side Scali...

whatever you do. dont belive him..pls

You call me asking somebody to "let something go" is the equivalent of dragging them to a dark side? What the hell dude. Joking or not, that is a weird thing to post.

 

[Madcatatlas]

hehe! it just looked so out of place on a forum, what you wrote Keys! so i just followed suit. Forgive moi!

 

[Zstream]

High leakage chips generally DON'T OC well, where the heck did you get that bit of 'wisdom' from? High leakage makes chips run into a thermal wall, like P4 did, especially on 90 nm.

I may not know about the super details of what goes into the process of creating a chip but you obviously lack the understanding of overclocking. A product that, under stress, shows to create a high level of leakage generally can handle more volts and power thus creating a better overclocking part.

Obviously you have to keep the dang thing cool...

 

[thilanliyan]

^This is what I have read as well.

 

[Janooo]

Using double-vias in circuit layout and design is as old as the technology of using metal interconnect for wiring together the circuits itself.

Naturally there is a balance to be struck between current density, lifetime requirements, yields, and everything else that goes into designing an IC when deciding where to use double-vias and where to just go with single-vias.

This may also strike some people as a surprise, except those folks who work in the industry, but it doesn't stop with double-vias either. You can have triple, quadruple, etc vias as well. I've seen layouts that had thousands of vias connecting two metal wires.

But leave it to AMD to figure out a way to make it into a marketing story, and boy did they ever.

Tapping into the sympathies of the fanbase is a tactic that is also probably as old as the concept of marketing itself, nothing new here either.

I must admit I have a new-found curiosity for AMD's marketing dept, I'm really interested to see what common industry practice they next intend to spin as innovation for their next parts. So far they've got "harvesting" and "double-via" on their innovation list...who knows, next time it could be "ZOMG we figured out how to use transistors and stuff to make electrical circuits! we are teh w00t!".

If it's old stuff then what took NV so long to use it? Did they have to wait for AMD to show them the way?

 

[Idontcare]

What was the first Nvidia IC to incorporate the use of more than single-via's?

 

[Scali]

I may not know about the super details of what goes into the process of creating a chip but you obviously lack the understanding of overclocking. A product that, under stress, shows to create a high level of leakage generally can handle more volts and power thus creating a better overclocking part.

Obviously you have to keep the dang thing cool...

Leakage means you can handle more volts... how creative.
Problem is that voltage isn't the measure of overclocking.
I mean, look at how CPUs have evolved. Back in the early days they ran on 5V. These days they run at around 1V. Now, does that make them clock lower? On the contrary.

And keeping the thing cool is exactly the problem with high leakage.
If you have more leakage, you pass more current through the chip without it actually doing anything. It just heats up the chip. That's why you run into that thermal wall as I said.

 

[jvroig]

If it's old stuff then what took NV so long to use it? Did they have to wait for AMD to show them the way?

Here we go again. Will this never end?

That's because you assume NV didn't use them before.

What really happened is AMD sat down with Anand and told him the story of the double-vias. NV didn't have the opportunity to sit down with Anand, and thus was not able to tell us of their double-vias.

From there (AMD talking about the RV770 story, etc, while NV not taking the time to tell their own story about pixie dusts and unicorns), you assumed that AMD invented double-vias, which is what kicked NV in the ass. In other words, despite GF100 being a much much bigger die and a new architecture, you assumed that the problem was just "non-use of double-vias".

Can we now drop this double-vias nonsense? It has gone on for several pages, you'd think people finally got some education and got over it.

 

[Janooo]

Here we go again. Will this never end?

That's because you assume NV didn't use them before.

What really happened is AMD sat down with Anand and told him the story of the double-vias. NV didn't have the opportunity to sit down with Anand, and thus was not able to tell us of their double-vias.

From there (AMD talking about the RV770 story, etc, while NV not taking the time to tell their own story about pixie dusts and unicorns), you assumed that AMD invented double-vias, which is what kicked NV in the ass. In other words, despite GF100 being a much much bigger die and a new architecture, you assumed that the problem was just "non-use of double-vias".

Can we now drop this double-vias nonsense? It has gone on for several pages, you'd think people finally got some education and got over it.

It was one of the problems.
What you guys don't realize is that double vias cost space. That's one thing that designers try to avoid.
Creative use of doubles in areas were singles were enough before made the trick. But whatever...

 

[jvroig]

What you guys don't realize is that double vias cost space. That's one thing that designers try to avoid.

What makes you think we don't realize that? Space is an obvious cost of double-vias. And even if somehow somebody didn't pick up that double-vias would need more space than single vias (or that triple-vias wouldn't take more space than double-vias), I'm sure Anand mentioned it in the article as well, probably for people who think the laws of physics can be broken routinely when GPU design and fabrication is the issue.


What I'm telling you is that this is wrong:

If it's old stuff then what took NV so long to use it? Did they have to wait for AMD to show them the way?

That is wrong because when you put it that way, you believe and imply that NV did not use double-vias until AMD did.

We've been over that story in this thread, and Keys came back to say that nVidia did respond about the double-vias issue, saying they did use double-vias. Implying otherwise is trying to resurrect the argument that is over.

Cliffs:
* nVidia did use double-vias, just like AMD.
* AMD did not invent double-vias or pioneer their use.
* All AMD did was get the story out first.
* In fact, duplicating vias (doube, triple, etc) is an old practice.