[SemiAccurate] Nvidia Pascal over a year ahead of 14/16nm competition

[Arachnotronic]

Much ado about nothing, but an entertaining read from Charlie

 

[raghu78]

A cleverly written satirical article. Nvidia has basically used 28nm placeholder GPUs to demo Drive PX2. But its nothing new. Slightly better than woodscrews Fermi. But still Nvidia is so dominant in GPUs right now that even if they are late by 3-4 months it won't matter one bit. AMD has to do all the running while Nvidia milk Maxwell for all its worth. Nvidia can release Pascal once AMD has laid its cards on the table. That is the advantage when you have 80% GPU market share and brand leadership.

I am sure we are going to see GP104 demoed at GTC 2016 in early April.
http://www.gputechconf.com/

videocardz has already found shipping manifests and its a good bet that GP104 samples are up and running.

http://videocardz.com/58101/nvidia-gp104-and-first-polaris-gpu-supposedly-spotted-on-zauba

We are likely to see GP104 in the market by early Q3 2016 looking at the current trends.

 

[AnandThenMan]

This is because they shared the same fab. With AMD now using Samsung, who knows what will happen to timing.

Fab tech runs up against the laws of physics so we are still going to see different fabs with very similar technology and performance. I'll be very surprised if Samsung and GloFo are ahead of TSMC for any significant amount of time.

 

[MrTeal]

A cleverly written satirical article. Nvidia has basically used 28nm placeholder GPUs to demo Drive PX2. But its nothing new. Slightly better than woodscrews Fermi. But still Nvidia is so dominant in GPUs right now that even if they are late by 3-4 months it won't matter one bit. AMD has to do all the running while Nvidia milk Maxwell for all its worth. Nvidia can release Pascal once AMD has laid its cards on the table. That is the advantage when you have 80% GPU market share and brand leadership.

I am sure we are going to see GP104 demoed at GTC 2016 in early April.
http://www.gputechconf.com/

videocardz has already found shipping manifests and its a good bet that GP104 samples are up and running.

http://videocardz.com/58101/nvidia-gp104-and-first-polaris-gpu-supposedly-spotted-on-zauba

We are likely to see GP104 in the market by early Q3 2016 looking at the current trends.

I find a few things really interesting about that.
First, the list of parts there seem like early bringup supplies as opposed to a functional chip. The reference to it as a thermal head along with things like a socket base and the power and ground probes make it seem like it might even be a socketed thermal power test system.

Second, 37.5x37.5 would be really small to get four stacks of HBM2 on there. I wonder if we might be seeing a lower tier GPU with GDDR5.

 

[HurleyBird]

I believe Charlie declared that Kepler had "won" vs GCN before it launched, so it's not unprecedented for him to deliver good news about Nvidia, although it is rare.

This is sarcasm though, and to be honest I agree with him here. Nvidia is playing fast and loose again. You shouldn't lie to the press, and Nvidia deserves to get mocked for demonstrating fake products until they stop doing it.

 

[Azix]

A cleverly written satirical article. Nvidia has basically used 28nm placeholder GPUs to demo Drive PX2. But its nothing new. Slightly better than woodscrews Fermi. But still Nvidia is so dominant in GPUs right now that even if they are late by 3-4 months it won't matter one bit. AMD has to do all the running while Nvidia milk Maxwell for all its worth. Nvidia can release Pascal once AMD has laid its cards on the table. That is the advantage when you have 80% GPU market share and brand leadership.

I am sure we are going to see GP104 demoed at GTC 2016 in early April.
http://www.gputechconf.com/

videocardz has already found shipping manifests and its a good bet that GP104 samples are up and running.

http://videocardz.com/58101/nvidia-gp104-and-first-polaris-gpu-supposedly-spotted-on-zauba

We are likely to see GP104 in the market by early Q3 2016 looking at the current trends.

nvidia would lose significant market share. The main reason they gained so much in the first place was lack of new product from AMD. They could have rebranded the 290x etc at the launch of the 970/980 and got reviewers promoting their cards but nope.

if they have new faster cards and nvidia doesn't, the same thing will happen to nvidia.

 

[HurleyBird]

nvidia would lose significant market share. The main reason they gained so much in the first place was lack of new product from AMD. They could have rebranded the 290x etc at the launch of the 970/980 and got reviewers promoting their cards but nope.

if they have new faster cards and nvidia doesn't, the same thing will happen to nvidia.

One would think, but for some reason it doesn't work that way. Remember evergreen and fermi?

 

[3DVagabond]

http://forums.anandtech.com/showpost.php?p=37443058&postcount=227



This is all you need to know about Charlie Dermejian, absolute loser at life.

Instead of attacking the person, point out what's inaccurate in the article. If nothing is inaccurate, then why would you want to discredit the author?

 

[gamervivek]

Playing fast and loose with the truth is what extremetech's Joel Hruska said of it as well.

http://www.extremetech.com/gaming/2...otype-allegedly-powered-by-maxwell-not-pascal

I was replying to someone who thought AMD were brazenly lying with their polaris demo while nvidia has displayed Pascal GPUs and amusingly enough found that AT had changed their earlier observations.

However looking at the board held up by NVIDIA CEO Jen-Hsun Huang, it appears to be a sizable card with 8 GDDR5 memory packages on the front. My gut instinct is that this may be the Pascal successor to GM206 with the 8 chips forming a 128-bit memory bus in clamshell mode, but at this point that’s speculation on my part. Update Kudos to our readers on this one. The MXM modules in the picture are almost component-for-component identical to the GTX 980 MXM photo we have on file. So it is likely that these are not Pascal GPUs, and that they're merely placeholders.

http://www.anandtech.com/show/9903/nvidia-announces-drive-px-2-pascal-power-for-selfdriving-cars

Never expected it to be posted this way on this forum though.

 

[3DVagabond]

Lisa Su held up Fury X and claimed it's overclockers dream. Business as usual.

Explain what Lisa Su had to do with this article? Also, I don't believe that's what happened. Can you provide the footage or quote where she stood on stage with the card (a real card not a fake, I assume?) and uttered those words.

I don't understand why people feel they want to defend nVidia all of the time? It's absolutely amazing the faith. Even when shown in black and white some people still excuse it. It's like a cult.

 

[3DVagabond]

Fab tech runs up against the laws of physics so we are still going to see different fabs with very similar technology and performance. I'll be very surprised if Samsung and GloFo are ahead of TSMC for any significant amount of time.

Note that even with Intel's manufaturing advantage 14nm was very problematic. So, it's not a shoe in by any stretch that one, or both, of these Fabs won't struggle.

 

[PhonakV30]

Wait What is story about Driver PX2? i thought it's Pascal Chip now they say it's 980M or whatever they call?


Edit : i read this article :
http://www.extremetech.com/gaming/2...otype-allegedly-powered-by-maxwell-not-pascal

I suppose Nvidia Fooled people?

 

[ShintaiDK]

Wait What is story about Driver PX2? i thought it's Pascal Chip now they say it's 980M or whatever they called?

The product will be pascal.

 

[Exophase]

I believe Charlie declared that Kepler had "won" vs GCN before it launched, so it's not unprecedented for him to deliver good news about Nvidia, although it is rare.

He followed that up pretty quickly to say it was only because nVidia got certain games to be more optimized for their GPUs. Which was admittedly a factor.

It's lame that nVidia shows a fake product again but given it's for some pretty niche/experimental automotive application it's hard to really care. But yeah they really should stop doing this.

Charlie looks kind of silly comparing its TDP to other automotive platforms that are not meant to address the same sort of self-driving and computer vision applications...

 

[MrTeal]

Wait What is story about Driver PX2? i thought it's Pascal Chip now they say it's 980M or whatever they call?


Edit : i read this article :
http://www.extremetech.com/gaming/2...otype-allegedly-powered-by-maxwell-not-pascal

I suppose Nvidia Fooled people?

It's a little ambiguous.
https://www.youtube.com/watch?v=2ZHxGyQqQw4
JHH didn't specifically claim that they had Pascal silicon done. He just held up the board and said "This is the Drive PX 2" which isn't really untrue, and said there were two chips on the front with Pascal and two chips on the back with Pascal. It is the Drive PX 2, it just doesn't have the Pascal GPU in the MXM slots (and likely not the new Tegra, though who knows) at this point in time. A little sketchy? Maybe. Definitely not wood screw levels of BS though, or even their nicely diagrammed Project Denver die shot that was really Fermi.

 

[railven]

Charlie did have a very pro-Nvidia article out before the GTX680 launch, saying how good of a part it was at the time.

I remember that. I think eve Charlie wouldn't openly threaten his credibility by saying the GTX 680 was a bad product. After sites like Anandtech gave Nvidia the tri-fecta win (power/cost/performance) Charlie just hoped on the band wagon.

Now the days of Fermi, that guy spewed so much venom.

 

[mrmt]

Don't know how this is remotely relevant. For all investors care what matters is shipping date and actual products on the market, JHH could be holding a watermelon instead of the fake ship and it wouldn't change a bit because Pascal is not imminent.

 

[Silverforce11]

Lisa Su held up Fury X and claimed it's overclockers dream. Business as usual.

Lies. Show me evidence where Lisu Su held up Fury X and claimed it's an overclockers dream.

Unless you are referring to one of the engineers, who was talking about the water cooler and how it can handle a ton of heat output...

 

[Silverforce11]

JHH held up a board with Maxwell GPUs on it, and called it Pascal. So yeah, business as usual, I guess. *shrug*

Wait, did he called it Pascal?

 

[AnandThenMan]

After sites like Anandtech gave Nvidia the tri-fecta win (power/cost/performance) Charlie just hoped on the band wagon.

I think things unfolded in reverse of this Charlie had his patented insider info and was talking about how good Kepler was before we had reviews.

 

[railven]

I think things unfolded in reverse of this Charlie had his patented insider info and was talking about how good Kepler was before we had reviews.

Charlie had no issues slamming Fermi as not fabable. That was based on what his insiders came up.

When his insiders gave him Kepler he basically knew he couldn't go and slam it like Fermi. Nvidia essentially hit a home run and he wasn't going to risk his credibility trying to dump on NV. He joined the bandwagon.

Considering Charlie's history, him giving prompts to Nvidia isn't a sign that he is neutral, or has hints of neutrality, it means he can't say jack without looking like a fool. (I use to love reading his blog at work before I realized he was (allegedly) in AMD's pocket. But as a red ant, I loved the mud he slung at Fermi )

EDIT: Not worth the infraciton

 

[maddie]

It's a little ambiguous.
https://www.youtube.com/watch?v=2ZHxGyQqQw4
JHH didn't specifically claim that they had Pascal silicon done. He just held up the board and said "This is the Drive PX 2" which isn't really untrue, and said there were two chips on the front with Pascal and two chips on the back with Pascal. It is the Drive PX 2, it just doesn't have the Pascal GPU in the MXM slots (and likely not the new Tegra, though who knows) at this point in time. A little sketchy? Maybe. Definitely not wood screw levels of BS though, or even their nicely diagrammed Project Denver die shot that was really Fermi.

Isn't this an outright lie?

 

[PPB]

Charlie had no issues slamming Fermi as not fabable. That was based on what his insiders came up.

When his insiders gave him Kepler he basically knew he couldn't go and slam it like Fermi. Nvidia essentially hit a home run and he wasn't going to risk his credibility trying to dump on NV. He joined the bandwagon.

Considering Charlie's history, him giving prompts to Nvidia isn't a sign that he is neutral, or has hints of neutrality, it means he can't say jack without looking like a fool. (I use to love reading his blog at work before I realized he was (allegedly) in AMD's pocket. But as a red ant, I loved the mud he slung at Fermi )

EDIT: Not worth the infraciton

I guess some people need these many words to say something as simple as "Charlie gives Nvidia credit when it is due, and bash them when they lie/are wrong or just have a bad product"

Wow, I just described honest journalism! Who would have known.

 

[kagui]

Isn't this an outright lie?

In the real world yes in the world of PR, you just need to say whetever you want more times

 

[railven]

I guess some people need these many words to say something as simple as "Charlie gives Nvidia credit when it is due, and bash them when they lie/are wrong or just have a bad product"

Wow, I just described honest journalism! Who would have known.

Charlie and honest journalism? Ha!