[SA] News of Nvidia’s Pascal tapeout and silicon is important

[3DVagabond]

Charlie writes with a persistent and overt bias, in particular, he will find a pro-AMD or anti-Nvidia angle for almost all his articles. The perceived validity of what he writes is a reflection of this.

There is barely any solid information regarding the current state of Pascal, so Charlie's criticism at the start of the piece is fair. However, Charlie's own conjecture is based on the same incomplete information; there is nothing to refute.

It's a misconception that Charlie is pro AMD. He simply hates nVidia. He's written plenty of negative AMD articles too though.

 

[3DVagabond]

This is definitely true, I expect to see big Pascal coming out very early with regards to the rest of the generation. What I don't expect is consumer big Pascal for a while, when they're working to get yields in order and their HPC business needs them. What's interesting is that they didn't have Pascal to put on that demo but still felt the need to say that it was rather than that it would be.

Their rival was showing working silicon. They didn't want to get up there and say they were still using last gen silicon.

 

[xthetenth]

Their rival was showing working silicon. They didn't want to get up there and say they were still using last gen silicon.

Right, I'm saying that they felt sufficient pressure to claim working silicon, but they didn't actually have working silicon. The question is whether that means no MXM module or that they're way behind or what.

 

[3DVagabond]

Right, I'm saying that they felt sufficient pressure to claim working silicon, but they didn't actually have working silicon. The question is whether that means no MXM module or that they're way behind or what.

We can only use deduction. If AMD shows working Polaris and nVidia is still showing Maxwell does that mean they are behind at this point? More than likely it does. We won't know for certain, but it makes it look that way.

 

[Silverforce11]

AMD will probably be a year+ behind on levels of DX12/Vulcan driver optimisation again on the next gen of gpu's.

What are you trying to say, AMD will be behind in DX12? lol

They help created the DX12/Vulkan standard. It's based on their own Mantle tech.

https://www.reddit.com/r/AdvancedMi...ide_games_made_a_post_discussing_dx12/cul9auq

Oxide effectively summarized my thoughts on the matter. NVIDIA claims "full support" for DX12, but conveniently ignores that Maxwell is utterly incapable of performing asynchronous compute without heavy reliance on slow context switching.

GCN has supported async shading since its inception, and it did so because we hoped and expected that gaming would lean into these workloads heavily. Mantle, Vulkan and DX12 all do. The consoles do (with gusto). PC games are chock full of compute-driven effects.

We may see Volta being great at DX12/Vulkan eventually.

 

[railven]

We may see Volta being great at DX12/Vulkan eventually.

This is why I can't take some of your opinions/predictions seriously. You act like NV's Pascal can't change things. You wrote it off before even seeing anything of it.

I understand you want AMD to end up ahead, but you're now acting like NV is going sit by and watch AMD devour them.

 

[Arachnotronic]

This is why I can't take some of your opinions/predictions seriously. You act like NV's Pascal can't change things. You wrote it off before even seeing anything of it.

I understand you want AMD to end up ahead, but you're now acting like NV is going sit by and watch AMD devour them.

Couldn't have said it better myself. There are some people just hoping and praying that NVIDIA will beef it to give AMD an edge.

 

[Arachnotronic]

The manifest is not about the Pascal dies and the roadmap is not legit, making a conclusion of an April launch based on those two sources will only make it false.

This is a very reasonable and IMO correct conclusion :thumbsup:

Roadmap looks homemade and the Zauba manifests do not point to actual GPU silicon. I have seen actual silicon shipments on Zauba and those ain't it.

 

[nvgpu]

AMD doesn't even support Feature Level 12_1, just like SM3.0 games didn't run on any AMD SM2.0 GPUs, all those current AMD GPUs will be outdated and won't run FL12_1 games at all and will have to be replaced in the end.

Async compute is pretty much irrelevant, Nvidia performs just fine without it as seen from the intentionally biased and poorly coded "async compute" games. Even Skylake does not support async compute and pretty much embarasses AMD with full FL 12_1 support.

 

[Silverforce11]

This is why I can't take some of your opinions/predictions seriously. You act like NV's Pascal can't change things. You wrote it off before even seeing anything of it.

What, like my claim that Maxwell could not do Async Compute from when it was released? Way before anyone found the truth and even before AnandTech's async article which they were fed FUD from NV PR? They still haven't fixed their article or offered an explanation why they ate up the PR lies.

I was attacked back then by many of you for making the correct conclusion about Maxwell's compute & graphics engine.

I'm not dismissing Pascal. I said Volta may do DX12 really well. That's entirely different than saying Pascal will be gimped at it which is what you are implying I said.

My post was in response to the ridiculous statement from @Dribble that AMD not be optimized for DX12/Vulkan. It's a little history lesson.

ps. We don't know much about Pascal besides its focus on Compute and mix-mode compute operations. It would be very silly to suggest much on it this early without any concrete leaks.

 

[Arachnotronic]

What are you trying to say, AMD will be behind in DX12? lol

They help created the DX12/Vulkan standard. It's based on their own Mantle tech.

https://www.reddit.com/r/AdvancedMi...ide_games_made_a_post_discussing_dx12/cul9auq



We may see Volta being great at DX12/Vulkan eventually.

Chess game? LOL! This is a company that is in very poor financial health and has lost substantial market segment share, and now they spin not having the right products in the market at the right time (one of the reasons they need to drop prices so much) as a "chess game" for "future-proofness."

This is pure, unadulterated marketing spin coming from, unsurprisingly, a person whose job it is to produce such marketing spin.

 

[3DVagabond]

Chess game? LOL! This is a company that is in very poor financial health and has lost substantial market segment share, and now they spin not having the right products in the market at the right time (one of the reasons they need to drop prices so much) as a "chess game" for "future-proofness."

This is pure, unadulterated marketing spin coming from, unsurprisingly, a person whose job it is to produce such marketing spin.

Well, you explain how their uarch was designed for the next gen API's before they were even announced then.

 

[Silverforce11]

Chess game? LOL! This is a company that is in very poor financial health and has lost substantial market segment share, and now they spin not having the right products in the market at the right time (one of the reasons they need to drop prices so much) as a "chess game" for "future-proofness."

This is pure, unadulterated marketing spin coming from, unsurprisingly, a person whose job it is to produce such marketing spin.

You don't think designing GCN with so much future-proofing is a strategic decision?

Designing next-gen API also cost them R&D $$, it was a risk to ensure it would be used in all the next API standards. That's pure dumb luck and not at all strategic according to you?

Listen to yourself. It's a bit ridiculous to now bash on AMD for powering next-gen API and having an uarch that's future-proofed them.. back then the likes of you were dissing AMD's Mantle and saying how it's useless. There were epic Mantle threads here. Guess what? Soon you all will be running Mantle's code underneath the DX12/Vulkan logo. -_-

At least be man enough to give credit where it is due.

You remind me of this: http://www.geforce.com/whats-new/gu...rmance-guide#rise-of-the-tomb-raider-purehair

PureHair** is Crystal Dynamics and Square Enix's hair rendering technology, which like our own HairWorks technology adds tens of thousands of hair strands to a character model.

** TressFX3.0, developed by AMD.

 

[railven]

You don't think designing GCN with so much future-proofing is a strategic decision?

Designing next-gen API also cost them R&D $$, it was a risk to ensure it would be used in all the next API standards. That's pure dumb luck and not at all strategic according to you?

Listen to yourself. It's a bit ridiculous to now bash on AMD for powering next-gen API and having an uarch that's future-proofed them.. back then the likes of you were dissing AMD's Mantle and saying how it's useless. There were epic Mantle threads here. Guess what? Soon you all will be running Mantle's code underneath the DX12/Vulkan logo. -_-

At least be man enough to give credit where it is due.

You remind me of this: http://www.geforce.com/whats-new/gu...rmance-guide#rise-of-the-tomb-raider-purehair



** TressFX3.0, developed by AMD.

Here we go again. Where has that gamble paid off? It's now 4 years since GCN launched and it's still lagging behind Nvidia. Future proofing? Tahiti (the first GCN card) is already obsolete. Even with the "future proofing GCN" uarch.

And how much royalties does AMD get from DX12? They future proofed themselves out of the market.

Where is this giant DX12 revolution? AtenRa basically confirmed Tomb Raider 2 would be DX12. It wasn't.

I just don't get how you can still put so much faith into a company that has essentially boiled away their credibility. Wait for Windows 8. Wait for Windows 10. Wait for DX12. You guys are still waiting.

 

[railven]

What, like my claim that Maxwell could not do Async Compute from when it was released? Way before anyone found the truth and even before AnandTech's async article which they were fed FUD from NV PR? They still haven't fixed their article or offered an explanation why they ate up the PR lies.

I was attacked back then by many of you for making the correct conclusion about Maxwell's compute & graphics engine.

And, yet, NV is still leading when using DX12 in the games/demos that have been shown. Again, it's taking DX12 to get AMD to parity of where NV was with DX11.

I'm not dismissing Pascal. I said Volta may do DX12 really well. That's entirely different than saying Pascal will be gimped at it which is what you are implying I said.

My post was in response to the ridiculous statement from @Dribble that AMD not be optimized for DX12/Vulkan. It's a little history lesson.

ps. We don't know much about Pascal besides its focus on Compute and mix-mode compute operations. It would be very silly to suggest much on it this early without any concrete leaks.

Are you kidding me? You completely ignored Pascal.

We may see Volta being great at DX12/Vulkan eventually.

You are implying that you aren't even confident that Volta would be "great at DX12/Vulkan." If you aren't confident Volta would be even "great" at it, you basically ruled out Pascal.

Again, for the sake of some folk around here, I hope AMD hits it out of the park. If they crash and burn, again, I get the feeling the console sub-section is going to have a lot of new faces.

 

[Silverforce11]

You are implying that you aren't even confident that Volta would be "great at DX12/Vulkan." If you aren't confident Volta would be even "great" at it, you basically ruled out Pascal.

Are you confident with specifics about Pascal, let alone Volta which is even further down the road?

 

[railven]

Are you confident with specifics about Pascal, let alone Volta which is even further down the road?

Not as confident as this poster:

We may see Volta being great at DX12/Vulkan eventually.

 

[Silverforce11]

Here we go again. Where has that gamble paid off? It's now 4 years since GCN launched and it's still lagging behind Nvidia. Future proofing? Tahiti (the first GCN card) is already obsolete. Even with the "future proofing GCN" uarch.

Tahiti is obsolete?

Someone forgot to tell Tahiti it's obsolete.

Tahiti why you so good?

Obsolete? Maybe for Kepler.

Oh, is that a Tahiti on the heels of the 970?!

Tahiti, why are you so fast? Poor 680/770...

Oi, Tahiti! Why are you matching Kepler Titan, you're obsolete!

Woah slow down Tahiti! You're supposed to be obsolete, why you destroying 680/770 and matching Kepler 780Ti!!

Looks to me like GCN, even the early Tahiti is excelling in modern games.

 

[nvgpu]

Tahiti/Pitcairn/Cape Verde doesn't even support variable refresh rate for gaming, you have far worse experience with stuttering than Kepler GPUs.

 

[Silverforce11]

Not as confident as this poster:

That's why I said it MAY be good at DX12. MAY is a language you use when you don't know or lack confidence.

Are you say it WILL be good at DX12? How can you say that with zero evidence?

 

[railven]

Tahiti is obsolete?

Good to know in 2016, you'll still be recommending Tahiti to buyers. AMD can count on you. :thumbsup:

 

[Silverforce11]

Good to know in 2016, you'll still be recommending Tahiti to buyers. AMD can count on you. :thumbsup:

280X is still a vastly better buy at that price segment for gaming than the gimped 960 Maxwell, so much for obsolete when new tech is so much slower. :thumbsup:

Seriously look at those charts carefully, Tahiti is matching Titan/780Ti and matching the 970 or on it's heels.

People still recommend the 970, right? It's still a good buy?

The gimping of Kepler is real.

 

[railven]

That's why I said it MAY be good at DX12. MAY is a language you use when you don't know or lack confidence.

Are you say it WILL be good at DX12? How can you say that with zero evidence?

I didn't say anything. I've yet to make a prediction/speculative post of either Pascal or Volta. Why would I?

But again, you're now debating the validity of Volta being "great at DX12/Vulkan eventually" completely skipping over Pascal. Do you not see it?

'I don't even have to worry about Pascal, we may see Volta be great for DX12/Vulkan eventually.'

 

[railven]

280X is still a vastly better buy at that price segment for gaming than the gimped 960 Maxwell, so much for obsolete when new tech is so much slower. :thumbsup:

It better be faster, it's using almost twice the power! D:

I also get the feeling based on your collection of images you are some how inferring I don't think Kepler is obsolete. It sure is.

 

[Silverforce11]

It better be faster, it's using almost twice the power! D:

You know the really funny thing, a gaming system with 960 and 280X, total system power will be 100W more for the 280X v 960 GPU power usage delta.

So it may be 200W vs 300W... and when we look at performance in modern games, Tahiti is killing the 960 by a huge margin.

The gaming rig is just as power efficient!

Old, but not obsolete! haha

A year from now, Hawaii, 390/X is going to routinely beat Maxwell 970/980 in modern games and will be matching or on the heels of the 980Ti. Just like Tahiti is doing to Kepler for the past year. When the time comes, at least we can agree, the gimping of Maxwell is real.