AMD's Future Chips & SoC's: News, Info & Rumours.

[juanrga]

Seriously though, I suspect a lot of the initial sales were BD/PD users upgrading to the new platform. We're now seeing the first indications of long-term platform demand, and from what I'm seeing? AMD really isn't putting that much of a dent in Intels overall market share.

This is what I said before (bolded the relevant part):

Numbers show that a minority of people are moving from Intel to AMD. Most sales for AMD products are coming from AMD people upgrading their old Phenom II and Bulldozer/Piledriver.

For next months, AMD predicts a double-digit reduction in sales/revenue for Q4.

AMD marketshare gains in PC market are in the 3% range with Zen. Much lower in the server space.

Iris Pro used an expensive memory solution to provide bandwidth to the iGPU. This new chip uses a cheaper HBM2 solution to feed the GPU.

What expensive memory solution?

eDRAM

 

[juanrga]

Raja Koduri is officially out AMD.

First Rori Read. Second Keller. Third Koduri. I guess Papermaster will be the next on leaving AMD.

 

[gamerk316]

Iris Pro used an expensive memory solution to provide bandwidth to the iGPU. This new chip uses a cheaper HBM2 solution to feed the GPU.

What expensive memory solution?

eDRAM. It's expensive, power hungry, but acts as a small super-fast cache that is responsible for hiding the latency involved in accessing main memory. HBM2 is expensive and power hungry as well, but offers a better performance mix across the board then Intels eDRAM solution.

 

[jdwii]

Raja Koduri is officially out AMD.

First Rori Read. Second Keller. Third Koduri. I guess Papermaster will be the next on leaving AMD.

WOW also Juan i had no idea man Newegg is having massive supply issues with coffee-lake

Anyways leaving Amd well vega was a massive failure i mean everyone try and think of the last time Amd was worse and more expensive in terms of GPU products? Sorry i can't i've been around since the radeon 4000HD series and when Amd offered a worse GPU they also priced it competitively VEGA is a massive flop just like bulldozer.

 

[-Fran-]

Raja Koduri is officially out AMD.

First Rori Read. Second Keller. Third Koduri. I guess Papermaster will be the next on leaving AMD.

There's an interesting take on his sabbatical and now confirmed departure: RTG exposure.

It seems his decision to get in bed with the devil was not something Lisa Su took lightly. I have zero clue how things unfolded or even started, but it all points out that Raja started it and Lisa opposed it (initially). I would imagine the rumours of Raja trying to sell RTG were somewhat true if this is the case.

Interesting rumour that kind of sucks for everyone involved.

Cheers!

 

[Gon Freecss]

Iris Pro used an expensive memory solution to provide bandwidth to the iGPU. This new chip uses a cheaper HBM2 solution to feed the GPU.

What expensive memory solution?

eDRAM

Oh, I see.

Iris Pro used an expensive memory solution to provide bandwidth to the iGPU. This new chip uses a cheaper HBM2 solution to feed the GPU.

What expensive memory solution?

eDRAM. It's expensive, power hungry, but acts as a small super-fast cache that is responsible for hiding the latency involved in accessing main memory. HBM2 is expensive and power hungry as well, but offers a better performance mix across the board then Intels eDRAM solution.

I see. I think Intel went with this move just to appease Apple. This is only a mobile chip. We'll still be seeing Intel integrated graphics in the upcoming generations. Btw, will Intel have access to AMD's IP and use them to further develop their iGPUs?

 

[jaymc]

Looks like it's Polaris for now anyway... Could be Vega in the future though.

Here's Adored's take on it... which does kinda paint an Interesting picture...
He reckons the reason they didn't just do it themselves is they always have trouble selling high end parts. This just doesn't make sense to me.. I think they should have done this themselves, but who knows exactly what is going on behind closed doors I guess.

It seem's to me that AMD had an opportunity to make a product that neither Intel nor Nvidia could make an they just threw it away.
I wonder what the deciding factor in this decision was it's a strange one that's for sure, financial pressure's ? Maybe a large injection of cash was needed to bolster up the books ?

https://www.youtube.com/watch?v=mNY5e5CFlbc

Edit: spelling and grammer

 

[gamerk316]

As I understand it, no IP is being exchanged. AMD is making a product for Intel, no different then their work with MSFT and Sony.

 

[adamsleath]

https://www.newegg.com/Processors-Desktops/SubCategory/ID-343?Order=BESTSELLING

try again;

r5 1600 is at number 1 bestseller according to my web browser

 

[juanrga]

Raja Koduri is officially out AMD.

First Rori Read. Second Keller. Third Koduri. I guess Papermaster will be the next on leaving AMD.

There's an interesting take on his sabbatical and now confirmed departure: RTG exposure.

It seems his decision to get in bed with the devil was not something Lisa Su took lightly. I have zero clue how things unfolded or even started, but it all points out that Raja started it and Lisa opposed it (initially).

It is not Koduri vs Lisa. It is CPU guys vs GPU guys. It is green (AMD) vs red (ATI)

https://www.hardocp.com/article/2016/05/27/from_ati_to_amd_back_journey_in_futility

 

[juanrga]

Raja Koduri is officially out AMD.

First Rori Read. Second Keller. Third Koduri. I guess Papermaster will be the next on leaving AMD.

WOW also Juan i had no idea man Newegg is having massive supply issues with coffee-lake

Some people has followed closely Newegg and their near doubling price scheme

https://hardforum.com/threads/intels-8th-generation-core-family-coffee-lake-lga-1151-6c-12t.1930226/page-81#post-1043296145
https://hardforum.com/threads/intels-8th-generation-core-family-coffee-lake-lga-1151-6c-12t.1930226/page-81#post-1043296596
https://hardforum.com/threads/intels-8th-generation-core-family-coffee-lake-lga-1151-6c-12t.1930226/page-81#post-1043296622
https://hardforum.com/threads/intels-8th-generation-core-family-coffee-lake-lga-1151-6c-12t.1930226/page-81#post-1043297930
https://hardforum.com/threads/intels-8th-generation-core-family-coffee-lake-lga-1151-6c-12t.1930226/page-81#post-1043299067

Stock last few hours each say.

Anyways leaving Amd well vega was a massive failure i mean everyone try and think of the last time Amd was worse and more expensive in terms of GPU products? Sorry i can't i've been around since the radeon 4000HD series and when Amd offered a worse GPU they also priced it competitively VEGA is a massive flop just like bulldozer.

And the responsible for that is Lisa Su.

 

[juanrga]

https://www.newegg.com/Processors-Desktops/SubCategory/ID-343?Order=BESTSELLING

try again;

r5 1600 is at number 1 bestseller according to my web browser

It is #2 just now, with #1 taken by Intel i7. And why Newegg is a bad place to look for this kind of info was explained above...

 

[adamsleath]

1600 is at number 1 right now.

.."as observed", and there's no reliable data to the contrary.
https://www.newegg.com/global/au/Processors-Desktops/SubCategory/ID-343?Order=BESTSELLING

 

[jdwii]

Either way Ryzen is doing fine it was never going to be in the top 10 forever anyways once Coffee-Lake becomes in stock at retail pricing Ryzen will fall further behind but as of today its in the top 10.

Doesn't even matter much anyways as that is just from people like us who buy those products not OEM's.

 

[-Fran-]

Haha, man... Mr Huang's response in the earnings call is a real treat. He not only acknowledges Raja as a great grab from Intel, but he is also saying "come at me bro" with no problems. You gotta respect that when talking to Intel.

Popcorn indeed. These next couple years, the leaks are going to be interesting.

Cheers!

 

[juanrga]

https://hothardware.com/news/amd-epyc-2-64-cores-128-threads-and-256mb-l3-cache

It seems my prediction of 6-core CCX for Zen2 has been just refuted. Either AMD goes for 8-core CCX modules or will maintain the same 4-core configuration and will add four CCX modules per die.

The quadrupling of L3 cache could help to mitigate the latency problems when accessing main RAM, but in the other hand, this seems to imply that Zen2 core size will be similar to Zen.

After to know the memory configuration for EPYC 2, I guess the hypothesis of 8-core CCX modules gains ground. So

• 
  ■ 8-core CCX module
  ■ up to 16-core RyZen 2 (single die and dual channel)
  ■ up to 32-core ThreadRipper 2 (dual die and quad-channel)
  ■ up to 64-core EPYC 2 (quad die and octo-channel)
  ■ up to 8-core APUs

 

[goldstone77]

Haha, man... Mr Huang's response in the earnings call is a real treat. He not only acknowledges Raja as a great grab from Intel, but he is also saying "come at me bro" with no problems. You gotta respect that when talking to Intel.

Popcorn indeed. These next couple years, the leaks are going to be interesting.

Cheers!

I love it! The gauntlet was thrown, and Huang picked it up and smacks it back in Intel's face! He calls Intel's architecture weak/useless, because they lack focus on creating a singular design. He goes on to say Nvidia's processor is the most significant and relevant technology in the world!

If you have four or five different architectures to support, that you offer to your customers, and they have to pick the one that works the best, you are essentially are saying that you don't know which one is the best [.....] If there's five architectures, surely over time, 80% of them will be wrong. I think that our advantage is that we are singularly focused.

"The amount of software engineering that goes on top of it is significant as well. So, if you look at the way that we do things, we plan our roadmap about five years out. It takes about three years to build a new generation, and we build multiple GPUs at the same time, and on top of that, there are some 5,000 engineers working on system software and numerics libraries, and solvers, and compilers, and graph analytics, and cloud platforms, and virtualization stacks in order to make this computing architecture useful to all of the people we serve.
So when you think about it from that perspective, it's just an enormous undertaking. Arguably the most significant undertaking of any processor in the world today. And that's why we are able to speed up applications by a factor of 100."

http://www.tomshardware.com/news/nvidia-intel-amd-gpu-koduri,35895.html

Edit: He also says they don't understand the needs of their customers!

you are essentially are saying that you don't know which one is the best

 

[jdwii]

Haha, man... Mr Huang's response in the earnings call is a real treat. He not only acknowledges Raja as a great grab from Intel, but he is also saying "come at me bro" with no problems. You gotta respect that when talking to Intel.

Popcorn indeed. These next couple years, the leaks are going to be interesting.

Cheers!

I love it! The gauntlet was thrown, and Huang picked it up and smacks it back in Intel's face! He calls Intel's architecture weak/useless, because they lack focus on creating a singular design. He goes on to say Nvidia's processor is the most significant and relevant technology in the world!

If you have four or five different architectures to support, that you offer to your customers, and they have to pick the one that works the best, you are essentially are saying that you don't know which one is the best [.....] If there's five architectures, surely over time, 80% of them will be wrong. I think that our advantage is that we are singularly focused.

"The amount of software engineering that goes on top of it is significant as well. So, if you look at the way that we do things, we plan our roadmap about five years out. It takes about three years to build a new generation, and we build multiple GPUs at the same time, and on top of that, there are some 5,000 engineers working on system software and numerics libraries, and solvers, and compilers, and graph analytics, and cloud platforms, and virtualization stacks in order to make this computing architecture useful to all of the people we serve.
So when you think about it from that perspective, it's just an enormous undertaking. Arguably the most significant undertaking of any processor in the world today. And that's why we are able to speed up applications by a factor of 100."

http://www.tomshardware.com/news/nvidia-intel-amd-gpu-koduri,35895.html

Edit: He also says they don't understand the needs of their customers!

you are essentially are saying that you don't know which one is the best

You know i say this as a 1080 owner and a proud 1080 owner Nvidia needs some real competition i agree with Jayztwocents from his recent techtalk someone needs to do something about Jensen even if that means Intel stepping in. Take a look at the new star wars edition Titan XP lol, that's what happens when a company has no real competition. VEGA was such a massive disappointment i can't even understand why he is saying its some great loss but whatever after what Kyle from hardocp reported a year ago Raja was basically overruling Lisa Su anyways.

I'd like to see Amd get a lot of money from this Intel deal but i feel like its going to be more of a niche market then anything else i could be wrong however. Nasty words like that from Jensen basically means Nvidia+Intel won't be happening any time soon.

 

[juanrga]

You know i say this as a 1080 owner and a proud 1080 owner Nvidia needs some real competition i agree with Jayztwocents from his recent techtalk someone needs to do something about Jensen even if that means Intel stepping in. Take a look at the new star wars edition Titan XP lol, that's what happens when a company has no real competition. VEGA was such a massive disappointment i can't even understand why he is saying its some great loss but whatever after what Kyle from hardocp reported a year ago Raja was basically overruling Lisa Su anyways.

Maybe Jensen knows that Vega's fiasco is a consequence of Lisa Su management... Maybe Jensen knows that without Koduri the fiasco had been about one order of magnitude greater.

 

[goldstone77]

1 year and 5 year stock price charts. You can see why Huang has the balls to say what he did about Intel. Intel's lack of innovation is shown in it's stock price that has been relatively stagnant to the point of depreciation for many years until recently. Nvidia on the other hand has been on a tear since 2016! Intel has little less than double Nvidia's market cap, but the stock price is less than 1/4 of Nvidia. The future is GPU's not CPU's, because the future is A.I. and AR/VR.

 

[goldstone77]

https://twitter.com/BitsAndChipsEng/status/928568572141621248

 

[YoAndy]

https://twitter.com/BitsAndChipsEng/status/928568572141621248

Nvidia is laughing at that,
With Nvidia’s Max-Q initiative they are providing OEMs with what they call Max-Q versions of the GTX 1060, 1070 and 1080 chips.
Let’s take a regular mobile GTX 1080 for instance, which runs at around 150 W. The Max-Q 1080 variant only needs between 90-110 W. This allows manufacturers to put the 1080 chip inside much slimmer and portable laptops, like the Asus Zephyrus or the Acer Predator Triton
https://www.ultrabookreview.com/16277-laptops-nvidia-max-q/
The new Acer Predator With nvidia GTX 1080 and the Core i7-7820HK https://www.ultrabookreview.com/15075-acer-predator-17x/

 

[-Fran-]

Spoiler

https://twitter.com/BitsAndChipsEng/status/928568572141621248

Nvidia is laughing at that,
With Nvidia’s Max-Q initiative they are providing OEMs with what they call Max-Q versions of the GTX 1060, 1070 and 1080 chips.
Let’s take a regular mobile GTX 1080 for instance, which runs at around 150 W. The Max-Q 1080 variant only needs between 90-110 W. This allows manufacturers to put the 1080 chip inside much slimmer and portable laptops, like the Asus Zephyrus or the Acer Predator Triton
https://www.ultrabookreview.com/16277-laptops-nvidia-max-q/
The new Acer Predator With nvidia GTX 1080 and the Core i7-7820HK https://www.ultrabookreview.com/15075-acer-predator-17x/

I don't know if laughing, since I'm pretty sure they'd rather be the ones partnering with Intel on that instead of AMD, but I can also understand why you're mentioning it.

This is a threat for them even if (so far) they hold a good thermal advantage over AMD and even if all their Notebooks will look nicer until this actually hits the market (keep in mind that is a NUC proto), but don't be confused or take lightly what Intel is doing; it is a real threat to nVidia and if Intel plays its cards right (even the ones hiding in the sleeve next to the sleeve-gun) could displace nVidia from the market with not much effort. The whole point of integrating the GPU and CPU into that package is to lower costs and attack thermals precisely. There's a reason Mr. Huang addressed it directly in the earnings call. He gave investors a very clear message: "yes, I am aware".

And that is not even thinking on whatever other wacky stuff they have in the pipeline. This is just the first known incarnation of their bizarre franken-MCMs.

Cheers!

 

[YoAndy]

Spoiler

https://twitter.com/BitsAndChipsEng/status/928568572141621248

Nvidia is laughing at that,
With Nvidia’s Max-Q initiative they are providing OEMs with what they call Max-Q versions of the GTX 1060, 1070 and 1080 chips.
Let’s take a regular mobile GTX 1080 for instance, which runs at around 150 W. The Max-Q 1080 variant only needs between 90-110 W. This allows manufacturers to put the 1080 chip inside much slimmer and portable laptops, like the Asus Zephyrus or the Acer Predator Triton
https://www.ultrabookreview.com/16277-laptops-nvidia-max-q/
The new Acer Predator With nvidia GTX 1080 and the Core i7-7820HK https://www.ultrabookreview.com/15075-acer-predator-17x/

I don't know if laughing, since I'm pretty sure they'd rather be the ones partnering with Intel on that instead of AMD, but I can also understand why you're mentioning it.

This is a threat for them even if (so far) they hold a good thermal advantage over AMD and even if all their Notebooks will look nicer until this actually hits the market (keep in mind that is a NUC proto), but don't be confused or take lightly what Intel is doing; it is a real threat to nVidia and if Intel plays its cards right (even the ones hiding in the sleeve next to the sleeve-gun) could displace nVidia from the market with not much effort. The whole point of integrating the GPU and CPU into that package is to lower costs and attack thermals precisely. There's a reason Mr. Huang addressed it directly in the earnings call. He gave investors a very clear message: "yes, I am aware".

And that is not even thinking on whatever other wacky stuff they have in the pipeline. This is just the first known incarnation of their bizarre franken-MCMs.

Cheers!

Displacing Nvidia from the market? 😀 Nvidia can do the same by using ARM processors actually Nvidia is already doing the same, Nvidia don't need Intel or AMD. The future could be GPU
From Nvidia:
The new Tegra X1 is the most advanced mobile processor we've ever created. Powerful new NVIDIA Maxwell™ architecture, 256 GPU cores, multi-core 64-bit CPUs, unbeatable 4K video capabilities, and more power-efficient performance than its predecessor make it perfect for even the most challenging mobile and automotive applications.

Nvidia been doing it for a long time now. Windows and Qualcom have been working together for a new Mobile CPU, for laptops and Tablets and Nvidia has his own and is the leader in Artificial Intelligence (AI)
Toyota
Audi
Volvo
Tesla
VW
HONDA
CITROEN
ROLLS ROYCE
PUGEAOT
LAMBORGINI
ASTON MARTIN
ALGA ROMEO
MERCEDES BENZ
MINI
BENTLEY
and more car manufacturers are working with Nvidia and you think Intel and AMD can make Nvidia go out of business 😀. The new flagship Audi A8 — which is the world’s first Level 3 autonomous driving car to go into production — features a multitude of high-tech splendors, all powered by NVIDIA.
,..http://www.nvidia.com/object/tegra-x1-processor.html
https://www.nvidia.com/en-us/self-driving-cars/
https://www.pcworld.com/article/3191401/computers/qualcomm-first-windows-10-arm-pc-coming-in-the-fourth-quarter.html

 

[-Fran-]

Displacing Nvidia from the market? 😀 Nvidia can do the same by using ARM processors actually Nvidia is already doing the same, Nvidia don't need Intel or AMD. The future could be GPU
From Nvidia:
The new Tegra X1 is the most advanced mobile processor we've ever created. Powerful new NVIDIA Maxwell™ architecture, 256 GPU cores, multi-core 64-bit CPUs, unbeatable 4K video capabilities, and more power-efficient performance than its predecessor make it perfect for even the most challenging mobile and automotive applications.

Nvidia been doing it for a long time now. Windows and Qualcom have been working together for a new Mobile CPU, for laptops and Tablets and Nvidia has his own and is the leader in Artificial Intelligence (AI)
Toyota
Audi
Volvo
Tesla
VW
HONDA
CITROEN
ROLLS ROYCE
PUGEAOT
LAMBORGINI
ASTON MARTIN
ALGA ROMEO
MERCEDES BENZ
MINI
BENTLEY
and more car manufacturers are working with Nvidia and you think Intel and AMD can make Nvidia go out of business 😀. The new flagship Audi A8 — which is the world’s first Level 3 autonomous driving car to go into production — features a multitude of high-tech splendors, all powered by NVIDIA.
,..http://www.nvidia.com/object/tegra-x1-processor.html
https://www.nvidia.com/en-us/self-driving-cars/
https://www.pcworld.com/article/3191401/computers/qualcomm-first-windows-10-arm-pc-coming-in-the-fourth-quarter.html

So you now have included Cars in a Notebook comment. Ok, I'll play ball.

Currently, nVidia is not the owner of the ARM spec, so they will depend a lot on how good ARM designs are (ISA wise) to continue developing whatever CPU backend they want. Now, on the upside, GPUs are beefed up RISC units in essence nowadays (hence his comment), but it will depend on a 100% on the ISA you're implementing on how good that promise fills in reality. Currently, as I'm aware, ARM is not "GPU-friendly" and the only "GPU" spec you have is OCL and CUDA for "CPU-like" tasks (big quote marks around it). So, for nVidia to really make good on their promise, they'll have to fork the ARM spec or just create one on their own.

Now, I do not see that happening soon, so nVidia will still depend on ARM and/or Intel for "CPU".

Also, I don't know where you got I mentioned "out of business". Are you reading correctly what I'm writing or maybe just assuming a bit too much beyond my words?

Cheers!