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

[goldstone77]

GLOBALFOUNDRIES is Hitting on all Cylinders
by Scotten Jones
Published on 09-29-2017 03:00 PM

12LP (leading performance) is a mid-life kicker for 14nm and offers 15% greater density and 10% better performance. 12LP is designed to be competitive with other 12nm foundry processes (TSMC recently announced a 12nm process and Samsung announced an 11nm process). Production is due to start in Q1-2018.

7LP (Leading performance) is GF's leading edge 7nm FinFET process. Whereas GF licensed Samsung's FinFET process at 14nm, at 7nm GF is developing their own process. Device performance at 7nm is >40% better than at 14nm and total power is >60% lower than 14nm. 7nm will provide 17 million gates/mm2 and a 30% die cost reduction versus 14nm with a >45% cost reduction for target segments. The PDK is available now and risk production is on track for the first half of 2018.

 

[juanrga]

It is curious why no Epyc-based HPC clusters are being built... I cannot find any news of any so far anyway, I cannot understand it.
Epyc is getting glowing reviews everywhere. For certain workloads it really excels plus it's cheaper 8 channel memory, more address space an way more PCI Express lanes. It wouldn't make much sense for it not to succeed..

Unless of coarse, I hope Intel have not resolved to using dirty tactics to defend their HPC market share... I can see them fighting tooth and nail for it. I wouldn't be surprised if they start offering financial rewards, rebates or other incentives for staying with Xeon.

People is choosing ARM for HPC clusters. Two or three ARM-based supercomputers are being built. So the older excuse of "dirty tactics" is not working here. The problem for AMD is that those people tests things by themselves. They don't make a purchase decision after reading a biased EPYC review from Anandtech where the performance of Xeons is crippled by 40%.

I said two years ago that EPYC was going to be crushed by both sides: ARM and Intel. And time is giving me the reason.

Well Yuka kindly pointed out that it is not being marketed as a HPC product so that explains that I guess.

Except that AMD tries to sell EPYC as a HPC product:

The AMD EPYC processor provides excellent performance for memory bound HPC workloads

The reasons why EPYC is being massively rejected in HPC are the same why is being massively rejected in servers.

 

[-Fran-]

It is curious why no Epyc-based HPC clusters are being built... I cannot find any news of any so far anyway, I cannot understand it.
Epyc is getting glowing reviews everywhere. For certain workloads it really excels plus it's cheaper 8 channel memory, more address space an way more PCI Express lanes. It wouldn't make much sense for it not to succeed..

Unless of coarse, I hope Intel have not resolved to using dirty tactics to defend their HPC market share... I can see them fighting tooth and nail for it. I wouldn't be surprised if they start offering financial rewards, rebates or other incentives for staying with Xeon.

People is choosing ARM for HPC clusters. Two or three ARM-based supercomputers are being built. So the older excuse of "dirty tactics" is not working here. The problem for AMD is that those people tests things by themselves. They don't make a purchase decision after reading a biased EPYC review from Anandtech where the performance of Xeons is crippled by 40%.

I said two years ago that EPYC was going to be crushed by both sides: ARM and Intel. And time is giving me the reason.

Well Yuka kindly pointed out that it is not being marketed as a HPC product so that explains that I guess.

Except that AMD tries to sell EPYC as a HPC product:

The AMD EPYC processor provides excellent performance for memory bound HPC workloads

The reasons why EPYC is being massively rejected in HPC are the same why is being massively rejected in servers.

That is a massive contradiction from all the presentations they've shown then.

I would imagine, trying to understand the reasoning, they cannot deny whatever appeal it might have to specific HPC type of workloads in their 2P configurations and still need to sell it using that angle.

Shame, really. I don't consider EPYC for HPC either. At least, not until they tune IF for lower latency.

 

[juanrga]

7nm will be not ready next year, otherwise Glofo wouldn't release 14LPP+ renamed to 12LP for marketing purposes next year. The reason why Glofo will release '12nm' next year is just to make the appearance that they are advancing, because other foundries have 10/12nm.

7LP starts RISK production in 1H 2018
12LP starts VOLUME production 1Q 2018

12LP risk production in 1H 2018

That slide is wrong, the slide mixed up 7nmLP risk production and 12nmLP volume production dates.
https://www.reddit.com/r/AMD_Stock/comments/71rn8k/mistake_gf_conference_12nm_production_starts_q1/

The source is Pcgameshardware.de which simply says that there is a mistake in the slide, but this is not demonstrated. The google translator also translates this part from the Pcgameshardware.de article:

The contractor promises higher packing density and better performance. In the end, 12LP is likely to be a renamed "14LPP" process based on TSMCs 12FFN.

The claim that Glofo 12LP is based on TSMC 12FFN is wrong. How can I know the rest they say is valid?

Being strict that slides says that production of 12LP starts in 2018. It doesn't say what kind of production is.

 

[juanrga]

It is curious why no Epyc-based HPC clusters are being built... I cannot find any news of any so far anyway, I cannot understand it.
Epyc is getting glowing reviews everywhere. For certain workloads it really excels plus it's cheaper 8 channel memory, more address space an way more PCI Express lanes. It wouldn't make much sense for it not to succeed..

Unless of coarse, I hope Intel have not resolved to using dirty tactics to defend their HPC market share... I can see them fighting tooth and nail for it. I wouldn't be surprised if they start offering financial rewards, rebates or other incentives for staying with Xeon.

People is choosing ARM for HPC clusters. Two or three ARM-based supercomputers are being built. So the older excuse of "dirty tactics" is not working here. The problem for AMD is that those people tests things by themselves. They don't make a purchase decision after reading a biased EPYC review from Anandtech where the performance of Xeons is crippled by 40%.

I said two years ago that EPYC was going to be crushed by both sides: ARM and Intel. And time is giving me the reason.

Well Yuka kindly pointed out that it is not being marketed as a HPC product so that explains that I guess.

Except that AMD tries to sell EPYC as a HPC product:

The AMD EPYC processor provides excellent performance for memory bound HPC workloads

The reasons why EPYC is being massively rejected in HPC are the same why is being massively rejected in servers.

That is a massive contradiction from all the presentations they've shown then.

Except it is not true because AMD guys as Scott Aylor, AMD corporate VP and GM of enterprise solutions business, have been trying to convince HPC people to use EPYC since first minute of launch...

 

[goldstone77]

AMD to launch 12nm Ryzen in February 2018, says mobo makers
Monica Chen, Taipei; Joseph Tsai, DIGITIMES [Wednesday 27 September 2017]

AMD has informed its partners that it plans to launch in February 2018 an upgrade version of its Ryzen series processors built using a 12nm low-power (12LP) process at Globalfoundries, according to sources at motherboard makers.

The company will initially release the CPUs codenamed Pinnacle 7, followed by mid-range Pinnacle 5 and entry-level Pinnacle 3 processors in March 2018, the sources disclosed. AMD is also expected to see its share of the desktop CPU market return to 30% in the first half of 2018.

AMD will launch the low-power version of Pinnacle processors in April 2018 and the enterprise version Pinnacle Pro in May 2018.

Their corresponding chipsets, the 400 series, will also become available in March 2018 with X470- or B450-based motherboards to be the first to hit the store shelves. The chipsets are still designed by ASMedia and its orders for the chipsets are expected to grow dramatically starting January 2018.

Thanks to stable chip orders for Microsoft's and Sony's game consoles, increased demand for graphics cards, growing sales for its Ryzen 7/5 processors, new Ryzen Pro product line for the enterprise sector and the top-end Ryzen Treadripper processors, AMD managed to achieve 19% sequential growth in second-quarter 2017 revenues and expects the amount to grow further by 23% in the third quarter.

AMD said it does not comment on products that have not been announced.

Visit our website for table, pictures and any remaining text.

 

[8350rocks]

https://www.youtube.com/watch?v=O98qP-FsIWo

Coffeelake is actually zero IPC gain from SKL/KL, reviews broken down.

What a con by intel.

 

[aldaia]

The google translator ...

Lmao, I have nothing else to add, lol

Other than you are wrong as usual but keep arguing for the shake of arguing to the point of being ridiculous.

 

[juanrga]

https://www.youtube.com/watch?v=O98qP-FsIWo

Coffeelake is actually zero IPC gain from SKL/KL, reviews broken down.

What a con by intel.

CoffeLake is a second optimization of Skylake. The core microarchitecture is the same, so the part of the IPC due to the core is the same. This has been known for years.

The IPC gains in CoffeLake come from larger L3 and the faster memory (2666 vs 2400), but this IPC gain only affects to memory-bound workloads that don't use all cores. It is also a very small increase. In some other part I estimate ~4% gains.

The main performance gains in CoffeLake come from higher clocks and 50--100% more cores (depending of the model).

 

[juanrga]

AMD to launch 12nm Ryzen in February 2018, says mobo makers
Monica Chen, Taipei; Joseph Tsai, DIGITIMES [Wednesday 27 September 2017]

AMD has informed its partners that it plans to launch in February 2018 an upgrade version of its Ryzen series processors built using a 12nm low-power (12LP) process at Globalfoundries, according to sources at motherboard makers.

The company will initially release the CPUs codenamed Pinnacle 7, followed by mid-range Pinnacle 5 and entry-level Pinnacle 3 processors in March 2018, the sources disclosed. AMD is also expected to see its share of the desktop CPU market return to 30% in the first half of 2018.

AMD will launch the low-power version of Pinnacle processors in April 2018 and the enterprise version Pinnacle Pro in May 2018.

Their corresponding chipsets, the 400 series, will also become available in March 2018 with X470- or B450-based motherboards to be the first to hit the store shelves. The chipsets are still designed by ASMedia and its orders for the chipsets are expected to grow dramatically starting January 2018.

Still no one has explained me why Pinnacle Ridge requires a new series of chipsets.

Also AMD could not inform its partners that it plans to use "a 12nm low-power (12LP) process", because LP means Leading Performance, not Low Power. So I wonder what part of the leak is real and what part is invented.

 

[goldstone77]

AMD to launch 12nm Ryzen in February 2018, says mobo makers
Monica Chen, Taipei; Joseph Tsai, DIGITIMES [Wednesday 27 September 2017]

AMD has informed its partners that it plans to launch in February 2018 an upgrade version of its Ryzen series processors built using a 12nm low-power (12LP) process at Globalfoundries, according to sources at motherboard makers.

The company will initially release the CPUs codenamed Pinnacle 7, followed by mid-range Pinnacle 5 and entry-level Pinnacle 3 processors in March 2018, the sources disclosed. AMD is also expected to see its share of the desktop CPU market return to 30% in the first half of 2018.

AMD will launch the low-power version of Pinnacle processors in April 2018 and the enterprise version Pinnacle Pro in May 2018.

Their corresponding chipsets, the 400 series, will also become available in March 2018 with X470- or B450-based motherboards to be the first to hit the store shelves. The chipsets are still designed by ASMedia and its orders for the chipsets are expected to grow dramatically starting January 2018.

Still no one has explained me why Pinnacle Ridge requires a new series of chipsets.

Also AMD could not inform its partners that it plans to use "a 12nm low-power (12LP) process", because LP means Leading Performance, not Low Power. So I wonder what part of the leak is real and what part is invented.

I can't find any information about the new chipsets. And the LP (low-power) I imagine it is a miss understanding, because of the previous 14nm LPP. We will find out soon enough if 12nm will be released in February.

 

[aldaia]

http://wccftech.com/amd-announces-2nd-gen-ryzen-vega-launching-12nm-2018/

AMD announced today at the GlobalFoundries Technology Conference that it will be transitioning its Ryzen CPUs and Vega GPUs to 12nm LP technology next year. The announcement came via the company’s Chief Technology Officer Mark Papermaster, who confirmed alongside GlobalFoundries that 12LP volume production will begin in the first quarter of next year. Second generation Ryzen and Vega products are expected to be available on the market soon thereafter.

 

[juanrga]

http://wccftech.com/amd-announces-2nd-gen-ryzen-vega-launching-12nm-2018/

AMD announced today at the GlobalFoundries Technology Conference that it will be transitioning its Ryzen CPUs and Vega GPUs to 12nm LP technology next year. The announcement came via the company’s Chief Technology Officer Mark Papermaster, who confirmed alongside GlobalFoundries that 12LP volume production will begin in the first quarter of next year. Second generation Ryzen and Vega products are expected to be available on the market soon thereafter.

Just immediately after the part that you quote, the article introduces the Papermaster slide that says clearly "Risk Production in H1 2018". The same event from Anandtech:

https://www.anandtech.com/show/11854/globalfoundries-adds-12lp-process-tech-amd-first-customer

GlobalFoundries expects to start risk production using the 12LP fabrication technology in Q1 2018. The company does not disclose any timeframes for HVM, but expect them to be different for various chip developers. For example, AMD is a very close partner of GlobalFoundries, so the company gets access to PDKs ahead of many other clients. That said, expect AMD to release its products made using the 12LP rather sooner than later. Mark Papermaster says that the aforementioned chips are to be released in 2018, but he does not elaborate whether to expect them in the first or the second half of the year.

I checked press releases and in no part Papermaster said what WCFTECH claims. The exact words of Papermaster were:

Our deep collaboration with GF has helped AMD bring a set of leadership high-performance products to market in 2017 using 14nm FinFET technology. We plan to introduce new client and graphics products based on GF’s 12nm process technology in 2018 as a part of our focus on accelerating our product and technology momentum.

NOTE: I also had fun reading the last part of the WCCFTECH article, where the author writes a table with specs of current and future RyZen chips. According to this WCCFTECH author, Summit Ridge, the current family of RyZen chips released this year, is the "RyZen 2000 series", and Pinnacle Ridge will be "RyZen 3000 series". This is the level of quality of WCCFTECH publications. 1800X, 1700X, 1700, 1600X,... are "2000 series"

 

[-Fran-]

Man... That place is a real hell hole for comments. I always laugh so much when reading them.

 

[YoAndy]

https://www.youtube.com/watch?v=O98qP-FsIWo

Coffeelake is actually zero IPC gain from SKL/KL, reviews broken down.

What a con by intel.

Yes, Intel hasn't change the microarchitecture so yeah IPC of course remains the same. They did optimize the memory controller and give the new chip a few other tweaks and of course higher clock speed.. why would they need to increase the IPC when they are ahead clock per clock and over 1GHz faster.

 

[8350rocks]

https://www.youtube.com/watch?v=O98qP-FsIWo

Coffeelake is actually zero IPC gain from SKL/KL, reviews broken down.

What a con by intel.

CoffeLake is a second optimization of Skylake. The core microarchitecture is the same, so the part of the IPC due to the core is the same. This has been known for years.

The IPC gains in CoffeLake come from larger L3 and the faster memory (2666 vs 2400), but this IPC gain only affects to memory-bound workloads that don't use all cores. It is also a very small increase. In some other part I estimate ~4% gains.

The main performance gains in CoffeLake come from higher clocks and 50--100% more cores (depending of the model).

watch the video, there are no IPC gains at same clockspeeds.

 

[jaymc]

AMD Preps AGESA 1.0.0.7 Firmware to support Raven Ridge APUs:

http://www.guru3d.com/news-story/amd-preps-agesa-1-7-firmware-to-support-raven-ridge-apus.html

 

[YoAndy]

https://www.youtube.com/watch?v=O98qP-FsIWo

Coffeelake is actually zero IPC gain from SKL/KL, reviews broken down.

What a con by intel.

CoffeLake is a second optimization of Skylake. The core microarchitecture is the same, so the part of the IPC due to the core is the same. This has been known for years.

The IPC gains in CoffeLake come from larger L3 and the faster memory (2666 vs 2400), but this IPC gain only affects to memory-bound workloads that don't use all cores. It is also a very small increase. In some other part I estimate ~4% gains.

The main performance gains in CoffeLake come from higher clocks and 50--100% more cores (depending of the model).

watch the video, there are no IPC gains at same clockspeeds.

I have no idea why some people on this Thread are more concerned about Intel than AMD .. we know that core per core Kabby and Coffe are the same......Kaby and Coffee Lake share the same microarchitecture, with Coffee Lake being effectively (ahem) caffeinated with a slightly refined manufacturing process (14nm++) over Kaby Lake (14nm+), as well as more cores and threads across the board, a different allocation of cache resources, and a few new overclocking knobs and levers. Learn more here:
http://www.tomshardware.com/news/intel-coffee-lake-kaby-lake,35549.html

 

[jaymc]

Look's like RavenRidge is almost here guy's...

"HP is readying an Envy x360 laptop with integrated AMD Vega mobile graphics"
http://www.pcgamer.com/hp-is-readying-an-envy-x360-laptop-with-integrated-amd-vega-mobile-graphics/

 

[-Fran-]

Look's like RavenRidge is almost here guy's...

"HP is readying an Envy x360 laptop with integrated AMD Vega mobile graphics"
http://www.pcgamer.com/hp-is-readying-an-envy-x360-laptop-with-integrated-amd-vega-mobile-graphics/

"Other specs include a 15.6-inch IPS display with a 1920x1080 resolution, 8GB of single-channel DDR4-2400"

HP is doing it on purpose... Or AMD is dropping the ball here not slapping HP in the face.

I hope AMD understands they cannot screw it up with reviews of laptops using single channel memory and other under-par configurations taking away that very important "initial impression" feeling from it's new line of APUs.

I will be reading about them on day 1 and expecting very eagerly the desktop versions of them.

Cheers!

 

[adamsleath]

i'm expecting clock speeds to be higher on zen2 (lower TDP) also some ipc improvements.

I also think AMD will add more cores to zen2, because they can.

 

[goldstone77]

i'm expecting clock speeds to be higher on zen2 (lower TDP) also some ipc improvements.

I also think AMD will add more cores to zen2, because they can.

I wouldn't expect a whole lot more than a slight clock improvement, and slight IPC improvement.

 

[YoAndy]

i'm expecting clock speeds to be higher on zen2 (lower TDP) also some ipc improvements.

I also think AMD will add more cores to zen2, because they can.

More cores? for what? Most programs and games can't even use 6 with 12 threads.

 

[juanrga]

I expect 6-core CCX for Zen2.

 

[goldstone77]

I expect 6-core CCX for Zen2.

It would fit for 48 core Starship. I think that is a reasonable expectation.