Asrock B550 PG Riptide review
be quiet! Dark Rock TF2 review
Basemark GPU v1.2 benchmarks with 36 GPUs
Corsair MP600 Pro XT PCIe 4.0 M.2 NVMe review
Teamgroup MP34Q 2TB NVMe SSD review
Lian Li SP750 (750W PSU) review
Teamgroup Cardea Z44Q 4TB NVMe SSD review
Asus ROG Swift PG32UQX monitor review
Plextor M10P 2TB NVMe PCIe 4.0 SSD review
ASRock RX 6600 XT Phantom Gaming D 8GB OC review
AMD Zen Enterprise Roadmap Shows 48-core CPUs in 2018
I'm not even sure where and how to begin with this information. But a slide-deck roadmap has leaked onto the web through the usual channels. The information in the slide-show reveals is rather gigantic.
The information based on the data-center thus enterprise segment, and is mentioning 14nm Naples With 32 Cores in 2017, a 7nm 48 Cores proc in 2018 – Zen 1 and Zen 2 Based Horned Own, Grey Hawk, Banded Kestrel and River Hawk CPUs. The AMD internal roadmap leaked reveals pretty much any AMD CPU that is available and will be available in the upcoming year or two.
The slides apparently dates back to February and includes proc overviews starting at 2016 running up-to 2018. For this timeframe obviously Naples With 32 Cores is up and coming for the enterprise market, but also 3 CPUs code-named A1100 ARM using ARM Cortex-A57 cores. Then there's Merlin Falcon using Excavator cores and a second Excavator based product called the Brown Falcon CPU.
| Enterprise Proc | AMD Snowy Owl | AMD Naples | AMD Starship |
|---|---|---|---|
| CPU Architecture | Zen 1 | Zen 1 | Zen 2 |
| Process Node | 14nm FinFET | 14nm FinFET | 7nm FinFET |
| Maximum Cores | 16 Cores | 32 Cores | 48 Cores |
| Maximum Threads | 32 Threads | 64 Threads | 96 Threads |
| Availability | Q2 2017 | Q2 2017 | 2018 |
Where your eyes will pop out though is slide 1, check the year 2018, where AMD has planned a 48 core server processor (that would be 96 threads) based on Zen 2 Cores. That product has code-name Starship, and would be based on 7nm Finfet. I do doubt that 7nm will be ready in 2018 though, especially with a product die this ginormous. TDPs would run anywhere from 30W up to 185W. Below the leaked slides, courtesy of Videocardz.
AMD Zen Octacore Summit Ridge engineering sample spotted at 3.2/3.5 GHz - 11/25/2016 11:55 AM
They are coming alright, a new Zen octacore Summit Ridge engineering sample has been spotted.The Engineering samples are internal test samples for qualification purposes. The new model that has been...
AMD ZEN 8-core Summit Ridge To launch January 17th 2017 - 11/20/2016 10:53 AM
You should mark that date in your agenda, 17-1-2017 as there now is a very strong indication that the 8-core ZEN processors from AMD is to belaunched. An AMD China partner called Maxsun shared some de...
AMD Zen to get three perf tiers and sits at the €200 marker - 11/17/2016 02:05 PM
An interesting slide surfaced at the Chiphell forum. As you guys know, AMD ZEN processors developed under code-name Summit Ridge (RS) will be released in Q1 of 2017. The initial wave will see 8-core ...
AMD ZEN 8-core Summit Ridge available in February 2017 - 09/12/2016 02:02 PM
As you guys know, the 8-core Summit Ridge processor series from AMD will be the first ZEN based product series released to the desktop consumer market. As it looks right now, Summit Ridge is to be rel...
Intel Kaby Lake and AMD Zen processors to only support Windows 10 and Unix - 09/02/2016 08:12 AM
News hits the web on multiple websites today that it seems likely that both AMD ZEN and Intel's Kaby lake processors are to only support Windows 10 and Unix compatible operating systems. That would b...
jortego128
Senior Member
Posts: 107
Joined: 2017-04-30
Senior Member
Posts: 107
Joined: 2017-04-30
#5431363 Posted on: 05/13/2017 03:22 PM
7nm for next year? I'll hold on my upgrade then!
Oh boy, can't wait!
That is a best case scenario-- I wouldn't hold my breath on that-- its almost entirely dependent on whether Global Foundries and perhaps TSMC have a working, viable 7nm process that is reliable at that point. I'd say the chances of that are probably less than 50%. A lot of things have to go right in a short amount of time to get chips ON THE MARKET in 2018.
Remember Vega? It has long been supposed to be on the market based on original product maps and we still have nothing.
Even Intel have expressed a lot of concern about lithography at 7nm and below. They cited strange physics issues that begin to arise with silicon transistors around 7nm and below.
7nm for next year? I'll hold on my upgrade then!
Oh boy, can't wait!
That is a best case scenario-- I wouldn't hold my breath on that-- its almost entirely dependent on whether Global Foundries and perhaps TSMC have a working, viable 7nm process that is reliable at that point. I'd say the chances of that are probably less than 50%. A lot of things have to go right in a short amount of time to get chips ON THE MARKET in 2018.
Remember Vega? It has long been supposed to be on the market based on original product maps and we still have nothing.
Even Intel have expressed a lot of concern about lithography at 7nm and below. They cited strange physics issues that begin to arise with silicon transistors around 7nm and below.
Timoo
Junior Member
Posts: 5
Joined: 2016-08-06
Junior Member
Posts: 5
Joined: 2016-08-06
#5431429 Posted on: 05/13/2017 07:35 PM
Why?
16 cores = 4 CCX
32 cores = 8 CCX
48 cores = 12 CCX
-> 64 cores = 16 CCX
Why should they change their CCX layout?
It seems perfectly reasonalbe to me that the consumer-grade top-notch CPU remains 2 CCX, whereas the starter for Professional workstations could take 4 CCX. Starting in priceline with something like 8 or 12 cores (4CCX with full bandwidth and disabled cores).
I don't see why they should change to a CCX of 6 cores.
I wonder if this means each CCX will move from 4 to 6 cores in 2018? The new version of the Ryzen 1800x would be a 12 core part?
Why?
16 cores = 4 CCX
32 cores = 8 CCX
48 cores = 12 CCX
-> 64 cores = 16 CCX
Why should they change their CCX layout?
It seems perfectly reasonalbe to me that the consumer-grade top-notch CPU remains 2 CCX, whereas the starter for Professional workstations could take 4 CCX. Starting in priceline with something like 8 or 12 cores (4CCX with full bandwidth and disabled cores).
I don't see why they should change to a CCX of 6 cores.
jortego128
Senior Member
Posts: 107
Joined: 2017-04-30
Senior Member
Posts: 107
Joined: 2017-04-30
#5431439 Posted on: 05/13/2017 08:04 PM
About GF's 7nm process, from the GF website...
TLDR; Read the last sentence-- sounds like it may actually be on track, but keep in mind the above quote is from Sept 2016.
About GF's 7nm process, from the GF website...
Santa Clara, Calif., September 15, 2016-GLOBALFOUNDRIES today announced plans to deliver a new leading-edge 7nm FinFET semiconductor technology that will offer the ultimate in performance for the next era of computing applications. This technology provides more processing power for data centers, networking, premium mobile processors, and deep learning applications.
GF's new 7nm FinFET technology is expected to deliver more than twice the logic density and a 30 percent performance boost compared to today’s 16/14nm foundry FinFET offerings. The platform is based on an industry-standard FinFET transistor architecture and optical lithography, with EUV compatibility at key levels. This approach will accelerate the production ramp through significant re-use of tools and processes from the company’s 14nm FinFET technology, which is currently in volume production at its Fab 8 campus in Saratoga County, N.Y. GF plans to make an additional mutli-billion dollar investment in Fab 8 to enable development and production for 7nm FinFET.
“The industry is converging on 7nm FinFET as the next long-lived node, which represents a unique opportunity for GF to compete at the leading edge,” said GF CEO Sanjay Jha. “We are well positioned to deliver a differentiated 7nm FinFET technology by tapping our years of experience manufacturing high-performance chips, the talent and know-how of our former IBM Microelectronics colleagues and the world-class R&D pipeline from our research alliance. No other foundry can match this legacy of manufacturing high-performance chips.”
“GF made a bold decision to jump directly from 14nm to 7nm--a decision that is now supported by several leading semiconductor companies as they see only marginal performance and power benefits for the high cost of the 10nm process node,” said Jim McGregor, founder and principal analyst at TIRIAS Research. “Much like the 28nm and 16/14nm process nodes, 7nm appears to be the next major process node that will be widely leveraged by the entire semiconductor industry for at least the next decade.”
“Leading-edge technologies like GF 7nm FinFET are an important part of how we deliver our long-term roadmap of computing and graphics products that are capable of powering the next generation of computing experiences,” said Dr. Lisa Su, president and CEO, AMD. “We look forward to continuing our close collaboration with GF as they extend the solid execution and technology foundation they are building at 14nm to deploy high-performance, low-power 7nm technology in the coming years.”
“IBM is committed to pushing the limits of semiconductor technology as part of its aggressive long term research agenda,” said Arvind Krishna, senior vice president and director of IBM Research. “IBM Research continues to collaborate with GF in developing new ideas, new skills and new technologies that will help accelerate our joint research in 7nm technology and beyond."
GF will deliver a comprehensive and competitive IP library, co-optimized with process development. To enable customers to accelerate adoption of 7nm FinFET technology, GF has expanded its strategic partnership with INVECAS beyond 14LPP and FDX™ processes to now include foundry IP development for 7nm process technologies. This will provide customers with a strong foundation to build early designs that meet their performance, power and area requirements.
“INVECAS specializes in providing unrivaled IP solutions, ASIC and design services to GF's customers that span the wide-range of GF's leading edge FinFET and FDX processes," said Dasaradha Gude, CEO, INVECAS. "Our strategic partnership with GF combined with our tailor-made foundry IP model allows us to develop a 7nm FinFET process foundation IP that meets the challenging performance requirements of 7nm customers’ leading-edge applications."
Building on the success of its 14LPP technology platform, GF's 7nm FinFET technology is positioned to enable next-generation computing applications that demand ultra-high performance, from high-end mobile SoCs to processors for cloud servers and networking infrastructure. The company’s high-performance offerings are complemented by its 22FDX® and 12FDX™ technologies, which have been developed to meet the ultra-low-power requirements of the next generation of intelligent connected devices, from mobile computing and 5G connectivity to artificial intelligence and autonomous vehicles.
GF's 7nm FinFET technology will be supported by a full platform of foundation and complex intellectual property (IP), including an application-specific integrated circuit (ASIC) offering. Test chips with IP from lead customers have already started running in Fab 8. The technology is expected to be ready for customer product design starts in the second half of 2017, with ramp to risk production in early 2018.
GF's new 7nm FinFET technology is expected to deliver more than twice the logic density and a 30 percent performance boost compared to today’s 16/14nm foundry FinFET offerings. The platform is based on an industry-standard FinFET transistor architecture and optical lithography, with EUV compatibility at key levels. This approach will accelerate the production ramp through significant re-use of tools and processes from the company’s 14nm FinFET technology, which is currently in volume production at its Fab 8 campus in Saratoga County, N.Y. GF plans to make an additional mutli-billion dollar investment in Fab 8 to enable development and production for 7nm FinFET.
“The industry is converging on 7nm FinFET as the next long-lived node, which represents a unique opportunity for GF to compete at the leading edge,” said GF CEO Sanjay Jha. “We are well positioned to deliver a differentiated 7nm FinFET technology by tapping our years of experience manufacturing high-performance chips, the talent and know-how of our former IBM Microelectronics colleagues and the world-class R&D pipeline from our research alliance. No other foundry can match this legacy of manufacturing high-performance chips.”
“GF made a bold decision to jump directly from 14nm to 7nm--a decision that is now supported by several leading semiconductor companies as they see only marginal performance and power benefits for the high cost of the 10nm process node,” said Jim McGregor, founder and principal analyst at TIRIAS Research. “Much like the 28nm and 16/14nm process nodes, 7nm appears to be the next major process node that will be widely leveraged by the entire semiconductor industry for at least the next decade.”
“Leading-edge technologies like GF 7nm FinFET are an important part of how we deliver our long-term roadmap of computing and graphics products that are capable of powering the next generation of computing experiences,” said Dr. Lisa Su, president and CEO, AMD. “We look forward to continuing our close collaboration with GF as they extend the solid execution and technology foundation they are building at 14nm to deploy high-performance, low-power 7nm technology in the coming years.”
“IBM is committed to pushing the limits of semiconductor technology as part of its aggressive long term research agenda,” said Arvind Krishna, senior vice president and director of IBM Research. “IBM Research continues to collaborate with GF in developing new ideas, new skills and new technologies that will help accelerate our joint research in 7nm technology and beyond."
GF will deliver a comprehensive and competitive IP library, co-optimized with process development. To enable customers to accelerate adoption of 7nm FinFET technology, GF has expanded its strategic partnership with INVECAS beyond 14LPP and FDX™ processes to now include foundry IP development for 7nm process technologies. This will provide customers with a strong foundation to build early designs that meet their performance, power and area requirements.
“INVECAS specializes in providing unrivaled IP solutions, ASIC and design services to GF's customers that span the wide-range of GF's leading edge FinFET and FDX processes," said Dasaradha Gude, CEO, INVECAS. "Our strategic partnership with GF combined with our tailor-made foundry IP model allows us to develop a 7nm FinFET process foundation IP that meets the challenging performance requirements of 7nm customers’ leading-edge applications."
Building on the success of its 14LPP technology platform, GF's 7nm FinFET technology is positioned to enable next-generation computing applications that demand ultra-high performance, from high-end mobile SoCs to processors for cloud servers and networking infrastructure. The company’s high-performance offerings are complemented by its 22FDX® and 12FDX™ technologies, which have been developed to meet the ultra-low-power requirements of the next generation of intelligent connected devices, from mobile computing and 5G connectivity to artificial intelligence and autonomous vehicles.
GF's 7nm FinFET technology will be supported by a full platform of foundation and complex intellectual property (IP), including an application-specific integrated circuit (ASIC) offering. Test chips with IP from lead customers have already started running in Fab 8. The technology is expected to be ready for customer product design starts in the second half of 2017, with ramp to risk production in early 2018.
TLDR; Read the last sentence-- sounds like it may actually be on track, but keep in mind the above quote is from Sept 2016.
bobdude
Senior Member
Posts: 1949
Joined: 2006-02-23
Senior Member
Posts: 1949
Joined: 2006-02-23
#5431456 Posted on: 05/13/2017 10:06 PM
cool then maybe microsoft can make a windows that will use 4 cores
cool then maybe microsoft can make a windows that will use 4 cores
Click here to post a comment for this news story on the message forum.
Member
Posts: 98
Joined: 2016-03-30
Proof of Zens scalability. it's on FAHR!!! Lower core count? "ain't nobody got time for that"