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

[juanrga]

4,56 Ghz
https://www.pcgamesn.com/amd-raven-ridge-overclocking

It is the older Ryzen bug again

https://www.pcgamesn.com/amd-raven-ridge-overclocking

Those overclocks aren't real.

Raven Ridge uses TIM. The Stilt delidded the APU and replaced with liquid metal. Temperatures dropped 12 ºC, but maximum overclock continued being 4.0GHz

https://forums.anandtech.com/threads/amd-ryzen-5-2400g-and-ryzen-3-2200g-apus-performance-unveiled.2533111/page-59#post-39302597

So, process limited I take it?

I think he got a bad chip. Others have pushed the chip to 4.1GHz and 4.2GHz. AMD also demoed 4.2GHz overclock

https://www.reddit.com/r/Amd/comments/7s56pd/the_upcoming_ryzen_apus_seem_to_already_be_the/dt2rseq/

So I think the new 14nm+ process (aka '12nm') is giving about 200MHz extra. It also agrees with this leak about Pinnacle Ridge showing a qualification sample with 200MHz above the Summit Ridge akin.

https://hothardware.com/news/amd-ryzen-5-2600-12nm-zen-cpu-asus-crosshair-vii-hero-x470-motherboard

14nm+ shouldn't be used as synonymous of 12 nm, unless you want to introduce even more confusion. Apparently there is an interim 14nm+ process in between the original 14nm and the 12nm used for the upcoming zen+. At least that is what toms reported.

According to AMD, its 14nm+ process is denser and more power-efficient than the 14nm node it was using previously. However, the company isn't sharing much beyond those claims. To be clear, this is not the GlobalFoundries 12nm LP process that AMD will transition to in April when the Zen+ processors are expected to launch. That new process will provide even more of a performance boost over the current 14nm+ LPP FinFET.

That may explain why it is easier to get in the 4.0 4.2 overclock range with raven ridge.

12nm is a marketing rebrand of 14nm+. As openly admitted by Glofo chief that 12nm is not a shrink but only an extension of 14nm.

 

[8350rocks]

4,56 Ghz
https://www.pcgamesn.com/amd-raven-ridge-overclocking

It is the older Ryzen bug again

https://www.pcgamesn.com/amd-raven-ridge-overclocking

Those overclocks aren't real.

Raven Ridge uses TIM. The Stilt delidded the APU and replaced with liquid metal. Temperatures dropped 12 ºC, but maximum overclock continued being 4.0GHz

https://forums.anandtech.com/threads/amd-ryzen-5-2400g-and-ryzen-3-2200g-apus-performance-unveiled.2533111/page-59#post-39302597

So, process limited I take it?

I think he got a bad chip. Others have pushed the chip to 4.1GHz and 4.2GHz. AMD also demoed 4.2GHz overclock

https://www.reddit.com/r/Amd/comments/7s56pd/the_upcoming_ryzen_apus_seem_to_already_be_the/dt2rseq/

So I think the new 14nm+ process (aka '12nm') is giving about 200MHz extra. It also agrees with this leak about Pinnacle Ridge showing a qualification sample with 200MHz above the Summit Ridge akin.

https://hothardware.com/news/amd-ryzen-5-2600-12nm-zen-cpu-asus-crosshair-vii-hero-x470-motherboard

14nm+ shouldn't be used as synonymous of 12 nm, unless you want to introduce even more confusion. Apparently there is an interim 14nm+ process in between the original 14nm and the 12nm used for the upcoming zen+. At least that is what toms reported.

According to AMD, its 14nm+ process is denser and more power-efficient than the 14nm node it was using previously. However, the company isn't sharing much beyond those claims. To be clear, this is not the GlobalFoundries 12nm LP process that AMD will transition to in April when the Zen+ processors are expected to launch. That new process will provide even more of a performance boost over the current 14nm+ LPP FinFET.

That may explain why it is easier to get in the 4.0 4.2 overclock range with raven ridge.

12nm is a marketing rebrand of 14nm+. As openly admitted by Glofo chief that 12nm is not a shrink but only an extension of 14nm.

Source? As I understand it, 12nm has tighter FEOL than 14nm+, though BEOL is mostly the same.

 

[aldaia]

12nm is a marketing rebrand of 14nm+. As openly admitted by Glofo chief that 12nm is not a shrink but only an extension of 14nm.

That is not what Paul Alcorn says
http://www.tomshardware.com/reviews/amd-ryzen-5-2400g-zen-vega-cpu-gpu,5467-3.html

According to AMD, its 14nm+ process is denser and more power-efficient than the 14nm node it was using previously. However, the company isn't sharing much beyond those claims. To be clear, this is not the GlobalFoundries 12nm LP process that AMD will transition to in April when the Zen+ processors are expected to launch. That new process will provide even more of a performance boost over the current 14nm+ LPP FinFET.

 

[goldstone77]

There is no doubt that the new Ryzen 5 2400G and Ryzen 3 2200G offer remarkable gaming performance for budget segments. I think we have to consider the price of 16 GB of RAM, now costing more than the processor, pushing these budget builds into the $500 range we should consider consoles like the Xbox 1X for purely gaming, which can offer playable resolutions up to 4k. That said computers still offer a wide array of functionality with business applications. I think it shows we still have a ways to go before consoles are completely edged out in favor of computers on a gaming price to performance ratio.

Really?? You think "it shows we still have a ways to go before consoles are completely edged".. If you try to build a PC equivalent to the XBOX ONE X you will end up spending more than twice the $495 USD price of the XBOX. The graphics card equivalent is the RX 580 or GTX 1060 and those retail for around $400.. The XBOX is a steal at that price. Building a budget PC for just gaming is a nonsense.
To match the XBOX performance you need:
1TB HDD
4K, HDR BLU-RAY, DVD Drive
RX 580 or GTX 1060
12GB RAM
Motherboard
Processor
A PSU to power all that
WINDOWS 10
Bluetooth 4.0
A Keyboard and Mouse.

Try to buy all that for $495. Sure we all love PC, if you want raw power at the end nothing beats PC, but trying to beat the XBOX ONE X in the budget section won't happen, is not even close.

That's exactly what I mean by that statement, if you take everything I said into context. You would be hard pressed to not to buy consoles vs. computer hardware for budget builds. I was just adding the business/productivity applications of a computer add greater functionality.

 

[pangolin_user]

4,56 Ghz
https://www.pcgamesn.com/amd-raven-ridge-overclocking

It is the older Ryzen bug again

https://www.pcgamesn.com/amd-raven-ridge-overclocking

Those overclocks aren't real

Nope
Initially I assumed it was a mistake. Pre-release platforms often display weird results in monitoring apps - part of the fun of putting together launch day reviews - so I figured there was nothing to it. But after testing and retesting it became obvious the overclock had stuck and this mighty chip was overclocking like a hero.

It's potentially down to the C-state settings in the BIOS I've disabled due to some issues I had getting 3DMark to run on the AMD test platform at the beginning. It's also quite possible it's the old Ryzen sleep timer bug appearing again.


But it’s completely repeatable. Every time I reboot and drop it into sleepy time mode for a heartbeat the BCLK setting pushes itself up to a heady 112.50MHz. With the x40.5 multiplier I had in place that meant it was sitting pretty at 4.56GHz when it woke up.

At that speed the performance numbers are incredible. The 2400G hits around 1,000 and 187 for Cinebench's multi and single-threaded tests, making the $100 more expensive Intel Core i5 8600K look a little foolish. And, with a healthy 1.5GHz clockspeed on the Vega 11 GPU, the gaming performance gets mighty playable at the top 1080p game settings. You do need some speedy, pricey DDR4 memory to get the most out of the graphics cores - this Vega chip has no HBM2 to call its own - so that does affect the overall platform costs.

But it's also possible to use the overclock with a discrete GPU in place too. That gives it a heroic level of graphics support from such a budget slice of silicon.

Btw Juan, x370 chipset support ryzen 2xxx

 

[goldstone77]

4,56 Ghz
https://www.pcgamesn.com/amd-raven-ridge-overclocking

It is the older Ryzen bug again

https://www.pcgamesn.com/amd-raven-ridge-overclocking

Those overclocks aren't real

Nope
Initially I assumed it was a mistake. Pre-release platforms often display weird results in monitoring apps - part of the fun of putting together launch day reviews - so I figured there was nothing to it. But after testing and retesting it became obvious the overclock had stuck and this mighty chip was overclocking like a hero.

It's potentially down to the C-state settings in the BIOS I've disabled due to some issues I had getting 3DMark to run on the AMD test platform at the beginning. It's also quite possible it's the old Ryzen sleep timer bug appearing again.


But it’s completely repeatable. Every time I reboot and drop it into sleepy time mode for a heartbeat the BCLK setting pushes itself up to a heady 112.50MHz. With the x40.5 multiplier I had in place that meant it was sitting pretty at 4.56GHz when it woke up.

At that speed the performance numbers are incredible. The 2400G hits around 1,000 and 187 for Cinebench's multi and single-threaded tests, making the $100 more expensive Intel Core i5 8600K look a little foolish. And, with a healthy 1.5GHz clockspeed on the Vega 11 GPU, the gaming performance gets mighty playable at the top 1080p game settings. You do need some speedy, pricey DDR4 memory to get the most out of the graphics cores - this Vega chip has no HBM2 to call its own - so that does affect the overall platform costs.

But it's also possible to use the overclock with a discrete GPU in place too. That gives it a heroic level of graphics support from such a budget slice of silicon.

Btw Juan, x370 chipset support ryzen 2xxx

Go back to the website and read the first line of the update:
Update February 13 2018: Turns out that it is a resurrection of the sleep timer bug that afflicted the original Ryzen CPUs. Dave has a sad face.

Check out the full review of the AMD Ryzen 5 2400G.

Yes, in all the excitement over the new Raven Ridge APUs I was fooled by Windows' timer settings going screwy when it went to sleep on the MSI board with the Ryzen 5 2400G. Windows speeds up its internal clock by 12.5% and thus every single benchmark then sees all workloads completing faster than before.

The bug didn't materialise when I tested the Ryzen 5 1400, so the timer issue isn't occurring with the original Ryzen processors, just the new Raven Ridge chips. So it is something to look out for.

We've tested with the most up-to-date BIOS we've been supplied for the MSI board and it still happens. I've had misreporting hardware before in pre-release reviews, but they've never affected the benchmarking results, especially not across the board. 3DMark is also meant to catch the timer issue, so when it came back as a valid score I assumed that's what it was. The higher temperatures and higher power draw also seemed consistent with an overclocked CPU.

Ah well, the Ryzen 5 2400G is still a fantastic little APU, with truly impressive low-end gaming performance without needing a discrete GPU. Even if it isn't quite such a hefty, sleepy overclocker.

Original story, February 12 2018:

 

[goldstone77]

12nm is a marketing rebrand of 14nm+. As openly admitted by Glofo chief that 12nm is not a shrink but only an extension of 14nm.

That is not what Paul Alcorn says
http://www.tomshardware.com/reviews/amd-ryzen-5-2400g-zen-vega-cpu-gpu,5467-3.html

According to AMD, its 14nm+ process is denser and more power-efficient than the 14nm node it was using previously. However, the company isn't sharing much beyond those claims. To be clear, this is not the GlobalFoundries 12nm LP process that AMD will transition to in April when the Zen+ processors are expected to launch. That new process will provide even more of a performance boost over the current 14nm+ LPP FinFET.

The 15% area reduction shows there was a change in density, so it's not truly just a 14+ revision. That said, the revisions to Ryzen are so few that 14+ isn't far away from the expected value added by the revisions.

 

[pangolin_user]

4,56 Ghz
https://www.pcgamesn.com/amd-raven-ridge-overclocking

It is the older Ryzen bug again

https://www.pcgamesn.com/amd-raven-ridge-overclocking

Those overclocks aren't real

Nope
Initially I assumed it was a mistake. Pre-release platforms often display weird results in monitoring apps - part of the fun of putting together launch day reviews - so I figured there was nothing to it. But after testing and retesting it became obvious the overclock had stuck and this mighty chip was overclocking like a hero.

It's potentially down to the C-state settings in the BIOS I've disabled due to some issues I had getting 3DMark to run on the AMD test platform at the beginning. It's also quite possible it's the old Ryzen sleep timer bug appearing again.


But it’s completely repeatable. Every time I reboot and drop it into sleepy time mode for a heartbeat the BCLK setting pushes itself up to a heady 112.50MHz. With the x40.5 multiplier I had in place that meant it was sitting pretty at 4.56GHz when it woke up.

At that speed the performance numbers are incredible. The 2400G hits around 1,000 and 187 for Cinebench's multi and single-threaded tests, making the $100 more expensive Intel Core i5 8600K look a little foolish. And, with a healthy 1.5GHz clockspeed on the Vega 11 GPU, the gaming performance gets mighty playable at the top 1080p game settings. You do need some speedy, pricey DDR4 memory to get the most out of the graphics cores - this Vega chip has no HBM2 to call its own - so that does affect the overall platform costs.

But it's also possible to use the overclock with a discrete GPU in place too. That gives it a heroic level of graphics support from such a budget slice of silicon.

Btw Juan, x370 chipset support ryzen 2xxx

Go back to the website and read the first line of the update:
Update February 13 2018: Turns out that it is a resurrection of the sleep timer bug that afflicted the original Ryzen CPUs. Dave has a sad face.

Check out the full review of the AMD Ryzen 5 2400G.

Yes, in all the excitement over the new Raven Ridge APUs I was fooled by Windows' timer settings going screwy when it went to sleep on the MSI board with the Ryzen 5 2400G. Windows speeds up its internal clock by 12.5% and thus every single benchmark then sees all workloads completing faster than before.

The bug didn't materialise when I tested the Ryzen 5 1400, so the timer issue isn't occurring with the original Ryzen processors, just the new Raven Ridge chips. So it is something to look out for.

We've tested with the most up-to-date BIOS we've been supplied for the MSI board and it still happens. I've had misreporting hardware before in pre-release reviews, but they've never affected the benchmarking results, especially not across the board. 3DMark is also meant to catch the timer issue, so when it came back as a valid score I assumed that's what it was. The higher temperatures and higher power draw also seemed consistent with an overclocked CPU.

Ah well, the Ryzen 5 2400G is still a fantastic little APU, with truly impressive low-end gaming performance without needing a discrete GPU. Even if it isn't quite such a hefty, sleepy overclocker.

Original story, February 12 2018:

Ok, thanks

 

[goldstone77]

https://twitter.com/KeithPlaysPC/status/963611815237378050

Not too bad for a $99 CPU.

 

[jdwii]

4,56 Ghz
https://www.pcgamesn.com/amd-raven-ridge-overclocking

It is the older Ryzen bug again

https://www.pcgamesn.com/amd-raven-ridge-overclocking

Those overclocks aren't real

Nope
Initially I assumed it was a mistake. Pre-release platforms often display weird results in monitoring apps - part of the fun of putting together launch day reviews - so I figured there was nothing to it. But after testing and retesting it became obvious the overclock had stuck and this mighty chip was overclocking like a hero.

It's potentially down to the C-state settings in the BIOS I've disabled due to some issues I had getting 3DMark to run on the AMD test platform at the beginning. It's also quite possible it's the old Ryzen sleep timer bug appearing again.


But it’s completely repeatable. Every time I reboot and drop it into sleepy time mode for a heartbeat the BCLK setting pushes itself up to a heady 112.50MHz. With the x40.5 multiplier I had in place that meant it was sitting pretty at 4.56GHz when it woke up.

At that speed the performance numbers are incredible. The 2400G hits around 1,000 and 187 for Cinebench's multi and single-threaded tests, making the $100 more expensive Intel Core i5 8600K look a little foolish. And, with a healthy 1.5GHz clockspeed on the Vega 11 GPU, the gaming performance gets mighty playable at the top 1080p game settings. You do need some speedy, pricey DDR4 memory to get the most out of the graphics cores - this Vega chip has no HBM2 to call its own - so that does affect the overall platform costs.

But it's also possible to use the overclock with a discrete GPU in place too. That gives it a heroic level of graphics support from such a budget slice of silicon.

Btw Juan, x370 chipset support ryzen 2xxx

We all knew it would manufactures said it would Amd said it would and guess what i bet most x370 boards will work with Zen 2 and Zen 3 just like Amd did in the past with AM2,AM2+,AM3,AM3+. A Phenom II X6 works in the same board a Athlon X2 4800 works in despite those CPU's being 5 years apart. Amd is known for compatibility i have no doubts that X370 boards will mostly work with upcoming Zen CPU's.

Obviously there will be features that don't work as well but in most cases users probably won't care much.

 

[juanrga]

4,56 Ghz
https://www.pcgamesn.com/amd-raven-ridge-overclocking

It is the older Ryzen bug again

https://www.pcgamesn.com/amd-raven-ridge-overclocking

Those overclocks aren't real

Nope
Initially I assumed it was a mistake. Pre-release platforms often display weird results in monitoring apps - part of the fun of putting together launch day reviews - so I figured there was nothing to it. But after testing and retesting it became obvious the overclock had stuck and this mighty chip was overclocking like a hero.

It's potentially down to the C-state settings in the BIOS I've disabled due to some issues I had getting 3DMark to run on the AMD test platform at the beginning. It's also quite possible it's the old Ryzen sleep timer bug appearing again.


But it’s completely repeatable. Every time I reboot and drop it into sleepy time mode for a heartbeat the BCLK setting pushes itself up to a heady 112.50MHz. With the x40.5 multiplier I had in place that meant it was sitting pretty at 4.56GHz when it woke up.

At that speed the performance numbers are incredible. The 2400G hits around 1,000 and 187 for Cinebench's multi and single-threaded tests, making the $100 more expensive Intel Core i5 8600K look a little foolish. And, with a healthy 1.5GHz clockspeed on the Vega 11 GPU, the gaming performance gets mighty playable at the top 1080p game settings. You do need some speedy, pricey DDR4 memory to get the most out of the graphics cores - this Vega chip has no HBM2 to call its own - so that does affect the overall platform costs.

But it's also possible to use the overclock with a discrete GPU in place too. That gives it a heroic level of graphics support from such a budget slice of silicon.

Btw Juan, x370 chipset support ryzen 2xxx

Read the update of the article or just the heading "AMD's Raven Ridge isn't such a heroic overclocker, it's suffering from the ol' timer bug". Those 4.56GHz were fake.

Yes, x370 supports Ryzen 2xxx, but as AMD's James commented in one of the links I have given: "The 400 Series motherboards are designed to improve memory stability and overclocking."

So it looks like 400-series mobos will be needed to extract all the potential from the new chips.

 

[juanrga]

If one checks the old roadmaps, all they mentioned a 14nm+ node for this year. Then the node was rebranded to "12nm" and the new roadmaps no longer mention 14nm+.

That 12nm is a rebrand of the formerly 14nm+ is also mentioned by Anandtech.

AMD's CPU and GPU roadmaps from earlier this year both listed a "14nm+" node, which at the time was a curious situation since GlobalFoundries didn't have a node between 14nm and 7nm. The obvious takeaway at the time being that GlobalFoundries was developing an interim node - an enhanced version of 14nm - which today's announcement finally confirms. In which case in accordance with Papermaster's official statement, it looks like we should be expecting both Vega and Zen products produced on 12LP in the next year.

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

There is nothing 12nm in the 12LP node. "12nm" is just a marketing term for an extension of 14nm. This is explained here

https://semiengineering.com/nodes-vs-node-lets/

where Glofo chief admits that 12nm is an extension of 14nm, not a true node shrink. Note he says extension of 14nm, not extension of 14nm+.

Those up to 15% better density and more than 10% performance claims for 12LP are using TSMC 16nm as baseline, not 14LPP

So the performance and density of 12LP will be very close to current 14LPP node.

 

[aldaia]

If one checks the old roadmaps

Time to move to new roadmaps.
https://www.anandtech.com/show/12233/amd-tech-day-at-ces-2018-roadmap-revealed-with-ryzen-apus-zen-on-12nm-vega-on-7nm/8

GF also stated that the 12LP process also offers a 15% area reduction against 14LPP. This is done, according to GlobalFoundries, by using 7.5T libraries rather than 9T libraries. This requires its customers to ‘recompile’ their 14LPP designs for the adjusted 12LP process.

 

[juanrga]

If one checks the old roadmaps

Time to move to new roadmaps.
https://www.anandtech.com/show/12233/amd-tech-day-at-ces-2018-roadmap-revealed-with-ryzen-apus-zen-on-12nm-vega-on-7nm/8

GF also stated that the 12LP process also offers a 15% area reduction against 14LPP. This is done, according to GlobalFoundries, by using 7.5T libraries rather than 9T libraries. This requires its customers to ‘recompile’ their 14LPP designs for the adjusted 12LP process.

From the link:

Most of the physical changes happen due to the process. Last year, AMD cited that its second generation Ryzen products will be on a ‘14+’ manufacturing node, but now it is going to be on a 12LP process. Some people argue that this is a clever marketing ploy as a name change, given nothing in the designs actually comes close to 14nm or 12nm. We were told that 12LP implements stricter design rules for the manufacturing process, requiring changes to library layout and the floorplan, but with an end result of a focus on performance per watt.

Just as I said above, 14nm+ was renamed to 12nm in the new roadmaps. It is a marketing rebrand. Just as Zen+ is a marketing rebrand for Zen.

Also that "15% area reduction" is over TSMC 16nm. Anandtech claims that Papermaster said "over 14LPP", but I gave the slide above and it says over "industry 16nm FINFET solutions". I give it again

Anandtech's authors are who are guessing a 14LPP baseline: "The company does not elaborate which process it used for comparison, but a naturally guess would be its own 14LPP which the company knows well." Glofo didn't mention 14LPP in any part so far as I know. What is more, this 'new' '12nm' node is not even mentioned in Glofo website. They only mention 14LPP and 7LP.

https://www.globalfoundries.com/technology-solutions/cmos/performance

 

[goldstone77]

Spoiler

If one checks the old roadmaps

Time to move to new roadmaps.
https://www.anandtech.com/show/12233/amd-tech-day-at-ces-2018-roadmap-revealed-with-ryzen-apus-zen-on-12nm-vega-on-7nm/8

GF also stated that the 12LP process also offers a 15% area reduction against 14LPP. This is done, according to GlobalFoundries, by using 7.5T libraries rather than 9T libraries. This requires its customers to ‘recompile’ their 14LPP designs for the adjusted 12LP process.

From the link:

Most of the physical changes happen due to the process. Last year, AMD cited that its second generation Ryzen products will be on a ‘14+’ manufacturing node, but now it is going to be on a 12LP process. Some people argue that this is a clever marketing ploy as a name change, given nothing in the designs actually comes close to 14nm or 12nm. We were told that 12LP implements stricter design rules for the manufacturing process, requiring changes to library layout and the floorplan, but with an end result of a focus on performance per watt.

Just as I said above, 14nm+ was renamed to 12nm in the new roadmaps. It is a marketing rebrand. Just as Zen+ is a marketing rebrand for Zen.

Also that "15% area reduction" is over TSMC 16nm. Anandtech claims that Papermaster said "over 14LPP", but I gave the slide above and it says over "industry 16nm FINFET solutions". I give it again

Anandtech's authors are who are guessing a 14LPP baseline: "The company does not elaborate which process it used for comparison, but a naturally guess would be its own 14LPP which the company knows well." Glofo didn't mention 14LPP in any part so far as I know. What is more, this 'new' '12nm' node is not even mentioned in Glofo website. They only mention 14LPP and 7LP.

https://www.globalfoundries.com/technology-solutions/cmos/performance

This is done, according to GlobalFoundries, by using 7.5T libraries rather than 9T libraries.

This would be a reduction in logic density, because 9T libraries are larger than 7.5T libraries. Without seeing the numbers their is no way to know for sure if there was any real node shrink. And GloFlo's 14nm transistors more resembles 20nm transistors, so the area reduction from "16nm" could be smaller than current "14nm". But technically 12nm, 14nm, and 16nm are all just branding name schemes anyway and not true nodes. IMO It's not worth arguing over such trivial details when everyone makes up their own node branding and redefines what constitutes density as it suits them. If AMD wants to call it 12nm let's just call it that. We called 14nm 14nm.

 

[jdwii]

4,56 Ghz
https://www.pcgamesn.com/amd-raven-ridge-overclocking

It is the older Ryzen bug again

https://www.pcgamesn.com/amd-raven-ridge-overclocking

Those overclocks aren't real

Nope
Initially I assumed it was a mistake. Pre-release platforms often display weird results in monitoring apps - part of the fun of putting together launch day reviews - so I figured there was nothing to it. But after testing and retesting it became obvious the overclock had stuck and this mighty chip was overclocking like a hero.

It's potentially down to the C-state settings in the BIOS I've disabled due to some issues I had getting 3DMark to run on the AMD test platform at the beginning. It's also quite possible it's the old Ryzen sleep timer bug appearing again.


But it’s completely repeatable. Every time I reboot and drop it into sleepy time mode for a heartbeat the BCLK setting pushes itself up to a heady 112.50MHz. With the x40.5 multiplier I had in place that meant it was sitting pretty at 4.56GHz when it woke up.

At that speed the performance numbers are incredible. The 2400G hits around 1,000 and 187 for Cinebench's multi and single-threaded tests, making the $100 more expensive Intel Core i5 8600K look a little foolish. And, with a healthy 1.5GHz clockspeed on the Vega 11 GPU, the gaming performance gets mighty playable at the top 1080p game settings. You do need some speedy, pricey DDR4 memory to get the most out of the graphics cores - this Vega chip has no HBM2 to call its own - so that does affect the overall platform costs.

But it's also possible to use the overclock with a discrete GPU in place too. That gives it a heroic level of graphics support from such a budget slice of silicon.

Btw Juan, x370 chipset support ryzen 2xxx

Read the update of the article or just the heading "AMD's Raven Ridge isn't such a heroic overclocker, it's suffering from the ol' timer bug". Those 4.56GHz were fake.

Yes, x370 supports Ryzen 2xxx, but as AMD's James commented in one of the links I have given: "The 400 Series motherboards are designed to improve memory stability and overclocking."

So it looks like 400-series mobos will be needed to extract all the potential from the new chips.

I somewhat think they are trying to hype Pinnacle Ridge and newer boards as AGESA 1.0.0.0a or whatever manufactures call it has improved memory compatibility for some including me with current Ryzen CPU's. Here is a Reddit thread of the discussion. I read other statements from different forums as well. In my case i can now run 3200mhz at default timings without having to touch CPU voltage or SOC voltage before it would crash or be unstable. Other's are getting 3600mhz from b-die memory and more are getting hynx chips to 3200mhz. I'm sure Amd did more improvements with pinnacle ridge itself to get higher memory support, time will tell.

https://www.reddit.com/r/Amd/comments/7vmi6e/all_msi_am4_motherboard_bios_updated_agesa_code/

 

[8350rocks]

If one checks the old roadmaps, all they mentioned a 14nm+ node for this year. Then the node was rebranded to "12nm" and the new roadmaps no longer mention 14nm+.

That 12nm is a rebrand of the formerly 14nm+ is also mentioned by Anandtech.

AMD's CPU and GPU roadmaps from earlier this year both listed a "14nm+" node, which at the time was a curious situation since GlobalFoundries didn't have a node between 14nm and 7nm. The obvious takeaway at the time being that GlobalFoundries was developing an interim node - an enhanced version of 14nm - which today's announcement finally confirms. In which case in accordance with Papermaster's official statement, it looks like we should be expecting both Vega and Zen products produced on 12LP in the next year.

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

There is nothing 12nm in the 12LP node. "12nm" is just a marketing term for an extension of 14nm. This is explained here

https://semiengineering.com/nodes-vs-node-lets/

where Glofo chief admits that 12nm is an extension of 14nm, not a true node shrink. Note he says extension of 14nm, not extension of 14nm+.

Those up to 15% better density and more than 10% performance claims for 12LP are using TSMC 16nm as baseline, not 14LPP

So the performance and density of 12LP will be very close to current 14LPP node.

15% area reduction = tighter FEOL. As I said...it is not a marketing rebrand.

 

[8350rocks]

If one checks the old roadmaps

Time to move to new roadmaps.
https://www.anandtech.com/show/12233/amd-tech-day-at-ces-2018-roadmap-revealed-with-ryzen-apus-zen-on-12nm-vega-on-7nm/8

GF also stated that the 12LP process also offers a 15% area reduction against 14LPP. This is done, according to GlobalFoundries, by using 7.5T libraries rather than 9T libraries. This requires its customers to ‘recompile’ their 14LPP designs for the adjusted 12LP process.

From the link:

Most of the physical changes happen due to the process. Last year, AMD cited that its second generation Ryzen products will be on a ‘14+’ manufacturing node, but now it is going to be on a 12LP process. Some people argue that this is a clever marketing ploy as a name change, given nothing in the designs actually comes close to 14nm or 12nm. We were told that 12LP implements stricter design rules for the manufacturing process, requiring changes to library layout and the floorplan, but with an end result of a focus on performance per watt.

Just as I said above, 14nm+ was renamed to 12nm in the new roadmaps. It is a marketing rebrand. Just as Zen+ is a marketing rebrand for Zen.

Also that "15% area reduction" is over TSMC 16nm. Anandtech claims that Papermaster said "over 14LPP", but I gave the slide above and it says over "industry 16nm FINFET solutions". I give it again

Anandtech's authors are who are guessing a 14LPP baseline: "The company does not elaborate which process it used for comparison, but a naturally guess would be its own 14LPP which the company knows well." Glofo didn't mention 14LPP in any part so far as I know. What is more, this 'new' '12nm' node is not even mentioned in Glofo website. They only mention 14LPP and 7LP.

https://www.globalfoundries.com/technology-solutions/cmos/performance

Could they be referring to the 12FDX node on GloFo's page?

https://www.globalfoundries.com/news-events/press-releases/globalfoundries-extends-fdxtm-roadmap-12nm-fd-soi-technology

Perhaps a hybrid substrate without using silicon on insulator?

 

[goldstone77]

2400G Delid FAIL | Single Channel Vs. Dual Channel DDR4 Memory for Ryzen OCed 2200G & 2400G APUs
Tech YES City
Published on Feb 14, 2018
[video="https://www.youtube.com/watch?v=a-WMpcTVykc&feature=youtu.be"][/video]
https://www.youtube.com/watch?v=a-WMpcTVykc&feature=youtu.be

Here are the results for the Single Channel DDR4 Vs. Dual Channel for Ryzen's New APUs, the 2400 and 2200G, the results are well... you will definitely want to stick with dual channel... Overclocking was really good on the 2200g to the point where it matched performance of the 2400g, though before I could properly test the 2400g overclocked... I ran into some problems - the delid failed... I did manage to get that 1350mhz however and 3.95 on the cpu core... (in 30 degree ambient temps c):

 

[aldaia]

Regarding GF 12nm and 7nm processes. Extremetech recently visited a GF plant in New York, and got some 1st hand information. https://www.extremetech.com/computing/263286-sitting-globalfoundries-talk-7nm-euv

Last week, we had the opportunity to tour GlobalFoundries’ Fab 8, in Malta, N.Y., and to sit down and talk about how the company’s roadmaps and timelines look through 2018 and beyond.

It is an interesting read, they mostly talk about EUV, and 7nm, but there is also some info about 12nm.

GF is working on introducing EUV at 7nm, but it’s only going to see limited use for contacts and vias, not as part of the critical path.

GlobalFoundries is skipping 10nm altogether and heading straight for 7nm, with an AMD Vega chip designed for machine intelligence workloads apparently serving as a so-called “pipe cleaner” to test the design and its capabilities.

Join that with AMD tech day at CES and it seems GF 7nm will really start risk production in 2018.

Also: For those of you curious about 12nm and how it compares with 14nm, GF confirmed its 12nm is a refinement and improvement to an existing process with some optimized layouts that offer up to 10 percent performance improvement or a 15 percent density increase.

The interesting part here is that the improvement in speed and density is respect to their own 14nm process. However, contrary to what previous slides state it's not 10% performance AND 15% density, but 10% performance OR 15% density. It remains to be seen if AMD will priorize performance or density for Zen+

 

[juanrga]

Yes, x370 supports Ryzen 2xxx, but as AMD's James commented in one of the links I have given: "The 400 Series motherboards are designed to improve memory stability and overclocking."

So it looks like 400-series mobos will be needed to extract all the potential from the new chips.

I somewhat think they are trying to hype Pinnacle Ridge and newer boards as AGESA 1.0.0.0a or whatever manufactures call it has improved memory compatibility for some including me with current Ryzen CPU's. Here is a Reddit thread of the discussion. I read other statements from different forums as well. In my case i can now run 3200mhz at default timings without having to touch CPU voltage or SOC voltage before it would crash or be unstable. Other's are getting 3600mhz from b-die memory and more are getting hynx chips to 3200mhz. I'm sure Amd did more improvements with pinnacle ridge itself to get higher memory support, time will tell.

https://www.reddit.com/r/Amd/comments/7vmi6e/all_msi_am4_motherboard_bios_updated_agesa_code/

I think the new board have been designed to run above the former spec.

 

[juanrga]

If one checks the old roadmaps, all they mentioned a 14nm+ node for this year. Then the node was rebranded to "12nm" and the new roadmaps no longer mention 14nm+.

That 12nm is a rebrand of the formerly 14nm+ is also mentioned by Anandtech.

AMD's CPU and GPU roadmaps from earlier this year both listed a "14nm+" node, which at the time was a curious situation since GlobalFoundries didn't have a node between 14nm and 7nm. The obvious takeaway at the time being that GlobalFoundries was developing an interim node - an enhanced version of 14nm - which today's announcement finally confirms. In which case in accordance with Papermaster's official statement, it looks like we should be expecting both Vega and Zen products produced on 12LP in the next year.

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

There is nothing 12nm in the 12LP node. "12nm" is just a marketing term for an extension of 14nm. This is explained here

https://semiengineering.com/nodes-vs-node-lets/

where Glofo chief admits that 12nm is an extension of 14nm, not a true node shrink. Note he says extension of 14nm, not extension of 14nm+.

Those up to 15% better density and more than 10% performance claims for 12LP are using TSMC 16nm as baseline, not 14LPP

So the performance and density of 12LP will be very close to current 14LPP node.

15% area reduction = tighter FEOL. As I said...it is not a marketing rebrand.

Up to 15% area reduction over TMSC 16nm.

The claim is that 12LP is a marketing rebrand of 14nm+, not of 16nm.

Could they be referring to the 12FDX node on GloFo's page?

https://www.globalfoundries.com/news-events/press-releases/globalfoundries-extends-fdxtm-roadmap-12nm-fd-soi-technology

Perhaps a hybrid substrate without using silicon on insulator?

For 12FDX, Glofo uses 22FDX as baseline. For 12LP Glofo uses TSMC 16nm as baseline or equivalent node.

 

[juanrga]

Regarding GF 12nm and 7nm processes. Extremetech recently visited a GF plant in New York, and got some 1st hand information. https://www.extremetech.com/computing/263286-sitting-globalfoundries-talk-7nm-euv

Last week, we had the opportunity to tour GlobalFoundries’ Fab 8, in Malta, N.Y., and to sit down and talk about how the company’s roadmaps and timelines look through 2018 and beyond.

It is an interesting read, they mostly talk about EUV, and 7nm, but there is also some info about 12nm.

GF is working on introducing EUV at 7nm, but it’s only going to see limited use for contacts and vias, not as part of the critical path.

GlobalFoundries is skipping 10nm altogether and heading straight for 7nm, with an AMD Vega chip designed for machine intelligence workloads apparently serving as a so-called “pipe cleaner” to test the design and its capabilities.

Join that with AMD tech day at CES and it seems GF 7nm will really start risk production in 2018.

Also: For those of you curious about 12nm and how it compares with 14nm, GF confirmed its 12nm is a refinement and improvement to an existing process with some optimized layouts that offer up to 10 percent performance improvement or a 15 percent density increase.

The interesting part here is that the improvement in speed and density is respect to their own 14nm process. However, contrary to what previous slides state it's not 10% performance AND 15% density, but 10% performance OR 15% density. It remains to be seen if AMD will priorize performance or density for Zen+

Sampling is not production. And I am not sure if Vega will use Glofo 7nm process or if it will use TSMC 7nm, which will be ready first. Check this post

The 15% density and 10% performance claims are using 16nm TSMC, as is explicit in the official slide that I have reproduced a dozen of times. There is no document, webpage, quote, or slide from Glofo where they mention 14LPP as baseline. And the above ExtremeTech quote don't mention 14LPP in any part.

Also there is no any change in the marketing description. Foundries marketing talk always use "AND" to mean "OR". For instance the marketing description for 14LPP is

14LPP technology can provide up to 55% higher device performance and 60% lower total power compared to 28nm technologies.

That "and" means OR. One simply has to check the associated transistor graph for the node

Either one obtains up to 55% higher performance maintaining the same power (horizontal line connecting both curves) OR
one obtains up to 60% lower power maintaining the same performance (vertical lines connecting both curve). The same happens with marketing claims about the future 7LP

Either >40% better performance at same power OR >55% power reduction at same performance.

 

[gamerk316]

Either >40% better performance at same power OR >55% power reduction at same performance.

Or some combination of the two. I'd bet something like 30% Power Reduction & 15% performance gain.

 

[goldstone77]

Either >40% better performance at same power OR >55% power reduction at same performance.

Or some combination of the two. I'd bet something like 30% Power Reduction & 15% performance gain.

I agree, if architectural/physical limits apply that would imply there will be a balance into the other direction to the most favorable outcome. Also, remember the marketing image: greater than incremental performance gains vs. competition, or something to that effect.