Discussion AMD Ryzen MegaThread! FAQ and Resources

[juanrga]

Someone sent me this online so i figured i'd post it here

https://videocardz.com/65913/how-fast-is-ryzen

Shows some 4/6 core Ryzen benchmarks


As said before the closer to launch the less salt you need for each leaked benchmark.

I don't understand why they give base clocks for Ryzen and turbo clocks for Intel. Also I don't understand why they give 1-core turbos for a multithreaded bench. The wattage for the 6C Ryzen model is also wrong. This is a 65W chip, not 95W+ as they report in the table.

 

[jwcrellin]

https://www.forbes.com/sites/antonyleather/2017/02/12/amd-ryzen-biggest-leak-yet-amazing-benchmarks-and-prices-starting-at-just-129/#5e2eb2162b5b
Great stuff, turn the intel atm pricing scheme on its head.

 

[Rookie_MIB]

One of the online retailers not only listed pricing, but they also listed this little bit: "expected availability Feb 28th."

 

[010TheMaster010]

http://schedule.gdconf.com/session/optimizing-for-amd-ryzen-cpu-presented-by-amd
There's also this

 

[jaymc]

check this out...

http://www.guru3d.com/news-story/amd-ryzen-4-2-ghz-processor-entry-surfaces-at-taobao-with-28th-feb-availibility.html

 

[jwcrellin]

4.2 turbo boost doesn't seem too far of a reach, the highest I've seen in most charts is 4.0. Still seems like clickbait at this point.

 

[jaymc]

maybe so mate. Jus taught I'd throw it up, it's chat bait now ;-)

Maybe this is the lesser spotted 1900x

I think that it's possible the reason AMD hid the single core IPC was to trick Intel or to lull them into a false sense of security...

Also imagine this.. what if the reason for the promise of instant availability and basically a non paper launch (or hard launch) was to not just catch them with their pants down on IPC, an only multi treaded benchmarks available etc...
But to also to catch them by undercutting them by half price, more or less, an to have loads of CPU's available at launch...
So as to steal wins on every front (price, performance & getting back market share) in the first blow of what is shaping up to be a brutal an viscous price war...

I was surprised AMD came in so cheap straight off the bat... At the end of the day Intel can afford to sell CPU's for cost price for a few years (or decades even actually)...

AMD on the other hand cannot !!

Unless they have a donkey load of them ready to go an are hoping to steal back a lot of the market share in the first battle, I think the pricing may have been too cheap straight off the bat to be honest, even a mistake Dr Su (or Lisa) did say at one point they were no longer the cheaper chip company... It's a genuine comeback we all want... not a few more death rattles... that may take a few more years...

It's competition in the market place we really need. Although I do think this will save the company.. Might even be intel thats in big trouble in ten years time ... oohhh s&*t did I say that. . . woops. 😀

 

[jdwii]

From what i've been reading about Intel i either expect them to be behind on performance per dollar when Ryzen comes out or Intel will do what they need to do and that is drop prices dramatically.

They are already rushing cannonlake out this year but sites are saying the 6 core won't be out to next year meaning IMO Ryzen has an advantage if the latest leaks are true. I expect a typical 5-7% improvement in IPC with cannonlake.

Amd said Zen is staying on 14nm to 2020 which gives Intel plenty of time to think of a strategy i'm not sure if Zen+ will be made on 14nm to at least give the core a 15% IPC boost throughout the years. Obviously expect better yields later down the line meaning higher clock rates and you bet Amd will push the limits.

Me and Juan agree on a lot of things but i still see no reason why Amd can't make a Enthusiast CPU aimed at 125-140TDP, i mean a 212+ Cooler master heat-sink could handle that.

Edit to add more to that Intel already sells CPU's with no heat-sink's included since they are aimed for overclockers.

 

[ScrewySqrl]

Me and Juan agree on a lot of things but i still see no reason why Amd can't make a Enthusiast CPU aimed at 125-140TDP, i mean a 212+ Cooler master heat-sink could handle that.

Edit to add more to that Intel already sells CPU's with no heat-sink's included since they are aimed for overclockers.

AMD can't make a 140W enthusiast Chip in the wake of the disaster that was the 95xx series. Which was a 220W CPU that people kept putting in boards with a 95W or 125W max.

 

[juanrga]

One of the online retailers not only listed pricing, but they also listed this little bit: "expected availability Feb 28th."

From what I heard only limited availability. Volume availability in March.

 

[8350rocks]

From what i've been reading about Intel i either expect them to be behind on performance per dollar when Ryzen comes out or Intel will do what they need to do and that is drop prices dramatically.

They are already rushing cannonlake out this year but sites are saying the 6 core won't be out to next year meaning IMO Ryzen has an advantage if the latest leaks are true. I expect a typical 5-7% improvement in IPC with cannonlake.

Amd said Zen is staying on 14nm to 2020 which gives Intel plenty of time to think of a strategy i'm not sure if Zen+ will be made on 14nm to at least give the core a 15% IPC boost throughout the years. Obviously expect better yields later down the line meaning higher clock rates and you bet Amd will push the limits.

Me and Juan agree on a lot of things but i still see no reason why Amd can't make a Enthusiast CPU aimed at 125-140TDP, i mean a 212+ Cooler master heat-sink could handle that.

Edit to add more to that Intel already sells CPU's with no heat-sink's included since they are aimed for overclockers.

Zen+ is targeting 7nm next year. I expect if Samsung/GloFo can get 7nm running by then, they will perhaps be ahead of Intel in process advantage if Intel keeps struggling with 10nm.

 

[juanrga]

From what i've been reading about Intel i either expect them to be behind on performance per dollar when Ryzen comes out or Intel will do what they need to do and that is drop prices dramatically.

They are already rushing cannonlake out this year but sites are saying the 6 core won't be out to next year meaning IMO Ryzen has an advantage if the latest leaks are true. I expect a typical 5-7% improvement in IPC with cannonlake.

Amd said Zen is staying on 14nm to 2020 which gives Intel plenty of time to think of a strategy i'm not sure if Zen+ will be made on 14nm to at least give the core a 15% IPC boost throughout the years. Obviously expect better yields later down the line meaning higher clock rates and you bet Amd will push the limits.

Me and Juan agree on a lot of things but i still see no reason why Amd can't make a Enthusiast CPU aimed at 125-140TDP, i mean a 212+ Cooler master heat-sink could handle that.

Edit to add more to that Intel already sells CPU's with no heat-sink's included since they are aimed for overclockers.

Zen+ is targeting 7nm next year. I expect if Samsung/GloFo can get 7nm running by then, they will perhaps be ahead of Intel in process advantage if Intel keeps struggling with 10nm.

This thread is about Zen which is a 14nm product; therefore, I will maintain my comments about Zen+ and 7nm concise and will not expand on the next material:

1) Lisa Su said that AMD will remain in the 14nm node for a long time:

AMD is establishing a number of products on 14-nm FinFET in 2016, Su said. “But it will be a long node. It will last three, four, five years. But within that node we can do a lot in optimization, and within that node, we can do a lot on power... once you’re in the node, it’s all about architecture.”

2) Papermaster just confirmed that AMD will be pushing a tock tock tock strategy:

"We're not going tick-tock... Zen is going to be tock, tock, tock."

Tick --> node shrink
Tock --> microarchitecture improvements.

3) Glofo expects 7nm risk production for late 2018. This means 7nm mass production for late 2019 or early 2020.

4) A final comment, Glofo "7nm" is in reality a true 14nm process node with a "7nm" marketing label. Recall that current Glofo "14nm" is in reality a 20nm node with a "14nm" marketing label.

 

[cdrkf]

@jdwii, I wonder if the reason AMD don't release a higher wattage 'enthusiast' part is due to xfr? *if* it functions in the way we expect it to (I.e. overclock up to the power and thermal limits available automatically), then the 1800x already is a 125w and even a 140w part, all you need to do is provide an appropriate motherboard and cooling (and possible set an increased power limit on the bios, it sounds a lot like the clock mechanism on Polaris)... the point is the feature kinda negates the need for higher rated parts...

 

[jdwii]

@jdwii, I wonder if the reason AMD don't release a higher wattage 'enthusiast' part is due to xfr? *if* it functions in the way we expect it to (I.e. overclock up to the power and thermal limits available automatically), then the 1800x already is a 125w and even a 140w part, all you need to do is provide an appropriate motherboard and cooling (and possible set an increased power limit on the bios, it sounds a lot like the clock mechanism on Polaris)... the point is the feature kinda negates the need for higher rated parts...

I could see that being a thing i still have no idea how they are going to do this and it would exceed the TDP obviously. I'd like to hear from others on how they could do this? Possibly they will do what modern GPUs do and ask the user about a target temperature like a target temperature of 70C and if you have better cooling it could turbo up to mid 4Ghz even as well as temps are OK. This whole thing makes me think that the chips can get to mid 4Ghz but they run super hot and binning isn't 100% yet.

 

[8350rocks]

From what i've been reading about Intel i either expect them to be behind on performance per dollar when Ryzen comes out or Intel will do what they need to do and that is drop prices dramatically.

They are already rushing cannonlake out this year but sites are saying the 6 core won't be out to next year meaning IMO Ryzen has an advantage if the latest leaks are true. I expect a typical 5-7% improvement in IPC with cannonlake.

Amd said Zen is staying on 14nm to 2020 which gives Intel plenty of time to think of a strategy i'm not sure if Zen+ will be made on 14nm to at least give the core a 15% IPC boost throughout the years. Obviously expect better yields later down the line meaning higher clock rates and you bet Amd will push the limits.

Me and Juan agree on a lot of things but i still see no reason why Amd can't make a Enthusiast CPU aimed at 125-140TDP, i mean a 212+ Cooler master heat-sink could handle that.

Edit to add more to that Intel already sells CPU's with no heat-sink's included since they are aimed for overclockers.

Zen+ is targeting 7nm next year. I expect if Samsung/GloFo can get 7nm running by then, they will perhaps be ahead of Intel in process advantage if Intel keeps struggling with 10nm.

This thread is about Zen which is a 14nm product; therefore, I will maintain my comments about Zen+ and 7nm concise and will not expand on the next material:

1) Lisa Su said that AMD will remain in the 14nm node for a long time:

AMD is establishing a number of products on 14-nm FinFET in 2016, Su said. “But it will be a long node. It will last three, four, five years. But within that node we can do a lot in optimization, and within that node, we can do a lot on power... once you’re in the node, it’s all about architecture.”

2) Papermaster just confirmed that AMD will be pushing a tock tock tock strategy:

"We're not going tick-tock... Zen is going to be tock, tock, tock."

Tick --> node shrink
Tock --> microarchitecture improvements.

3) Glofo expects 7nm risk production for late 2018. This means 7nm mass production for late 2019 or early 2020.

4) A final comment, Glofo "7nm" is in reality a true 14nm process node with a "7nm" marketing label. Recall that current Glofo "14nm" is in reality a 20nm node with a "14nm" marketing label.

The 14nm node from Samsung, that GloFo is using, is a hybrid node with 20nm BEOL.

 

[jdwii]

http://wccftech.com/amd-ryzen-processors-am4-official-launch-2-march/

7AM on the 2nds we will have a launch

 

[ththth]

Any chance we will see an update to Toms creative benchmarks comparing Ryzen performance in After Effects and Premiere with 6700K, 7700K, 6950K etc?

http://www.tomshardware.com/charts/cpu-charts-2015/-30-Adobe-After-Effects-CC,3721.html

http://www.tomshardware.com/charts/cpu-charts-2015/-31-Adobe-Premiere-Pro-CC,3722.html

Would like to see a bit more aggressive test for premiere compared with the previous encoding test and if possible Blackmagic Davinci Resolve and Blackmagic Fusion - these two applications are fast becoming more popular than Premiere/After Effects in my industry.

 

[ComputerSecurityGuy]

Just speculating, there's a chance that AMD will fab CPUs on a different process from another foundry. Anand says that
"Under this new agreement, AMD has been granted a “limited waiver” to work with other foundries “with respect to certain products in the 14nm and 7nm technology nodes.” As a result, while AMD still has to meet their overall wafer buy targets over the 5 year period, AMD also will continue working with other foundries on unspecified products where they see a need to have them manufactured at somewhere other than GlobalFoundries."
However, I'd guess that the core CPU biz sits on Glofo's 14nm and it's other things like GPUs, APUs, console chips and chipsets that might be done on other fabs. The new Xbox SoC, for example, might very well be on TSMC's 16nm process as the Xbox S was.

 

[juanrga]

@jdwii, I wonder if the reason AMD don't release a higher wattage 'enthusiast' part is due to xfr? *if* it functions in the way we expect it to (I.e. overclock up to the power and thermal limits available automatically), then the 1800x already is a 125w and even a 140w part, all you need to do is provide an appropriate motherboard and cooling (and possible set an increased power limit on the bios, it sounds a lot like the clock mechanism on Polaris)... the point is the feature kinda negates the need for higher rated parts...

Note that XFR is a 1-core turbo; a 125W chip would have higher base clocks and higher all-core turbo

 

[jdwii]

@jdwii, I wonder if the reason AMD don't release a higher wattage 'enthusiast' part is due to xfr? *if* it functions in the way we expect it to (I.e. overclock up to the power and thermal limits available automatically), then the 1800x already is a 125w and even a 140w part, all you need to do is provide an appropriate motherboard and cooling (and possible set an increased power limit on the bios, it sounds a lot like the clock mechanism on Polaris)... the point is the feature kinda negates the need for higher rated parts...

Note that XFR is a 1-core turbo; a 125W chip would have higher base clocks and higher all-core turbo

Isn't that what Intel already does what makes that technology so great?

 

[illegaloperation]

Note that XFR is a 1-core turbo; a 125W chip would have higher base clocks and higher all-core turbo

Source?

 

[Crumpet 1]

Yeah, sauce please.

 

[jdwii]

Yeah, sauce please.

http://wccftech.com/amd-ryzen-architecture-detailed/

Not sure this says "XFR in combination with two other features, Pure Power and Precision Boost, work in tandem to ensure that all active cores are running at the highest clock speeds they are capable of without exceeding the default power and thermal limits"

We are very close to launch i'm sure we will know soon. The site i linked earlier said 2nd of march at 7am that could be when the NDA is up and we can have fun reading 100 of reviews ha ha or perhaps just me.

 

[Crumpet 1]

Yeah, sauce please.

http://wccftech.com/amd-ryzen-architecture-detailed/

Not sure this says "XFR in combination with two other features, Pure Power and Precision Boost, work in tandem to ensure that all active cores are running at the highest clock speeds they are capable of without exceeding the default power and thermal limits"

We are very close to launch i'm sure we will know soon. The site i linked earlier said 2nd of march at 7am that could be when the NDA is up and we can have fun reading 100 of reviews ha ha or perhaps just me.

Every time I see a wccf article I die a little inside.

But "work in tandem to ensure that all active cores are running at the highest clock speeds they are capable of" sounds less like single core, and more like all cores, though potentially all cores individually?

 

[8350rocks]

We have a nickname for wccf in this thread...

WTFtech.