AMD CPUs, SoC Rumors and Speculations Temp. thread 2

[8350rocks]

He also has a good record.

My record has been good as well...

 

[8350rocks]

True, point was I don't care about how power efficient their top end 8 core cpu is, as I don't pay the power bill. If zen ends up with 150W tdp and overclocks to use 200W I wont complain.

I think that's what intel was thinking from their top end parts as well. most people with 6700K's overclock them to use 140W and if you noticed their stock tdp has increased in the last generation. desktops generally don't need efficiency as most people with i7 6700K's are running 600W plus power supplies and can easily run the power.

even amd's last gen 8350 took over 8 years for the average consumer to pay off the price difference in power draw compared to owners of the 3770K.

https://www.youtube.com/watch?v=fBeeGHozSY0


EDIT: point is power efficiency is irrelivent in the desktop market for 90% of gamers and power users.

U series and laptops... different story, desktop no.

This is a point I have been making for a while...and the argument about efficiency evaporates at HEDT...

 

[jimmysmitty]

True, point was I don't care about how power efficient their top end 8 core cpu is, as I don't pay the power bill. If zen ends up with 150W tdp and overclocks to use 200W I wont complain.

I think that's what intel was thinking from their top end parts as well. most people with 6700K's overclock them to use 140W and if you noticed their stock tdp has increased in the last generation. desktops generally don't need efficiency as most people with i7 6700K's are running 600W plus power supplies and can easily run the power.

even amd's last gen 8350 took over 8 years for the average consumer to pay off the price difference in power draw compared to owners of the 3770K.

https://www.youtube.com/watch?v=fBeeGHozSY0


EDIT: point is power efficiency is irrelivent in the desktop market for 90% of gamers and power users.

U series and laptops... different story, desktop no.

This is a point I have been making for a while...and the argument about efficiency evaporates at HEDT...

I don't disagree but what I always hated was the flip flop of people on what matters. Not you but there are a few who will jump to whatever suits them. Be it better power numbers, performance per core, performance per clock etc.

I think everything matters.

With FX the power was just too high to begin with and overclocking made it worse. Zen will be better, it is a more efficiency based uArch. How much better is yet to be seen.

 

[8350rocks]

https://www.semiwiki.com/forum/content/5180-globalfoundries-visit-%C2%96-part-2-%C2%96-waking-sleeping-giant.html

As interesting as all of that is, they then went on to describe their new FX-14 ASIC offering utilizing the 14LLP process manufactured at Fab 8 in Malta. As an outside observer one concern I had for GF was how they would differentiate themselves from Samsung and become more than a second source when they are licensing the 14nm process from Samsung. While we were in Burlington we heard the AMD has committed to 14LPP giving GF a volume first source customer to drive yield learning.

An interesting segment from a semiwiki article in October about the 14FF node:

Daniel commented that Qualcomm and others are doing server chips. Will a foundry do a very high performance process for server chips. This led to a discussion about the IBM chip business acquisition and whether IBM’s 14nm FinFET on SOI process will be available to outside customers. Global is committed to support IBM’s SOI technology for 10 years but beyond that they can’t comment on IBM technology plans although they did say they think it is a game changer.

Interestingly...someone said SOI was dead...

Look at that...14FF on FDSOI...IP from IBM...

Interesting when the targets for production are up as well...

GF has 14FF making full production scale by Q4 2016, but 14FF FDSOI hits that scale 3-6 months later.

AMD has not announced a firm date yet...

GF has not announced firm dates yet either...

Just saying...no predictions...no prognostications...

 

[juanrga]

If 10-core BDW is ~140W, then 8-core BDW would be ~ 112W. Zen likely has less L3 cache (20MB vs 16MB) which reduces power consumption. Reduce clocks by ~10% and we are on 95W easily.

95W for a quad-core Zen on 14nm would be a complete disappointment.

Maybe to you I plan to overclock the living F#$% out of mine id be happy to hit 200W when its all said and done.

I refer to stock chip wattage rating, not overclocking.

 

[juanrga]

https://www.semiwiki.com/forum/content/5180-globalfoundries-visit-%C2%96-part-2-%C2%96-waking-sleeping-giant.html

As interesting as all of that is, they then went on to describe their new FX-14 ASIC offering utilizing the 14LLP process manufactured at Fab 8 in Malta. As an outside observer one concern I had for GF was how they would differentiate themselves from Samsung and become more than a second source when they are licensing the 14nm process from Samsung. While we were in Burlington we heard the AMD has committed to 14LPP giving GF a volume first source customer to drive yield learning.

An interesting segment from a semiwiki article in October about the 14FF node:

Daniel commented that Qualcomm and others are doing server chips. Will a foundry do a very high performance process for server chips. This led to a discussion about the IBM chip business acquisition and whether IBM’s 14nm FinFET on SOI process will be available to outside customers. Global is committed to support IBM’s SOI technology for 10 years but beyond that they can’t comment on IBM technology plans although they did say they think it is a game changer.

Interestingly...someone said SOI was dead...

Look at that...14FF on FDSOI...IP from IBM...

Interesting when the targets for production are up as well...

GF has 14FF making full production scale by Q4 2016, but 14FF FDSOI hits that scale 3-6 months later.

AMD has not announced a firm date yet...

GF has not announced firm dates yet either...

Just saying...no predictions...no prognostications...

AMD doesn't use SOI at 14nm.

And we have known for several years now that only ~5% of total volume production foundry was using SOI process at 14nm (with FinFETs) and IBM finally sold the foundries because couldn't sustain them in the long run.

 

[Vogner16]

This is a point I have been making for a while...and the argument about efficiency evaporates at HEDT...

I don't disagree but what I always hated was the flip flop of people on what matters. Not you but there are a few who will jump to whatever suits them. Be it better power numbers, performance per core, performance per clock etc.

I think everything matters.

With FX the power was just too high to begin with and overclocking made it worse. Zen will be better, it is a more efficiency based uArch. How much better is yet to be seen.

I have been and always will be more concerned with the performance per $. IF the amd part uses sooo much more power that I could have paid off the more expensive intel part in 1 year with my use case scenario and is just as fast, I would go to intel.

Having said that Intel has only recently produced chips that create a better performance per $ ratio for my use case scenario. In games intel generally gets 10-20 better fps over my 8350, and only recently has intel managed to produce performance in rendering benchmarks that could beat out my 8350 for the price of the part and the 4 year life span I expect for a desktop pc.

unless one can find a used 8350 for under $100 it is a worse deal than the new skylake i7 in the 4 year plan.

Only reason im not upgrading to skylake today is the promise of zen and my current pc still able to hit over 35 fps in all games I own. If zen gets delayed to a point far out into the future such as mid to late 2017 and my pc is no longer able to hit 30 fps on new games, then I may be forced to get kaby lake, as it will surely be the better deal.

 

[jimmysmitty]

This is a point I have been making for a while...and the argument about efficiency evaporates at HEDT...

I don't disagree but what I always hated was the flip flop of people on what matters. Not you but there are a few who will jump to whatever suits them. Be it better power numbers, performance per core, performance per clock etc.

I think everything matters.

With FX the power was just too high to begin with and overclocking made it worse. Zen will be better, it is a more efficiency based uArch. How much better is yet to be seen.

I have been and always will be more concerned with the performance per $. IF the amd part uses sooo much more power that I could have paid off the more expensive intel part in 1 year with my use case scenario and is just as fast, I would go to intel.

Having said that Intel has only recently produced chips that create a better performance per $ ratio for my use case scenario. In games intel generally gets 10-20 better fps over my 8350, and only recently has intel managed to produce performance in rendering benchmarks that could beat out my 8350 for the price of the part and the 4 year life span I expect for a desktop pc.

unless one can find a used 8350 for under $100 it is a worse deal than the new skylake i7 in the 4 year plan.

Only reason im not upgrading to skylake today is the promise of zen and my current pc still able to hit over 35 fps in all games I own. If zen gets delayed to a point far out into the future such as mid to late 2017 and my pc is no longer able to hit 30 fps on new games, then I may be forced to get kaby lake, as it will surely be the better deal.

You were not here for what I am talking about. Most normal people focus on what matters to them. For me it is what I am willing to spend and what there is that fits that. I bought a GTX 980Ti because at $650 it was the better performer and it was easier to find than the Fury X. Plus I didn't want the AiO cooler. I wont go cheaper just because I plan everything out to a certain price.

However when Athlon 64 was king it was the performance per clock per watt because that is what it was best at. When Phenom failed to beat Core 2 it was IMC, monolithic and highly threaded that matered, because Phenom was failing to deliver on the performance per clock per watt. Then with FX the same people touted the price after it dropped for moar cores.

I really don't miss that and hope it doesn't show up with Zen. I just want to see Zen as it is, not nit picked to be whatever makes it look better. That is also why hype is annoying. It could be a great CPU. Or it could be crap. Hype does not allow it to be a decent CPU.

 

[Vogner16]

You were not here for what I am talking about. Most normal people focus on what matters to them. For me it is what I am willing to spend and what there is that fits that. I bought a GTX 980Ti because at $650 it was the better performer and it was easier to find than the Fury X. Plus I didn't want the AiO cooler. I wont go cheaper just because I plan everything out to a certain price.

However when Athlon 64 was king it was the performance per clock per watt because that is what it was best at. When Phenom failed to beat Core 2 it was IMC, monolithic and highly threaded that matered, because Phenom was failing to deliver on the performance per clock per watt. Then with FX the same people touted the price after it dropped for moar cores.

I really don't miss that and hope it doesn't show up with Zen. I just want to see Zen as it is, not nit picked to be whatever makes it look better. That is also why hype is annoying. It could be a great CPU. Or it could be crap. Hype does not allow it to be a decent CPU.

true. I would also get a 980ti over a fury x but I would probabaly end up watercooling it anyway and would have to do some research on fury overclocking now that voltages have been unlocked. the story is always changing.

I like to read the hype but like to see the hype turn into reality. fx was a cherry picked benchmark disaster, but I waited for it to drop price to $200 then bought the 8150. it was decent for the price. then waited for 8350 to drop to $190 and upgraded to that when I found out steamroller wasn't happening and popped that upgrade in using the same mobo and ram. saved me a lot of money.

Im just a cheap pc builder 😀

 

[juanrga]

This is a point I have been making for a while...and the argument about efficiency evaporates at HEDT...

If AMD offers to HEDT users

A) 95W FX-98370E CPU or

B) 95W Zen octo-core CPU

will most HEDT users chose the less efficient Piledriver CPU or will prefer the more efficient Zen CPU?

The next equation always holds

Performance = Efficiency * Power

The higher the efficiency the higher the performance for a given wattage, independently of if it it 2W (mobile) or 400W (high-end server/HPC).

 

[juanrga]

This is a point I have been making for a while...and the argument about efficiency evaporates at HEDT...

I don't disagree but what I always hated was the flip flop of people on what matters. Not you but there are a few who will jump to whatever suits them. Be it better power numbers, performance per core, performance per clock etc.

I think everything matters.

With FX the power was just too high to begin with and overclocking made it worse. Zen will be better, it is a more efficiency based uArch. How much better is yet to be seen.

I have been and always will be more concerned with the performance per $. IF the amd part uses sooo much more power that I could have paid off the more expensive intel part in 1 year with my use case scenario and is just as fast, I would go to intel.

Having said that Intel has only recently produced chips that create a better performance per $ ratio for my use case scenario. In games intel generally gets 10-20 better fps over my 8350, and only recently has intel managed to produce performance in rendering benchmarks that could beat out my 8350 for the price of the part and the 4 year life span I expect for a desktop pc.

unless one can find a used 8350 for under $100 it is a worse deal than the new skylake i7 in the 4 year plan.

Only reason im not upgrading to skylake today is the promise of zen and my current pc still able to hit over 35 fps in all games I own. If zen gets delayed to a point far out into the future such as mid to late 2017 and my pc is no longer able to hit 30 fps on new games, then I may be forced to get kaby lake, as it will surely be the better deal.

You were not here for what I am talking about. Most normal people focus on what matters to them. For me it is what I am willing to spend and what there is that fits that. I bought a GTX 980Ti because at $650 it was the better performer and it was easier to find than the Fury X. Plus I didn't want the AiO cooler. I wont go cheaper just because I plan everything out to a certain price.

However when Athlon 64 was king it was the performance per clock per watt because that is what it was best at. When Phenom failed to beat Core 2 it was IMC, monolithic and highly threaded that matered, because Phenom was failing to deliver on the performance per clock per watt. Then with FX the same people touted the price after it dropped for moar cores.

I really don't miss that and hope it doesn't show up with Zen. I just want to see Zen as it is, not nit picked to be whatever makes it look better. That is also why hype is annoying. It could be a great CPU. Or it could be crap. Hype does not allow it to be a decent CPU.

The same flip-flop is happening with Zen. When Keller rejoined AMD the original hype was that Zen would beat Skylake performance, that Zen would be the new K8 and left Intel in the dust. When the 40% figure was given by Papermaster and he confirmed that AMD will not beat Intel in raw performance, the hype changed to Haswell-like single thread performance but moar cores. Now they know that Intel Broadwell will offer two more cores: 10 vs 8, then the hype has changed to Haswell-like octo-core at Intel quad-core prices... despite Lisa Su has made clear that "AMD is no longer the cheap company" and that Zen will be priced accordingly.

 

[Vogner16]

The same flip-flop is happening with Zen. When Keller rejoined AMD the original hype was that Zen would beat Skylake performance, that Zen would be the new K8 and left Intel in the dust. When the 40% figure was given by Papermaster and he confirmed that AMD will not beat Intel in raw performance, the hype changed to Haswell-like single thread performance but moar cores. Now they know that Intel Broadwell will offer two more cores: 10 vs 8, then the hype has changed to Haswell-like octo-core at Intel quad-core prices... despite Lisa Su has made clear that "AMD is no longer the cheap company" and that Zen will be priced accordingly.

I remember the preprepre hype of better than skylake but never believed it. how could a company beat its competition that is 5 generations ahead in IPC in one jump? nah ~cheef keef

at Intel quad-core prices... Where do you see this?

Ive seen Lisa Su's "AMD is no longer the cheap company" and have said over and over to fully expect lga 2011 type prices. haswell 8 core perf for haswell 8 core price.

EDIT: If you read the little study I did with the downclocking of an 8350 and applying claims from AMD to IPC in cinebench it shows between Ivy and haswell perf when turbo clocks are accounted for. ~~ $550 - $700 for the big daddy zen core

 

[jimmysmitty]

The same flip-flop is happening with Zen. When Keller rejoined AMD the original hype was that Zen would beat Skylake performance, that Zen would be the new K8 and left Intel in the dust. When the 40% figure was given by Papermaster and he confirmed that AMD will not beat Intel in raw performance, the hype changed to Haswell-like single thread performance but moar cores. Now they know that Intel Broadwell will offer two more cores: 10 vs 8, then the hype has changed to Haswell-like octo-core at Intel quad-core prices... despite Lisa Su has made clear that "AMD is no longer the cheap company" and that Zen will be priced accordingly.

I remember the preprepre hype of better than skylake but never believed it. how could a company beat its competition that is 5 generations ahead in IPC in one jump? nah ~cheef keef

at Intel quad-core prices... Where do you see this?

Ive seen Lisa Su's "AMD is no longer the cheap company" and have said over and over to fully expect lga 2011 type prices. haswell 8 core perf for haswell 8 core price.

EDIT: If you read the little study I did with the downclocking of an 8350 and applying claims from AMD to IPC in cinebench it shows between Ivy and haswell perf when turbo clocks are accounted for. ~~ $550 - $700 for the big daddy zen core

A 5 generation jump is not that hard. Core 2 came out after the Pentium 4 130nm, P4 90nm, PD 65nm and all the Pentium Extremes got their buts handed to them but the K8 uArch and beat K8 and even K10 (Phenom). Was a pretty big jump.

Considering how far behind PD is, Zen can make a pretty massive jump.

 

[Vogner16]

A 5 generation jump is not that hard. Core 2 came out after the Pentium 4 130nm, P4 90nm, PD 65nm and all the Pentium Extremes got their buts handed to them but the K8 uArch and beat K8 and even K10 (Phenom). Was a pretty big jump.

Considering how far behind PD is, Zen can make a pretty massive jump.

that's not 5 gens.

K8 was ahead then core 2 beat k8 and k10. that's 2 gens jump (claimed best cpu since K8)

p4 was ahead 65nm then K8 jumped 3 gens to beat it. (claimed the best cpu in years)

 

[juanrga]

This is a point I have been making for a while...and the argument about efficiency evaporates at HEDT...

If AMD offers to HEDT users

A) 95W FX-98370E CPU or

B) 95W Zen octo-core CPU

will most HEDT users chose the less efficient Piledriver CPU or will prefer the more efficient Zen CPU?

The next equation always holds

Performance = Efficiency * Power

The higher the efficiency the higher the performance for a given wattage, independently of if it it 2W (mobile) or 400W (high-end server/HPC).

Who cares about power consumption? Which one is outright the best performer??? I would imagine Zen by a landslide.

I didn't ask about power consumption. I asked about efficiency and you just selected the more efficient chip (B).

However, I present you this case as a case and point:

You can choose between the following for the same money:

1.) FX9590 with unlimited supply on LN2 and stable overclocked rig at 8.0 GHz, total build consumes 1200W

2.) i7-5970x that was bad silicon and only clocks to 4.0 GHz, total build consumes 700W for 60-70% performance of the 8.0 GHz 9590 build.

The money is the same for both over the course of 5 years including electricity costs if you run the AMD machine at full load for 24 hours/day 365 days/year.

Which would you choose? 8.0 GHz FX9590 or 4.0 GHz Intel?

I gave you two realistic options to choose. You don't:

• 
  ■ The 5970x doesn't exist.
  ■ You assume one is very bad silicon, whereas the other is golden chip. Most 9590 cannot hit 8GHz, less still run stable 24/365.
  ■ One uses LN2 the other doesn't?
  ■ Unlimited supply of LN2? LOL
  ■ I don't buy your electricity costs
  ■ 1200W @8GHz on 32nm vs 700W @4GHz on 22nm? I don't buy the 700W claim. Even @4.2GHz the i7-5960X is under 500W.
  ■ 60--70% performance? How? Assuming linear scaling for both chips, the i7@4GHz would be still faster than FX@8GHz.

I used realistic configurations to illustrate a basic law of computer science. If we are allowed to build fantasies that violate the laws of physics then I prefer a car can hit 2x the speed of light and a 32-core CPU can hit 10GHz on air.

 

[juanrga]

The same flip-flop is happening with Zen. When Keller rejoined AMD the original hype was that Zen would beat Skylake performance, that Zen would be the new K8 and left Intel in the dust. When the 40% figure was given by Papermaster and he confirmed that AMD will not beat Intel in raw performance, the hype changed to Haswell-like single thread performance but moar cores. Now they know that Intel Broadwell will offer two more cores: 10 vs 8, then the hype has changed to Haswell-like octo-core at Intel quad-core prices... despite Lisa Su has made clear that "AMD is no longer the cheap company" and that Zen will be priced accordingly.

I remember the preprepre hype of better than skylake but never believed it. how could a company beat its competition that is 5 generations ahead in IPC in one jump? nah ~cheef keef

at Intel quad-core prices... Where do you see this?

Ive seen Lisa Su's "AMD is no longer the cheap company" and have said over and over to fully expect lga 2011 type prices. haswell 8 core perf for haswell 8 core price.

EDIT: If you read the little study I did with the downclocking of an 8350 and applying claims from AMD to IPC in cinebench it shows between Ivy and haswell perf when turbo clocks are accounted for. ~~ $550 - $700 for the big daddy zen core

I remember all the hype: "faster than skylake", "redesigned CMT architecture", "4ALU+3AGU", "base clock of 4.0--4.1GHz", "AVX512", "2x256 bit FMAC",... It was all wrong, except the 4 ALUs.

As mentioned multiple times I am expecting integer IPC somewhat between Sandy and Haswell. And floating IPC similar to Sandy/Ivy. The big unknown are clocks, but I expect ~3.5GHz turbo or so.

I agree on that Zen will priced high. I expect $800 or more. Recall the FX-9590 was launched with a price close to $900.

 

[Vogner16]

I remember all the hype: "faster than skylake", "redesigned CMT architecture", "4ALU+3AGU", "base clock of 4.0--4.1GHz", "AVX512", "2x256 bit FMAC",... It was all wrong, except the 4 ALUs.

As mentioned multiple times I am expecting integer IPC somewhat between Sandy and Haswell. And floating IPC similar to Sandy/Ivy. The big unknown are clocks, but I expect ~3.5GHz turbo or so.

I agree on that Zen will priced high. I expect $800 or more. Recall the FX-9590 was launched with a price close to $900.

All good stuff. exactly what I am expecting.

http://wccftech.com/amd-zen-launch-q4-2016/

zen news "wccftech exclusive" LOL

 

[jimmysmitty]

I remember all the hype: "faster than skylake", "redesigned CMT architecture", "4ALU+3AGU", "base clock of 4.0--4.1GHz", "AVX512", "2x256 bit FMAC",... It was all wrong, except the 4 ALUs.

As mentioned multiple times I am expecting integer IPC somewhat between Sandy and Haswell. And floating IPC similar to Sandy/Ivy. The big unknown are clocks, but I expect ~3.5GHz turbo or so.

I agree on that Zen will priced high. I expect $800 or more. Recall the FX-9590 was launched with a price close to $900.

All good stuff. exactly what I am expecting.

http://wccftech.com/amd-zen-launch-q4-2016/

zen news "wccftech exclusive" LOL

That is a site I only look at everything as rumors.

I mean even their commentor base seems to be a bit off. One guy thinks the Intel i7 6900K is a Skylake part that is a quad core with HT based on Kaby Lake when it is clearly a 8 core w/HT Broadwell-E part for LGA 2011v3.

That's why WCCFTech is best kept to a minimum for any news and take with a grain of salt the size of Texas.

 

[-Fran-]

That site... Man, when I feel angry go to that page and read the comments. They make me feel like I'm a wonderful human being, hahaha.

In any case, Juan, 3.5Ghz *turbo* sound very low for an high performing part... Unless they want to go full circle with clocks... I mean, the IPC gains will be dwarfed in both stock clocks and OC (don't have much faith in OC wth LPP). That would be sad, really. Where are you getting that number from? I kinda forgot 😛

Cheers!

 

[juanrga]

I find very interesting that their "exclusive sources" are giving them the same information that AMD gave to everyone else at FADD 2015. Moreover, they add a final "AMD can change the plans". Their diagram about Zen is plain wrong. There are four half FP pipes, two FMUL plus to FADD; I explained them this in the comments of an old article when they pretended that Zen has 2x256bit FMAC pipes. They didn't get still, because now pretend that there is two 128bit FMAC pipes plus two "complementary" pipes.

 

[juanrga]

The same flip-flop is happening with Zen. When Keller rejoined AMD the original hype was that Zen would beat Skylake performance, that Zen would be the new K8 and left Intel in the dust. When the 40% figure was given by Papermaster and he confirmed that AMD will not beat Intel in raw performance, the hype changed to Haswell-like single thread performance but moar cores. Now they know that Intel Broadwell will offer two more cores: 10 vs 8, then the hype has changed to Haswell-like octo-core at Intel quad-core prices... despite Lisa Su has made clear that "AMD is no longer the cheap company" and that Zen will be priced accordingly.

I remember the preprepre hype of better than skylake but never believed it. how could a company beat its competition that is 5 generations ahead in IPC in one jump? nah ~cheef keef

at Intel quad-core prices... Where do you see this?

On another forum where AMD fanboys post fake slides like this

More concretely they are expecting octo-core Zen CPU to be priced at Skylake i5 levels.

 

[juanrga]

That site... Man, when I feel angry go to that page and read the comments. They make me feel like I'm a wonderful human being, hahaha.

In any case, Juan, 3.5Ghz *turbo* sound very low for an high performing part... Unless they want to go full circle with clocks... I mean, the IPC gains will be dwarfed in both stock clocks and OC (don't have much faith in OC wth LPP). That would be sad, really. Where are you getting that number from? I kinda forgot 😛

Estimations from certain forum poster that has access to 14LPP node internal data and power consumption of tests chips. He got access to target Carrizo clocks before anyone else and he was very accurate. That is why I am trusting him.

He suggests ~2.8GHz base clock and 3.5GHz turbo clock for the octo-core Zen CPU. The number look credible.

Note that 10-core Broadwell-E is expected to have ~3.5GHz turbo as well.

 

[-Fran-]

Estimations from certain forum poster that has access to 14LPP node internal data and power consumption of tests chips. He got access to target Carrizo clocks before anyone else and he was very accurate. That is why I am trusting him.

He suggests ~2.8GHz base clock and 3.5GHz turbo clock for the octo-core Zen CPU. The number look credible.

Note that 10-core Broadwell-E is expected to have ~3.5GHz turbo as well.

That last bit is an interesting piece of information. BW-E will use skylake's 14nm process, right? The it would be a nice "bench" against Zen when it finally launches. We could start gathering numbers for when both launch. I know Cannonlake should be the contender (or the next one in the pipeline from Intel), but given speed and manufacturing process, it would make for an interesting comparison. Especially since Zen should be aimed to the server market eventually. I kind of not remember if it was going to be launched for desktop first or last.

In any case, how much headroom would both have? 95W is very neat, since it makes it sound, if the process can cope with it, that 125W would be totally feasible with 3.5Ghz *base*. I would love to see some torture numbers at least, haha.

Cheers!

 

[jimmysmitty]

Estimations from certain forum poster that has access to 14LPP node internal data and power consumption of tests chips. He got access to target Carrizo clocks before anyone else and he was very accurate. That is why I am trusting him.

He suggests ~2.8GHz base clock and 3.5GHz turbo clock for the octo-core Zen CPU. The number look credible.

Note that 10-core Broadwell-E is expected to have ~3.5GHz turbo as well.

That last bit is an interesting piece of information. BW-E will use skylake's 14nm process, right? The it would be a nice "bench" against Zen when it finally launches. We could start gathering numbers for when both launch. I know Cannonlake should be the contender (or the next one in the pipeline from Intel), but given speed and manufacturing process, it would make for an interesting comparison. Especially since Zen should be aimed to the server market eventually. I kind of not remember if it was going to be launched for desktop first or last.

In any case, how much headroom would both have? 95W is very neat, since it makes it sound, if the process can cope with it, that 125W would be totally feasible with 3.5Ghz *base*. I would love to see some torture numbers at least, haha.

Cheers!

The headroom will highly depend on a lot of factors such as the process, the uArch and others. So far we have not seen Samsungs 14nm in anything other than their Exynos ARM low power chip. Zen would be the first x86 high powered chips to use this node.

 

[de5_Roy]

ARM on AMD: The A1100 Seattle Silicon at SuperComputing 15
http://www.anandtech.com/show/9807/arm-on-amd-the-a1100-seattle-silicon-at-supercomputing-15
The AMD A8-7670K APU Review: Aiming for Rocket League
http://www.anandtech.com/show/9763/the-amd-a8-7670k-apu-review-rocket-league
Globalfoundries might become Chinese takeaway
http://www.fudzilla.com/news/processors/39346-gloflo-might-become-chinese-takeaway