[Techpowerup] AMD "Zen" CPU Prototypes Tested, "Meet all Expectations"

[Fjodor2001]

Are you mad Fjodor?
<BS cut>

No, are you? I was just using your own argument against you.

 

[Arachnotronic]

I'm not sure why are you getting defensive. You always ask for sources and solid proof about any remotely positive news article related to I&#822;n&#822;t&#822;e&#822;l&#822; AMD products, but apparently the same doesn't apply to A&#822;M&#822;D&#822;'&#822;s&#822; Intel's products.

Intel has had a very strong track record of execution and has been putting out top-notch PC/server processors for about a decade straight, widening the gap against AMD's offerings pretty much every year throughout that decade.

There is every reason to be skeptical about AMD, while there is far less reason to be skeptical of Intel in PCs/servers. AMD is a company that has been promising big for years but it does not deliver, and it is only getting worse as its best engineers/managers have long since jumped ship.

I think you really want/hope to see AMD "beat" Intel for whatever reason (you seem to have a dislike for Intel from your posting history), and that's fine, but don't get mad when people provide you with a reality check.

p.s. I am not part of the "Intel Sales Team" as you put it -- have never been on Intel payroll and don't expect that I ever will be given my line of work -- but I do own Intel stock.

 

[Abwx]

Intel has had a very strong track record of execution and has been putting out top-notch PC/server processors for about a decade straight, widening the gap against AMD's offerings pretty much every year throughout that decade.

There is every reason to be skeptical about AMD, while there is far less reason to be skeptical of Intel in PCs/servers. AMD is a company that has been promising big for years but it does not deliver, and it is only getting worse as its best engineers/managers have long since jumped ship.

I think you really want/hope to see AMD "beat" Intel for whatever reason (you seem to have a dislike for Intel from your posting history), and that's fine, but don't get mad when people provide you with a reality check.

p.s. I am not part of the "Intel Sales Team" as you put it -- have never been on Intel payroll and don't expect that I ever will be given my line of work -- but I do own Intel stock.

So much for reality check, you should apply it to yourself.

If Intel was that good they wouldnt had relied to bribing the OEMs when AMD had a significantly better offering on the server and PC market...

They wouldnt had relied to contra revenues to keep their competitor out of the tablet and generaly low power markets..

 

[TechGod123]

Intel has had a very strong track record of execution and has been putting out top-notch PC/server processors for about a decade straight, widening the gap against AMD's offerings pretty much every year throughout that decade.

There is every reason to be skeptical about AMD, while there is far less reason to be skeptical of Intel in PCs/servers. AMD is a company that has been promising big for years but it does not deliver, and it is only getting worse as its best engineers/managers have long since jumped ship.

I think you really want/hope to see AMD "beat" Intel for whatever reason (you seem to have a dislike for Intel from your posting history), and that's fine, but don't get mad when people provide you with a reality check.

p.s. I am not part of the "Intel Sales Team" as you put it -- have never been on Intel payroll and don't expect that I ever will be given my line of work -- but I do own Intel stock.

Reliable track record?

You mean the one where they paid OEMs to use their chips cause AMD was wrecking them?

 

[mysticjbyrd]

Reliable track record?

You mean the one where they paid OEMs to use their chips cause AMD was wrecking them?

Hey now, say what you will, but if there is one thing a monopoly is known for it is reliability! They reliably stick it to the consumers every chance they get.

 

[frozentundra123456]

Almost as reliable as AMD fans bringing up old news.

 

[DrMrLordX]

Netburst was awhile back, you know?

More recently, we've seen Larrabee, Cherry Trail, 14nm rollout, 10nm rollout have, uh, issues. Intel has its little (or not so little) hiccups and delays. They aren't perfect.

 

[nismotigerwvu]

Netburst was awhile back, you know?

More recently, we've seen Larrabee, Cherry Trail, 14nm rollout, 10nm rollout have, uh, issues. Intel has its little (or not so little) hiccups and delays. They aren't perfect.

Yeah, it's almost as if making something out of billions of low-double-digit nanometer transistors was REALLY freaking hard. People like to the point the finger at the failings of tech companies (GloFo, AMD, Nvidia, TSMC IBM, Intel....ect) but it's downright amazing what the industry at large has achieved.

 

[Dresdenboy]

Yeah, it's almost as if making something out of billions of low-double-digit nanometer transistors was REALLY freaking hard. People like to the point the finger at the failings of tech companies (GloFo, AMD, Nvidia, TSMC IBM, Intel....ect) but it's downright amazing what the industry at large has achieved.

In many brains (maybe due to evolutionary roots) the better is the enemy of the good.

 

[tential]

Netburst was awhile back, you know?

More recently, we've seen Larrabee, Cherry Trail, 14nm rollout, 10nm rollout have, uh, issues. Intel has its little (or not so little) hiccups and delays. They aren't perfect.

I've never claimed intel to be perfect personally. Personally, I'm expecting delays on BOTH sides as it's been shown these latest nodes are very hard.

But, if AMD is putting their trusts in GloFlo... then I mean, I'm expecting something bad for intel. But it's astronomically worse for anyone who trusts GloFlo.....

 

[Arachnotronic]

Netburst was awhile back, you know?

More recently, we've seen Larrabee, Cherry Trail, 14nm rollout, 10nm rollout have, uh, issues. Intel has its little (or not so little) hiccups and delays. They aren't perfect.

Intel isn't perfect, but its execution has been substantially better than AMD's.

The best FX chip you can buy today is based on the 2012 Piledriver core...

 

[AtenRa]

Intel isn't perfect, but its execution has been substantially better than AMD's.

Intel
SB 2011
Ivy 2012
Haswell 2013
Haswell Refresh 2014
Broadwell 2015 (only two SKUs)
Skylake 2015 (3 months after official release and still in short supply)

AMD
FX Bulldozer 2011
Llano 2011
FX Vishera 2012
Trinity 2012
Ritchland 2013
Kaveri 2014
Carrizo 2015 (mobile only still in short supply)

Where did Intel do better in Execution ??

The best FX chip you can buy today is based on the 2012 Piledriver core...

And this is relevant in what way exactly ?? It was an AMD decision not to produce any new Bulldozer based high-end SKUs until ZEN.

 

[bystander36]

Intel
SB 2011
Ivy 2012
Haswell 2013
Haswell Refresh 2014
Broadwell 2015 (only two SKUs)
Skylake 2015 (3 months after official release and still in short supply)

AMD
FX Bulldozer 2011
Llano 2011
FX Vishera 2012
Trinity 2012
Ritchland 2013
Kaveri 2014
Carrizo 2015 (mobile only still in short supply)

Where did Intel do better in Execution ??



And this is relevant in what way exactly ?? It was an AMD decision not to produce any new Bulldozer based high-end SKUs until ZEN.

You can go back to Nehalem too, but Intel's chips were significantly faster throughout that whole list. At least as far as the CPU side goes. And as far as notebooks go, they were also much more power efficient, which hurts AMD's mobile options and the APU's.

 

[Arachnotronic]

Intel
SB 2011
Ivy 2012 & Sandy Bridge-E
Haswell 2013 & Ivy Bridge-E
Haswell Refresh 2014 & Haswell-E
Broadwell 2015 (only two SKUs)
Skylake 2015 (3 months after official release and still in short supply)

AMD
FX Bulldozer 2011
Llano 2011
FX Vishera 2012
Trinity 2012
Ritchland 2013
Kaveri 2014
Carrizo 2015 (mobile only still in short supply)

Where did Intel do better in Execution ??

In each of those years, Intel's processors have been much better than the comparable AMD part. Intel has also been more successful at building out many different product SKUs to address many different market segments (reflected in the additions I made above in bold).

I honestly can't believe I have to defend the statement that Intel's execution in PC processors has been better than AMD's.

 

[AtenRa]

You can go back to Nehalem too, but Intel's chips were significantly faster throughout that whole list. At least as far as the CPU side goes. And as far as notebooks go, they were also much more power efficient, which hurts AMD's mobile options and the APU's.

The context was about execution, delays etc , not performance or perf/watt.

 

[Arachnotronic]

The context was about execution, delays etc , not performance or perf/watt.

Product quality and competitiveness is part of "execution."

 

[AtenRa]

In each of those years, Intel's processors have been much better than the comparable AMD part. Intel has also been more successful at building out many different product SKUs to address many different market segments (reflected in the additions I made above in bold).

I honestly can't believe I have to defend the statement that Intel's execution in PC processors has been better than AMD's.

Ok, your "execution" is not in the same context as my "execution"

 

[squirrel dog]

Having the cpu market cornered is good only for intel and it's share holders . Users end up with over priced product without market competition.

 

[BigDaveX]

The context was about execution, delays etc , not performance or perf/watt.

Even if you go via that logic, Intel still has AMD beat out via virtue of the fact that AMD had to pull the plug on two entire product lines - four-module incarnations of Steamroller and Excavator - because they wouldn't have been competitive with what Intel had to offer. The worst Intel has done was reduce the number of desktop Broadwell SKUs.

 

[DrMrLordX]

Yeah, it's almost as if making something out of billions of low-double-digit nanometer transistors was REALLY freaking hard. People like to the point the finger at the failings of tech companies (GloFo, AMD, Nvidia, TSMC IBM, Intel....ect) but it's downright amazing what the industry at large has achieved.

It is amazing what they have achieved thus far. Expectations are running high, and investors are still pouring billions of <insertnameofcurrencyhere> into the industry. They expect further progress. Those of us who were spoiled by continuous node shrinks have similar expectations.

Even if you go via that logic, Intel still has AMD beat out via virtue of the fact that AMD had to pull the plug on two entire product lines - four-module incarnations of Steamroller and Excavator - because they wouldn't have been competitive with what Intel had to offer.

We'll never know exactly why they did that, though a lot of it has to do with the fact that they couldn't effectively/efficiently carry out node shrinks of their SOI process. 28nm shp was not suitable for 4m chips (for whatever reason) back when they had to make decisions like that. IBM's 22nm SOI was presumably available, yet GF/AMD didn't bite on that.

So it really makes you wonder, who exactly failed to execute here? GF is a separate entity despite the cloying WSA. About the only thing I can think of really stopping AMD from moving forward with 4m SR/XV chips was the issue of Piledriver's large n' slow L3. AMD may not have had a suitable "fix" for that problem on future processors. L3-less chips would have been suitable for 1p systems, but not mp systems. HT Assist wants/needs L3.

The worst Intel has done was reduce the number of desktop Broadwell SKUs.

Let's not forget the Cannonlake/10nm delay. Officially it hasn't happened yet since it's only 2015, but we're getting 14nm Kabylake for a reason.

. . .

So that one vague presser from AMD sure generated a lot of banter, as usual. Will Zen measure up? I don't know. There's a lot riding on this release, and AMD is sure going to a lot of trouble to SOUND confident about it. I for one would welcome GF stepping up to the plate with a working sub-28nm node. Even XV on 14nm would be exciting. The prospect of something better than that, and on a new node, is very interesting. I don't think there's much more we can discuss here of substance until AMD releases something more concrete.

 

[dark zero]

Having the cpu market cornered is good only for intel and it's share holders . Users end up with over priced product without market competition.

MS and the developers will be tired (if they are not already tired) and they will switch to ARM in order to get to more people.

 

[Fjodor2001]

I think you really want/hope to see AMD "beat" Intel for whatever reason (you seem to have a dislike for Intel from your posting history), and that's fine, but don't get mad when people provide you with a reality check.

You're misinterpreting me. I don't care who beats who. But I'd like to see some more progress and competition than what we're seeing right now in the desktop CPU segment. And one thing is for sure, that won't happen unless AMD has competitive products to offer.

Also, I look from it from a pure technological perspective. And I have to say it will be interesting to learn more about both the Zen CPU and the Zen APUs. The former because it's a completely new big core x86 uArch, something we have not seen in a long time. And the latter because the combination of "new x86 uArch + fast GPU cores + HBM + 14 nm" can potentially be a very desirable combination.

p.s. I am not part of the "Intel Sales Team" as you put it -- have never been on Intel payroll and don't expect that I ever will be given my line of work -- but I do own Intel stock.

In that case I don't see why you so strongly express your hopes that AMD will fail with Zen. If they do, what good will come out of that for you? Are you more interested in your Intel stocks rising than having progress and competition in the CPU segment?

 

[BigDaveX]

We'll never know exactly why they did that, though a lot of it has to do with the fact that they couldn't effectively/efficiently carry out node shrinks of their SOI process. 28nm shp was not suitable for 4m chips (for whatever reason) back when they had to make decisions like that. IBM's 22nm SOI was presumably available, yet GF/AMD didn't bite on that.

So it really makes you wonder, who exactly failed to execute here? GF is a separate entity despite the cloying WSA. About the only thing I can think of really stopping AMD from moving forward with 4m SR/XV chips was the issue of Piledriver's large n' slow L3. AMD may not have had a suitable "fix" for that problem on future processors. L3-less chips would have been suitable for 1p systems, but not mp systems. HT Assist wants/needs L3.

To be fair, AMD were probably expecting Haswell and Skylake to be bigger performance leaps than they actually ended up being. Even with what we got however, Steamroller and Excavator would have only maintained the status quo (about level in anything that used all eight cores, but slaughtered in anything single-threaded or lightly multi-threaded) that Piledriver already had against Ivy Bridge, which was clearly getting them nowhere.

Let's not forget the Cannonlake/10nm delay. Officially it hasn't happened yet since it's only 2015, but we're getting 14nm Kabylake for a reason.

True enough, though if that node's providing difficulties for Intel chances are it will for GF as well, seeing how the last time that AMD/GF produced a better process than Intel was all the way back with the 90nm node. We'll just have to see whether or not GF can deliver with 14nm first.

 

[DrMrLordX]

To be fair, AMD were probably expecting Haswell and Skylake to be bigger performance leaps than they actually ended up being. Even with what we got however, Steamroller and Excavator would have only maintained the status quo (about level in anything that used all eight cores, but slaughtered in anything single-threaded or lightly multi-threaded) that Piledriver already had against Ivy Bridge, which was clearly getting them nowhere.

It would be interesting to compare overall revenues and profits for AMD from launch of Bulldozer to, let's say, Q4 2013 and then look at AMD's revenues from Q1 2014 to the present. I like Kaveri, buuuuut honestly I think had they prioritized the "big core" server/enthusiast FX chips over APUs, they probably would have been better off financially speaking. They sort of gave up on that and tried going a different way, which fizzled out since they didn't have the expansive software toolset necessary to help developers make full use of APUs. In their defense, I do not think they really had a process available - save IBM's 22nm SOI - that would have let them continue pushing the envelope on the Opteron/FX front. So they did the best they could with 28nm shp and the rest is history.

True enough, though if that node's providing difficulties for Intel chances are it will for GF as well, seeing how the last time that AMD/GF produced a better process than Intel was all the way back with the 90nm node. We'll just have to see whether or not GF can deliver with 14nm first.

We'll just have to see for ourselves. Samsung seems to be doing okay with 14nm LPE (isn't that what they've used for the A9x?), but Samsung != GlobalFoundries.

 

[ShintaiDK]

6-8 core FX CPUs was never sold in big numbers. Despite failing, APUs is still the bread and butter.