[AMD] K12 will be on 28nm

[Phynaz]

Stubborness is not an argument, i guess that thoses numbers hurts your brand preference, it s you that are relying on myths and deflections as a mean to patheticaly negates numbers..

As for the bolded part good luck for compensating, at equal competences with the competition, what amount to 20% lower perf/watt due to process

You know what's really amazing? If we take your argument (design can barely lead to a 10% efficiency improvement) and apply it to AMD it actually makes AMD look even worse. You're saying all the efficiency improvements AMD has delivered over the last couple of years - the improvements over Bulldozer - aren't actually do to all the work AMD has done to improve their designs.

You're argument says the improvements have actually come from GloFo's process improvements and not AMD's engineering efforts. I wonder how you reconcile that with your prior posts. Or even with the existence of the cat cores.

Kinda stepped in it, haven't you?

 

[inf64]

I highly doubt k12 and Zen will be on 28nm.

 

[Abwx]

You know what's really amazing? If we take your argument (design can barely lead to a 10% efficiency improvement) and apply it to AMD it actually makes AMD look even worse. You're saying all the efficiency improvements AMD has delivered over the last couple of years - the improvements over Bulldozer - aren't actually do to all the work AMD has done to improve their designs.

You're argument says the improvements have actually come from GloFo's process improvements and not AMD's engineering efforts. I wonder how you reconcile that with your prior posts. Or even with the existence of the cat cores.

Kinda stepped in it, haven't you?

They surely optimised as much as possible for low power but if you did take attention you would had noticed that they got very competitive in this segment once their chip was fabbed at GF, 19% and 38% less leakage at the CPU and GPU levels compared to TSMC s fabbed chips, so yes, GF process had a lot to do with their Beema/Mullins competitivness.

 

[positivedoppler]

Ok, since this is pretty much proven false, is there still any reason for any type of debate in this thread?

 

[jdubs03]

Nope, not really.

 

[AtenRa]

For certain scenarios, it's not worth to use AMD processors even if someone gives them for free.

Same applies to every hardware from every manufacturer.

 

[piesquared]

Ok, since this is pretty much proven false, is there still any reason for any type of debate in this thread?

Isn't that status quo for these forums? There's troll post after troll post after troll thread after troll thread. The title of this thread isn't even accurate and completely taking out of context what the speaker said, he did not say "will". That was just the fabrications and prayers of a shill.

Not surprising though considering the author.

 

[jdubs03]

Isn't that status quo for these forums? There's troll post after troll post after troll thread after troll thread. The title of this thread isn't even accurate and completely taking out of context what the speaker said, he did not say "will". That was just the fabrications and prayers of a shill.

Not surprising though considering the author.

I think you're being a bit extreme; there is quite a bit of trolling, but there are some good contents in some of the threads, though some content does get recycled. But, based on what the AMD dude said, it was confusing and did not make it seem like they would be going to a new node. Blame his incompetence in creating a coherent response rather than forum members who are just trying to question what the guy said.

 

[Phynaz]

They surely optimised as much as possible for low power but if you did take attention you would had noticed that they got very competitive in this segment once their chip was fabbed at GF, 19% and 38% less leakage at the CPU and GPU levels compared to TSMC s fabbed chips, so yes, GF process had a lot to do with their Beema/Mullins competitivness.

When was the FX line ever not fabbed at GloFo?

And you tell me to pay attention.

 

[Abwx]

When was the FX line ever not fabbed at GloFo?

And you tell me to pay attention.

That s right, you should pay attention to what is said, otherwise your posts will be redundants, or more precisely even more redundants.

if you did take attention you would had noticed that they got very competitive in this segment once their chip was fabbed at GF, 19% and 38% less leakage at the CPU and GPU levels compared to TSMC s fabbed chips, so yes, GF process had a lot to do with their Beema/Mullins competitivness.

FX is using 32nm while Beema is at 28nm and the leakage comparison was between GF and TSMC 28nm s of course, for the 32nm process i already gave some precisions in this very thread.

Currently the only advantage of Intel is said lower voltage, that is, process, a FX need 17% more voltage at 3.5 than a 4770K, this induce 37% more TDP, and despite this a FX8370E manage to be at same power efficency than a 4670K on high throughput softs, at equal process the FX would had better efficency than a 4770K, let alone if it s a Kaveri based FX.

Some think that Intel is getting to smaller nodes early (at all costs!!) because of costs, actualy the truth is that without a node advantage their CPUs wouldnt hold the perf/watt ratios crowns, no wonder that they are pouring huge amounts of money in process RD and capital expenses.

 

[Enigmoid]

Some think that Intel is getting to smaller nodes early (at all costs!!) because of costs, actualy the truth is that without a node advantage their CPUs wouldnt hold the perf/watt ratios crowns, no wonder that they are pouring huge amounts of money in process RD and capital expenses.

Sandy Bridge in servers destroyed Bulldozer.

And before you start pulling nonsense die sizes and performance charts (84 W in fritz - 10% VRM - etc. nonsense when what people care about is draw from the wall socket) that is the reason why AMD is virtually nonexistent in the server space.

And the some is pretty much exclusively you.

 

[AtenRa]

Sandy Bridge in servers destroyed Bulldozer.

If you are talking about sales yes, but not performance.

If you see SPECint and SPECfp they where very close.

 

[Abwx]

Sandy Bridge in servers destroyed Bulldozer.

And before you start pulling nonsense die sizes and performance charts (84 W in fritz - 10% VRM - etc. nonsense when what people care about is draw from the wall socket) that is the reason why AMD is virtually nonexistent in the server space.

And the some is pretty much exclusively you.

The hurry to dismiss any bench that do not support your preference just showed how cluless you are, Haswell i5 has 9% higher IPC than an IB i5 in Fitzchess and has better perfs than a 8370E, but you are right, next time i ll use Winrar or 7zip where the i5 did get nothing in IPC and get trounced by 35-40% by said 8370E, notice that Anand use 7 zip to check the servers Integer IPC, it is indeed more relevant than Fritz to compare servers chips.

 

[Arachnotronic]

If you are talking about sales yes, but not performance.

If you see SPECint and SPECfp they where very close.

What about performance per watt?

 

[Phynaz]

That s right, you should pay attention to what is said, otherwise your posts will be redundants, or more precisely even more redundants.



FX is using 32nm while Beema is at 28nm and the leakage comparison was between GF and TSMC 28nm s of course, for the 32nm process i already gave some precisions in this very thread.



Some think that Intel is getting to smaller nodes early (at all costs!!) because of costs, actualy the truth is that without a node advantage their CPUs wouldnt hold the perf/watt ratios crowns, no wonder that they are pouring huge amounts of money in process RD and capital expenses.

I never mentioned Beema. I specifically said the big core line.

But I just checked, and the Vcore (and therefore power consumption) went up between Zambezi and Vishera.

So I guess you're right, AMD is not making efficiency improvements to their designs.

Lol, more voltage and higher efficency is just complete non sense

 

[ShintaiDK]

If you are talking about sales yes, but not performance.

If you see SPECint and SPECfp they where very close.

Performance wise AMD got beaten so badly.
http://www.anandtech.com/show/5553/the-xeon-e52600-dual-sandybridge-for-servers

Bulldozer failed everywhere for everything.

 

[AtenRa]

What about performance per watt?

Depends on the application. Vishera based Opterons where extremely close in many workloads vs SB.

It was 22nm Ivy that raised the bar of perf/watt against Bulldozer, not SB.

 

[Phynaz]

What about performance per watt?

It's not relevant when AMD doesn't have an advantage :biggrin:

 

[Abwx]

But I just checked, and the Vcore (and therefore power consumption) went up between Zambezi and Vishera.

Because 10% higher frequency will require something like 4.6% higher voltage, not that easy to extract conclusions without a minimum of, accurate, thoughts, isnt it, let alone with some huge bias.

 

[positivedoppler]

This thread needs to be put out of its miserly.

 

[Abwx]

This thread needs to be put out of its miserly.

It was a troll thread anyway, and the OP has completely disappeared of the scene for some reason....

 

[Phynaz]

Because 10% higher frequency will require something like 4.6% higher voltage, not that easy to extract conclusions without a minimum of, accurate, thoughts, isnt it, let alone with some huge bias.

Except the 8350 doesn't have a 10% higher frequency than 8150. And the 8320 doesn't have a 10% higher frequency than 8120. When they are both at maximum voltage they have 100Mhz additional frequency.

AMD bumped voltage and power consumption to get an extra 100Mhz out of the chips. Using your numbers they bumped voltage by 4.6% to get a 2.4% increase in frequency.

We also know that power consumption increases by the cube of the voltage and the square of the frequency. I'll leave you to do the math.

Like you said, AMD hasn't improved the efficiency of their big core designs. I was wrong when I thought they had. So again you were correct when you said the following:

Lol, more voltage and higher efficency is just complete non sense

I don't know why you are arguing with me. I said you are correct!

Anyway, I'm done replying to your circular arguments. Your agenda is blatantly obvious, and any more of this back and forth is going to get the mods attention.

 

[rainy]

Except the 8350 doesn't have a 10% higher frequency than 8150.

It have - in fact, base clock is 11 percent higher (4.0 GHz vs 3.6).

http://www.cpu-world.com/CPUs/Bulldozer/AMD-FX-Series FX-8350.html
http://www.cpu-world.com/CPUs/Bulldozer/AMD-FX-Series FX-8150.html

 

[Phynaz]

It have - in fact, base clock is 11 percent higher (4.0 GHz vs 3.6).

http://www.cpu-world.com/CPUs/Bulldozer/AMD-FX-Series FX-8350.html
http://www.cpu-world.com/CPUs/Bulldozer/AMD-FX-Series FX-8150.html

Full load clocks are 100Mhz higher. Who cares about clocks for an idle CPU? AMD tried tried that one when they intentionally lowered the base clocks to be able to say they had more turbo. Nobody fell for that marketing tactic.

 

[rainy]

Full load clocks are 100Mhz higher. Who cares about clocks for an idle CPU? AMD tried tried that one when they intentionally lowered the base clocks to be able to say they had more turbo. Nobody fell for that marketing tactic.

Quite obviously you're never wrong, isn't?