AMD "Summit Ridge" Silicon Reserved for 8-core CPUs Initially

[cdawall]

Well, it's not based off of nothing: AMD has been promising better performance for years and constantly underdelivered.
But my educated guess is also: we don't know at this point.

That's called stipulations. You know nothing and you are guessing. Remember amds low power isn't based off of fx either and offers 3-4x the performance per clock and a 50% performance per clock increase over phenom ii. Not everything is an fx chip. Intel promised better performance from netburst as well. That's how we got the p4 3.6ghz.

This chip is a ground up redesign with arguably one of the best chip designers to grace this earth engineering it. Will it be great? Fuck if I know, but I doubt it sucks.

 

[Fx]

I can't freaking wait. I am getting one.

 

[bug]

That's called stipulations. You know nothing and you are guessing. Remember amds low power isn't based off of fx either and offers 3-4x the performance per clock and a 50% performance per clock increase over phenom ii. Not everything is an fx chip. Intel promised better performance from netburst as well. That's how we got the p4 3.6ghz.

This chip is a ground up redesign with arguably one of the best chip designers to grace this earth engineering it. Will it be great? Fuck if I know, but I doubt it sucks.

Didn't we hear that for K10 and Bulldozer? We all hope for the best, but you completely ignore AMD's past track record?
And it's not stipulations, it's history. History may change, but more often then not it tends to repeat itself.
Just like you, I'm thinking the addition of Jim Keller must change things this time around. And probably unlike you, I'm afraid the gap to close may be too big even for Jim.
In the end, what can you do? We can't all keep quiet until Zen is released. So we'll just speculate instead. Based on our hopes, on AMD's track record, on our affinities to either AMD or intel...

 

[cdawall]

Didn't we hear that for K10 and Bulldozer? We all hope for the best, but you completely ignore AMD's past track record?
And it's not stipulations, it's history. History may change, but more often then not it tends to repeat itself.
Just like you, I'm thinking the addition of Jim Keller must change things this time around. And probably unlike you, I'm afraid the gap to close may be too big even for Jim.
In the end, what can you do? We can't all keep quiet until Zen is released. So we'll just speculate instead. Based on our hopes, on AMD's track record, on our affinities to either AMD or intel...

AMD has good and bad just like Intel. Remember Intel stuck with netburst just as long as amd stuck with fx

 

[rruff]

Today AMD only makes sense if you're scraping the bottom of the barrel. Here's hoping Zen will change that, even if I'm not holding my breath.

AMD is only claiming a 40% performance increase per core clock, yes? And surely that is best case? Intel is already ahead of that. That doesn't mean Zen won't perform better per price compared to current Intel, particularly if the midrange FX chips have 8c/16t. But I very much doubt they will be rivaling Intel at the top end, especially since we won't be seeing them anytime soon.

The last rumor I heard is that we will see consumer Zen in about a year. The server chips will come first. Resources at AMD are still really tight. There is only so much they can do.

 

[medi01]

AMD is only claiming a 40% performance increase per core clock, yes? And surely that is best case?

That was one of the goals for Zen, although I don't recall what they used as a baseline.
"Surely" is nothing. Although I am be more than happy if AMD rolls out competitive mid range, and fuck high end.

Heck, you can't close gap this wide in one go, what high end.

Semicustom (mostly consoles) is about 50% of AMD business and that is what keeps it afloat. (let's bash them more for buying ATI, shall we?)

 

[cdawall]

AMD is only claiming a 40% performance increase per core clock, yes? And surely that is best case? Intel is already ahead of that. That doesn't mean Zen won't perform better per price compared to current Intel, particularly if the midrange FX chips have 8c/16t. But I very much doubt they will be rivaling Intel at the top end, especially since we won't be seeing them anytime soon.

The last rumor I heard is that we will see consumer Zen in about a year. The server chips will come first. Resources at AMD are still really tight. There is only so much they can do.

40% per clock would make them at least viable again. Add in the 8c/16t combo and keep clocks up and it should be pretty powerful for a normal desktop chip.

 

[HD64G]

This 40% advance for Zen over Excavator architecture is without calculationg the advance in performance coming from lowering manufacturing from 28nm to 14nm. So, even if 40% is proven to be 30-35%, we need to add close to 80% more. Then we have a good estimation of how it will perform compared to Excavator and by this to Intel's CPUs.

 

[cdawall]

This 40% advance for Zen over Excavator architecture is without calculationg the advance in performance coming from lowering manufacturing from 28nm to 14nm. So, even if 40% is proved to be 30-35%, we need to add close to 80% more. Then we have a good estimation of how it will perform compared to Excavator and by this to Intel's CPUs.

Shrinking a die doesn't make it faster so I have literally no idea what you are saying.

 

[HD64G]

Shrinking a die doesn't make it faster so I have literally no idea what you are saying.

Keeping the die in same size when going from 28nm to 14nm though is making the die twice more powerful in the same TDP by having twice the transistor count in it, eh? Logical speaking, that's why I spoke about 80% more powerful cpu through more advanced manufacturing process as they will try to make Zen less greedy in power consumtion than last FXs.

Zen will be a big core design compared to the modulated one of BD for sure, so, not so hard to make the above hypothesis me thinks. After all Intel had much bigger cores than AMD since 1st iX gen of them to have room for HT in it by seperating it in 2 threads. Somewhat like SMT of Zen's.

 

[cdawall]

Keeping the die in same size when going from 28nm to 14nm though is making the die twice more powerful in the same TDP by having twice the transistor count in it, eh? Logical speaking, that's why I spoke about 80% more powerful cpu through more advanced manufacturing process as they will try to make Zen less greedy in power consumtion than last FXs.

Zen will be a big core design compared to the modulated one of BD for sure, so, not so hard to make the above hypothesis me thinks. After all Intel had much bigger cores than AMD since 1st iX gen of them to have room for HT in it by seperating it in 2 threads. Somewhat like SMT of Zen's.

No. Want proof? Northwood vs Prescott, ivy vs Sandy or better yet another halfing of size 130nm clawhammer/Newcastle vs 65nm Brisbane/lima. There was no doubling of performance. Performance increase is based off of design changes not die size changes. Otherwise the current i7 would be what 2-3x as fast as the current models based off of die shrink alone.

 

[doxology]

People are speculating because of AMD's past. Granted it is because their recent moves haven't been that great, but here is the thing. This was developed ground up using the newest technology available and developed by someone that knows what he is doing.

The past means nothing as this is a new chip. AMD has been limping by with their CPU's for years. As a company this product has to pull through. It is why they went back to a more traditional chip design. As people have seen this die shrink can be a huge asset in power consumption. People that have seen the newest GPU's from AMD have been amazed what it can do with its power draw. The below example isn't the best video I have seen on this but was the first one I found with a simple google search on Polaris power draw.

So all things taken into consideration so far it would seem AMD should have the intention and ability to make a competitive CPU. Even more so if you consider the fact that Intel hasn't really made any super ground breaking moves with their CPU's for a while now. They have just been chugging along. Granted you would have to expect that Intel themselves might not be that far away from their next CPU line as the Core line has been going for a while now and maybe they have been waiting on that to see what AMD has to offer.

It is easy to doubt AMD, but considering what is on the line for AMD as a company at this point it would be unwise to underestimate them, because they can't afford another Bulldozer, in fact another Bulldozer would probably be the end of AMD, and they know that, so you have to expect that if they where going to put the resources into designing a new chip from the ground up that they would make sure it is the best it can be.

 

[PP Mguire]

One reason that makes me think it might be half decent is because they aren't going crazy with PR bullshit trying to hype this thing. Basically completely silent.

 

[cdawall]

One reason that makes me think it might be half decent is because they aren't going crazy with PR bullshit trying to hype this thing. Basically completely silent.

Same thing I have been thinking

 

[rruff]

One reason that makes me think it might be half decent is because they aren't going crazy with PR bullshit trying to hype this thing. Basically completely silent.

Frankly I don't know what the hype was ever meant to accomplish. Maybe they just got smart and laid off the PR team to save money.

 

[xvi]

Jim Keller designed zen, the same guy who designed athlon 64

I was just going to say something along the lines of "Didn't AMD re-hire the guy who made AMD great back in the day?"

 

[newtekie1]

I was just going to say something along the lines of "Didn't AMD re-hire the guy who made AMD great back in the day?"

He also designed the latest Apple chips in their phones and tablets.

 

[cdawall]

He also designed the latest Apple chips in their phones and tablets.

Which also didn't suck

 

[newtekie1]

Which also didn't suck

Correct, in fact he took them from sucking a lot to not sucking.

 

[Fluffmeister]

Problem is they need to go from sucking to exceptional*

Or maybe not, people seem happy to wait for performance they can already get today. /shrugs

*Great porn title.

 

[rruff]

Problem is they need to go from sucking to exceptional*

And then keep not sucking. That's going to be tough for a company in AMD's position.

 

[cdawall]

And then keep not sucking. That's going to be tough for a company in AMD's position.

Having more R&D doesn't always mean better products unluckily.

 

[Fluffmeister]

Having more R&D doesn't always mean better products unluckily.

Unluckily Intel can afford to unluckily luckily.

 

[cdawall]

Unluckily Intel can afford to unluckily luckily.

Yep they could release a shit gen and no one would bat an eye...

 

[Prima.Vera]

I was just going to say something along the lines of "Didn't AMD re-hire the guy who made AMD great back in the day?"

I think it was this guy?

Dirk Meyer

Yep they could release a shit gen and no one would bat an eye...

But if AMD does that, everyone looses their minds!