[Austin Statesman] AMD sees a way forward (with new Zen design)

[VR Enthusiast]

No, they won't be, which is why I said "never". If your profit is $13 billion a year, losing $4 billion still makes you profitable. Practically they cannot lose $4 billion a year because the CEO would be ousted eventually.

If your competitors spend 11 billion or more on fabs, how do you stay ahead of them while spending 8 billion? CapEX counts and TSMC and GlobalFoundries continue to spend more money on fabs while Intel is forced to spend less and less every year because of hair-brained contra revenue schemes.

As nice as the A9X is, I suspect when we see Surface Pro 4 14" come out with a Skylake mobile processor, it will destroy the iPad Pro in benchmarks.

Like that'll matter, Apple will still outsell it by 100-1. Nobody is safe from them and least of all Intel.

 

[Boze]

Like that'll matter, Apple will still outsell it by 100-1. Nobody is safe from them and least of all Intel.

On that... we can agree.

 

[DrMrLordX]

You're really latching onto Cinebench I see, even after I said I threw it out there for the hell of it.

But you aren't answering the question. If I could show you that number, would you accept it as plausible? Yes or no?

FYI, Cinebench R10 is an interesting bench since:

Cinebench is generally held to (at least slightly) favor Intel processors
Cinebench scales pretty well with core/thread count. It certainly goes as far as 8c/16t, such as is provided by the 5960x
Cinebench is mostly/all FP, which has clasically been a weak point for AMD and a strong point for Intel
Cinebench R10 is old enough that newer ISAs won't affect its performance. It's "legacy" code that shows how older FP software might perform, and how newer unoptimized code might perform.
All three Cinebenches are frequently used to demonstrate the power of new processors anyway (including Skylake, most recently).

Regardless, the question still stands. What say ye?

 

[CHADBOGA]

You're assuming Intel will keep on making the same $13 billion for the next 10yrs, suffice to say that's a prediction I won't back.

Clearly they will make more than that in the next 10 years,

Then comes the real downer, custom ARM cores are getting to the point where they can replace Intel Xeons at the low end

No more successfully than Opteron has done.

& we certainly know that virtually every tech behemoth, like Google & FB, are looking to go this route in the future.

With AMD being no option, of course all these companies are "looking" at ARM, otherwise how do you think the pricing negotiations they do with Intel would go?

 

[CHADBOGA]

Finally, everyone trying to hem and haw and massage whatever numbers they can to make Zen look attractive, you're wasting your time, because all of you consistently try to compare Zen against what Intel offers right now.

Zen is a year away. If you think Intel will sit on the sidelines and let AMD catch up, you are sorely mistaken.

More likely it will be 18 months away.

 

[R0H1T]

Clearly they will make more than that in the next 10 years,

Are you counting that for each year or 10yrs as a whole :sneaky:

No more successfully than Opteron has done.

Totally different kettle of fish, so not comparable.

With AMD being no option, of course all these companies are "looking" at ARM, otherwise how do you think the pricing negotiations they do with Intel would go?

Really, is that why Apple's going full steam ahead with their custom cores & Samsung's rumored to have gone with Mongoose next year? There is no logical reason to think that these firms can't (or won't) replace their own severs with custom ARM cores, this isn't just a fluke & the billions spent in developing them will eventually be geared towards high performance (in house?) usage. You really think this is just smoke & mirrors to get cheap x86 from Intel :hmm:

 

[frozentundra123456]

More likely it will be 18 months away.

Definitely, maybe even longer for any kind of real availability.

 

[CHADBOGA]

Are you counting that for each year or 10yrs as a whole :sneaky:
Totally different kettle of fish, so not comparable.
Really, is that why Apple's going full steam ahead with their custom cores & Samsung's rumored to have gone with Mongoose next year? There is no logical reason to think that these firms can't (or won't) replace their own severs with custom ARM cores, this isn't just a fluke & the billions spent in developing them will eventually be geared towards high performance (in house?) usage. You really think this is just smoke & mirrors to get cheap x86 from Intel :hmm:

Why have you gone from Facebook & Google, to mentioning Apple & Samsung?

Try to stay focused FFS.

 

[R0H1T]

Why have you gone from Facebook & Google, to mentioning Apple & Samsung?

Try to stay focused FFS.

Because these two already have custom ARM cores which could be deployed in servers, in house at least initially. If they can you can bet Google & FB certainly will, spend a few billion dollars now & save many more in the future & that's what everyone can learn from Apple.

 

[Arachnotronic]

@ R0H1T

No way that Apple gets into the business of building server processors. This is simply absurd.

Samsung had an in-house server effort, but it was completely canned a while back. A lot of the folks Samsung had hired to work on these chips have been redirected to Samsung's (server) memory businesses.

 

[R0H1T]

@ R0H1T

No way that Apple gets into the business of building server processors. This is simply absurd.

Samsung had an in-house server effort, but it was completely canned a while back. A lot of the folks Samsung had hired to work on these chips have been redirected to Samsung's (server) memory businesses.

What happens next if/when they replace Intel for the Macbooks? The desktops are next & servers, maybe just in house, a logical followup to that. Intel process lead will vanish at 10~7nm & then it's open season, that's when I predict ARM to hit them hard. The major reason why Intel's sustained their server dominance is partly due to their vastly superior foundries, however ARM is the king of low power & they're expanding quickly (more cores albeit just in phones atm) & improving tremendously. So, logically (at least for me) it makes sense to get a high IPC, relatively low power core in your servers. It's not as far fetched as some think it is, & just to reiterate Intel isn't as infallible as they're made out to be ^_^

Just to add a few things here ~ I don't see Intel going forward with their next full node shrink(10nm) unless they make serious inroads into the mobile/tablet market or better still they produce large volumes of chips for other fabless chipmakers like AMD

So if neither of these conditions are fulfilled it'll be doubly hard for them to pursue their tick-tock strategy any further than ~2016 ! Not saying that they won't go to 10nm rather their next full node shrink will be delayed by atleast a year or two

One prediction that I did get right

 

[Qwertilot]

@ R0H1T

No way that Apple gets into the business of building server processors. This is simply absurd.

Absurd is rather strong Apple have clearly got the capability to roll a plausibly effective (for some uses) server chip should they feel the need. Also building a fair few data centers etc.

The motivation would be just like the Ax stuff - cost saved on each processor/the extra control they got over precisely what goes into it vs the sunk cost of designing/building it.

I'm sure they've been doing those sums/evaluating. Have to imagine the really huge scale people like Google/FB etc will have been too. What the answer to that is/if anything will ever happen, is of course another matter.

 

[frozentundra123456]

Well this thread is about Zen, and those who seem take some satisfaction at seeing Intel theoretically replaced in servers by ARM, had better be careful what they wish for, because if ARM replaces Intel, the odds against Zen being a success become even higher. Bad enough that they have to compete with one company with vastly superior resources. How are they going to compete with the equally powerful ARM makers who have vastly superior resources as well, and are even more cutthroat than Intel and will undoubtedly sell for lower margins.

 

[Boze]

I love how the AMD fanboys are predicting the end of Intel now that AMD is CTD, with the company's only possible hope of life support being Zen. Which need to arrive sooner rather than later.

And I got some news for you guys who are so far up AMD's a** that you think trying to predict the demise of Intel will be a successful countertactic. It won't. AMD is going away one way or another. They've been a complete failure at everything they've attempted. Desktop. Mobile. Server.

Fail. Fail. Fail.

Intel will do what its always done. Spend more money on R&D and crush its opposition.

 

[PPB]

More R and D than apple for example? Intel is the elephant in the x86 room, against the other heavywheights in the ARM space, not so much.

 

[frozentundra123456]

More R and D than apple for example? Intel is the elephant in the x86 room, against the other heavywheights in the ARM space, not so much.

If you are referring to my post, obviously the comparison was Intel vs AMD and AMD vs ARM, both of which could use AMD's research budget as petty cash. If Apple gets into servers too, that is just a third company that will dwarf AMD in terms of resources, both for R and D and actual implementation of the product. So the point is, AMD had better hope x86 retains its dominance of the server and high end cpu market if Zen is to have one iota of a chance of success.

 

[R0H1T]

Well this thread is about Zen, and those who seem take some satisfaction at seeing Intel theoretically replaced in servers by ARM, had better be careful what they wish for, because if ARM replaces Intel, the odds against Zen being a success become even higher. Bad enough that they have to compete with one company with vastly superior resources. How are they going to compete with the equally powerful ARM makers who have vastly superior resources as well, and are even more cutthroat than Intel and will undoubtedly sell for lower margins.

Theoretically? I did say 2.5yrs back that the next die shrink from Intel would be late & what'd they do in the meantime, sink some $8 billion dollars to prop up Atom to the point where they don't even mention contra revenues in the balance sheet. Likewise it's only one's imaginations that's self defeating or limiting & if you don't aim higher you'll never reach the pinnacle. If I were Apple I'd make sure to end all of my current reliance on an external entity, like Intel & x86, for obvious reasons & the same goes for Google & FB.

As for Zen, it's first & foremost a high performance (also server oriented IIRC) part that'll compete in the high margin low volume segment of the market where it'd be launched. It's fate will be decided by market forces & not what Intel wishes for.

 

[Arachnotronic]

If I were Apple I'd make sure to end all of my current reliance on an external entity, like Intel & x86, for obvious reasons & the same goes for Google & FB.

Have you ever run a company like Apple, Google, or Facebook?

 

[R0H1T]

Have you ever run a company like Apple, Google, or Facebook?

Definitely not but they're certainly on my acquisition list :biggrin:

Back to the point, I see cyclical (or seasonal as I'd like to call it) & then secular trends. I did say back in the day that x86 user base is shrinking & that goes without saying, however the biggest enemy of Intel is Intel itself & their egregious margins. I'm sure there will come a time when the likes of Google & FB, even Apple with its cloud services, will refuse to pay the Intel tax. As it stands today Intel is standing on its two feet, one being the Wintel duopoly & the other obviously servers. I don't see why as ARM erodes one that it can't do the same to the other, if custom ARM core makers are determined to get a viable x86 replacement then there's no real barrier stopping them from doing so. Like I said, it's only a matter of time & vision & the $$ spent will be well worth it in the end, as Apple has shown us.

Intel saying no to Apple a decade back may well turn out to be their biggest blunder yet, as for AMD there's still some hope left. Their GPU division being spun off could make it an attractive target for say Apple, now depending on how Zen performs in the market the rest of the company may or may not survive past 2017.

 

[DrMrLordX]

I love how the AMD fanboys are predicting the end of Intel now that AMD is CTD, with the company's only possible hope of life support being Zen. Which need to arrive sooner rather than later.

And I got some news for you guys who are so far up AMD's a** that you think trying to predict the demise of Intel will be a successful countertactic. It won't. AMD is going away one way or another. They've been a complete failure at everything they've attempted. Desktop. Mobile. Server.

Fail. Fail. Fail.

Intel will do what its always done. Spend more money on R&D and crush its opposition.

Le sigh.

Intel isn't going anywhere. Apple will not kill them. Zen won't kill them either.

But here's one thing it might do . . .

http://cdn.overclock.net/0/05/052237f1_CZ-CB10-3.4G-Static.png

From:

http://www.overclock.net/t/1560230/jagatreview-hands-on-amd-fx-8800p-carrizo/400_100#post_24310470

And the Carrizo in question was running @ 3.4 GHz (static no throttling) during the Cinebench R10 run:

http://www.overclock.net/t/1560230/jagatreview-hands-on-amd-fx-8800p-carrizo/400_100#post_24314113

So what can we tell from a Cinebench R10 score of 13146 from a 2m/4t Excavator @ 3.4 GHz?

13146 * 1.4 * 4 = 73617.6 (round to 73618)

That's assuming 8c/16t Zen, 40% increase in IPC, and a clockspeed of 3.4 GHz. Now let's look at Skylake running at a static clockspeed. How about 4.8 GHz Skylake:

http://www.overclockers.com/wp-content/uploads/2015/07/intel_i76700K-46.jpg

From:

http://www.overclockers.com/intel-skylake-i7-6700k-cpu-review/

40731. Nice score right? Assuming an 8c/16t Skylake-E could hit the same clockspeed, you'd be looking at a score of 81462.

Normalizing for clockspeed, Skylake has a performance of ~2121 CB per GHZ per core. Kabylake should be no different. Zen (according to AMD's estimate of a 40% IPC increase) should have a performance of ~2707 per GHz per core.

Hmmmmm!

 

[Boze]

Its really sad that this forum doesn't allow "bad language" (Christ that's pathetic..) because there are choice words to use to describe the level of delusion I'm seeing...

You honestly think Zen is going to be an 8 core / 16 thread part... okay.

 

[Arachnotronic]

Its really sad that this forum doesn't allow "bad language" (Christ that's pathetic..) because there are choice words to use to describe the level of delusion I'm seeing...

You honestly think Zen is going to be an 8 core / 16 thread part... okay.

Zen will be an 8C/16T part at the top. There are plenty of unknowns though. Will AMD hit its perf/clock goals? What kind of clocks will AMD hit? What will AMD charge for the thing?

 

[Boze]

Again you're fixated on Cinebench, when I've said once already that I just drew a name out of a hat...

I'll give you the same deal I offered What'sHisFace. If a $370 Zen can compete against a $370 Core i7, and if a $600 Zen can compete against a $600 Skylake-E, and if a $1000 Zen can compete against an i7 6960X or i7 7960X, then I'll never post here again.

Take the deal. If you're convinced that AMD will release some magical savior product. If not, keep blowing smoke.

 

[CHADBOGA]

One prediction that I did get right

Even a blind chicken finds the odd kernel of corn.

 

[Arachnotronic]

Definitely not but they're certainly on my acquisition list :biggrin:

Back to the point, I see cyclical (or seasonal as I'd like to call it) & then secular trends. I did say back in the day that x86 user base is shrinking & that goes without saying, however the biggest enemy of Intel is Intel itself & their egregious margins.

And what, pray tell, is so "egregious" about Intel's margins?