Leaked Patch Confirms AMD Zen Will Have 32 Cores Per Socket?

[a3venom]

This is precisely why there are no real and clear winners especially when either company has different roadmap schedules. If they had a similar release timeframe for their products (such as HD 4870 and GTX 280) then it would be a more valid comparison but while Nvidia may be the first to market, AMD does make a counter and introduce a similar competitive offering. So it's really equal or similar with no real clear winner unless you look primarily at performance. For example with price/performance the RX480 is currently holding the performance crown (but may be displaced with the GTX 1060) whereas for performance the 1080 holds the performance crown (at the expense of price/performance).

Price/Perf isn't the issue when just saying who has the more advanced technology - Nvidia has the upper hand without any argument as AMD hasn't come up with a performance king (that beats Nvidia by atleast 25%) since a few years now. AMD is always playing catch up, which is not really a big deal but it clearly makes them the underdog/company with the inferior tech.
it doesnt matter if the 1080 was 1000$ and rx480 was $9.99 - we are just talking about technological advancements, not hating on any products from either side.

 

[Zarathustra[H]]

I'm looking forward to Zen in October. Can't come soon enough.

I'm starting to suspect my 5 years of overvolting my i7-3930k to unsafe levels in order to hit 4.8-5.0Ghz (granted, always with adequate cooling) is finally starting to degrade the chip. It seems to be getting hotter and spinning up the fans more frequently than it used to.

(Either that, or the cooler has come loose, I should probably check that)

I just hope it lasts me until October!

 

[Tonkatsu]

I'm looking forward to Zen in October. Can't come soon enough.

Sorry to say, but Zen release in October is a media myth. You'll be lucky to see it in December.

SemiAccurate Forums - View Single Post - Official AMD Zen uarchitecture thread

I mean, come on, look at the timespan between Bulldozer samples (Q3 2010) and Bulldozer release (Oct 12, 2011), and compare to when Zen samples are known to have come out (Q3 2016, i.e. now). And AMD's resources have only dwindled since 2011.

 

[Pieter3dnow]

Sorry to say, but Zen release in October is a media myth. You'll be lucky to see it in December.
SemiAccurate Forums - View Single Post - Official AMD Zen uarchitecture thread
I mean, come on, look at the timespan between Bulldozer samples (Q3 2010) and Bulldozer release (Oct 12, 2011), and compare to when Zen samples are known to have come out (Q3 2016, i.e. now). And AMD's resources have only dwindled since 2011.

Yeah and Bulldozer was delayed by several years the architecture was around 3 years late. So why not put that on Zen too Zen will be 3 years late because that happened to Bulldozer .

Did you know what message you were quoting? The person Bridgman is talking about says his inside sources say November .

 

[Zarathustra[H]]

Sorry to say, but Zen release in October is a media myth. You'll be lucky to see it in December.

SemiAccurate Forums - View Single Post - Official AMD Zen uarchitecture thread

I mean, come on, look at the timespan between Bulldozer samples (Q3 2010) and Bulldozer release (Oct 12, 2011), and compare to when Zen samples are known to have come out (Q3 2016, i.e. now). And AMD's resources have only dwindled since 2011.

Yeah, you are right. The official word now is very limited release late in Q4 2016, followed by volume early in Q1 2017.

 

[doz]

And Intel will be releasing yet another processor with minor gains, albeit gains that AMD will have to catch up to this year.... If the theory that AMD is releasing an 8 core that is equivalent to IvyBridge, they had better price it below $200 for the overclockable/high clocked part to have a fighting chance. I would hope they can come up with a bit more than IB power......

 

[Zarathustra[H]]

And Intel will be releasing yet another processor with minor gains, albeit gains that AMD will have to catch up to this year.... If the theory that AMD is releasing an 8 core that is equivalent to IvyBridge, they had better price it below $200 for the overclockable/high clocked part to have a fighting chance. I would hope they can come up with a bit more than IB power......

Meanwhile I'm still running a Sandy-E at 4.8Ghz and can still keep up with the latest and greatest Intel can offer. My cores are just as fast as overclocked Skylake cores.

Gains have been minimal since sandy, and overclocking potential has gone down, meaning that once overclocked to its potential anything Sandy or newer from Intel is more or less the same speed, unless you got a bad outcome in the silicone lottery.

Besides. Even if you don't ocerclock, anything Sandy or newer is fast enough. Having the fastest CPU no longer matters, as long as it is fast enough. Zen - if it lives up to the pre-market claims - will be fast enough, it doesn't matter if it keeps up with Intel's latest and greatest.

 

[doz]

My 2600k wont hold 4.9ghz anymore, nor 4.2ghz for that matter There IS a difference overclocked vs. stock and its noticeable without benchmarks.

That said, I know it does not need to keep up with Intel's latest but at the same time, 8 core IB power does not equate to 7600k pricing (future).

 

[xorbe]

My 2600k wont hold 4.9ghz anymore, nor 4.2ghz for that matter

Same, my 2500k lost all kinds of MHz at the end. It's why I only use 4.5 on the 4790k system.

 

[Decko87]

Same, my 2500k lost all kinds of MHz at the end. It's why I only use 4.5 on the 4790k system.

2500k still holding steady at 4.5ghz, never had the push it and it's still rewarding me daily.

 

[Tonkatsu]

Yeah and Bulldozer was delayed by several years the architecture was around 3 years late. So why not put that on Zen too Zen will be 3 years late because that happened to Bulldozer .

That is a distinct possibility, yes. Past behavior is not a perfect indicator of future behavior, but it is the best indicator that exists. You know that the typical time between a new microarchitecture sampling and launch is one year, right? Sandy Bridge samples shipped in Q1 2010 while Sandy Bridge launch was Jan 5, 2011. Are you going to call Sandy Bridge a horribly-delayed chip too?

Did you know what message you were quoting? The person Bridgman is talking about says his inside sources say November .

No. Think about it. If his inside sources had said November, why would he have been saying October until John Bridgman had to correct him? The disappointing truth is he misinterpreted what, if anything, he heard. AMD never said Zen would be launched in Q4 2016 for sure, it just said it would be "available", by which it means that it would be selling Zen to OEMs and distributors and have Zen revenue show up on AMD's balance sheets, not that Zen would be available to you or me.

AMD Zen Reportedly Delayed Until Early Next Year - Phoronix Forums

Ok, so availability means "not available". I'll give you that, English isn't my native language so nuances escape me from time to time.
But how do you get a "full year of revenue" if you product isn't available since the beginning of the year?

Yeah, I think we are talking about two different things - one is when we start shipping for revenue, and the other is the actual launch announcement. Typically we start shipping parts before the actual launch so that they can be built into systems/boards/whatever by the time the launch happens.

Case in point: Intel has already started shipping final Kaby Lake processors now, but Kaby Lake launch isn't for months yet. If you want to consider a real arrival in the October timeframe, Kaby Lake is your chip.

The corrected bitsandchips article says "We have to update this article, because we have misread John Bridgman's words. He didn't say that Zen will be available in 4Q16 for sure or something like this, as we reported. He just pointed out that the online magazines and IT journalists often misunderstand the terms 'shipping', 'scheduling' and 'sampling'".
http://www.bitsandchips.it/52-engli...zen-is-on-track-and-will-be-available-in-4q16

As such, I believe Gian Maria Forni's "inside sources" are just John Bridgman, and Gian is still making a leap by saying Zen will "hit the market in November/December 2016", when "towards the end of Q4" as said by Lisa Su really means December, and that is for Zen shipping and not Zen launch. Really, the primary source you should be listening to is John Bridgman, who as a highly-placed engineer in AMD counts as a primary source in his own right.

Speaking of which, here's John Bridgman talking about SMT in the Pentium 4.

AFAICS [As far as I can see] most of the suckage with HT [hyperthreading] came from combining SMT with a narrow-issue core, where there weren't enough integer resources to go around.

 

[Pieter3dnow]

That is a distinct possibility, yes. Past behavior is not a perfect indicator of future behavior, but it is the best indicator that exists. You know that the typical time between a new microarchitecture sampling and launch is one year, right? Sandy Bridge samples shipped in Q1 2010 while Sandy Bridge launch was Jan 5, 2011. Are you going to call Sandy Bridge a horribly-delayed chip too?


No. Think about it. If his inside sources had said November, why would he have been saying October until John Bridgman had to correct him? The disappointing truth is he misinterpreted what, if anything, he heard. AMD never said Zen would be launched in Q4 2016 for sure, it just said it would be "available", by which it means that it would be selling Zen to OEMs and distributors and have Zen revenue show up on AMD's balance sheets, not that Zen would be available to you or me.

AMD Zen Reportedly Delayed Until Early Next Year - Phoronix Forums


Case in point: Intel has already started shipping final Kaby Lake processors now, but Kaby Lake launch isn't for months yet. If you want to consider a real arrival in the October timeframe, Kaby Lake is your chip.

The corrected bitsandchips article says "We have to update this article, because we have misread John Bridgman's words. He didn't say that Zen will be available in 4Q16 for sure or something like this, as we reported. He just pointed out that the online magazines and IT journalists often misunderstand the terms 'shipping', 'scheduling' and 'sampling'".
http://www.bitsandchips.it/52-engli...zen-is-on-track-and-will-be-available-in-4q16

As such, I believe Gian Maria Forni's "inside sources" are just John Bridgman, and Gian is still making a leap by saying Zen will "hit the market in November/December 2016", when "towards the end of Q4" as said by Lisa Su really means December, and that is for Zen shipping and not Zen launch. Really, the primary source you should be listening to is John Bridgman, who as a highly-placed engineer in AMD counts as a primary source in his own right.

Speaking of which, here's John Bridgman talking about SMT in the Pentium 4.

I have my reservations on how and why you can defer from past experience that AMD will do the same thing on a whim.
Inside sources and who they are don't bother me that much. On the point where Gian knows what he publishes would not backfire on him as the stuff he posts being false.

 

[Tonkatsu]

Yeah, you are right. The official word now is very limited release late in Q4 2016, followed by volume early in Q1 2017.

Yup, that is an accurate understanding of Lisa Su's statements from the latest Q2 earnings call. And if we're talking about the 32-core versions (the topic of this thread, ha), those are almost certainly server-only and won't be arriving until later in Q2 2017.

AMD Zen May Arrive In 2016, But Mass Availability Slated For 2017

“We have been very focused on the server launch for first half of 2017. Desktop should launch before that. In terms of true volume ability, I believe it will be in the first quarter of 2017. We may ship some limited volume towards the end of the fourth quarter, based on how bring up goes and the customer readiness,” said Lisa Su, AMD’s CEO.

Hopefully, good things come to those who wait!

 

[Tonkatsu]

I have my reservations on how and why you can defer from past experience that AMD will do the same thing on a whim.
Inside sources and who they are don't bother me that much. On the point where Gian knows what he publishes would not backfire on him as the stuff he posts being false.

My point is, it is not a whim, it is standard operating procedure for AMD. And don't get me started on what "tech journalists" will or won't do.

 

[Pieter3dnow]

My point is, it is not a whim, it is standard operating procedure for AMD.

So make a detailed post on the launch of each and every cpu AMD made since Bulldozer and prove your point ...

 

[Zarathustra[H]]

I'm looking forward to Zen in October. Can't come soon enough.

I'm starting to suspect my 5 years of overvolting my i7-3930k to unsafe levels in order to hit 4.8-5.0Ghz (granted, always with adequate cooling) is finally starting to degrade the chip. It seems to be getting hotter and spinning up the fans more frequently than it used to.

(Either that, or the cooler has come loose, I should probably check that)

I just hope it lasts me until October!

Looks like I was calling the death of my i7-3930k prematurely.

Looks like the combination of accidentally setting the wrong voltage offset in bios and having a failing fan splitter resulted in higher temps than I was used to, and fans spinning out of control giving me the impression it was on the way out.

Fixed that up, and it appears to be running as well as it ever has now.

I'm still going to be looking at Zen when it launches, but it's no longer a immediate need for me.

 

[Tonkatsu]

So make a detailed post on the launch of each and every cpu AMD made since Bulldozer and prove your point ...

Well, ok. AMD hasn't really announced sampling dates for their chips since the Bulldozer fiasco, but here's the data I could come up with that isn't rumor-based.

Trinity (first Piledriver chip) Sample Oct 2011, Launch Oct 2, 2012, 12 months

Seattle (ARMv8) Sample March 2014, Launch Jan 14, 2016: 22 months

Like I said, typical time between AMD sampling chips and AMD launch dates is at least a year for new architectures.

 

[Tonkatsu]

I'm still going to be looking at Zen when it launches, but it's no longer a immediate need for me.

Cool beans. Glad your system is still healthy. It is good not to be under the gun for hardware replacement!

 

[Zarathustra[H]]

Well, ok. AMD hasn't really announced sampling dates for their chips since the Bulldozer fiasco, but here's the data I could come up with that isn't rumor-based.

Trinity (first Piledriver chip) Sample Oct 2011, Launch Oct 2, 2012, 12 months

Seattle (ARMv8) Sample March 2014, Launch Jan 14, 2016: 22 months

Like I said, typical time between AMD sampling chips and AMD launch dates is at least a year for new architectures.

I'd imagine that AMD might try to accelerate this one given where they are Fina coat with their burn rate and upcoming bond maturations. I have never worked with a chip company though, so I don't know how much something like this CAN be accelerated,a and how much of it is a kind of "mythical man month" type situation.

I'm inclined to believe the statements made at their shareholder meeting - however - with limited shipments near the end of Q4 2016, and full shipments early Q1 2017, merely because you don't lie to your shareholders. People go to jail for that shit.

Now, we don't know what kind of plan lies behind those faulted though. Are they trying to accelerate launch and hoping to make limited release in late Q4 2016, but not sure if they are going to make it? Or do they have a built in pad with internal deadlines and a gap until the external ones hit so they don't miss them, and they might actually be done early?

Who knows.

I'm just hoping g for an as early as possible release, without any significant corner cutting.