AMD Confirms Zen Launch

depends on what they are doing, lower clocked isn't bad if they can do more per clock. Lower clocks would be beneficial if you can lower voltage and thus lower power.

Clocks are a direct influence of the architecture, so its all depends.
 
Man, I hope this does not turn out like the Polaris 11 and GTX 960 comparison B.S. shown last year. I say we will have to really wait to see the real chip in a real motherboard tested before we can make any semi sure decisions on buying Zen. I also hope their power targets are actual too - a good mITX motherboard with great performing Zen (no gpu!) could give AMD a leg up on that platform - packing more cores being taking up by the internal gpu that gets unused since a superior graphics card is used can give AMD a good advantage for small more powerful builds. That is if Zen is actually powerful.
 
noko said:
Man, I hope this does not turn out like the Polaris 11 and GTX 960 comparison B.S. shown last year. I say we will have to really wait to see the real chip in a real motherboard tested before we can make any semi sure decisions on buying Zen. I also hope their power targets are actual too - a good mITX motherboard with great performing Zen (no gpu!) could give AMD a leg up on that platform - packing more cores being taking up by the internal gpu that gets unused since a superior graphics card is used can give AMD a good advantage for small more powerful builds. That is if Zen is actually powerful.
Click to expand...

Internal GPU when disabled is dark silicon. Very nice for cooling. You can get HEDT MiniITX boards. So Zen offers nothing new there. Assuming it can even compete with LGA1151 to begin with.
 
Shintai said:
Internal GPU when disabled is dark silicon. Very nice for cooling. You can get HEDT MiniITX boards. So Zen offers nothing new there. Assuming it can even compete with LGA1151 to begin with.
Click to expand...
Well I just built a mITX system using Fractal Node case (10.2L), I5 6500 and have coming a GTX EGVA 1060 SC coming. The GPU in the cpu is pointless other then a potential backup if the discrete GPU dies or has issues. I rather that part of the CPU just have more cores and no GPU. So Zen offers something unique on the SFF platform if it performs well and is not too power hungry. My Nano SFF system uses a I7 6700K another wasted space gpu in the cpu. So basically saying AMD could have an area in SFF which will be more useful then Intel. Still depends upon performance and price. Intel does have the E processors and there are available mITX that support them but some significant $ goes with them as well.
 
noko said:
Well I just built a mITX system using Fractal Node case (10.2L), I5 6500 and have coming a GTX EGVA 1060 SC coming. The GPU in the cpu is pointless other then a potential backup if the discrete GPU dies or has issues. I rather that part of the CPU just have more cores and no GPU. So Zen offers something unique on the SFF platform if it performs well and is not too power hungry. My Nano SFF system uses a I7 6700K another wasted space gpu in the cpu. So basically saying AMD could have an area in SFF which will be more useful then Intel. Still depends upon performance and price. Intel does have the E processors and there are available mITX that support them but some significant $ goes with them as well.
Click to expand...

Quicksync for example?

But I dont see your issue with the IGP die area, seems you are fixated on it and somehow trying to demonize it. Zen isn't unique, Zen just doesn't have an IGP to begin with so dont try and glorify it. Instead you can focus on the "moar cores" that may or may not be useful. One thing is certain tho, the 1-4 core performance will be lower with Zen, a lot lower.
 
I think it will be a huge upgrade compared to current AMD chips.....against intel not so much. I personally dont expect to be tempted to go full AMD setup again...but who knows? Well see.....Id rather be surprised than disappointed
 
40% increase in single threaded applications over the latest bulldozer offering.
People will buy it cause AMD fanboys are like that I think alot of them switched to Intel since 2011.
 
crosspost from Overclock.net: [Anand] New Zen engineering sample numbers - 8 core tops out at 3.6GHz
A poster on anandtech who leaked previous stuff has more:
New Zen microarchitecture details

Hi all! I went into the forest recently and birds have tweeted something again. They spoke about 2 new SKUs that have just hit the post-boxes of the mainboard manufacturers. They are both A0 revisions and Engineering Samples, so there is no improvement since my last post.

The first one is an 8-core design with AMD's HT implementation and it's got a 3150 MHz base clock, it's all-core turbo is 3300 MHz and the max turbo for 1 core is 3600 Mhz. Yes, here are some improvements regarding the previous 8-core SKU under the same TDP envelope.

The second SKU is a 4-core one with AMD's HT. It's got a 65W TDP and the base clock is still 2900 MHz. All-core turbo is 3100 MHz, max turbo is 3400 MHz. I don't know if it's only an SKU for testing mainboards or something is not okay with the clock-wattage correlation. I mean on higher clocks the 4-core SKU steps into the 95W TDP envelope, AMD can't keep the wattage low. Maybe GloFo's 14nm process needs some maturing... Frankly I don't have a clue what's in the background.

Retail AM4 mainboards are under production. The whole platform will be ready to have a paper-launch at the end of the year with a real availability in February of 2017. Performance wise the Zen uarch will be around Haswell and Broadwell (except for FMA), it seems it won't catch Skylake clock for clock. It's not a big deal, but if the clocks can't go higher until the start it won't fulfill the expectations. And we all know that expectations in this case are very high...
Click to expand...

rumor....bag of salt...remember.
 
Q1 '17.... AKA Paper launch 11:59PM April 30th with retail availability "2H17"
 
primetime said:
yea but tops out at 3.6 seems more than believable on top of what we have already heard anyway (not to mention their latest gpus dont give much reason to think otherwise) well see but i wouldnt be shocked
Click to expand...

For a lot of [H]ardOCP people I don't think stock clocks really matter that much anyway. I mean the highest clocked Sandy Bridge at launch (2600K) was 3.4-3.8ghz, but most of them could hit upper 4ghz range and some even hit 5ghz. There's no way to know for sure until it's actually available of course but "low" stock clocks aren't necessarily a disaster for enthusiasts if they have lots of headroom. Lower clock speeds on a smaller process should help AMD in the power/heat department as well, which is what has really been killing them trying to get design wins in laptops and pre-builts.
 
travbrad said:
For a lot of [H]ardOCP people I don't think stock clocks really matter that much anyway. I mean the highest clocked Sandy Bridge at launch (2600K) was 3.4-3.8ghz, but most of them could hit upper 4ghz range and some even hit 5ghz. There's no way to know for sure until it's actually available of course but "low" stock clocks aren't necessarily a disaster for enthusiasts if they have lots of headroom. Lower clock speeds on a smaller process should help AMD in the power/heat department as well, which is what has really been killing them trying to get design wins in laptops and pre-builts.
Click to expand...

True, but the low base clock is another indication that Zen is potentially very power hungry north of 3 GHz, like some of us have expected for some time. This explains why a higher turbo is allowed for a single core, where the max turbo for all cores is "only" 3.3 GHz.

This farther implies Zen won't OC well. This is pretty much matching my expectations so far.
 
gamerk2 said:
True, but the low base clock is another indication that Zen is potentially very power hungry north of 3 GHz, like some of us have expected for some time. This explains why a higher turbo is allowed for a single core, where the max turbo for all cores is "only" 3.3 GHz.

This farther implies Zen won't OC well. This is pretty much matching my expectations so far.
Click to expand...

About half wrong or right however you want to put it ;) .

The problem is that even if it would be power hungry (north of 3 ghz) would it scale enough. Lets say something silly: Zen would clock to 5 ghz and use 300Watt would that be acceptable to "many" overclockers , my estimated guess would be yes , as long as the performance scales the extra power is "trivial" . That was the problem with Bulldozer even if you got it to 5 GHZ in older DX11 games it would barely scale.

For the people who can't or won't overclock what would they care as long as stock performance is good for them at the rated price performance level.
 
Shintai said:
Quicksync for example?

But I dont see your issue with the IGP die area, seems you are fixated on it and somehow trying to demonize it. Zen isn't unique, Zen just doesn't have an IGP to begin with so dont try and glorify it. Instead you can focus on the "moar cores" that may or may not be useful. One thing is certain tho, the 1-4 core performance will be lower with Zen, a lot lower.
Click to expand...
Demons you say, hmmm I better watch that. Really I rather have more cores in that space instead of an inactive CPU gpu. In other words a 95w 8core/16 thread processor vice anemic 4core/8threaded one. Rather simple concept :rolleyes: and no demons.
 
The link indicates A0 samples and still engineering. So an A1 version could be the real McCoy and reason Zen did not hit late this year. So clock speeds are probably not set in stone yet but board manufacturers do need Zen cpu's for testing. Base clock 3.4ghz turbo of 3.8ghz for single core, 3.6ghz all. Max OC 4ghz+. I probably could live with that if the price is right for an 8core/16thread design. Over time that will most likely go up like all of AMD processors.
 
noko said:
Demons you say, hmmm I better watch that. Really I rather have more cores in that space instead of an inactive CPU gpu. In other words a 95w 8core/16 thread processor vice anemic 4core/8threaded one. Rather simple concept :rolleyes: and no demons.
Click to expand...

But it doesn't work that way. More interconnect and uncore needed and cores use more power than IGP.
 
psy81 said:
Past practice (Phenom II 955, 965, etc.. and FX-8150, 8320, 8350, etc...) You really think they're gonna come out with one CPU and that's it???
Click to expand...

:D How about the past history of just about every processor ever made. Of course they're going to come out with more. Totally with you on that.

Unless they fold as a company I guess... Hopefully that won't happen though. Love my Intel chips, but I would LOVE to see AMD put out a competitive CPU again.
 
Have we heard anything regarding AM3+ cooler compatability? I think I remember seeing the mounting position would be identical, but not the depth?
 
I hope the engineers do some hand tuning for Zen and not just rely on the standard optimizing tools that net them a small percentage most times. They really need this to be a winner.
 
Pieter3dnow said:
About half wrong or right however you want to put it ;) .

The problem is that even if it would be power hungry (north of 3 ghz) would it scale enough. Lets say something silly: Zen would clock to 5 ghz and use 300Watt would that be acceptable to "many" overclockers , my estimated guess would be yes , as long as the performance scales the extra power is "trivial" . That was the problem with Bulldozer even if you got it to 5 GHZ in older DX11 games it would barely scale.

For the people who can't or won't overclock what would they care as long as stock performance is good for them at the rated price performance level.
Click to expand...


The reason why Bulldozer didn't scale with DX11 is because its core utilization bottleneck wouldn't allow it to. If you over clock Bulldozer on a DX11 game what you expect of all 8 cores overclocking you only get 1/8 or 1/4 of that! Doesn't really matter the CPU either, even Intel's will behave the same way if the CPU was causing a bottleneck to begin with and the DX11 core utilization being subpar on most apps. But Because of Intel's increased IPC per core it is harder to notice the differential at times.

Bowman15 said:
I hope the engineers do some hand tuning for Zen and not just rely on the standard optimizing tools that net them a small percentage most times. They really need this to be a winner.
Click to expand...


There is no such thing as hand tuning or standard optimizing tools, if you are talking about hand laid out transistors, that isn't hand tuning, that is actually analyzing transistor layouts during design of chip prior to functionality of the architecture. Its not tuning that they are doing, its well before tuning. "Standard" optimizing tools, there is is no such thing either. If you are talking about automated layouts that GPU's use, that is not the same thing as optimization tools. There really are no tools that optimize for you.
 
Last edited:
britjh22 said:
Have we heard anything regarding AM3+ cooler compatibility? I think I remember seeing the mounting position would be identical, but not the depth?
Click to expand...


Now, this should be the latest info on compatibility, but what I'm taking from the information I've reviewed, it seems to be still in flux how AM3+ is gonna work with AM4, if its the same, different, will there be upgrade kits provided by the makers, etc, etc.
 
Comixbooks said:
40% increase in single threaded applications over the latest bulldozer offering.
People will buy it cause AMD fanboys are like that I think alot of them switched to Intel since 2011.
Click to expand...

I was loyal to AMD (3 AMD builds in a row) and was one of the ones that made that switch. I mean 5 years is a long time to wait...
 
I couldnt wait, grabbed a 6800k this weekend. I hope AMD does well. This will mean my next upgrade will be a ZEN+ or 2011-v3 8 core since intel will bring the price down due to age/competition.
 
Shadohh said:
I couldnt wait, grabbed a 6800k this weekend. I hope AMD does well. This will mean my next upgrade will be a ZEN+ or 2011-v3 8 core since intel will bring the price down due to age/competition.
Click to expand...
Same. Waiting on my 1070 and 1TB SSD currently. Have all my other parts (6700k, 16gb ram, gene mobo, fractal node matx case, h100i).
 
I think they'll be released at super high clock speeds based on the fact that I overclocked my butt and used it to make this post.
 
doz said:
Same. Waiting on my 1070 and 1TB SSD currently. Have all my other parts (6700k, 16gb ram, gene mobo, fractal node matx case, h100i).
Click to expand...
we will have very similar systems. I got the 6800k at 4ghz, evga ftw d 1070, 32gb 2666mhz, gigabyte phoenix mobo, noctua cpu cooler. bf1/dx11 butter smooth 3440x1440 @ 60fps.

I hope AMD can deliver the same.
 
The first one is an 8-core design with AMD's HT implementation and it's got a 3150 MHz base clock, it's all-core turbo is 3300 MHz and the max turbo for 1 core is 3600 Mhz. Yes, here are some improvements regarding the previous 8-core SKU under the same TDP envelope.
Click to expand...

Base clock about what I expected (3.2 GHz), turbo is a bit on the low side (only 150 Mhz for all cores). The fact a single core can ramp significantly higher (3x the single core boost) indicates thermals might be a limiting factor. This is farther evidence Zen may not OC well.

The second SKU is a 4-core one with AMD's HT. It's got a 65W TDP and the base clock is still 2900 MHz. All-core turbo is 3100 MHz, max turbo is 3400 MHz. I don't know if it's only an SKU for testing mainboards or something is not okay with the clock-wattage correlation. I mean on higher clocks the 4-core SKU steps into the 95W TDP envelope, AMD can't keep the wattage low. Maybe GloFo's 14nm process needs some maturing... Frankly I don't have a clue what's in the background.
Click to expand...

Bolded for emphasis. This is even more evidence toward what I've suspected for some time: Above 3 GHz, power draw becomes a significant issue. This would explain why the 65W four-core part is actually clocked below the 8-core part (which I assume is 95W).

I can now safely assume that clocks for Zen tap out at about 3.2 non-turbo, and that GloFo's 14nm process results in significant power draw above 3 GHz as some of us have suspected. Taken together, I can also conclude Zen will not be a good OCing design.

So far, everything I've seen leaked indicates my suspicions about Zen have been spot on.
 
guru3d reports that Zen has improved their clock speeds.Anyone else heard of this?
 
GF 14lpp had some metal layer defects that hurt performance/power until just recently, very possible we already have a1 and a2 revision Zen chips, their is a new revision of polaris 10 that is 50% better on power consumption, very likely same situation for zen.
 
luminousone said:
GF 14lpp had some metal layer defects that hurt performance/power until just recently, very possible we already have a1 and a2 revision Zen chips, their is a new revision of polaris 10 that is 50% better on power consumption, very likely same situation for zen.
Click to expand...

I'd actually love to put together a Zen+Polaris system just to mix things up again. I don't see it supplanting my Intel+Pascal system or whatever I have at the time of Zen release, but there's no reason not to have an extra system in the house. This is typically how I test the waters, and this is the first time I've felt like it would be fun since maybe the first gen of Phenom. (which was in my first proper GHTPC)
 
Last edited by a moderator:
luminousone said:
GF 14lpp had some metal layer defects that hurt performance/power until just recently, very possible we already have a1 and a2 revision Zen chips, their is a new revision of polaris 10 that is 50% better on power consumption, very likely same situation for zen.
Click to expand...


Metal layer defects won't cause that much of a change in power consumption. The only time this would happen if the design of the metal layer is screwed up and leakage is happening all over the place and tell you what if that was the case, it would happen even at very low clocks and that isn't happening here, because that is easily noticeable your general power curves, and thermals, will not look normal they will be all over the place. And the chip will not even be put into mass production, because its impossible to validate such a chip if there were such errors in the node and this is why when there have been sever problems at this level (node), no chip manufacturer has never went into mass production with that node because its really is a wasted investment, even more than not producing the chip to begin with.

Also each metal respin will eat up 1 q of time so if they go to A1 then Q1 2017 is kinda in jeopardy (still have validation to do), and if they go to A2, yeah they are not going to have ready till 2nd half of next year.
 
Last edited:
Honestly I don't know enough to say much about the fab process, completely possible issue is else where and GF would rather not talk straight about it.!, or could be parts held over time, that binned better.
 
Actually look up antenna effect pretty much is what you were talking about.
 
It better be good. My FX 6300 needs an update, asap. I actually switched to Sandy Bridge when it came out but it fried in 2014, so I built something fast and cheap with the already out of date 6300. Currently waiting to see KBL vs ZEN and see who gives me more bang for buck. Then in February at the latest, it's new mobo/cpu/ram for me, and right after that, new GPU and monitor too. It's always exciting to do a full upgrade!
 
You must log in or register to reply here.
Back
Top