AMD Zen Rumours Point to Earlier Than Expected Release

I bought my Bulldozer motherboard in July 2011 and the processors didn't release until October 2011. Back then we could plug our old AMD processors into the mobo without issue though.

Same here.


I got the AM3+ MB and put a 955 in there to tide me over until bulldozer hit.... It was so bad I skipped the 8150. I did eventually put an 8350 in there and that is what I still have.



Going forward, before I make any MB purchases, I am waiting for actual zen results first. It does not even have to beat intels latest and greatest, it just has to be in the general ballpark with a slight price reduction. If by some miracle it does just as well if not better, I'll pay the same intel prices. If it's garbage, I'll reluctantly jump to intel. On the gpu side though, amd all the way.
 
Same here.


I got the AM3+ MB and put a 955 in there to tide me over until bulldozer hit.... It was so bad I skipped the 8150. I did eventually put an 8350 in there and that is what I still have.



Going forward, before I make any MB purchases, I am waiting for actual zen results first. It does not even have to beat intels latest and greatest, it just has to be in the general ballpark with a slight price reduction. If by some miracle it does just as well if not better, I'll pay the same intel prices. If it's garbage, I'll reluctantly jump to intel. On the gpu side though, amd all the way.

I would agree with the AMD GPU side except that I would have to have a new case to upgrade to them. (Fury air cooled or Fury X.) Therefore, I went ahead and got a EVGA 980Ti instead since it fits my Define r3 without issue. Otherwise, I would have purchased the R9 Fury XFX one. (In fact, I did and then cancelled the order when I found it would not fit.)
 
Same here.

I got the AM3+ MB and put a 955 in there to tide me over until bulldozer hit.... It was so bad I skipped the 8150. I did eventually put an 8350 in there and that is what I still have.

Going forward, before I make any MB purchases, I am waiting for actual zen results first. It does not even have to beat intels latest and greatest, it just has to be in the general ballpark with a slight price reduction. If by some miracle it does just as well if not better, I'll pay the same intel prices. If it's garbage, I'll reluctantly jump to intel. On the gpu side though, amd all the way.

I always skip things until reviews are out. Actually, more than likely I tend to buy things once they're about a year old and the price has plummeted. On my upgrade cycle it works really well.
 
I REFUSE to get my hopes up.
I CANNOT afford another bulldozer-load of emotional damage.

There some factors that are totally different now, the process is a lot lower then what it used to be the difference this time is smaller.
The Bulldozer was a way different beast al together, if you would benchmark old bulldozer vs old Intel on modern API then you could see that it would have made a dent for gaming. Hopelessly screwed by design instructions per cycle could not catch up in a way it would have made a difference in DX9 DX11 or OpenGL.

I'm not going to tell you to run to the edge of a cliff and take a jump but check out what Jim Keller has been doing before he went back to AMD, the design of the cpu he did.

And no one is putting a gun to your head and told you to buy anything for that matter just be a little bit more patient, if you don't like it don't buy it.....

http://dresdenboy.blogspot.de/2015/11/amd-k12-looks-to-be-at-least-4-wide.html
 
Last edited:
Does CPU matter for gaming that much anymore? I would figure a lower cost amd product would do about as well as an Intel of 50% higher price... Only until you get into server caliber stuff that you start to see a difference.

I use AMD almost exclusively for my server stuff at home... Hard to balk at 12 core opteron cpus for $8 each
 
http://www.planet3dnow.de/cms/21487-geruecht-launcht-amd-zen-deutlich-eher-als-bisher-angenommen/

^ The original source of the articles linked doesn't seem too strong. One motherboard manufacturer contact said March 2016. If it were really launching in less than 4 months, there would be much more confirmation. It's probably a mistake or misunderstanding between release and of a sample schedule.

AMD typically takes about 1.5 years between tape out and release, if there are no significant problems, and Zen supposedly taped out last quarter. As the link I posted above suggests, late 2016 to early 2017 is a much more realistic time frame.
 
It seems like people are confusing the AM4 platform coming out as meaning Zen will launch at the same time.
 
Does CPU matter for gaming that much anymore?
Right now not too much. Major difference today that a higher performance CPU will give you in game is less dips in frame rate + a smoother and more consistent frame rate.

In the future DX12 will actually allow game developers to start making use of 6-8 cores without too much trouble. You probably won't actually --NEED-- more than 4 cores for a long time mind you. If you want top fps with everything maxed, like lots do, then 6-8 cores with a highish clock speed will probably become necessary over the next 2-4 years.
 
My testing of Ashes of the Singularity (current 0.64 build) shows no significant bottlenecking with a FX-9590 at 4.8 GHz.

High preset:
T86MM4b.jpg


Extreme preset:
zjoumQ2.jpg


Crazy preset:
geZmPdd.jpg


Only the least intensive situations (High + Normal batches) has the GPU bound bar starting to drop from 97%+. This is at 1200p so higher resolutions should be GPU bound earlier on. The CPU framerate only drops from 76.1 FPS (High) to 70.1 FPS (Crazy) while the actual framerate drops 46% from the same changes. Current Intel offerings with 4c/8thr should have a few extra frames here and there but I doubt many people could tell the difference.

I am assuming that because of the heavy use of multi-threaded AI this benchmark will represent a higher CPU / GPU load ratio than your typical Action / Shooter game would in DX12.
 
Right now not too much. Major difference today that a higher performance CPU will give you in game is less dips in frame rate + a smoother and more consistent frame rate.

In the future DX12 will actually allow game developers to start making use of 6-8 cores without too much trouble. You probably won't actually --NEED-- more than 4 cores for a long time mind you. If you want top fps with everything maxed, like lots do, then 6-8 cores with a highish clock speed will probably become necessary over the next 2-4 years.

It's actually the opposite, CPU performance will matter even less because the GPU batches will be spread out more between the chips.
 
Utilizing more than 4 cores doesn't mean that it won't be able to make use of faster CPU's. There are lots of things where only a fast CPU will do because some things in the game engine just don't scale well past single thread.
 
Utilizing more than 4 cores doesn't mean that it won't be able to make use of faster CPU's. There are lots of things where only a fast CPU will do because some things in the game engine just don't scale well past single thread.

I think the correlation is/has gotten muffled and confused. In DX11 it always was one core to GPU could communicate at a time. What confuses is likely that while the main thread, the one doing near 100%, was talking the other 2-3 threads, even 8 in some games, had to wait till it was done to speak to the GPU. This way is still a WAIT IN LINE approach but still better than all on one thread. The DX12 approach allows each of those threads to speak to the GPU, pending available lanes, at one time not needing to wait for the other.

The easiest way to look at this is DX11 is a one lane road, DX12 is a 8 lane (in one direction) highway. In DX11 the time to frame was thread one [total time to render and send] plus time to send for each additional thread (the time to render is likely shorter than the main thread so they get done in the time it takes the main thread to render therefore they don't impact the time to frame). In DX12 the mas time to frame is just the slowest thread and in this case it is the time to render and send for the main thread.

At least that is the easiest way to put it.

http://www.amd.com/en-us/innovations/software-technologies/directx12

That page has most of the descriptions. Watch that video it is real helpful and easy to understand.
 
It seems like people are confusing the AM4 platform coming out as meaning Zen will launch at the same time.

agreed. bristol ridge is our launch cpu for AM4.
it exists already, it's called carizzo with its DDR4 controller enabled and a more generous TDP.
 
So, AMD just confirmed Zen hasn't even taped out yet:

http://seekingalpha.com/article/374...y-investors-conference-transcript?part=single

Zen was a clean sheet design that started few years ago. We are in the final figure of executing and the milestone that you want hear us talk about is Zen tapping out, which should be over the next several months. And then putting samples in the hands of our customer and then starting portfolio of revenue in 2017.

Also confirmed a 2017 release.
 
What he said does not confirm a 2017 release. I seem to recall them saying that 2017 will be the first full year of revenue for Zen, not that 2017 will be the release year of Zen. I am sick of guessing and am hoping for some real, concrete information, not generalizations.

But consider the time necessary from tape out to actual release is about 12 months, assuming no problems. Add three additional months for every necessary re-spin. Given Zen hasn't taped yet, your looking, best case, at a December 2016 release, if Zen were to tape sometime this month, and has no problems whatsoever. Zen is slipping, this pretty much confirms as much.
 
But consider the time necessary from tape out to actual release is about 12 months, assuming no problems. Add three additional months for every necessary re-spin. Given Zen hasn't taped yet, your looking, best case, at a December 2016 release, if Zen were to tape sometime this month, and has no problems whatsoever. Zen is slipping, this pretty much confirms as much.

AMD released Bulldozer in October, which would be perfect timing for the first run of chips.
 
If Zen was slipping why would they already patch stuff for compilers :) . Still it will take a long time :)
 
But consider the time necessary from tape out to actual release is about 12 months, assuming no problems. Add three additional months for every necessary re-spin. Given Zen hasn't taped yet, your looking, best case, at a December 2016 release, if Zen were to tape sometime this month, and has no problems whatsoever. Zen is slipping, this pretty much confirms as much.

What are you talking about? Zen has been confirmed to have tapped out already, along with AMD's new GPU family...GloFlo announced that they entered validation testing for AMD at the beginning of November, so I think we will see Zen launch right before or right after back to school season...

AS for the Spring release, I think it's pretty clear that it's pointing to the AM4 platform launch, with Carrizo based APU's alongside as launch partners, with Zen CPUs being able to be dropped in later in the year.
 
But consider the time necessary from tape out to actual release is about 12 months, assuming no problems. Add three additional months for every necessary re-spin. Given Zen hasn't taped yet, your looking, best case, at a December 2016 release, if Zen were to tape sometime this month, and has no problems whatsoever. Zen is slipping, this pretty much confirms as much.

Zen initial tapeout - engineering samples and test runs. The final production tapouts to fix anything found is going to occur in the next several months. You can have more then one tape out for a processor. He also talks about revenue from Zen in 2016. Kinda hard to make money from something not available. Still anything can happen up to launch. AMD is moving in the right direction and are cornering some markets, growing others, recovering as well lost markets.
 
I don't know anyone that wishes AMD ill will. Everyone in this industry wants competition. But you can only complete for so long on an aging infrastructure with price only. Sooner or later they need something that is at least near Intel in IPC.

x86 theory isn't something new. I realize Intel's R&D and manufacturing process is much ahead of Global Foundries but that doesn't account for the severe lack of IPC. That's all AMD's fault. That is something AMD should be able to address even with inferior silicon. Maybe not catch up or exceed but they should be better than where they are.
 
Still braced for disappointment.

Exactly, same here. This especially since Intel is coming out with 4 Broadwell - E processors, 2 6 core ones, one eight core and a 10 core processor. I am an AMD fan but, I will not go with a processor that is far behind what is already out again.
 
Zen initial tapeout - engineering samples and test runs. The final production tapouts to fix anything found is going to occur in the next several months. You can have more then one tape out for a processor. He also talks about revenue from Zen in 2016. Kinda hard to make money from something not available. Still anything can happen up to launch. AMD is moving in the right direction and are cornering some markets, growing others, recovering as well lost markets.

I am curious to know if AMD is talking about the fiscal or regular 2016 year. If it's fiscal, that means it would be out before October.
 
Exactly, same here. This especially since Intel is coming out with 4 Broadwell - E processors, 2 6 core ones, one eight core and a 10 core processor. I am an AMD fan but, I will not go with a processor that is far behind what is already out again.

i don't think there is any indication that amd is intending zen to compete with broadwell-e/skywell-e.

Socket AM4 is an analogue of s1151, i.e. a single platform that covers the £45-£270 price range.
 
Last edited:
i don't think there is any indication that amd is intending zen to compete with broadwell-e/skywell-e.

Socket AM4 is an analogue of s1151, i.e. a single platform that covers the £45-£270 price range.

If they can make something as fast as a 2600k/3770k/2500k/3570k and sell it for a much better price than skylake or whatever gen intel will have by then, they will be the winners in my book.
 
I expect Zen in 2016.
It hasn't even taped out yet and it isn't expected to for several more months. AMD has yet to release a chip earlier than 18 months between it taping out and release. AMD's 12-14 months schedule for a new clean design is ridiculously optimistic, and something the company has never accomplished. Your expectations are unrealistic.

Zen is a mid*- to late 2017 chip at this point, with a good chance of never if AMD splits itself into profitable and unprofitable segments, as it's principle investors have apparently been doing for months.

* lol, no
 
If they can make something as fast as a 2600k/3770k/2500k/3570k and sell it for a much better price than skylake or whatever gen intel will have by then, they will be the winners in my book.

Intel prices have crept up lately ($149 i5 750, $179/199 i5 4670K or 4690K, $229 6600K).
So yes - the price opening is there to hit at $199 or less.

Ideally, OC vs OC, Zen is at least 90-95% of an overclocked IvyBridge - even then I may be too generous. Additionaly, AMD has normally needed an iteration or two just to get clocks up.

See TBred A vs TBred B. Clawhammer vs Venus. Phenom I vs Phenom II.
 
4 u fan boys Intel.isnt doing anything new except renaming shit. We have reached the peak of Moores law. Now we need to move from silicon. The industry will not admit it due to money.
 
4 u fan boys Intel.isnt doing anything new except renaming shit. We have reached the peak of Moores law. Now we need to move from silicon. The industry will not admit it due to money.

Not so sure about that since we are not privy to their cpu details they are developing but have not released. Unless Zen hard core info is released in a timely fashion or I cannot afford to upgrade my work computer, I will be probably switching to Broadwell E from my FX 8350 with 32GB of ram. We will see though since I am in no hurry with this one.
 
4 u fan boys Intel.isnt doing anything new except renaming shit. We have reached the peak of Moores law. Now we need to move from silicon. The industry will not admit it due to money.

I don't agree with that, silicon is far from done, and will be around for a long time to come. Could optimization from another manufacturing process occur??? Absolutely, but it isn't going to happen overnight and is going to take time.
 
It hasn't even taped out yet and it isn't expected to for several more months. AMD has yet to release a chip earlier than 18 months between it taping out and release. AMD's 12-14 months schedule for a new clean design is ridiculously optimistic, and something the company has never accomplished. Your expectations are unrealistic.

Zen is a mid*- to late 2017 chip at this point, with a good chance of never if AMD splits itself into profitable and unprofitable segments, as it's principle investors have apparently been doing for months.

* lol, no

From a monetary perspective, AMD has enough cash to make it through 2018 even if they continue to lose north of $100 Million per quarter. 2019-2020 is the period where their debt turns over though, and where AMD simply doesn't have the cash to pay. Zen is pretty much AMDs last shot to turn a profit.

Farther, given the lack of news, it's highly likely K12 has been delayed in order to focus on Zen. So there isn't any immediate followup is Zen doesn't perform.

As for release, I expect limited sampling very late Q4 2016, with consumer availability late Q1/early Q2 2017. Mid-tier chips to round out the lineup likely won't launch to Q3 2016 at the earliest. Note all this assumes no delays, which is probably optimistic.

From a design standpoint, Zen looks designed close to what Intel is currently doing, but I'm expecting IB/Haswell performance at best. So AMD better be careful how it prices it's chips, since no one will buy them if they insist selling them north of $500 because they are "new".
 
I don't agree with that, silicon is far from done, and will be around for a long time to come. Could optimization from another manufacturing process occur??? Absolutely, but it isn't going to happen overnight and is going to take time.

Wish it would go faster though--I'd love to have a diamond semiconductor CPU/motherboard, if only because it'd look badass (not that there aren't much better reasons to use it. ;).

Barring that, I'll try my darnedest to wait for Zen and see how that goes. If I get impatient I can just break out the i5 I have sitting in the closet in the meanwhile.
 
Back
Top