AMD says that their Ryzen CPU's will last 4 years.

[Trixanity]

If people would actually read the thread and read the source, they'd realize what this means and by extension how ambitious AMD is. They'll be shipping IPC improvements every year for the next four years. That's how AMD operates. Intel and Nvidia do something similar. You don't reinvent your architecture every year; you improve it with a fairly quick cadence while another team starts working on a brand new one which takes at least 3 years. Mark Papermaster even says Ryzen is on a "tock-tock-tock" schedule meaning architecture improvements till Zen is replaced.

 

That some of you even think AMD could release a product in this kind of market and then wait four years to release a new one is ridiculous. OEMs want product refreshes annually. AMD will provide that. Zen+ will be released in 2018 along side Navi. 

Ryzen would fall behind fairly quickly if it would stand still for four years in any case.

[othertomperson]

1 hour ago, Princess Cadence said:

I mean, Intel has a lot of CPU that are up-to-date and are 4 or more years old, and a CPU should be fully functional if well cared for 20 or more years, I honestly feel AMD is going a bit too far on their marketing trying to squeezing every little thing into their hype, the more they try to make this a huge deal while it isn't the more disappointment consumers are likely to have by the time this hits the shelves

It doesn't sound like they are saying the CPUs will still be decent in 4 years. It sounds like they're going to treat Ryzen like they did Bulldozer. They made IPC improvements on Bulldozer and called them Piledriver, Excavator... but they were largely the same architecture. Improvements from one to the next were modest.

 

Largely Intel and Nvidia have been following a two-generation cycle. Intel's architecture-die shrink cycle, and Nvidia's two generations of Fermi, two generations of Kepler, Maxwell followed by die-shrunk-Maxwell... AMD, whether it be GCN or Bulldozer, tend to run their architectures for years upon years will little real innovation.

[Princess Luna]

3 minutes ago, othertomperson said:

It doesn't sound like they are saying the CPUs will still be decent in 4 years. It sounds like they're going to treat Ryzen like they did Bulldozer. They made IPC improvements on Bulldozer and called them Piledriver, Excavator... but they were largely the same architecture. Improvements from one to the next were modest.

 

Largely Intel and Nvidia have been following a two-generation cycle. Intel's architecture-die shrink cycle, and Nvidia's two generations of Fermi, two generations of Kepler, Maxwell followed by die-shrunk-Maxwell... AMD, whether it be GCN or Bulldozer, tend to run their architectures for years upon years will little real innovation.

I understand now, though I dont see how is this supposed to be any good for the costumers, you will still have to buy a whole new CPU to have such polishing to the architecture while it will most likely not be a reasonable deal to who bought the first ryzen processors anyways, this really seems like going to be no different than what Intel done with skylake to kaby lake so this news are quite irrelevant in my point of view

[Mira Yurizaki]

People still use Sandy/Ivy Bridge and don't feel a need to upgrade, or are just starting to feel the need. And there's a noticeable performance gap between that and Skylake: http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/11

 

But if it's just Ryzen will be the foundation of improvements, then I don't really see that being an issue either. Intel's architectures haven't really changed much since Sandy Bridge either.

[Sauron]

This is nonsense and nobody should take this sort of statement seriously. They have no way to know what will happen in the next 4 years, what people will need and when they will need it. Buy what you need now.

[Guest Darth Revan]

3 hours ago, TechGod said:

This is such a stupid comment. You really think AMD hasn't thought about the future? You're delusional. 

No, you're delusional thinking that AMD thought about the future.

Ryzen is all they have. They went all in with this. Hoping that it will be a success and they can later on have the funds to make another architecture. And honestly I do want them to succeed, I really do. 

3 hours ago, Energycore said:

Here, you'll need this

*hands tinfoil hat*

Thank you very much.

 

[Humbug]

LOL I knew people will take this the wrong way and panic.

 

chill guys it just means that in the next 4 years the CPUs they launch will be improvements based on the original design of Zen. They will still get faster and faster. In the same way that Intel CPUs have got faster since Sandy Bridge... It's just that within that time you won't see in one generation any massive 40% IPC gains in one go like you did from excavator-->ryzen or from core2-->sandy bridge.

 

[MadyTehWolfie]

If it hands intel their ass, expect intel with their R&D budget to just develop something faster after a year or two let alone 4 years. SIgh then back to the same old same old slight improvement each year for the consumer grade stuff.

[Humbug]

1 hour ago, MadyTehWolfie said:

If it hands intel their ass, expect intel with their R&D budget to just develop something faster after a year or two let alone 4 years. SIgh then back to the same old same old slight improvement each year for the consumer grade stuff.

intel cannot develop and roll out an architecture within 1-2 years. Some people seem to think that intel is some kind of God company which can respond immediately to competition or which has some secret uber powerful architecture waiting to launch when needed.

 

the fact that things are not improving as fast as they did in early-mid 2000s doesn't mean that intel isn't trying. It's natural for any industry that that level of growth and improvement cannot be sustained forever.

 

if zen is competitive intel can adjust their pricing. Maybe after 1-2 years they can launch six cores as mainstream etc... But don't expect them to suddenly jump their ipc by 20% in one shot.

[Zodiark1593]

1 hour ago, Humbug said:

intel cannot develop and roll out an architecture within 1-2 years. Some people seem to think that intel is some kind of God company which can respond immediately to competition or which has some secret uber powerful architecture waiting to launch when needed.

 

the fact that things are not improving as fast as they did in early-mid 2000s doesn't mean that intel isn't trying. It's natural for any industry that that level of growth and improvement cannot be sustained forever.

 

if zen is competitive intel can adjust their pricing. Maybe after 1-2 years they can launch six cores as mainstream etc... But don't expect them to suddenly jump their ipc by 20% in one shot.

*looks sideways at Core 2* 

 

AMD is simply hyping up something utterly meaningless for the consumer. If one were to measure the success of their architecture by how "long it lasts", Bulldozer had already surpassed the 4 year mark, and I'm rather certain that Bulldozer was a resounding success. *sarcasm*

 

Come to think of it, was Bulldozer not hyped up to such extent as well? Would be a shame if that massive hype train derailed. 

 

[Coldfuson]

7 hours ago, M.Yurizaki said:

People still use Sandy/Ivy Bridge and don't feel a need to upgrade, or are just starting to feel the need. And there's a noticeable performance gap between that and Skylake: http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/11

 

But if it's just Ryzen will be the foundation of improvements, then I don't really see that being an issue either. Intel's architectures haven't really changed much since Sandy Bridge either.

Yeah I'm still on Ivy Bridge here I think... 4930k? what codename is that? I know it has a 4XXX sku, but the architecture is a year behind. 3XXX skus were ivy bridge right?

[TechGod]

7 hours ago, Darth Revan said:

No, you're delusional thinking that AMD thought about the future.

Ryzen is all they have. They went all in with this. Hoping that it will be a success and they can later on have the funds to make another architecture. And honestly I do want them to succeed, I really do. 

Thank you very much.

 

Did you used to work at AMD? Is this how you know that? 

[LAwLz]

4 hours ago, Humbug said:

intel cannot develop and roll out an architecture within 1-2 years. Some people seem to think that intel is some kind of God company which can respond immediately to competition or which has some secret uber powerful architecture waiting to launch when needed.

 

the fact that things are not improving as fast as they did in early-mid 2000s doesn't mean that intel isn't trying. It's natural for any industry that that level of growth and improvement cannot be sustained forever.

 

if zen is competitive intel can adjust their pricing. Maybe after 1-2 years they can launch six cores as mainstream etc... But don't expect them to suddenly jump their ipc by 20% in one shot.

But they would not need to roll out another über architecture to stomp AMD. Even if we assume that such an architecture do/don't exist (personally I think Intel has some yet-to-be-implemented performance increasing things in their sleeves), all they would have to do to boost performance is to release 6 core chips on the mainstream socket. All of a sudden you got 50% more "raw performance".

 

If you want to go even further, implement AVX512 (currently only available on some Skylake EP/EX chips, and Xeon Phi if I recall correctly). For some specific workloads that could potentially increase performance by way more than 20%.

 

Remove the GPU on the 6-core SKUs if you are worried about cost.

 

 

I am really hoping that AMD makes a good comeback, but it's foolish to assume that Intel will have no way of responding if Zen turns out to be great.

[Eroda]

16 hours ago, wrathoftheturkey said:

Well, shit. I mean, that's true for many intel processors too, but saying it like that seems to foreshadow what they did before (not make another competitive architecture for four years)

well they might just do an iterive process node jump etc in 18 months that gives a small performance gain  because its not like intel are making big gains every 12 months anymore. plus 8C16T it should solidly slay every intel mainstream CPU its kinda like buying a broadwell E 8 Core its not something you will need to upgrade from for 2-3 years unless you want to

[MrDynamicMan]

17 hours ago, wrathoftheturkey said:

(not make another competitive architecture for four years)

Intel says hi

[samuel]

This is consumer friendly IMO. I can keep my FX8320 for another 4 years and then just upgrade everything.

[othertomperson]

11 hours ago, Princess Cadence said:

I understand now, though I dont see how is this supposed to be any good for the costumers, you will still have to buy a whole new CPU to have such polishing to the architecture while it will most likely not be a reasonable deal to who bought the first ryzen processors anyways, this really seems like going to be no different than what Intel done with skylake to kaby lake so this news are quite irrelevant in my point of view

I'm not too bothered tbh? The processors won't be unusable in 4 years, they just won't be current. It'll be like Ivy Bridge is now, it's still decent it's just that there's better out there

[Humbug]

2 hours ago, LAwLz said:

But they would not need to roll out another über architecture to stomp AMD. Even if we assume that such an architecture do/don't exist (personally I think Intel has some yet-to-be-implemented performance increasing things in their sleeves), all they would have to do to boost performance is to release 6 core chips on the mainstream socket. All of a sudden you got 50% more "raw performance".

 

If you want to go even further, implement AVX512 (currently only available on some Skylake EP/EX chips, and Xeon Phi if I recall correctly). For some specific workloads that could potentially increase performance by way more than 20%.

 

Remove the GPU on the 6-core SKUs if you are worried about cost.

 

 

I am really hoping that AMD makes a good comeback, but it's foolish to assume that Intel will have no way of responding if Zen turns out to be great.

Yep I agree they will respond well. And they can also increase their stock clockspeeds. The advantage is still very much with them regardless of how competitive or not ryzen is...

 

I was just detecting a tone on this forum that people thought the reason for slowed improvements in IPC since Sandy Bridge was because of the non-competition from AMD.

[Guest]

16 hours ago, Trixanity said:

If people would actually read the thread and read the source, they'd realize what this means and by extension how ambitious AMD is. They'll be shipping IPC improvements every year for the next four years. That's how AMD operates. Intel and Nvidia do something similar. You don't reinvent your architecture every year; you improve it with a fairly quick cadence while another team starts working on a brand new one which takes at least 3 years. Mark Papermaster even says Ryzen is on a "tock-tock-tock" schedule meaning architecture improvements till Zen is replaced.

 

That some of you even think AMD could release a product in this kind of market and then wait four years to release a new one is ridiculous. OEMs want product refreshes annually. AMD will provide that. Zen+ will be released in 2018 along side Navi. 

Ryzen would fall behind fairly quickly if it would stand still for four years in any case.

I'm concerned that they will hit the same brick wall that Intel has, before realizing it. It is a bit of hype and borderline exaggeration to claim this. Marketing once again, at its finest. They are 1. Extremely good at forecasting or 2. wrong. 

 

It will be interesting to follow - I am more excited what this means for the development of AI and other technologies than I am for, specifically, the gaming industry. Although the VR and graphical advancements for primary and alternative devices, in the upcoming future, should be some of the most interesting advancements in tech that we have seen for quite some time. 

[Mira Yurizaki]

4 hours ago, othertomperson said:

I'm not too bothered tbh? The processors won't be unusable in 4 years, they just won't be current. It'll be like Ivy Bridge is now, it's still decent it's just that there's better out there

People should really learn to shift their idea of what "obsolete" means. It's not "there's something better", because technically we can sue the pants off of every hardware company out there for selling "obsolete" stuff. Like Intel release an i7 and an i5, well that i5 is "obsolete" because the i7 is better.

 

No, it should mean "when the thing cannot do what I want it to."

[captain cactus]

I feel like they could've said it better, both AMD and all the other publications quoting AMD.

 

The Ryzen CPUs won't "last" 4 years (they will physically of course), the Zen architecture will. After that we'll have a new architecture from AMD and until then they'll improve upon the Zen architecture (with Zen+ already confirmed by AMD themselves). So we'll have Ryzen, Zen+, Zen something else and Zen still something else before we'll have a new architecture. AMD won't pull another Excavator on us. They've learned that that was a massive mistake which cost them a lot of money and customers.

 

[Sniperfox47]

9 hours ago, Zodiark1593 said:

Come to think of it, was Bulldozer not hyped up to such extent as well? Would be a shame if that massive hype train derailed. 

Yeah. Bulldozer was a shitshow. But there is a difference here with Ryzen. Bulldozer was hyped to heck but nobody got to play with it before release. Ryzen has chips being used in live demos in front of major industry figures where it's presented so that they can open it up, look at it, touch it.

 

And to be fair to bulldozer, it was great for certain server workloads, it just wasn't particularly useful for consumers or anything that used floating point calculations.

[Doobeedoo]

Is it really that hard to understand what AMD meant by this?!

[Trixanity]

52 minutes ago, Alexokan said:

I'm concerned that they will hit the same brick wall that Intel has, before realizing it. It is a bit of hype and borderline exaggeration to claim this. Marketing once again, at its finest. They are 1. Extremely good at forecasting or 2. wrong. 

 

It will be interesting to follow - I am more excited what this means for the development of AI and other technologies than I am for, specifically, the gaming industry. Although the VR and graphical advancements for primary and alternative devices, in the upcoming future, should be some of the most interesting advancements in tech that we have seen for quite some time. 

Silicon is hitting a brick wall on the CPU side. That's just how it is. I do believe AMD can deliver but it's probably to the tune of single digit percentages. Zen+ might be double digits but any subsequent release I'd expect to be single digit. Keep in mind they're already working on these things and have plans for at least the next five years. They already know what to work on to improve on the Zen design. Design improvements don't come out of a vacuum.

 

The advantages to subsequent releases is likely to be new features like with Intel; at least more so than performance I'd wager. Ryzen will be good enough for most users for the foreseeable future. What's most important for AMD is to get Ryzen out to all markets including servers and laptops and Vega as well. 2017 can make or break AMD.

[othertomperson]

56 minutes ago, M.Yurizaki said:

People should really learn to shift their idea of what "obsolete" means. It's not "there's something better", because technically we can sue the pants off of every hardware company out there for selling "obsolete" stuff. Like Intel release an i7 and an i5, well that i5 is "obsolete" because the i7 is better.

 

No, it should mean "when the thing cannot do what I want it to."

This principle is why I'm still on Sandy Bridge-E. When new games are bottlenecked I'll upgrade, and not before. Surprisingly I've started to see (due to Dx12) people being bottlenecked by an i7 6700K, while I am not. Maybe I should give Ryzen more credit after all.