AMD CPU speculation... and expert conjecture

[8350rocks]

AMD will be reentering the large scale x86 server market and produce a HEDT solution from the new uarch. Keller is heading it up. Things will be good...but not yet.

 

[Zircoben]

Does anyone ever see AMD pulling out of CPUs entirely and focusing on GPUs?

We would all hate to see Intel then have a monopoly (high prices, slower technology, etc.) , but that really seems like what could very well happen.

 

[8350rocks]

Negative, there would be absolutely no way the government would allow that...the anti-trust ramifications alone would be horrendous.

 

[james pinnix]

But the anti trust would effect intel more then amd

 

[de5_Roy]

amd a57 based seattle details (by that, i mean finally!! useful promo slides and all.)
http://www.tomsitpro.com/articles/amd-opteron-seattle-arm-soc,1-2105.html
http://www.anandtech.com/show/8362/amds-big-bet-on-arm-powered-servers-a1100-revealed

apparently, it was designed by famous xzibit who put an soc inside the soc.... goes to show how much i know about servers.....
from the diagrams, though, the chip seems to be aimed at storage servers and similar stuff, not compute. amd was trumpeting better compute/watt than kyoto (or something) yet the features that actually attractive aren't the arm cores.

openGL NG (not good? )
http://www.anandtech.com/show/8363/khronos-announces-next-generation-opengl-initiative
i don't know what clean break means.

 

[gamerk316]

Negative, there would be absolutely no way the government would allow that...the anti-trust ramifications alone would be horrendous.

CPU <> X86. X86 is dwarfed by MIPS. There's plenty of competition there.

Granted, desktop users are screwed, by they're in the minority when it comes to CPU sales.

 

[gamerk316]

openGL NG (not good? )
http://www.anandtech.com/show/8363/khronos-announces-next-generation-opengl-initiative
i don't know what clean break means.

http://www.extremetech.com/gaming/187796-opengl-4-5-released-next-gen-opengl-unveiled-cross-platform-mantle-killer-dx12-competitor

"Clean Break" = Not Backward Compatible. Its a brand new API. Which I've been calling for over a decade now. Essentially, we're finally getting what OGL 3.0 should have been.

So as I predicted, we now have the worst thing possible for games: A graphical API war.

EDIT

That being said, this officially kills Mantle. The need for a vendor specific cross platform middle-level API goes away when you have a open cross platform middle-level API. You'll see a few more titles already in development use it, but with both DX12 and OGL NG fixing the problems Mantel dealt with, its essentially an obsolete API. Its got a year or so left in it, at most.

 

[blackkstar]

I think that some of you will change your tune about AMD HEDT after you watch Carrizo go mobile only and then proceed to only show up in $500 craptops with 15in or great screens at 768p resolution with TN panels and 5400rpm hard drives.

I don't want to see this happen to AMD, but if we see it go just like Jaguar, I can finally dub the whole thing "the I love mobile AMD experience" It goes something like this.

New architecture looks great on paper and looks competitive with Intel

AMD releases prototype that looks great

Product is released by AMD, shows up in a bunch of awful OEM products

After a few months, we get a design "win" that is mediocre at best but AMD backs it because it's the only remotely appealing device to use AMD's new product

The mediocre design win fails (acer V5) and AMD ignores it

Right before the replacement AMD parts shows up and by the time Intel launches new chips, AMD product gets some alright design wins that fill specific niches (MSI A4-5000 laptop, Thinkpad x140e)

Everyone who went "AMD mobile is the future!" complains that Intel is playing dirty and that next time things will be different. They also say HEDT is dead anyways and that mobile is the only future for AMD. They then view the entire scenario as the death of x86 and then run around going "tablets, phones, and ARM are killing AMD!"

It's going to happen with Carrizo and it's going to be fun to watch how the usual crowd goes through the usual stages of coping with being so wrong. The whole thing is always made even sweeter by the fact that most AMD mobile announcements on social websites are full of comments asking for an HEDT part.

 

[jdwii]

AMD will be reentering the large scale x86 server market and produce a HEDT solution from the new uarch. Keller is heading it up. Things will be good...but not yet.

You hear the news from juan? About the 16 core fx? That would mean Amd took keller and forced him to produce blue meth to take over Walter whites business, and then forced him to make a quick design and his partner which is probably juan made the new x86 design.

Came to think that maybe if we wish for it its nothing more than a 4-6-8 core fx with SMT.
But man if what juan is saying is correct then this just sucks and i do not expect many people buying some weak CPU when will Amd learn? They could make a 40 core fx and no one would care if its single threaded performance is around a Pentium 4.

 

[jdwii]

I think that some of you will change your tune about AMD HEDT after you watch Carrizo go mobile only and then proceed to only show up in $500 craptops with 15in or great screens at 768p resolution with TN panels and 5400rpm hard drives.

I don't want to see this happen to AMD, but if we see it go just like Jaguar, I can finally dub the whole thing "the I love mobile AMD experience" It goes something like this.

New architecture looks great on paper and looks competitive with Intel

AMD releases prototype that looks great

Product is released by AMD, shows up in a bunch of awful OEM products

After a few months, we get a design "win" that is mediocre at best but AMD backs it because it's the only remotely appealing device to use AMD's new product

The mediocre design win fails (acer V5) and AMD ignores it

Right before the replacement AMD parts shows up and by the time Intel launches new chips, AMD product gets some alright design wins that fill specific niches (MSI A4-5000 laptop, Thinkpad x140e)

Everyone who went "AMD mobile is the future!" complains that Intel is playing dirty and that next time things will be different. They also say HEDT is dead anyways and that mobile is the only future for AMD. They then view the entire scenario as the death of x86 and then run around going "tablets, phones, and ARM are killing AMD!"

It's going to happen with Carrizo and it's going to be fun to watch how the usual crowd goes through the usual stages of coping with being so wrong. The whole thing is always made even sweeter by the fact that most AMD mobile announcements on social websites are full of comments asking for an HEDT part.

To be honest i will always do most of my gaming on a custom build not a laptop. And the FX APU's Amd has are great for laptops and plenty for my gaming needs, my Llano A8 3520M is doing ok need to upgrade though to the next best APU from Amd on laptops been quite a bit of improvements.
One thing i love is the power consumption this laptop uses its very low and i have a SSD in it to as all laptops should. That really has to be the biggest bottleneck for most people even my dad noticed a big difference with a SSD over a hard drive and that is on a old Athlon 4800+ CPU with 4GB of ram and a 6150se from Nvidia.

 

[8350rocks]

If OGL uses HLSL it is over for DX...

Just saying...

 

[esrever]

AMD will be reentering the large scale x86 server market and produce a HEDT solution from the new uarch. Keller is heading it up. Things will be good...but not yet.

You hear the news from juan? About the 16 core fx? That would mean Amd took keller and forced him to produce blue meth to take over Walter whites business, and then forced him to make a quick design and his partner which is probably juan made the new x86 design.

Came to think that maybe if we wish for it its nothing more than a 4-6-8 core fx with SMT.
But man if what juan is saying is correct then this just sucks and i do not expect many people buying some weak CPU when will Amd learn? They could make a 40 core fx and no one would care if its single threaded performance is around a Pentium 4.

No one will care anyways because intel will market better than AMD. It doesn't matter what AMD make, people will never see something good until they beat intel at everything and that isn't going to happen. The physical impossibility of it is just based on how much intel can fund every sector more than AMD. AMD makes a 8 core sandybridge performing chip, they will still end up still be behind on marketing, people still won't buy it. You can just check AMD's gpu sales vs nvidia, AMD is getting Tom's monthly GPU recommendation for the price top to bottom but nvidia still wins the sales. Flip it to CPUs it would be even worse since keeping up with intel is on a completely different level than keeping up with nvidia. Add to that the fact that intel is cozy with all the OEMs the AMD designs never get put in the places they shine. Its not even necessarily intel's or the OEM's fault for that, you go to bestbuy and all the consumers want intel iSomthings when they buy an laptop, they don't read reviews, they don't know about kaveri or beema, add to that the salesman is always going to push intel anyways because they get more marketing from intel. People get told at bestbuy how AMD laptops overheat, under perform and lag, which is true in some sense but the real answer is just that AMD chips are quite competitive in mobile but its very hard to find a laptop that can show it off. The only people who end up buying AMD is the ones who buy the cheap ones since they don't want to spend a lot of money and their is a sea of cheap AMD laptops with cheap specs and nothing special about them.

That was a lot more than I wanted to rant about but people keep saying AMD should come out with this this and that on a forum and it just won't work. AMD could have the best CPU and intel would still sell a lot more. AMD will never have the best CPU because intel sinks more money into R&D than AMD is worth every quarter. Its a lose lose situation to cater to these markets. Although I would like AMD to return to the HEDT, I don't think it would even matter. I would be one of the hand full of people who would actually buy AMD HEDT and that just isn't going to be profitable. I would much rather AMD just make what they make work and continue to make things they are actually able to.

People saying Keller will be a turn around in HEDT space are just lying to themselves. Keller will make a nice design, it will be good for what AMD intend it to be but it will be outsold by intel 4 to 1 even if it is the best thing ever.

 

[szatkus]

No one will care anyways because intel will market better than AMD. It doesn't matter what AMD make, people will never see something good until they beat intel at everything and that isn't going to happen. The physical impossibility of it is just based on how much intel can fund every sector more than AMD. AMD makes a 8 core sandybridge performing chip, they will still end up still be behind on marketing, people still won't buy it. You can just check AMD's gpu sales vs nvidia, AMD is getting Tom's monthly GPU recommendation for the price top to bottom but nvidia still wins the sales. Flip it to CPUs it would be even worse since keeping up with intel is on a completely different level than keeping up with nvidia. Add to that the fact that intel is cozy with all the OEMs the AMD designs never get put in the places they shine. Its not even necessarily intel's or the OEM's fault for that, you go to bestbuy and all the consumers want intel iSomthings when they buy an laptop, they don't read reviews, they don't know about kaveri or beema, add to that the salesman is always going to push intel anyways because they get more marketing from intel. People get told at bestbuy how AMD laptops overheat, under perform and lag, which is true in some sense but the real answer is just that AMD chips are quite competitive in mobile but its very hard to find a laptop that can show it off. The only people who end up buying AMD is the ones who buy the cheap ones since they don't want to spend a lot of money and their is a sea of cheap AMD laptops with cheap specs and nothing special about them.

You are so right.

"- AMD had better CPUs in the past.
- But only for short time.
- 6 years...
- What?"

"- Only GeForce is fully compatible with DX11.
- WTF?"

 

[de5_Roy]

"- Only GeForce is fully compatible with DX11.
- WTF?"

lol, haven't heard that one.

i am hoping that the advent of 4K and wider availability of 1080p videos will get more budget-conscious people (the ones that still buy pcs) to upgrade from ye olde intel igps in oem pcs or locally sold prebuilt pcs to amd apus especially in b.r.i.c. countries. amd made a good move launching am1 platform and the new apus in european countries and china. if microsoft hadn't botched windows 8, amd coulda seen a bit more sales.

 

[gamerk316]

Broadwell info:

http://www.extremetech.com/computing/187770-intels-core-m-mobile-offensive-broadwell-y-delivers-huge-performance-increases-dramatically-better-power-consumption

About as expected: Limited CPU gains, potentially big GPU gains.

 

[de5_Roy]

^^ the unveiling is msotly about (over)hyping the core M cpus/socs and bragging about their 14nm process. s/a says compared to intel's claims, no other foundries have "real"(as i read it) shrink to 14nm process. i am waiting to see if intel can deliver and how sam/glofo/tsmc react.

 

[juanrga]

People saying Keller will be a turn around in HEDT space are just lying to themselves. Keller will make a nice design, it will be good for what AMD intend it to be but it will be outsold by intel 4 to 1 even if it is the best thing ever.

Recent news about AMD increasing the core number from 16 to 20 cores are an evident sign that this will be a smaller and weaker core, and "nice" is not the word that I would use to describe the details about the core. I would use "weird".

 

[8350rocks]

People saying Keller will be a turn around in HEDT space are just lying to themselves. Keller will make a nice design, it will be good for what AMD intend it to be but it will be outsold by intel 4 to 1 even if it is the best thing ever.

Recent news about AMD increasing the core number from 16 to 20 cores are an evident sign that this will be a smaller and weaker core, and "nice" is not the word that I would use to describe the details about the core. I would use "weird".

I would not take that rumor too seriously...the maximum projected number I have heard for HEDT is just a tad over half that figure (and that was not a certainty at that)...16 would be aiming high...

Now, for servers? Possibly...though you would likely be looking at something on a massive monolithic die. That would not surprise me there...since they already had a 16 core MCM package Opteron.

 

[juanrga]

After the deception at knowing details of the new AMD arch (the K in K12 doesn't mean rocKs but sucKs), Nvidia finally presents the long-time awaited CPU

http://blogs.nvidia.com/blog/2014/08/11/tegra-k1-denver-64-bit-for-android/

As expected it is a highly innovative design for a CPU. I think that it deserves the name of GCPU or CPU made from the perspective of a GPU maker!

I was expecting a VLIW design with transmeta-like translation from armv8 and Nvidia has presented an in-order design (yes, as cores in GPUs) but then has added an innovative DCO (Dynamic Code Optimization) technique to simulate OoO logic. The White paper claims a much larger instruction window (up to 10x more than usual OoO hardware).

This could open the doors for a new kind of higher-ILP CPUs beyond the traditional OoO superscalar designs. In fact Nvidia claims that Denver can execute up to 7 instructions per clock. For the sake of comparison, the ARM core used in Seattle peaks at 3 instructions. Haswell designs are 4-wide.

With the larger caches and wide issue operation, Denver can achieve IPC rates comparable to an Intel Core processor, but consuming only the power of a mobile processor. Amazing!

NVIDIA's performance goals for Denver are quite aggressive - it can actually be compared to desktop CPUs, with some benchmarks demonstrating performance on par with an Intel Haswell-based Celeron processor (Figure 4). It’s also extremely fast for a mobile processor as seen when compared to the Apple A7 processor on those same benchmarks. The performance levels shown will make Denver a snappy processor, suitable for super-smartphones, tablets, and Chromebooks.

Now waiting for the Boulder core... which will be a scaled up version of this.

 

[-Fran-]

And to add a little more gas to the fire: http://techreport.com/news/26906/nvidia-claims-haswell-class-performance-for-denver-cpu-core

Let's see if nVidia is right about their claims when they release it.

Cheers!

 

[8350rocks]

First of all, NVidia will not get any design wins to speak about...second, Qualcomm and Samsung are still dominating market share by a wide margin.

Third, if this is reminiscent of the last batch of hype NVidia put out over Tegra 4i, then reality will show the benchmarks are half what they said, and it will still end up nowhere.

NVidia is hemorrhaging money from this project, at least when this one fails maybe they will finally give up. I mean, the only design win they had last generation to speak of was their own SHIELD that flopped...

 

[palladin9479]

openGL NG (not good? )
http://www.anandtech.com/show/8363/khronos-announces-next-generation-opengl-initiative
i don't know what clean break means.

http://www.extremetech.com/gaming/187796-opengl-4-5-released-next-gen-opengl-unveiled-cross-platform-mantle-killer-dx12-competitor

"Clean Break" = Not Backward Compatible. Its a brand new API. Which I've been calling for over a decade now. Essentially, we're finally getting what OGL 3.0 should have been.

So as I predicted, we now have the worst thing possible for games: A graphical API war.

EDIT

That being said, this officially kills Mantle. The need for a vendor specific cross platform middle-level API goes away when you have a open cross platform middle-level API. You'll see a few more titles already in development use it, but with both DX12 and OGL NG fixing the problems Mantel dealt with, its essentially an obsolete API. Its got a year or so left in it, at most.

This is a very good thing. I'm predicting it'll be similar to OpenGL ES but with vastly expanded features. Also looks like they are finally saying "f*ck it" to the CAD/CAM industry folks since you don't see too many of them supporting the new redesign. OpenGL should of been split into two sets a long time ago, one focused on the quality, realism and accuracy of CAD/CAM the other for performance and ease of development for desktops.

 

[palladin9479]

First of all, NVidia will not get any design wins to speak about...second, Qualcomm and Samsung are still dominating market share by a wide margin.

Third, if this is reminiscent of the last batch of hype NVidia put out over Tegra 4i, then reality will show the benchmarks are half what they said, and it will still end up nowhere.

NVidia is hemorrhaging money from this project, at least when this one fails maybe they will finally give up. I mean, the only design win they had last generation to speak of was their own SHIELD that flopped...

The Tegra 4 is an amazing chip, it just gets hot so it'll throttle if placed inside any phone. Shield didn't flop, its also an amazing device but it's for a very niche market segment. I own one and I constantly use it, got a ton of games along with lots of emulators and what not loaded on it. The Shield isn't a smart phone and doesn't compete with them, it's competing with the Nintendo handhelds and doing a pretty remarkable job considering it's nVidia's first attempt into the marketplace and doesn't have any exclusive titles. It's one of those devices you really don't know what to do with until you have one, then you start getting all these ideas. My one complaint is that they are trying to to do too much with it as a platform vs focusing on mobile gaming. It's caused people to get distracted with it thinking it's only for desktop streaming instead of using it as a standalone portable gaming device.

 

[Cazalan]

Well Microsoft made something that will eat weeks of compute time.

http://www.extremetech.com/extreme/187778-microsofts-awesome-hyperlapse-tech-finally-makes-first-person-gopro-videos-watchable

Bring on those 20 core APUs, but I think they mean 4 CPU 16 GPU cores.

 

[gamerk316]

openGL NG (not good? )
http://www.anandtech.com/show/8363/khronos-announces-next-generation-opengl-initiative
i don't know what clean break means.

http://www.extremetech.com/gaming/187796-opengl-4-5-released-next-gen-opengl-unveiled-cross-platform-mantle-killer-dx12-competitor

"Clean Break" = Not Backward Compatible. Its a brand new API. Which I've been calling for over a decade now. Essentially, we're finally getting what OGL 3.0 should have been.

So as I predicted, we now have the worst thing possible for games: A graphical API war.

EDIT

That being said, this officially kills Mantle. The need for a vendor specific cross platform middle-level API goes away when you have a open cross platform middle-level API. You'll see a few more titles already in development use it, but with both DX12 and OGL NG fixing the problems Mantel dealt with, its essentially an obsolete API. Its got a year or so left in it, at most.

This is a very good thing. I'm predicting it'll be similar to OpenGL ES but with vastly expanded features. Also looks like they are finally saying "f*ck it" to the CAD/CAM industry folks since you don't see too many of them supporting the new redesign. OpenGL should of been split into two sets a long time ago, one focused on the quality, realism and accuracy of CAD/CAM the other for performance and ease of development for desktops.

Oh, don't get me wrong, re-writing OGL from scratch is a good thing. Very good in fact. But remember, NVIDIA/AMD could be supporting up to five different graphical API's in their drivers now: DX11, OGL 4.x, DX12, OGL NG, and for AMD, Mantle. Which begs the question: Which API's are going to get shafted due to difficulty writing the drivers?

Hence why both MSFT and Khronos better be careful designing the API, since implementation will likely determine its success.