AMD Post-Excavator

Discussion of all forms of processors, from AMD to Intel to VIA.

Moderators: Flying Fox, morphine

AMD Post-Excavator

Postposted on Tue Jan 14, 2014 9:20 pm

Hey guys, I partially read Anandtech's take on the new Kaveri chips and some of the things he said caught my interest. Here's the article.

On the first page, Anand said this:

Kaveri is the launch vehicle for AMD's Steamroller CPU architecture, the 3rd iteration of the Bulldozer family (and second to last before moving away from the architectural detour)


Architectural detour. Hmmm. So is he implying that Bulldozer is just a 'temporary' architecture AMD's just trying to live with until they come up with a new/resurrected core the way Pentium 4 was for Intel?

And then this:

I suspect we'll eventually see an AMD return to the high-end, but that'll come once we're done with the Bulldozer family.


These are all fine and dandy... IF Anand turns out to be right. There are also rumors that AMD is secretly gong back to their K10 core and tweaking it to make a comeback. This is all speculation and rumors, of course.
The three pillars of my digital life: AMD FX-8350, Google Nexus 7 (Qualcomm Snapdragon S4 Pro), Intel Core i5-2450M
ronch
Gerbil Elite
 
Posts: 641
Joined: Mon Apr 06, 2009 7:55 am
Location: C:\Program Files\

Re: AMD Post-Excavator

Postposted on Tue Jan 14, 2014 10:29 pm

Wishful thinking on Anand's part. I could see them reviving K10 but the shrinking PC market doesn't have enough money to justify Intel or AMD developing a new CPU architecture for a very long time.
NovusBogus
Gerbil Elite
 
Posts: 512
Joined: Sun Jan 06, 2013 12:37 am

Re: AMD Post-Excavator

Postposted on Tue Jan 14, 2014 10:32 pm

I think it is somewhat optimistic. I'd say odds are currently stacked against them ever re-entering the high-end desktop market. My guess is that they will go with a mid-performance APU-type architecture for the desktop. If they have high-end x86 chips at all they will be server (Opteron) parts with lots of cores, moderate clock speeds, and middle-of-the-road single-thread performance.

They want to be a major player in ARM. This will be a long, uphill climb for them and is a risky bet as there are already a lot of players in this space. Maybe they can leverage their expertise in server platforms to push ARM into the enterprise, or maybe an ARM-based APU will get them some design wins in devices where more graphics performance is needed than existing embedded GPUs can deliver. Time will tell.

I give 1 in 4 odds that 4 years from now they are a GPU vendor only.
(this space intentionally left blank)
just brew it!
Administrator
Gold subscriber
 
 
Posts: 37672
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: AMD Post-Excavator

Postposted on Tue Jan 14, 2014 10:38 pm

just brew it! wrote:I give 1 in 4 odds that 4 years from now they are a GPU vendor only.

Going full circle and be ATI again? :o
Image
The Model M is not for the faint of heart. You either like them or hate them.

Gerbils unite! Fold for UnitedGerbilNation, team 2630.
Flying Fox
Gerbil God
 
Posts: 24378
Joined: Mon May 24, 2004 2:19 am

Re: AMD Post-Excavator

Postposted on Thu Jan 16, 2014 4:02 pm

just brew it! wrote:maybe an ARM-based APU will get them some design wins in devices where more graphics performance is needed than existing embedded GPUs can deliver.

But then they're competing with the Nvidia Tegra K1 which is already on the market. Day late and a dollar short.
i5-3570K, ASRock Z77 Pro4-m, Asus GTX660 TOP, 120 GB Vertex 3 Max IOPS, 2 TB Samsung EcoGreen F4, 8GB G-Skill @1.25V, Silverstone PS07B
DPete27
Gerbil Jedi
Silver subscriber
 
 
Posts: 1635
Joined: Wed Jan 26, 2011 12:50 pm
Location: Madison, Wisconsin

Re: AMD Post-Excavator

Postposted on Mon Jan 20, 2014 2:03 pm

To me, it's simple why Ian and Rahul made that statement. If AMD doesn't fix it's CPU performance problem, they won't be in the APU business beyond that point regardless of how good the graphics are.

As they stated, Haswell parts today are more power efficient, offer much better performance, and cost significantly less. That 4330 was $50 less, to be exact.

The 6750 GPU they used in the graphics performed significantly better than Kaveri, and it's modern day equivalent the 7750 can be had for around $80. The point they were making is, why buy a Kaveri when ya can have a faster, cooler processor and use the money saved to almost outright buy a 7750 for >3-4x the GPU performance anyway? A i3 4330 + 7750 costs only ~$30 more than the A10-7850k by itself.

However, NVIDIA is expected to offer true shared memory in the Maxwell generation


I loved reading about HSA and I hope it gains industry adoption. But this one tidbit sure got my attention though...
Kougar
Gerbil Team Leader
 
Posts: 277
Joined: Tue Dec 02, 2008 2:12 am
Location: Texas

Re: AMD Post-Excavator

Postposted on Mon Jan 20, 2014 4:18 pm

Mantle doesn't work on an i3 4330, the i3 to i7 don't come with a DSP block, and from all reviews only the $400 crystal well model can hope to match kaveri.
So indeed you need a discreet GPU to compete with kaveri. Now add that GPU to your power usage, and the 45w kaveri start to look amazingly good.

For a desktop, its no second coming. For an htpc, its the chip to get, and for laptop you finnaly get good gaming without a discreet GPU.
For that htpc, the hope is that the vce2, the same as the ps4 , get some good use in the future.
Also adding a gnc discreet card will not disable the onchip gcn cores. Mantle is prime for HSA and distributed compute.

AMD can't make a chip that rule all aspect of computing, but the < 50w models are very compelling.
We might have to wait a while, but 900+MHz gcn in kaveri should happen if gf hire people that know how to run a foundry.

The bencharmk to watch is bf4 on kaveri... Using a discreet GPU. I wouldn't be surprised that kaveri is twice faster then a i7 4770k if both system are using a 7750.
sschaem
Gerbil Team Leader
 
Posts: 247
Joined: Tue Oct 02, 2007 11:05 am

Re: AMD Post-Excavator

Postposted on Mon Jan 20, 2014 5:04 pm

sschaem wrote:Mantle doesn't work on an i3 4330, the i3 to i7 don't come with a DSP block

You'll want to be careful with such comparisons. Why are you assuming that Mantle or TrueAudio will be noticeable assets from the outset? They may well end up being ignored, driven into niches, or take a long time to blossom.

See also: Nvidia's PhysX and 3D Vision.
There is a fixed amount of intelligence on the planet, and the population keeps growing :(
morphine
Gerbil Khan
Silver subscriber
 
 
Posts: 9955
Joined: Fri Dec 27, 2002 8:51 pm
Location: Portugal (that's next to Spain)

Re: AMD Post-Excavator

Postposted on Mon Jan 20, 2014 5:15 pm

morphine wrote:You'll want to be careful with such comparisons. Why are you assuming that Mantle or TrueAudio will be noticeable assets from the outset? They may well end up being ignored, driven into niches, or take a long time to blossom.

See also: Nvidia's PhysX and 3D Vision.

Mantle already has plenty of support (normally I'd say unlike PhysX, but Warframe supports it so heh.)

And TrueAudio will be supported by third party audio engines, meaning game devs won't need to support it directly.
Meow.
Savyg
Gerbil Elite
Silver subscriber
 
 
Posts: 639
Joined: Thu Aug 26, 2004 6:18 am
Location: Between desert and tundra

Re: AMD Post-Excavator

Postposted on Mon Jan 20, 2014 5:20 pm

Savyg wrote:Mantle already has plenty of support.
And TrueAudio will be supported by third party audio engines, meaning game devs won't need to support it directly.

No, no, no.

Sorry guys, but you cannot count your eggs before the chicken lays them. The effect of those two techs is, at this point, speculation. I'll believe when the Mantle-enabled games are out and we can look at what kind of practical gains they have from using it, and I'll believe TrueAudio when it's actually implemented in practice and working in the majority of relevant games.

Again, I'm having flashbacks of 3D-Vision and PhysX. It was exactly the same kind of thing. Like I said before, I hope that both succeed since they're beneficial technologies, but I'm not putting any money in them before they're even out and about.
There is a fixed amount of intelligence on the planet, and the population keeps growing :(
morphine
Gerbil Khan
Silver subscriber
 
 
Posts: 9955
Joined: Fri Dec 27, 2002 8:51 pm
Location: Portugal (that's next to Spain)

Re: AMD Post-Excavator

Postposted on Mon Jan 20, 2014 5:25 pm

Maybe Anand knows things.
tipoo
Gerbil
 
Posts: 33
Joined: Tue Mar 19, 2013 5:43 pm

Re: AMD Post-Excavator

Postposted on Mon Jan 20, 2014 5:36 pm

Kougar wrote:The 6750 GPU they used in the graphics performed significantly better than Kaveri, and it's modern day equivalent the 7750 can be had for around $80. The point they were making is, why buy a Kaveri when ya can have a faster, cooler processor and use the money saved to almost outright buy a 7750 for >3-4x the GPU performance anyway? A i3 4330 + 7750 costs only ~$30 more than the A10-7850k by itself.


3-4x the GPU performance? How does that work? I'm kind of surprised by the number of posts such as this claiming that an HD 7750 is so much faster than the A10-7800. I mean, it is, of course, but not by anywhere near such margins. Neither the 7750, nor the 7800k provide anywhere close to flawless gaming at 1080p. From the limited comparisons I've been able to find, it seems like an i3 + HD 7750 would get, at most, about 50% more GPU performance. And in many cases, much less.

The point is, while the 7750 system does provide measurably better gaming performance, the difference in FPS such a system provides over a top end Kaveri is not likely to make or break gaming, other than perhaps in a few select titles. The extra cost (espeically after Kaveri prices settle a bit) and complexity in configuration involved in adding a discrete GPU, almost makes the Kaveri vs i3 + HD 7750 a bit of a toss up (but, yes, I'd still give the nod to the discrete graphics). Now if spending another $50 and going with something like an R7 260x, or 650Ti, that's where discrete graphics become worthwhile, at least in my opinion.
cynan
Gerbil Elite
Gold subscriber
 
 
Posts: 812
Joined: Thu Feb 05, 2004 2:30 pm

Re: AMD Post-Excavator

Postposted on Mon Jan 20, 2014 6:46 pm

morphine wrote:Sorry guys, but you cannot count your eggs before the chicken lays them. The effect of those two techs is, at this point, speculation. I'll believe when the Mantle-enabled games are out and we can look at what kind of practical gains they have from using it, and I'll believe TrueAudio when it's actually implemented in practice and working in the majority of relevant games.

The reality of current APIs is well known and exactly what is purported to be addressed by several world class game developers. And yes, I trust their word over yours.

I don't recall any game developers being half as excited about PhysX, ever.
Meow.
Savyg
Gerbil Elite
Silver subscriber
 
 
Posts: 639
Joined: Thu Aug 26, 2004 6:18 am
Location: Between desert and tundra

Re: AMD Post-Excavator

Postposted on Mon Jan 20, 2014 6:47 pm

morphine wrote:
sschaem wrote:Mantle doesn't work on an i3 4330, the i3 to i7 don't come with a DSP block

You'll want to be careful with such comparisons. Why are you assuming that Mantle or TrueAudio will be noticeable assets from the outset? They may well end up being ignored, driven into niches, or take a long time to blossom.

See also: Nvidia's PhysX and 3D Vision.



Semiaccurate had a interesting rumor buried in their Kaveri article:

A little birdie also told SemiAccurate that this unit is the PS4 sound system directly lifted without changes. If you were wondering who would support TrueAudio, the PS4 is the main dev target for this generation of consoles and it should be an easy port from there to Hawaii, CI, and now Kaveri. This is the long way of saying everyone should support it.


That gives me a little hope about TrueAudio catching on. That, and the fact that AudioKinect and Firelight intend to support it in their middleware.
Hz so good
Gerbil Elite
 
Posts: 534
Joined: Wed Dec 04, 2013 5:08 pm

Re: AMD Post-Excavator

Postposted on Mon Jan 20, 2014 6:57 pm

Savyg wrote:The reality of current APIs is well known and exactly what is purported to be addressed by several world class game developers. And yes, I trust their word over yours.

I believe you misunderstood me.

I am very much aware (and have for several years) of how DirectX architecture does not relate or make use of GPU+CPU power. I am aware that Mantle wishes to solve this (and how it's going to accomplish it), and I too claim that it is a Good Thing (tm), and would like for devs to kiss the old-and-busted DX architecture goodbye. I was not questioning Mantle's usefulness or potential in any way.

What I'm trying hard to point out is that a lot of people are taking Mantle as a done deal, the bee's femurs, the third coming of Cthulhu, or any variation thereof. "It's going to be awesome!". But there are absolutely zero Mantle games out there yet, and one (1) game that has stated a release date for their Mantle version (it's EA, so it's "this month" but at their usual patch/QA rate, nobody knows the year/millennium. But I digress.)

I really want to throw a party, but I'm not going to do that before the actual birthday. The person in question might get hit by a bus until then :)
There is a fixed amount of intelligence on the planet, and the population keeps growing :(
morphine
Gerbil Khan
Silver subscriber
 
 
Posts: 9955
Joined: Fri Dec 27, 2002 8:51 pm
Location: Portugal (that's next to Spain)

Re: AMD Post-Excavator

Postposted on Mon Jan 20, 2014 7:07 pm

Fair enough.

I don't know if it'll be a gigantic improvement, but at this point I'd take a solid 10% over existing APIs, and it sounds like it'll deliver much more than that.

We shall see.
Meow.
Savyg
Gerbil Elite
Silver subscriber
 
 
Posts: 639
Joined: Thu Aug 26, 2004 6:18 am
Location: Between desert and tundra

Re: AMD Post-Excavator

Postposted on Mon Jan 20, 2014 7:30 pm

Savyg wrote:Fair enough.

I don't know if it'll be a gigantic improvement, but at this point I'd take a solid 10% over existing APIs, and it sounds like it'll deliver much more than that.

We shall see.


A "solid 10%" isn't even going to be noticeable except maybe in benchmarks (and barely).
.
Deanjo
Gerbil XP
 
Posts: 395
Joined: Tue Mar 03, 2009 11:31 am

Re: AMD Post-Excavator

Postposted on Mon Jan 20, 2014 7:46 pm

Deanjo wrote:A "solid 10%" isn't even going to be noticeable except maybe in benchmarks (and barely).

I'd still take it over a 0% increase. Though that's not even relevant to anything.
Meow.
Savyg
Gerbil Elite
Silver subscriber
 
 
Posts: 639
Joined: Thu Aug 26, 2004 6:18 am
Location: Between desert and tundra

Re: AMD Post-Excavator

Postposted on Tue Jan 21, 2014 12:19 pm

cynan wrote:3-4x the GPU performance? How does that work? I'm kind of surprised by the number of posts such as this claiming that an HD 7750 is so much faster than the A10-7800. I mean, it is, of course, but not by anywhere near such margins. Neither the 7750, nor the 7800k provide anywhere close to flawless gaming at 1080p. From the limited comparisons I've been able to find, it seems like an i3 + HD 7750 would get, at most, about 50% more GPU performance. And in many cases, much less.

The point is, while the 7750 system does provide measurably better gaming performance, the difference in FPS such a system provides over a top end Kaveri is not likely to make or break gaming, other than perhaps in a few select titles. The extra cost (espeically after Kaveri prices settle a bit) and complexity in configuration involved in adding a discrete GPU, almost makes the Kaveri vs i3 + HD 7750 a bit of a toss up (but, yes, I'd still give the nod to the discrete graphics). Now if spending another $50 and going with something like an R7 260x, or 650Ti, that's where discrete graphics become worthwhile, at least in my opinion.


Yeah I admit I got that completely wrong, went and re-looked at the graphics benches. 30-100% better performance, but still under 30FPS in some benches. Still, the point still stands that the $50 in savings from buying a i3 4330 (which is a faster CPU) instead of the A10-7850k is half the cost of the R7 260x right there. Kaveri is targeting a niche in an already shrinking desktop market.

AMD can't afford to price their flagship model to Intel's bottom-model midrange parts, nor can they continue as an ongoing concern when they target only a small percentage of the server & desktop markets. AMD does sell a lot of laptop processors but those are all low margin budget models already. Kaveri will help there a little, but again it's only targeting a thin sliver of the market, between budget laptops and laptops with discrete GPUs. I'm curious to see the actual power consumption difference between Haswell and Kaveri when in a laptop, as Haswell should easily do better there.
Kougar
Gerbil Team Leader
 
Posts: 277
Joined: Tue Dec 02, 2008 2:12 am
Location: Texas

Re: AMD Post-Excavator

Postposted on Fri Feb 14, 2014 1:07 am

This is what AMD needs to do to get ahead.

Phenom III specs:

-14nm
-TRUE 8 cores
-8MB of L2 and 16MB of L3
-Make it consume less than ~80W
-Stock frequency 4.2Ghz, turbo up to 4.6Ghz.
-Leave room for higher overclocking :D
-Ram support for 2133mhz+
4960X
Gerbil In Training
 
Posts: 3
Joined: Mon Feb 10, 2014 6:14 pm

Re: AMD Post-Excavator

Postposted on Fri Feb 14, 2014 7:08 am

4960X wrote:This is what AMD needs to do to get ahead.

Phenom III specs:

-14nm
-TRUE 8 cores
-8MB of L2 and 16MB of L3
-Make it consume less than ~80W
-Stock frequency 4.2Ghz, turbo up to 4.6Ghz.
-Leave room for higher overclocking :D
-Ram support for 2133mhz+

...and a pony. :wink:
(this space intentionally left blank)
just brew it!
Administrator
Gold subscriber
 
 
Posts: 37672
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: AMD Post-Excavator

Postposted on Fri Feb 14, 2014 7:59 am

My only wish is that AMD go back to using true cores...Based on an architecture that is performance-focused.

This Bulldozer family has gone on for long enough!
stmok
Gerbil
 
Posts: 17
Joined: Tue Aug 31, 2004 6:57 am
Location: Sydney, Australia

Re: AMD Post-Excavator

Postposted on Fri Feb 14, 2014 9:41 am

4960X wrote:-Stock frequency 4.2Ghz, turbo up to 4.6Ghz.

Uh-huh. 4.6Ghz and room for overclocking? I take it you'll expect to get a couple bottles of LN2 in the package? :)
There is a fixed amount of intelligence on the planet, and the population keeps growing :(
morphine
Gerbil Khan
Silver subscriber
 
 
Posts: 9955
Joined: Fri Dec 27, 2002 8:51 pm
Location: Portugal (that's next to Spain)

Re: AMD Post-Excavator

Postposted on Sun Mar 23, 2014 11:48 pm

stmok wrote:My only wish is that AMD go back to using true cores...Based on an architecture that is performance-focused.

This Bulldozer family has gone on for long enough!


I won't say those Bulldozer cores aren't true cores. If AMD was simply able to clock them higher and say, hit >6.0GHz while sticking to 65w and beating the hell out of Intel, I don't think anyone would even complain about Bulldozer's resource-sharing.
The three pillars of my digital life: AMD FX-8350, Google Nexus 7 (Qualcomm Snapdragon S4 Pro), Intel Core i5-2450M
ronch
Gerbil Elite
 
Posts: 641
Joined: Mon Apr 06, 2009 7:55 am
Location: C:\Program Files\

Re: AMD Post-Excavator

Postposted on Mon Mar 24, 2014 12:11 am

ronch wrote:
stmok wrote:My only wish is that AMD go back to using true cores...Based on an architecture that is performance-focused.

This Bulldozer family has gone on for long enough!


I won't say those Bulldozer cores aren't true cores. If AMD was simply able to clock them higher and say, hit >6.0GHz while sticking to 65w and beating the hell out of Intel, I don't think anyone would even complain about Bulldozer's resource-sharing.



65Watts would take 11-14 nm.. and under .8? Volts? (Failed ohms law section) while neat in idea form. It is 100% beyond the processor tech they have or exists.. not ecen the 14nm intel test parts scale much over 4ghz
Cybert said: Capitlization and periods are hard for you, aren't they? I've given over $100 to techforums. I should have you banned for my money.
maxxcool
Gerbil Elite
Silver subscriber
 
 
Posts: 645
Joined: Thu Sep 12, 2002 8:40 am
Location: %^&*%$$

Re: AMD Post-Excavator

Postposted on Mon Mar 24, 2014 1:05 am

Voluntarily removed.
Last edited by codedivine on Tue Mar 25, 2014 1:07 am, edited 1 time in total.
codedivine
Gerbil Elite
Silver subscriber
 
 
Posts: 702
Joined: Sat Jan 24, 2009 8:13 am

Re: AMD Post-Excavator

Postposted on Mon Mar 24, 2014 8:13 am

I have been saying this for a while now....I wish they "AMD " would do a die shrink of the phenom 2 thuban core and with that shrink make it a true 8 core and drastically improve the on die memory controller. It just baffles me that AMD invented the onboard mem controller but can not get the bandwidth needed out of it in this day and age.

I hate that they keep adding steps in the pipeline and increasing the MHZ of the CPU to make up for the extra steps the chip has to do to to perform a operation. Intel reduced the amount of steps to complete a operation.

I think AMD needs to take a bunch of steps back to the Phenom 2 shrink and optimize it along with its memory controller. It really astonished me that they went the way of Intel's prescott when they designed bulldozer, I do not know what they were smoking when they made that decision.

I still would love to see a 28nm phenom 2 thuban core since I do not think it would cost much to make a bunch of engineering samples since the blueprint is available just to see what kind of performance just a die shrink would bring. I think the results would be interesting and shed some light on what they could have done just shrinking from 45nm and improving a old but good design with new instructions etc.
2600k HT on@4705mhz 8gb Cas9 1600 mem 2x EVGA GTX770 4gb Classified cards in SLI @1320 mhz core and 2003 mhz mem,mounted in CM HAF922 with a TX-850 PSU 2xHTPC's 2xi3 2120 3.3ghz dual core,1xasus LP HD6570 1xHIS hd7750@1150core1325mem,55"PanyVT30
vargis14
Graphmaster Gerbil
 
Posts: 1147
Joined: Fri Aug 20, 2010 6:03 pm
Location: philly suburbs

Re: AMD Post-Excavator

Postposted on Mon Mar 24, 2014 8:52 am

NovusBogus wrote:Wishful thinking on Anand's part. I could see them reviving K10 but the shrinking PC market doesn't have enough money to justify Intel or AMD developing a new CPU architecture for a very long time.


Going back to the 2nd post in the thread, this is totally wrong. Is the assumption above that AMD is going to develop a new architecture after the next one or two iterations of the old one? If so, that's false. Both AMD and Intel and Nvidia and any other processor design house work on these things for years before anyone hears anything about them - and my guess is that the timetable is longer for CPUs than for GPUs. AMD likely had a rough roadmap for their next architecture before the first Bulldozer hit the shelves. Then, I'd wager that roadmap saw some changes after Dirk was shown the door, based upon the weaknesses of the Bulldozer architecture. Since those early days, I imagine that roadmap has been continuously refined and is no longer a roadmap, rather it is a full technical specification at this point. Much of the next architecture is already designed at this point, major features are implemented and "frozen", as the engineers continually close in on the smaller and smaller details.

The PC market is still huge. So is the laptop market and the server market. These markets have always used variations of the same architecture.

Intel seems to have reached a point where it has a fundamental architecture that it may stick with for a very long time and will just continually refine it unless and until some breakthrough is discovered.

AMD's fundamental architecture has some big issues: power consumption, max attainable performance, even performance, etc., etc. AMD will bring out another X86/X64 architecture or it will eventually have to exit the X86 market, because Intel won't stop until AMD is no longer in the market.

The words from the Anandtech article don't sound plucked out of the nearest a-hole. They're also not extremely specific. The article says that one more major revision of Bulldozer will launch in 2015 and that it will be the last, but that does not present a date-certain for the launch of the next architecture. It sounds like the earliest we could expect that would be 2016, which would be a full five years after the launch of the FX 8150. But it could stretch to 2017, too - AMD is good at stretching like that!

The big question is whether AMD's next architecture will save anything from the Bulldozer era. There has been a lot of APU work done. It seems strange to imagine "back-porting" all that APU work to the Deneb/Thuban architecture, but that's not to say it is impossible. What is interesting is that I always saw Bulldozer as an attempt to deliver hyperthreading on steroids, so if AMD ditches the module approach, will it implement some form of hyperthreading? From a marketing standpoint, it would be a big disadvantage not to have some kind of counter to Intel's hyperthreading.

Finally, it should go without saying, that's all just a bunch of guesswork and speculation. My main point is that I think the comments in Anandtech's article were not speculation, though.
flip-mode
Gerbil Khan
Silver subscriber
 
 
Posts: 9084
Joined: Thu May 08, 2003 12:42 pm
Location: Cincinnati, OH

Re: AMD Post-Excavator

Postposted on Fri Jun 06, 2014 8:52 am

Ok, I guess we should revive this thread, especially now that AMD has revealed that they ARE working on a new, big x86 core to succeed the Bulldozer lineage.

The post before this by Flip-Mode is pretty much as spot-on as it could ever get without the announcement from AMD last May 5. Quite interesting. While many had hoped and were half-expecting that AMD is secretly working on something, I don't recall anybody going out and saying it with more conviction than Flip.

So now that it's out in the open, what are your thoughts about this new x86 core? I have two, at this point:

1. AMD obviously took one step forward and two steps back in designing Bulldozer. I don't know what they were smoking back then either, but to me it's a sound architecture. Not the best, but hey, it can add, right? So, having said that, AMD now obviously needs to focus on single-thread performance and the way there is apparently not through achieving blazing clock speeds, especially with fabrication processes out of their control since 2008. GF's apparently holding the Bulldozer lineage back to this day and AMD would be foolish to hope that GF will get their act together just in time for their next x86 or ARM core. So, I would think this new x86 core will be a wide architecture similar to Intel's current big cores and will also have SMT. About time, too. Intel's had SMT since the Pentium 4 14 years ago.

2. Many have credited Jim Keller for K8 (sometimes, even K7) and now he's also being seen as the man behind AMD's new K12 ARM/x86 cores. Ok, two things:

First, Jim, prior to leaving AMD for the first time, worked at AMD from around 1999 to 2000, if I'm right. We can assume AMD took about 5 years to put K8 together and Jim obviously didn't stay that long at AMD. So PERHAPS he just defined the specifications for K8 or made a few changes to what AMD had so far when he came in, told the other engineers what to do, then hurried out the door. How can he have been the father of the K8 then? I've read somewhere that when Jim left, Fred Webber (who was brought in when AMD acquired Nexgen) took over. I reckon Fred has long since left AMD.

Second, as Flip-mode has said earlier, the specifications and architectural features of K12X and K12A must've been defined even before Bulldozer hit the streets, given how microprocessor design schedules work which AMD must surely be familiar with. You can't start your next CPU project only when your new CPU gets released. So obviously, this would imply that AMD's new CPU projects were started some time in 2010 or perhaps even earlier, way before Jim even re-joined AMD. AMD announced Jim's re-employment at AMD only last August 2012 (heck, even Mark Papermaster joined AMD only in October of 2011). So when Jim came in, the engineers were already working on something and Jim's contributions are probably in the area of trying to figure out how to actually pull off what they set out to do. So, it could be that Jim didn't really define the specifications or just added to them. It's the exact opposite of K8 where Jim probably came in during the project's beginning, drew the specs, then left. Either way, Jim deserves credit, but so do the other engineers who have been toiling away on this project, especially the engineers who defined the specs of K12 before Jim came in. They deserve credit the same way Jim got credit for K8.
The three pillars of my digital life: AMD FX-8350, Google Nexus 7 (Qualcomm Snapdragon S4 Pro), Intel Core i5-2450M
ronch
Gerbil Elite
 
Posts: 641
Joined: Mon Apr 06, 2009 7:55 am
Location: C:\Program Files\

Re: AMD Post-Excavator

Postposted on Fri Jun 06, 2014 11:13 am

I still would not get my hopes up for a triumphant return to the desktop.

The ARM-based K12 is clearly a server-oriented part, and they've stated that they plan to leverage the same socket/chipset infrastructure for their next-gen x86 CPU. That says to me that the x86 part is going to be aimed squarely at the server market as well. They probably figure the ARM chip will see significant uptake by cost-conscious enterprise users who run Linux in their datacenters, while the new x86 part will be aimed at users who are inextricably tied to MS operating systems for their server deployments.

I have not seen anything that leads me to believe that AMD is planning to make a serious attempt at winning back significant market share in the performance desktop/workstation segment (though the x86 server parts may see some use in special-purpose workstations that benefit from a large number of cores). They are focusing most of their attention on mobile and server, and implicitly ceding the performance desktop segment to Intel. IMO this is a sensible strategy; mobile and server are the growth segments, and I do not think that AMD has the manufacturing resources to compete head-to-head with Intel in the high-end desktop space.
(this space intentionally left blank)
just brew it!
Administrator
Gold subscriber
 
 
Posts: 37672
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Next

Return to Processors

Who is online

Users browsing this forum: No registered users and 2 guests

cron