AMD Cancels 28nm APUs, Starts From Scratch At TSMC

MrSeb writes "According to multiple independent sources, AMD has canned its 28nm Brazos-based Krishna and Wichita designs that were meant to replace Ontario and Zacate in the second half of 2012. The company will likely announce a new set of 28nm APUs at its Financial Analyst Day in February — and the new chips will be manufactured by TSMC, rather than its long-time partner GlobalFoundries. The implications and financial repercussions could be enormous. Moving 28nm APUs from GloFo to TSMC means scrapping the existing designs and laying out new parts using gate-last rather than gate-first manufacturing. AMD may try to mitigate the damage by doing a straightforward 28nm die shrink of existing Ontario/Zacate products, but that's unlikely to fend off increasing competition from Intel and ARM in the mobile space."

Comments

Take your time, let software catch up.

[Kenja]

So far I have been totally unable to tax my current CPU past 40% utilization. I think we can take a break and let software catch up and older systems fall off the support map before the next generation of CPUs hit.

Re:Take your time, let software catch up.

[Dunbal]

Don't worry, the next OS version should do it...

Re:Take your time, let software catch up.

[CSMoran]

So far I have been totally unable to tax my current CPU past 40% utilization. I think we can take a break and let software catch up and older systems fall off the support map before the next generation of CPUs hit.

Just because your usage scenario is not CPU-bound does not mean everyone else's is.

Re:Take your time, let software catch up.

[marcosdumay]

The change in feature size won't just be usefull to get faster processors (altough servers could use some of them), it is also important to reduce the power footprint of the chips (that being AMD, it means both CPU and GPU will use less power) and to reduce the price of those chips.

Re:Take your time, let software catch up.

[Anonymous Coward]

I salute you, mythical IT-worker who manages to get an overclocked computer work-approved.

Re:Take your time, let software catch up.

[0123456]

I salute you, mythical IT-worker who manages to get an overclocked computer work-approved.

Who said it was approved? In a previous job a friend inherited a computer from someone who'd left and never understood why it would crash every few days and hit bugs that no-one else seemed to see until he looked in the BIOS and discovered the previous user had overclocked it.
 

Re:Take your time, let software catch up.

[gstoddart]

So far I have been totally unable to tax my current CPU past 40% utilization.

Well, DfrgNtfs.exe is using 25% of my quad-core, and I'm not doing much else. I've gone well into 70% more more at times if I'm actually doing something intensive.

I'm using 7GB out of 8GB of RAM, and if I had 16GB I could probably put a hell of a dent in it too.

I don't even consider what I'm doing to be much of a load, and in the past I've been on machines where something literally was CPU bound for as much as an hour and I needed to walk away.

I don't even find it tough to use up that much resources ... hell, I stopped using Mozilla because it would expand to well over 1GB of RAM overnight (with the same # of windows and tabs that used to fit in 300MB).

I think the software has already caught up ... especially if you're like me and open something and leave it open.

Re:Take your time, let software catch up.

[Bengie]

With multi-core CPUs, just because you can't reach 100% usage doesn't mean your not CPU limited.

Re:Take your time, let software catch up.

[Skarecrow77]

Exactly. Too bad I already posted in the thread and can't mod you up anymore.

Nobody pays much attention to single-core performance anymore, and I have no idea why. There are tons of programs that people use on a regular basis that are single-core limited.

  Intel has made only modest gains in performance-per-clock-cycle since the core 2 duo. AMD I'm pretty sure is actually going backwards if I am correctly remembering some of the bulldozer vs thurban reviews.

Looking at forthcoming offerings, AMD especially seems to be assuming that we're all constantly using our CPUs to run handbrake 24/7 or batch encode a couple hundred wavs to mp3 at a time, and thus would love 12 cores.

Re:Take your time, let software catch up.

[Joce640k]

Vista? Ack.

At least have the decency to install Windows 7.

Re:Take your time, let software catch up.

[PRMan]

Seriously, this.

In building computers for my wife and my brother, I just went with lower end I3 and Phenom X2(4) processors. Why? Because the effective performance difference between the two for the applications they are running is .001%. And the price difference between those and say, an I7 is 1000%.

But I made sure to get both systems SSD drives. Price difference? About 200% (500GB HDD $60 vs 128GB SSD $125). But the performance difference is about 700%.

Re:Take your time, let software catch up.

[jd]

Software isn't the bottleneck. Caches are *tiny* compared to the size of even single functions in modern programs, which means they get flooded repeatedly, which in turn means that you're pulling from main memory a lot more than you'd like. Multi-core CPUs aren't (as a rule) fully independent - they share caches and share I/O lines, which in turn means that the effective capacity is slashed as a function of the number of active cores. Cheaper ones even share(d) the FPU, which was stupid. The bottleneck problem is typically solved by increasing the size of the on-chip caches OR by adding an external cache between main memory and the CPU. After that, it depends on whether the bottleneck is caused by bus contention or by slow RAM. Bus contention would require memory to be banked with each bank on an independent local bus. Slow RAM would require either faster RAM or smarter (PIM) RAM. (Smart RAM is RAM that is capable of performing very common operations internally without requiring the CPU. It's unpopular with manufacturers because they like cheap interchangeable parts and smart RAM is neither cheap nor interchangeable.)

Really, the entire notion of a CPU - or indeed a GPU - is getting tiresome. I liked the Transputer way of doing things (System-on-a-Chip architecture) and I still like that way of doing things. The Transputer had some excellent ideas - it's a shame it took Inmos so long to design an FPU (and a crappy one at that) and given that the T400 had a 20MHz bus at a time most CPUs were running at 4MHz, it's a damn shame they failed to keep that lead through to the T9000.

What I'd like to see is a SoC where instead of discrete cores (uck!) you have banks of independent registers, pools of compute elements and hyperthreading such that the software can dynamically configure how to divide up the resources. There's nothing to stop you moving all the GPU logic you like into such a system. It's merely more pools of compute elements. Microcode is already in use and microcode is nothing more than software binding of compute elements to form instructions. (Hell, microcode was already common on some architectures back in the 80s and was available for microprocessors within a decade of their being invented.) There's nothing that says microcode HAS to be closed firmware from the manufacturer - let the OS do the linking. It's the OS' job to partition resources and it can do so on-the-fly as needs dictate - something a manufacturer firmware blob can't do. Put the first 4 gigs onto the SoC and have one MMU per core plus one spare, so that each core can independently access memory (provided they don't try to access the same page). The spare is for direct access to memory from the main bus without going through any CPU (required for RDMA, which most peripherals should be capable of these days).

Such a design, where the OS converts the true primitives into the primitives (ie: instruction set) useful for the tasks being performed, would allow you to add in any number of other true primitives. Since any microcode-driven CPU is essentially a software processor anyway, you can afford to put extra compute elements out there. Any element not needed would not be routed to. Real-estate isn't nearly as expensive as is claimed, as evidenced by the number of artistic designs chip manufacturers etch in. Those designs are dead space that can magically be afforded, but there's nothing to stop you from replacing them with the necessary inter-primitive buffering to build ever-more complex instructions from primitives without loss of performance. I'm willing to bet HPC would look a whole lot more impressive if BLAS and LAPACK functions were specifically in hardware rather than being hacked via a GPU.

Of course, SoC means larger chips. So? Intel was talking about wafer-scale processors several years back (remember their 80-core boast?) and production has only improved since then. The yield is high enough quality that this is practical and since the idea is to software-wire the internals it becomes trivial to bypass defects. T

Re:Take your time, let software catch up.

[Anonymous Coward]

Nobody pays much attention to single-core performance anymore, and I have no idea why. There are tons of programs that people use on a regular basis that are single-core limited.

There's a very simple reason: physical limitations. The current processor technology is more or less maxed out for single-thread performance. There's probably some gains available by completely changing the instruction set or completely giving up on multi-thread performance, but nothing that Intel can put into a chip they can sell. They can't up clock speed anymore due to the speed of light (except a little bit when doing a die shrink). The obsession with multi-core isn't because Intel and AMD think everyone wants to run more threads; software is moving towards using more threads because Intel and AMD simply can't improve single-thread performance but they, at least for a little while longer, can keep adding more cores.

Re:Take your time, let software catch up.

[Kjella]

Looking at forthcoming offerings, AMD especially seems to be assuming that we're all constantly using our CPUs to run handbrake 24/7 or batch encode a couple hundred wavs to mp3 at a time, and thus would love 12 cores.

I think it's quite obvious that AMD didn't have the resources to hit many targets, so they picked two:

1) Laptops/Low-end PCs with Bobcat cores (Fusion/Llano APUs)
2) Servers with Bulldozer cores (Valencia/Interlagos)

Sadly the latter seems to have misfired a bit even in the server arena, but it's no question IMHO that the high-end desktop market was intentionally abandoned. Either that or they've missed their design targets by many miles, they can't have been that off on single core performance. I can sort of understand, Intel was already dominating and the Atom threatened their low end (remember, CPU designs have a 2-3 years lead time) and they couldn't afford to lose their bread and butter machines. So they aimed Bobcat low (power), Bulldozer wide (cores) and left Intel to compete with themselves. Not to be too much of a cynic, but it's better for AMD to win some markets than being a loser in all of them.

Re:Take your time, let software catch up.

[hkultala]

Software isn't the bottleneck. Caches are *tiny* compared to the size of even single functions in modern programs, which means they get flooded repeatedly, which in turn means that you're pulling from main memory a lot more than you'd like.

Wrong.

The code size of average function is much smaller than instruction cache for any modern processor.
And then there are L2 and L3 caches.

Instruction fetch needing to go to main memory is quite rare.

And then about data.. depends totally on what the program does.

Multi-core CPUs aren't (as a rule) fully independent - they share caches and share I/O lines, which in turn means that the effective capacity is slashed as a function of the number of active cores. Cheaper ones even share(d) the FPU, which was stupid.

None one of the CPU's sharing FPU with multiple HW threads are cheap.

Sun Niagara I had slow shared FPU, but the chip was not cheap

AMD Bulldozer, which usually has sucky performance, sucks less on code which uses the shared FPU.

FPU operations just have long latencies and there are always lots of data dependencies, so in practice you cannot
utilize FPU well from one threads, you need to feed instructions from multiple treads.

Intel uses HyperThreading for this, AMD Bulldozer it's CMT/shared FPU/module.
GPU's are barrel processors for the same reason.

The bottleneck problem is typically solved by increasing the size of the on-chip caches OR by adding an external cache between main memory and the CPU.

Much more often the bottleneck is between the levels of the chip's caches.
The big outer level caches are slow and processors spend quite often small time waiting for data coming from them. And if you increase the size of the last level caches, you make them even slower.

One of the reason's for bulldozer's sucky performance is because it has small L1 caches(so it needs to fetch data deom L2 cache often), but big and slow L2 cache. So there is this relatively long L2 latency happening quite often.

External cache.. has not been been used for about 10 years by Intel or AMD. It's either slow or expensive, and usually both. Now when even internal caches can easily be made with sizes over 10 megabytes, the external cache has to be very expensive in order to compete with internal caches, and still it only makes sense on some server workloads.

After that, it depends on whether the bottleneck is caused by bus contention or by slow RAM. Bus contention would require memory to be banked with each bank on an independent local bus. Slow RAM would require either faster RAM or smarter (PIM) RAM. (Smart RAM is RAM that is capable of performing very common operations internally without requiring the CPU. It's unpopular with manufacturers because they like cheap interchangeable parts and smart RAM is neither cheap nor interchangeable.)

Smart RAM is a dream, and a research topic in universities. It's uncommon because it does not (yet) exist.

And most of the problems/algorithms are not solveable by "simple" smart ram that can only operation on data near each others. And it you try to make it even smarter, then you end up making it costlier and slower, it will become just chip with multicore processor and memory on same chip.

There are some computational tasks where smart ram would improve the performance by great magnitude, but for the >90% of all the other problems, it has quite little use.

Re:Take your time, let software catch up.

[bill_mcgonigle]

So far I have been totally unable to tax my current CPU past 40% utilization.

Oh, you should try Firefox sometime!

Re:Take your time, let software catch up.

[hairyfeet]

No but he IS touching on something that we retailers could have told you is one of the biggest reasons for the slowdown in PC sales, and that is for the vast majority PCs are good enough for the jobs they have.

Look at one of the big sellers around here which is backed up by AMD having trouble filling all the orders...brazos. is brazos gonna compete with some Ivy bridge desktop replacement? not a chance in hell. Then why is it selling like crazy? For the same reason i sold my laptop and bought a Brazos EEE PC, and that is the jobs people have on the go aren't that computationally heavy and therefor the battery life and price make a bigger difference. in my own case i'm not transcoding video on the road, i'm accessing my webmail, watching HD movies, listening to music, maybe some light gaming. What in that list needs a monster PC?

I've found with my years of working PC retail that I'd be considered a "hardcore user" compared to most since i have a Deneb quad at home and actually DO play shooters and transcode as well as multitrack audio editing AT HOME but most of my customers, what do THEY do with a PC? they go to Facebook, play Farmville, check their webmail, watch YouTube, maybe do a little MS Word editing or play some game they got off the Walmart "300 games for Windows" rack. Now what there needs a giant CPU? Not a damned thing, in fact even the Brazos chip while running a full Windows 7 HP spends most of its time idle. hell i found playing full HD videos the CPU was barely hitting 15% with the GPU roughly the same depending on the action. Having the decoding in silicon drops the hell out of power usage.

So while the guys that run gamer sites or live for benchmarks will scoff frankly the average user, which outnumbers them by a 100,000 to one (last number on hardcore PC gamers I saw put the number at 30 million) and they won't give a crap that Brazos is 'long in the tooth" or that Thuban isn't king of the hill because "Will you look at that price? And look at how nicely videos play, woo hoo!"

This is why I really wasn't surprised when I walked into my local Walmart, a place that just a couple of years ago you were lucky to find a single Sempron in the back, to find that more than 2/3rds of the units had bright red AMD Fusion stickers. Hell I paid $350 for A Brazos EEE that gets 6 hours watching HD video, plays L4D or TF2, has 320Gb HDD to hold my music and movies, and that is INCLUDING an 8Gb RAM upgrade and a nice carrying case to put it in. Hell if AMD can keep prices THAT low nobody but the niche hardcore users will give a shit.

I know I can't keep the AMD desktops and netbooks in simply because the price is so much lower. For the jobs the average Joe has the AMD platforms are more than "good enough" and even someone like me who thought I'd always lug a 20 pound desktop replacement has found that I don't frankly miss it. 3 pounds, 6 hours 720P HD, light gaming and all for $350? Sold AMD, thanks for taking my money.

Re:Take your time, let software catch up.

[Kjella]

So while the guys that run gamer sites or live for benchmarks will scoff frankly the average user, which outnumbers them by a 100,000 to one (last number on hardcore PC gamers I saw put the number at 30 million)

Okay I heard Earth has an overpopulation problem, but did I doze off there for a while? Because I seem to have missed some recent developments...

Competition ?

[unity100]

AMD has no competition in APU arena. It is dominating it.

http://techreport.com/articles.x/21730/8

its actually possible to game with acceptable detail and fps with entry-mid level laptops without paying a fortune now.

Global Foundries

[Anonymous Coward]

The description is somewhat misleading in that Global Foundries is not a "long-time partner," but what were AMD's own internal wafer fabs until Global Foundries was spun out as a separate company in 2009.

Re:AMD = Stagnated.

[dc29A]

I hope you like $500 celerons...

If this was 1995, I'd believe it. In 2011, Intel competes with itself. If they drive up CPU prices, they won't be able to make more and more profits because people do *NOT* need to upgrade. The vast majority of the population is doing fine on a dual core 4+ year old CPU running a browser and IM program and watching videos. Since people do not need to upgrade, but Intel has to sell more and more CPUs, their profits would collapse and then the stock and then ... hilarity ensues.

Long-time partner? Really?

[WilliamBaughman]

Calling Global Foundries AMD's "long-time partner" really dates "MrSeb", he must have started reporting tech news in the last three years. Global Foundries isn't just a "partner" to AMD, it's part-owned by AMD, and was spun out of AMD's manufacturing and merged with Chartered Semiconductor.

Re:AMD = Stagnated.

[GreatBunzinni]

Your assumption that you can simply ignore AMD's influence in the CPU market and still end up with a relevant model to explain and predict its outcome is both naive and disingenuous. AMD does have products which outperform equivalent Intel products, even when not accounting with Intel shenanigans such as relying on funny compiler tricks, and AMD happens to price them quite attractively. If you haven't considered any AMD offering on any budget for any serious desktop and instead opted to rely only on Intel products then you are both clueless and economically-challenged.

Re:Long-time partner? Really?

[confused one]

All true; but, they're down to 9% ownership and according to the articles no longer have rights to appoint someone to the GloFlo board. Looks like the relationship is becoming increasingly sour.

Re:AMD = Stagnated.

[Guppy]

In 2011, Intel competes with itself.

That's part of the problem. One of the speculated reasons the Atom processor is so far behind, is that Intel was afraid it would cannibalize more profitable segments of its mobile CPU market. As a result, they launched it with a bunch of contractual restrictions on it (customers had to agree not to use it in any notebook larger than 10"-form factor), while using pricing models that discouraged 3rd party graphics (Atoms bundled with Intel's chipset were sometimes actually cheaper than solo Atoms, making nVidia ION combos uneconomical).

Since AMD had no strong CPUs in the netbook segment, everyone had to simply accept these restrictions at first, until AMD introduced their Ontaria and Zacate series.

Re:AMD = Stagnated.

[GreatBunzinni]

Intel i5 661: http://www.newegg.com/Product/Product.aspx?Item=N82E16819115217&Tpk=i5%20661
According to these benchmarks, we have:

• AMD Phenom II X4 965 4,291 $129.99*
• Intel Core i5 661 @ 3.33GHz 3,286 $175.66*

And this doesn't account for the money spent on a motherboard, which adds a hefty price to any intel offering.

So, looks like you botched your careful number check.

Re:AMD = Stagnated.

[mrchaotica]

I got an AMD Phenom II X4 840 for $59.99 a few days ago (at Microcenter); I'm sure it's more than half as fast as a 965, so it's an even better value. I got a new motherboard (AMD 760G chipset) with it too; it was also $59.99. Not bad, I think -- would I have been able to find an Intel solution for that price/performance?