Why have AMD APUs failed on the market?

[BSim500]

For $90 (A8-7600) what CPU + dGPU can you buy that will offer the same or better performance (both CPU and iGPU) ??

You often won't be paying "only" $90 though as many pre-built Kaveri's have a silly price premium in many areas (because they aren't selling very well and don't benefit from economies of scale).

Example - EBuyer UK:-

£190 Cheapest Pentium G3240, 4GB RAM, 500GB HDD, no ODD, Win 8.1
http://www.ebuyer.com/643320-zoostorm-desktop-pc-7260-8001

£350 Cheapest A8 7600, 8GB RAM, 1TB HDD, DVD Writer, Win 8.1
http://www.ebuyer.com/662321-zoostorm-desktop-pc-7260-2024

Component balancing:-
£26 4GB RAM
£12 DVD writer
£ 8 difference between 1TB vs 500GB HDD
£46 Sub-Total

£350 - £190 - £46 = £114 Kaveri premium (enough for a "free" 2GB VRAM 260X with 2.5x the A8's performance with change to spare and native 1080p capability at decent +40-60fps on Medium in many games). And that's ignoring that for many games you've posted (eg, BI), 4GB RAM + separate 1-2GB VRAM is entirely playable, but you still need 8GB with an iGPU no matter what... Others have posted similar examples from other countries. Simply shouting "what can you buy for $90" over & over doesn't change anything for most people, since:-

1. Most non-tech savvy people buy computers not CPU's and the nearest consistent dGPU-less price equivalent to cheap Pentium "pre-builts" in many countries are stuff like the A4 / A6's 4xxx/6xxx, not the A8 / A10's 7xxx.

2. Most tech-savvy people interested in building their own gaming rig that they want to last for the next 3-4 years will have figured out a dGPU is better long-term value (perf-per-$ / perf-per-watt) even if it isn't the lowest absolute price, ie, +$70 in Year 1 may also = -$70 in Year 3 due to not having to upgrade so often whereas APU iGPU's are outdated almost every 12-18 months. (Hint : Look at the number of people on Youtube with older A10-5800K CPU's who've ended up adding a dGPU anyway for 2014 games...) "Buy cheap, buy twice..."

3. $90 isn't everyone's "hard" budget limit. (Why this obsession with "exactly $90" over & over? Before the A8 was released, and prior to AMD's price-cuts you were happily promoting $180 A10-7850K's, so your own budget seems to be remarkably "flexible" rather than being any serious "budget"...)

4. It's 2015 and many people simply aren't interested in gaming PC's with sub-console performance for heavier games even to save $50-$70 (and for lighter flash games, it's irrelevant). If an A8-7600 is "good enough", then why didn't AMD stick one in the PS4? I think we both know the answer to that...

More on the point, IGPU computing will not catch on because it's usefulness is just as limited as that of many cores in general,people just don't care about luxmark or whatever fringe software that can utilize it.

^ This. The whole hype over HSA simply hasn't been matched in reality. Why? If you want software coders to rewrite everything for your 15% market share, then you have to put serious money into it (on the software support side). So APU's (on a market level) aren't actually that "cheap" to persuade everyone to buy at a premium vs a Pentium for the bulk of non-gaming usage and then demand everyone rewrite their whole software suites which doesn't need doing for the +85% of the other brand, and is irrelevant to most people who don't have giant +5,000 row complex spreadsheets or whose "photo editing" is mostly simple crop / resize / red-eye reduction / color adjustment on photo's taken on their mobile phone rather than perform complex blurs on multi-layered RAW images taken on their semi-pro "Four-Thirds" dSLR. In fact, AMD had more support for 3DNow! from major games devs than they've had with attempts at forcing APU "compute" / HSA specific code optimizations on a primarily disinterested public.

 

[TheELF]

Can you overclock it on a locked Haswell part?

Yes you can.
IGPU, RAM ,and even the CPU cache, are all unlocked on all and any haswell.
I used to have my g1820's IGPU at 1500Mhz (from 1050Mhz) with an additional 0.2V ,at all times.

 

[AtenRa]

Yes you can.
IGPU, RAM ,and even the CPU cache, are all unlocked on all and any haswell.
I used to have my g1820's IGPU at 1500Mhz (from 1050Mhz) with an additional 0.2V ,at all times.

Only Z- boards can OC memory and voltage. H81-B85 can only OC the iGPU. And believe me, even HD4600 at 1450MHz is not enough for many many games.

 

[Yuriman]

Only Z- boards can OC memory and voltage. H81-B85 can only OC the iGPU. And believe me, even HD4600 at 1450MHz is not enough for many many games.

Very true. A4/A6 iGPUs are not enough for many games either, though you can definitely play more of them. Minecraft, for instance, sits in that gap. A8/A10 iGPUs seem to play most games reasonably well if you turn off all of the eye candy and in some cases knock the resolution down.

384GCN cores is a reasonable approximation to a HD2900XT / 8800GT, right?

 

[ElFenix]

Kaveri uses some silly Application Power Management and adjust CPU and GPU clocks accordingly to workload and bottleneck. But I'm old school and would love to set my clocks to certain value and keep it there stable. More direct control on overclocking. I guess the new method caters more towards casual users thou.

If it worked well it'd dynamically change the load during usage, which is way more fine control than you can exercise directly.
What you want is akin to wanting a carburetor when efi is around.

 

[Enigmoid]

Hmm . . . okay. Can we at least compare GFlops between two different processors running the same application?

http://www.aida64.com/support/faq :

And . . .

http://forums.guru3d.com/showpost.php?p=4743967&postcount=28

Haswell coming up big on 32-bit fp thanks to AVX2 + HT . . . but it still falls short of 384 shaders. A locked i5 won't turn in that kind of performance, either.

Some boards let you disable APM. Mine is disabled, I think. It doesn't seem to make much of a difference really, once you start overclocking things.

http://forums.aida64.com/topic/1539-opencl-gpgpu-benchmarks/#entry8636

Hmm. Is that a driver problem or a hardware problem? Doesn't seem to be helping HD4600's bandwidth much.

Indeed. Hmm . . . has anyone done bandwidth-sensitivity comparisons between different GPGPU applications? How many can you think of are available for download that could be tested? I wouldn't mind taking a crack at it.

Here is the Aida64 benchmarks between the two.

i5 CPU + igp pumps out more 32 bit FLOPS than Kaveri 384 shader CPU + igp.

Kaveri only wins 24 bit integer IOPS.

HD 4600 has no problems hitting fairly close to peak efficiency.

 

[pablo87]

I think AMD invested way too much in APU's, it was the wrong vision.

They basically dumbed their technology down to Intel's level, focusing the battle on the CPU where Intel dominated. They should have played to their strength, GPU. In De Bono's terminology, they competed instead of Sur-Peting (book is Sur-Petition: Going beyond Competion).

In the mid 90's a chip that could do it all - video (2d then), audio, modem, etc.. - was the in thing. Nvidia got its start like that (ironic to AMD's strategic mistake, NVIDIA failed fast and focused on 3D graphics instead), even IBM had an offering (MPACT if I recall).

One of the reasons this fad came to naught is once the price of a desktop fell below $1,000, Intel really started to focus on integrating PC components like sound, vga, ethernet into their chipset in order to maintain their profit margins as PC prices came down - their POR must have been to get a bigger piece of a shrinking pie. One of the key by products of course is that it dumbed down those parts of the PC in favor of the microprocessor. Plus it helped fill Intel's older generation fully depreciated fabs. Brilliant. Absolutely brilliant.

Prior to the acquisition of ATI by AMD, ATI's founder and CEO KY Ho had begun investing in integrated graphics chipsets (they had an Intel license) but it wasn't a focus. I don't think KY believed in it then.

AMD on the other hand was probably feeling the heat from Intel's design wins (at that time, AMD was completely reliant on SIS, VIA, etc...for those integrated chipsets) and decided to acquire ATI.

AMD never thought of themselves as a GPU company, they thought of themselves as a CPU company that acquired GPU technology to compete with Intel. The key word here is compete. And over time, they never changed that vision.

But today its quite clear that at some point 4-5 years ago when they came up with Bobcat, they should have made that U-turn.

First, they never had the infrastructure, culture, credibility and reputation to compete in the Enterprise space. They sold processors there very well for a while in spite of this because they had superior product. Once Intel fixed their Netburst mistake, AMD's server market share quickly eroded. Even in client, they were always weak.

Second, dumbing down these other PC components served Intel well in Enterprise but it didn't do much ultimately for the Consumer except for price - and you know the old saying, there's always somebody cheaper than you. In this case, ARM and Android.

What would a U-turn have looked like for AMD? Well, it would have been an emphasis on GPU and not CPU. For AMD, CPU should have become the integrated aspect of the PC, the part that is almost given away, built on trailing fabs, costed down as much as possible not only in the actual part cost, but the R&D invested in it and of course, the rest of the BOM. Turn the table on Intel if you will.

By definition, it meant no shared memory well because that's a bottle neck to better GPU performance. It also means no APU at all because as we've seen with AMD's APU's, the GPU technology therein is always at least 1 generation behind.

What would the end product have looked like? Well, a motherboard with a BGA processor with USB, sound etc.. integrated, with the lowest possible BOM, with a slot for memory, some power circuitry, some sheet metal wrapped around it and of course, the PCI-Express slot. Sub $100 for everything except OS, storage and of course, GPU. Entry level probably 7770 2 years ago, probably 7850 today.

Same thing in Mobile. And technology wise would this separation preclude them from having hooks that essentially advantage AMD GPU's over NVidia's on their own platform? Probably not.

And no option for OEM's to buy anything else from AMD because the positioning would have been fundamental to who the company is.

Result? AMD would have owned the mainstream Consumer market. Of course "real" gamers would still be going for Intel processors as they are today. And the bottom of the market would still be the bottom of the market. But for not much more (in the case of the Chromebooks) and a lot less (in the case of the Gaming market), you could choose AMD.

And ironically, this approach would have probably yielded better results in the Enterprise markets that are not exactly Enterprise (like Education, SMB), with minimal investment, as a by-product if you will.

In fairness to AMD, I myself strongly believed in APU's back 4-5 years. I thought it would work. But the bottlenecks, the delayed cadence, the mixed message, turned out to be a failure.

 

[cbn]

My guess is that AMD wants low-end users to buy 1M + GCN chips and rely on HSA/OpenCL for acceleration often enough that the AMD chips can still compete with Pentiums and Celerons effectively. There are probably some opportunities out there for such a solution to work occasionally, but it'll be pretty inconsitant.

For that same cost of the 1M chip with large iGPU, I'll bet a person could also have a 2M chip with small iGPU (and we know the 2M chip with high enough clocks should do well enough against a locked Celeron/Pentium even without HSA/open cl acceleration). This, of course, assuming AMD changed its die layouts.

 

[DrMrLordX]

i5 CPU + igp pumps out more 32 bit FLOPS than Kaveri 384 shader CPU + igp.

Glad you finally found an HD4600 bench. I could not find one at a moment's notice. It's close . . . then you throw in the 7850k:

http://wccftech.com/kaveri-a107850k-tested-maxwell-dgpus-compute-gpgpu-battle-fares-remarkably/

Assuming the 7850k can also put out ~115-120 GFlops with its Steamroller cores (which it can, same cores as in the 7700k), the combined 32 fp for the 7850k beats the i5-4670. And that's stock! Overclocked:

http://cdn3.wccftech.com/wp-content/uploads/2014/01/A10-7850K-GPGPU-635x420.jpg

HD 4600 has no problems hitting fairly close to peak efficiency.

Good, I look forward to what Gen8 brings to the table. Broadwell Iris Pro may well be a monster in GPGPU.

For that same cost of the 1M chip with large iGPU, I'll bet a person could also have a 2M chip with small iGPU (and we know the 2M chip with high enough clocks should do well enough against a locked Celeron/Pentium even without HSA/open cl acceleration). This, of course, assuming AMD changed its die layouts.

AMD isn't going to do that, though, and they have signaled that quite clearly. Besides, at that price point, I'll restate: OEMs won't want it. They want E1/E2 Jaguars and crap.

 

[SPBHM]

384GCN cores is a reasonable approximation to a HD2900XT / 8800GT, right?

384GCN can be faster than those cards easily, in the form of the r7 250 GDDR5, but with 128bit DDR3 I can see it being slower for old games?
the GPU is not the main problem, the memory is.

 

[cbn]

For that same cost of the 1M chip with large iGPU, I'll bet a person could also have a 2M chip with small iGPU (and we know the 2M chip with high enough clocks should do well enough against a locked Celeron/Pentium even without HSA/open cl acceleration). This, of course, assuming AMD changed its die layouts.

AMD isn't going to do that, though, and they have signaled that quite clearly. Besides, at that price point, I'll restate: OEMs won't want it. They want E1/E2 Jaguars and crap.

If a steamroller quad core with reduced DDR3 PHY (GDDR5 removed, etc) and small iGPU were ~120mm2 on 28nm it could be in a low price price bracket.

In fact, 120mm2 on 28nm is just barely above the size of cat core SOC (~100mm2). The major difference would be that SR chip needs the 65nm southbridge while the cat core SOC obviously does not.

However, 20% extra die size + southbridge chip is small price to pay for such a massive increase in cpu performance over cat cores IMO.

Looking retrospectively, I have to wonder what would have happened if AMD followed a small iGPU approach compared to what they have done since Llano with the large iGPU approach? My guess is that volumes would have been higher and the selling price per mm2 silicon would have also been higher.

 

[cbn]

With that ~120mm2 steamroller quad core small iGPU mentioned, I still think a construction core hexcore with small iGPU released on either FM2+ or FM3 would be an even better idea.

One reason is that the hexcore is only slightly larger, but it also should not yield any dual cores (only hexcores and quad cores). Lack of dual core construction core SKU means less overlap with any potentially higher clocked cat quad core desktop chips.

As far as overlap goes with FM3, a hexcore SR chip is only a six thread processor while Zen is supposed to have SMT. Therefore 6C/6T does not overlap with 8C/16T to any major extent. Likewise a hexcore SR released late on FM2+ shouldn't enter into an overlap situation with Vishera either as that chip will likely be gone by then replaced by 8C/16T Zen.

 

[NTMBK]

With that ~120mm2 steamroller quad core small iGPU mentioned, I still think a construction core hexcore with small iGPU released on either FM2+ or FM3 would be an even better idea.

One reason is that the hexcore is only slightly larger, but it also should not yield any dual cores (only hexcores and quad cores). Lack of dual core construction core SKU means less overlap with any potentially higher clocked cat quad core desktop chips.

As far as overlap goes with FM3, a hexcore SR chip is only a six thread processor while Zen is supposed to have SMT. Therefore 6C/6T does not overlap with 8C/16T to any major extent. Likewise a hexcore SR released late on FM2+ shouldn't enter into an overlap situation with Vishera either as that chip will likely be gone by then replaced by 8C/16T Zen.

The increased market for such a processor vs. a 6 core AM3+ Piledriver chip or a 4 core Steamroller chip evidently wasn't significant enough to justify the R&D costs of producing the new chip. Just give up already, it's not going to happen!

 

[cbn]

The increased market for such a processor vs. a 6 core AM3+ Piledriver chip or a 4 core Steamroller chip evidently wasn't significant enough to justify the R&D costs of producing the new chip. Just give up already, it's not going to happen!

By the time AMD launched a hexcore construction core with small iGPU on FM2+ or FM3, Vishera would likely be done.

So comparing to 6 core AM3+ (or any Vishera) is not the idea. Please read my post again.

 

[NTMBK]

By the time AMD launched a hexcore construction core with small iGPU on FM2+ or FM3, Vishera would likely be done.

So comparing to 6 core AM3+ (or any Vishera) is not the idea. Please read my post again.

Why would Vishera be "done"? If we follow your logic and there is still a niche use case for CMT even after Zen comes out (unlikely, but let's go with it), they would keep churning out Vishera chips/selling off their existing stocks. It still wouldn't make sense for them to make a new 3 module Steamroller chip.

 

[cbn]

Why would Vishera be "done"? If we follow your logic and there is still a niche use case for CMT even after Zen comes out (unlikely, but let's go with it), they would keep churning out Vishera chips/selling off their existing stocks. It still wouldn't make sense for them to make a new 3 module Steamroller chip.

Having old stock of Vishera is one thing, but if rumors are true I can't imagine AMD making AM3+ processors and chipsets alongside octocore Zen. (re: The rumored Summit Ridge Zen on FM3 serves the same purpose as Vishera: A large octocore chip (with Zen also rumored to have SMT) with L3 cache and no iGPU.)

And as far as CMT existing alongside Zen, the rumored Bristol Ridge APU (Q3 2016) on FM3 is supposed to have four excavator cores and 512sp iGPU. If true, I am thinking there must be some benefit left for CMT even in late 2016. My guess is that construction cores while weaker than Zen in Multi-thread will still be stronger in Single thread.

Now regarding Vishera vs. FM2+/FM3 hexcore (with iGPU) here are my criticisms:

1. Vishera with its very large L3 cache on 32nm takes up too much silicon area to be economical. In contrast, a hexcore construction core chip (without L3 cache) + small iGPU on 28nm is going to be much smaller.

2. Vishera does not have iGPU for OEM Pre-built desktop. (Apparently there was an AM3+ chipset with integrated graphics but I do not see it on the pre-builts for whatever reason)

3. The AM3+ chipsets are beginning to show their age (lack of usb 3.0 etc.) and are also apparently expensive to produce.

4. 32nm for processors in 2016? Late 2016? I'm guessing AMD will have moved on to use 32nm for chipsets instead.

 

[mrmt]

Why would Vishera be "done"? If we follow your logic and there is still a niche use case for CMT even after Zen comes out (unlikely, but let's go with it), they would keep churning out Vishera chips/selling off their existing stocks. It still wouldn't make sense for them to make a new 3 module Steamroller chip.

I think vishera is *already* done, in fact, not only Vishera, but the entire CMT line. AMD today is not willing to stock up more CMT chips and instead are wrangling with GLF for not fulfill their take-or-pay commitments of the WSA. If AMD was seeing good market prospects on its CMT line it could just stock up more chips and be done with it, the fact that they don't tells everything one needs to know about their future viability on the market place.

To put things in perspective, AMD expects to sell its last Richland chip in Q415, 6 quarters after the initial Kaveri launch, and they are having to sell it below cost. I don't think Kaveri is that much better than Richland to not suffer the same fate a few quarters down the road. Vishera is in a better situation because it is now a low volume part, but if they stock up more of it they will have an even bigger write down, because of the die size.

 

[DrMrLordX]

reduced DDR3 PHY (GDDR5 removed, etc)

Neither you nor I (nor anyone else here, I suspect) really know why the DDR3 PHY on Kaveri is as large as it is. Does it need an expanded PHY for HSA compliance (partial or otherwise)?

I don't know when AMD will replace the dual-core cat chips with something better. My guess is 1M Carrizo will be a strong candidate; however, much to your chagrin, it will still have a "big" iGPU.

I think vishera is *already* done, in fact, not only Vishera, but the entire CMT line.

Vishera, Trinity, and Richland have their heads on the chopping block, provided they haven't already been whacked. Parts are still in the channel, and retailers have been trying to dump them like mad. Steamroller/Kaveri is still with us 'til mid 2016.

I don't know what to expect from the cat cores, since Carrizo seems to have rendered most of the cat family obsolete (we'll know more once we see some benches). Carrizo quads @ 12W TDP would obviate the need for any of the AM1 Kabinis, and it is not outside the realm of possibility that the cat dual cores could be replaced by 1M Carrizo.

As much as I like the cat cores, I don't see any future mention of their development on roadmaps or anything. They may be just as dead as the Piledriver chips and AM3+.

That would be a pretty significant advantage for AMD, to bring their large and confused line of processors down to two chips: Kaveri for socketed desktop, and Carrizo for BGA desktop/mobile. Of course that sort of ignores the tablet space but let's not talk about that, we'll start another contra revenue spat.

 

[mrmt]

Vishera, Trinity, and Richland have their heads on the chopping block, provided they haven't already been whacked. Parts are still in the channel, and retailers have been trying to dump them like mad. Steamroller/Kaveri is still with us 'til mid 2016.

AMD CPU business shrank 70% since 2011, and they forecast a 15% drop in revenue, focused on the CPU business. I expect that 15% drop in total revenue to reflect in 20-25% drop in their CPU business, and even more than that on the CMT chips.

I don't know what to expect from the cat cores, since Carrizo seems to have rendered most of the cat family obsolete (we'll know more once we see some benches). Carrizo quads @ 12W TDP would obviate the need for any of the AM1 Kabinis, and it is not outside the realm of possibility that the cat dual cores could be replaced by 1M Carrizo.

The cat cores are much more competitive in terms of costs than the CMT cores. I think AMD isn't extending their cat line up because Zen will replace it, not because they don't have plans for that market. Think about it, a small, cheap, synthesizable core is much easier to get semi-custom deals than a complex, high performance, high cost core.

 

[erunion]

A I think AMD isn't extending their cat line up because Zen will replace it

More and more the evidence points to Zen as cat core replacement. Even though a handful people still hold out hope that Zen is a bulldozer replacement, I'm confident that won't be the case.

 

[monstercameron]

More and more the evidence points to Zen as cat core replacement. Even though a handful people still hold out hope that Zen is a bulldozer replacement, I'm confident that won't be the case.

hmmm cat core replacement in that it is rumored to be a small core but AMD is also rumored to have just one x86 core for all their legacy markets. In a sense it is replacing both small core and big core designs.

 

[erunion]

hmmm cat core replacement in that it is rumored to be a small core but AMD is also rumored to have just one x86 core for all their legacy markets. In a sense it is replacing both small core and big core designs.

AMD has 2 cores today but only one in the works.

You said that the new core is replacing both, or you could say that one core is being replaced and the other isn't.

I guess what really matters is the products that they bring to market. We'll see.

 

[USER8000]

You often won't be paying "only" $90 though as many pre-built Kaveri's have a silly price premium in many areas (because they aren't selling very well and don't benefit from economies of scale).

Example - EBuyer UK:-

£190 Cheapest Pentium G3240, 4GB RAM, 500GB HDD, no ODD, Win 8.1
http://www.ebuyer.com/643320-zoostorm-desktop-pc-7260-8001

£350 Cheapest A8 7600, 8GB RAM, 1TB HDD, DVD Writer, Win 8.1
http://www.ebuyer.com/662321-zoostorm-desktop-pc-7260-2024

Component balancing:-
£26 4GB RAM
£12 DVD writer
£ 8 difference between 1TB vs 500GB HDD
£46 Sub-Total

£350 - £190 - £46 = £114 Kaveri premium (enough for a "free" 2GB VRAM 260X with 2.5x the A8's performance with change to spare and native 1080p capability at decent +40-60fps on Medium in many games). And that's ignoring that for many games you've posted (eg, BI), 4GB RAM + separate 1-2GB VRAM is entirely playable, but you still need 8GB with an iGPU no matter what... Others have posted similar examples from other countries. Simply shouting "what can you buy for $90" over & over doesn't change anything for most people, since:-

1. Most non-tech savvy people buy computers not CPU's and the nearest consistent dGPU-less price equivalent to cheap Pentium "pre-builts" in many countries are stuff like the A4 / A6's 4xxx/6xxx, not the A8 / A10's 7xxx.

2. Most tech-savvy people interested in building their own gaming rig that they want to last for the next 3-4 years will have figured out a dGPU is better long-term value (perf-per-$ / perf-per-watt) even if it isn't the lowest absolute price, ie, +$70 in Year 1 may also = -$70 in Year 3 due to not having to upgrade so often whereas APU iGPU's are outdated almost every 12-18 months. (Hint : Look at the number of people on Youtube with older A10-5800K CPU's who've ended up adding a dGPU anyway for 2014 games...) "Buy cheap, buy twice..."

3. $90 isn't everyone's "hard" budget limit. (Why this obsession with "exactly $90" over & over? Before the A8 was released, and prior to AMD's price-cuts you were happily promoting $180 A10-7850K's, so your own budget seems to be remarkably "flexible" rather than being any serious "budget"...)

4. It's 2015 and many people simply aren't interested in gaming PC's with sub-console performance for heavier games even to save $50-$70 (and for lighter flash games, it's irrelevant). If an A8-7600 is "good enough", then why didn't AMD stick one in the PS4? I think we both know the answer to that...



^ This. The whole hype over HSA simply hasn't been matched in reality. Why? If you want software coders to rewrite everything for your 15% market share, then you have to put serious money into it (on the software support side). So APU's (on a market level) aren't actually that "cheap" to persuade everyone to buy at a premium vs a Pentium for the bulk of non-gaming usage and then demand everyone rewrite their whole software suites which doesn't need doing for the +85% of the other brand, and is irrelevant to most people who don't have giant +5,000 row complex spreadsheets or whose "photo editing" is mostly simple crop / resize / red-eye reduction / color adjustment on photo's taken on their mobile phone rather than perform complex blurs on multi-layered RAW images taken on their semi-pro "Four-Thirds" dSLR. In fact, AMD had more support for 3DNow! from major games devs than they've had with attempts at forcing APU "compute" / HSA specific code optimizations on a primarily disinterested public.

Since I am uk based I liked your twisting of the ebuyer prices. The amd rig has win7 pro and the Intel one had win8 with bing.

But again you also ignored the cheaper a8 7600 rig with proper win8 for £320;http://www.ebuyer.com/662329-zoostorm-desktop-pc-7260-0039

It gets better if you bother to actually look at this model:http://www.ebuyer.com/662337-zoostorm-desktop-pc-7260-3017

It's the same model without is for £240. Add £70 for win8 that means it costs £310.

Here is an a10-5700 zoostorm for £235:
http://www.amazon.co.uk/gp/aw/d/B00...ostorm+a10&dpPl=1&dpID=41AQL5j3miL&ref=plSrch

Has win8 with bing.

Plus using your logic of going cheap why would you be getting a pentium dual core to last a few years anyway?

Plus for games like lol,dota and Minecraft I know plenty of people running them fine on the apu graphics. Remember most comparisons of the apus are to core i3 CPUs which have faster igps than Pentium's. Plus Intel gaming drivers are improving but are still a bit meh.

The thing is what is stupid is that none of the a8 7600 zoostorms use win8 with bing meaning the extra cost is the windows license.

 

[mrmt]

hmmm cat core replacement in that it is rumored to be a small core but AMD is also rumored to have just one x86 core for all their legacy markets. In a sense it is replacing both small core and big core designs.

That depends on how you look AMD legacy markets. Are we talking about markets they *say* they have presence, like servers or high end desktops, or are we talking about markets they have *actual* presence?

I don't doubt Zen will be a nice fit on the latter but not the former. The former would make Zen to have a broader scope than Broadwell, and given AMD R&D funding crunch of lately, I don't think it's remotely possible.

Since AMD accomplished what people thought it was impossible, reducing the gross profits of their CPU business a smaller size of their semi-custom gross profits, and that AMD is chasing 80MM/quarter businesses (very low cost/profit), what we have to ask is whether the consumer tail is going to wag the semi-custom dog.

 

[cbn]

I don't know when AMD will replace the dual-core cat chips with something better.

Its not the fact there are dual core cat chips that bothers me. It is just that there is no high speed dual core cat chip (eg, the fastest dual core cat chip that I know of is clocked 1.45 GHz).

So why are there no high clocked dual core cat chips? I suspect this is because there are too many low clocked quad core cat chips (Sempron 3850, A4-5000, etc). Why are there too many low clocked quad core cat chips? I suspect this is because clocking quad core cat chips higher (without a price increase) could cause overlap with the cheap dual core FM2 chips (like A4-6300, A4-7300, etc). So notice what we have been seeing is fast clocked quad core cat chips (like Athlon 5350), but they are priced at a premium over the FM2 dual cores.

Now granted the overlap I am speaking about is not a total overlap (re: the dual core FM2 chips have higher single thread and higher TDP), but I believe it is close enough.

My guess is 1M Carrizo will be a strong candidate; however, much to your chagrin, it will still have a "big" iGPU.

Carrizo won't work as a replacement for the current cat cores because the die size is too big. Yes, it will have the performance, but not the low cost to manufacture.