Why have AMD APUs failed on the market?

[cbn]

They have an Excavator part coming in 2016 for desktop: Bristol Ridge. If they really want to dick around with 128, 256, or even 384 shader variants of Bristol Ridge, they've given no indication of that. Expanding the Excavator lineup shortly before the Zen launch just seems like a poor idea all around.

I don't think quad core is going to be optimal for gaming in late 2016.

Even today there are many games the older Vishera hexcore beats the newer steamroller quad core:

http://forums.anandtech.com/showpost.php?p=37167707&postcount=25
http://forums.anandtech.com/showpost.php?p=37167710&postcount=26)

And as far as Summit Ridge for FM3 goes, I am skeptical about its single thread performance.

 

[cbn]

It can be about raising clocks if you're going to keep the bin counts the same. But that is irrelevant. It is important to see what the market will and won't tolerate. I've done my best to show you what OEMs are interested in selling today. 25W Kabini quads in BGA form or AM1 socket - either one - would have been just dandy for AiOs. We know the die sizes are tiny, we know they run cool . . . OEMs just like to be dicks sometimes.

It's almost painful watching people go through gymnastics to try and figure out how AMD can tune their APUs to fit lower price points, when it's already obvious that the OEMs would just gobble up the savings and stiff consumers.

Or they could try to sell them to DIY buyers, but they'd probably wind up in the same position as AM1 Kabini: somewhat respected some of the time, but generally unloved.

My point was simply these cheap cat core SOCs have more potential left in them for desktop. AMD just needs to make room for the clock increase by getting rid of dual core FM2/FM2+/FM3 chips that potentially overlap. This means getting rid of Bristol Ridge as the quad core die will yield harvested dual cores whereas a FM2+/FM3 hexcore will not.

Certainly I don't believe cooling would hold back any higher TDP/faster clocked cat core in AIOs. In the past I have seen lots of cheap laptops rated for "35 watts", so I don't think that type of cooling is cost prohibitive for the larger AIO chassis.

 

[DrMrLordX]

If AMD can make a 95 watt hexcore on 32nm, they can surely make a 95 watt hexcore on 28nm.

Furthermore, AMD is already making quad core steamroller with 512sp iGPUs at 65 watts. Why would a hexcore Steamroller with small iGPU be tough at 95 watts? Answer: It isn't

Tell you what, you go argue with the source while I grab some popcorn and a lawnchair.

AMD just needs to make room for the clock increase by getting rid of dual core FM2/FM2+/FM3 chips that potentially overlap. This means getting rid of Bristol Ridge as the quad core die will yield harvested dual cores whereas a FM2+/FM3 hexcore will not.

AMD just put Bristol Ridge on their roadmap. They have made many project cancellations and other sacrifices just to keep what few products they have in the pipe, in the pipe. And now you want them to axe Bristol Ridge and replace it with cat cores that may not even have a future in AMD's chip R&D and a yet-to-be-designed Excavator hexcore that would come out at about the same time as Zen?

I suppose you want them to axe Kaveri refresh as well, since that's where all the Steamroller 1M chips come from, too (as die harvested chips).

That would be a criminally-insane waste of time, money, and effort.

Certainly I don't believe cooling would hold back any higher TDP/faster clocked cat core in AIOs. In the past I have seen lots of cheap laptops rated for "35 watts", so I don't think that type of cooling is cost prohibitive for the larger AIO chassis.

Obviously, the OEMs do not agree with you. Otherwise, you would be able to get a Kabini quad at a reasonable price and a decent clockspeed in an AiO right now, and basically, you can not.

 

[Erenhardt]

That iGPU is smaller than the one on the A10-7850K (384 sp vs 512 sp), and the cpu cores are locked and clocked lower than A10-7850K or Athlon x4 860K. It is also comes with a smaller, lower TDP heatsink.

Also, another thing to consider (with any APU value assessment) is the higher price of fast RAM for iGPU gaming:

-- 2 x4GB DDR3 2133 is $10 more than 2 x 4GB DDR3 1600.

-- 2 x4GB DDR3 2400 is $13 more than 2 x 4GB DDR3 1600.

Not only the smaller igpu is performing almost the same as 7850k, but the locked cpu is not a problem for the grand majority of users. How many people overclock their PCs?

The truth is, you can put 7850k on the price/perf board to show how apu sux, or you can go with budget option for budget comparisons. Whatever fits your bill.

 

[mrmt]

And as far as Summit Ridge for FM3 goes, I am skeptical about its single thread performance.

If AMD is really bringing Excavator on the market for 2016, what does it tell you about the performance of Zen parts? I really don't think AMD will bring excavator at all to the desktop, especially because even if Zen cannot decisively beat Excavator performance wise, AMD should not have any business on the consumer market anymore.

 

[Insert_Nickname]

I was under the impression that the 6800k, at least for "enthusiast" use, was a bit slower than Kaveri in most things. Hmm, tell you what. Do you OC your 6800K? I would like to know how it compares to my 7700k.

Here you go:

http://www.anandtech.com/show/7677/amd-kaveri-review-a8-7600-a10-7850k/10
http://www.anandtech.com/show/7677/amd-kaveri-review-a8-7600-a10-7850k/11

I'd expect the 7700K to be between the 7600 and 7850K, judging by specs.

As you can see, its basically a toss up between the stock 6800K and 7850K. Which one is faster depends on the test. The 6800K does have a significant frequency advantage, so a Kaveri clocked similar will be faster.

 

[cbn]

If AMD can make a 95 watt hexcore on 32nm, they can surely make a 95 watt hexcore on 28nm.

Furthermore, AMD is already making quad core steamroller with 512sp iGPUs at 65 watts. Why would a hexcore Steamroller with small iGPU be tough at 95 watts? Answer: It isn't

Tell you what, you go argue with the source while I grab some popcorn and a lawnchair.

TDP for a six core Steamroller with same clocks as 7850K would be 125W MINIMUM.

The clocks on a hexcore Steamroller don't need to be as high as the clocks on a 7850K, therefore 95 watts is easily possible at a lower frequency.

 

[cbn]

Certainly I don't believe cooling would hold back any higher TDP/faster clocked cat core in AIOs. In the past I have seen lots of cheap laptops rated for "35 watts", so I don't think that type of cooling is cost prohibitive for the larger AIO chassis.

Obviously, the OEMs do not agree with you. Otherwise, you would be able to get a Kabini quad at a reasonable price and a decent clockspeed in an AiO right now, and basically, you can not.

There is no such thing as a higher TDP, faster clocked cat core SOC at the moment....so the OEMs couldn't put such a chip in a AIO even if they wanted to.

And just because the Intel chips (J1800, J1900, 2955U, N2830) in these new $300 to $400 AIOs have low TDPs doesn't mean cooling for a higher TDP would be difficult. (In fact, I have already noticed 19.5" 1600 x 900 AIOs that have 35 watt Intel processors)

And that is just our discussion about AIOs. Of course, going beyond that there are also the regular desktops that use inexpensive SOCs like J1800/J1900, etc. In these cases it is much the same. The J1800 or J1900 is chosen not because cooling is the issue, but rather the OEM wants to make a cheap desktop.

 

[Shivansps]

Actually, they could put in a 2650 instead of a J1800, and a 3850 instead of a J1900 for the same or less money, but it will have worse cpu perf, and they need active cooling.

There is a reason for AM1 being active cooling, its not like AMD wanted to waste money on a fan, even the 15W soldered Kabinis has both a bigger heatsink and a fan.

Look at the "9W E1-2100" http://www.ecs.com.tw/ECSWebSite/Pr...oryID=1&DetailName=Feature&MenuID=106&LanID=0

then compare it to the Intel ones http://www.ecs.com.tw/ECSWebSite/Pr...yID=1&TypeID=46&MenuID=17&childid=M_7&LanID=0

 

[cbn]

If AMD is really bringing Excavator on the market for 2016, what does it tell you about the performance of Zen parts? I really don't think AMD will bring excavator at all to the desktop, especially because even if Zen cannot decisively beat Excavator performance wise, AMD should not have any business on the consumer market anymore.

Regarding Excavator on desktop, High density libraries is something I am concerned about.

Some info here:

http://www.anandtech.com/show/6201/amd-details-its-3rd-gen-steamroller-architecture/2

The power savings comes from not having to route clocks and signals as far, while the area savings are a result of the computer automated transistor placement/routing and higher density gate/logic libraries.

The tradeoff is peak frequency. These heavily automated designs won’t be able to clock as high as the older hand drawn designs. AMD believes the sacrifice is worth it however because in power constrained environments (e.g. a notebook) you won’t hit max frequency regardless, and you’ll instead see a 15 - 30% energy reduction per operation. AMD equates this with the power savings you’d get from a full process node improvement.

We won’t see these new libraries and automated designs in Steamroller, but rather its successor in 2014: Excavator.

The area and power savings sounds nice for space and thermal constrained scenarios like notebook, but what about performance on desktop?

Is Excavator on desktop (Bristol Ridge) going to have enough IPC to counteract the reduction in frequency? If not, its going to be slower than Steamroller.

Or maybe AMD is more concerned with CPU cores downclocking under an APU's iGPU load? (And AMD sees the maximum single thread reduction as a small price to pay for better efficiency/single thread under desktop iGPU load)

In any event, I am thinking Bristol Ridge might be potentially better as a desktop APU compared to Kaveri desktop APU (on the cpu side), but potentially worse as an Athlon x4 chip compared to Kaveri Athlon x4 860K.

 

[cbn]

I suppose you want them to axe Kaveri refresh as well, since that's where all the Steamroller 1M chips come from, too (as die harvested chips)

AMD really needed (needs) a better update for FM2+ than Kaveri refresh. According to that link you provided me back in post #678:

All of the current APU designs support only up to 4 (Trinity / Richland) or 6 (Kaveri) cores."

So yes, I would have loved (and would still love) to see a 6 core Kaveri APU design fielded, but with the large iGPU cut down for obvious reasons.

And this especially if cat cores end up on 20nm sometime in 2015.

1M Kaveri vs. 20nm cat core quad <----That is potential for even more performance overlap, which is a shame because I would like to see AMD push the cat core silicon harder on the desktop, instead of holding back. (mobile cat core is obviously is a different story as those chips need to have lower TDPs)

P.S. Regarding 3M SR APU overlap with AM3+, some would exist but AM3+ lifespan is coming to an end. AM3+ also a low volume part (due to expense of the big L3 cache and less integrated chipset, not to mention lack of modern features), so I don't know how much impact there would be anyway? With this mentioned, we also have to consider that a 3M SR APU would overlap less than a Kaveri refresh with existing 2M Trinity/Richland/Kaveri APUs already in the pipeline. Overall I am thinking having that 3M SR APU (rather than Kaveri refresh) would benefit reducing overlap more than it increases it.

 

[DrMrLordX]

As you can see, its basically a toss up between the stock 6800K and 7850K. Which one is faster depends on the test. The 6800K does have a significant frequency advantage, so a Kaveri clocked similar will be faster.

Boo! I'm not asking for Anandtech's old review. I'm asking about your 6800k. You brought it up, so let's see how fast it goes. My 7700k is way out of spec and already faster than a stock 7850k.

The clocks on a hexcore Steamroller don't need to be as high as the clocks on a 7850K, therefore 95 watts is easily possible at a lower frequency.

. . . and I'm sure people would just love a 3.1 ghz Steamroller hexcore that turbos to 3.4-3.8 ghz rather inconsistently. Oh, wait, no they wouldn't. 'cuz a 6300 would wind up being faster much of the time, not to mention the 8310/8320E and all that. Le sigh.

There is no such thing as a higher TDP, faster clocked cat core SOC at the moment....so the OEMs couldn't put such a chip in a AIO even if they wanted to.

Uh, in comparison to what? An E1-2500? There's the 5350, which is not present in BGA or AM1 form in any AiO at a big box store that I can find. What I did find was the E2-3800.

And just because the Intel chips (J1800, J1900, 2955U, N2830) in these new $300 to $400 AIOs have low TDPs doesn't mean cooling for a higher TDP would be difficult. (In fact, I have already noticed 19.5" 1600 x 900 AIOs that have 35 watt Intel processors)

Okay, are any of those 35W Intel CPUs in AiOs at the $400 price point?

And that is just our discussion about AIOs. Of course, going beyond that there are also the regular desktops that use inexpensive SOCs like J1800/J1900, etc. In these cases it is much the same. The J1800 or J1900 is chosen not because cooling is the issue, but rather the OEM wants to make a cheap desktop.

OEMs want to make cheap everything.

Regarding Excavator on desktop, High density libraries is something I am concerned about.

Is Excavator on desktop (Bristol Ridge) going to have enough IPC to counteract the reduction in frequency? If not, its going to be slower than Steamroller.

They don't have to use HDL if they don't want to. Bristol Ridge is a desktop part that is (apparently) distinct from Carrizo. That is, it may not have the FCH on-die, though honestly we have no way of knowing whether or not that is the case. Also, keep in mind that, once Bristol Ridge comes out, the 28nm planar process should be more mature, as should the Excavator design. The engineers on that project will probably squeeze out some more performance here and there.

So yes, I would have loved (and would still love) to see a 6 core Kaveri APU design fielded, but with the large iGPU cut down for obvious reasons.

Well, sorry, we won't get one, and I doubt you would have liked the clockspeeds on it anyway.

And this especially if cat cores end up on 20nm sometime in 2015.

What indicator is there that the cat cores will wind up on 20nm? Does GF have a 20nm process available for even low-power CPUs right now? AMD hasn't even moved their GPUs to 20nm yet. Hell, the only future cat core that shows up on their roadmaps now is Carrizo-L(Puma, not Puma+ as was initially hoped). All signs point to a stagnation in the cat core lineup.

 

[mrmt]

The area and power savings sounds nice for space and thermal constrained scenarios like notebook, but what about performance on desktop?

When the Bulldozer architecture was being showcased by AMD engineers there was a slide that said "higher throughput without significant loss on single threaded performance", which was what Bulldozer was really about.

Today what I'm reading in this message is "HDL designs are not suitable for vanilla, high TDP desktop, only low TDP parts will be viable once it is implemented", and that means no desktop Excavator. And given the low volumes AMD is selling today, I don't expect two Carrizo dies on two different processes.

 

[DrMrLordX]

Today what I'm reading in this message is "HDL designs are not suitable for vanilla, high TDP desktop, only low TDP parts will be viable once it is implemented", and that means no desktop Excavator. And given the low volumes AMD is selling today, I don't expect two Carrizo dies on two different processes.

AMD's roadmaps clearly show an Excavator quad in Q3 2016 named Bristol Ridge with 512 shaders. It probably isn't Carrizo at all, but an Excavator APU with no on-die FCH on 28nm planar (no HDL).

It's just a theory, but I'm guessing a few months ago when AMD was promising us that Carrizo would work on FM2+, some engineers discovered to their chagrin that the HDL Carrizo designs had major clockspeed problems and that they would not be much faster than desktop Kaveri in x86 workloads. Without compute/graphics focus, desktop Carrizo would be pretty bad, or at best a sidegrade. So they canned Carrizo on FM2+ and began work on an Excavator design with 512 shaders using no HDL libraries and no on-die FCH (which they would have to remove to get the die size down). The 2016 Q3 launch date is about how long it would take to get such a product to market given a start date of maybe last October/November, especially given that most of their R&D money is being sucked up by Zen.

 

[cbn]

. . . and I'm sure people would just love a 3.1 ghz Steamroller hexcore that turbos to 3.4-3.8 ghz rather inconsistently. Oh, wait, no they wouldn't. 'cuz a 6300 would wind up being faster much of the time, not to mention the 8310/8320E and all that. Le sigh.

1.) I'm sure a hexcore SR could have a higher frequency than that considering AMD is able to clock the 65 watt quad core A10-7800 at 3.5 base/3.9 turbo.

But even at that lower clock, It would still do a lot better than a 95 watt Kaveri quad core in games that scale to hexcore:

http://forums.anandtech.com/showpost.php?p=37167707&postcount=25
http://forums.anandtech.com/showpost.php?p=37167710&postcount=26)

And, of course, it would do better than a 65 watt A8-7600 which is also clocked low, but only has four cores.

2.) Hexcore SR could always overclocked to 4.5 Ghz+ with the proper board.

3.) FX-6300/FX-8310/FX-8320E, etc won't drop into a FM2+ board

Uh, in comparison to what? An E1-2500? There's the 5350, which is not present in BGA or AM1 form in any AiO at a big box store that I can find. What I did find was the E2-3800.

Unfortunately Athlon 5350 is about the price of a 35 watt Pentium G3250T (http://ark.intel.com/products/83539/Intel-Pentium-Processor-G3250T-3M-Cache-2_80-GHz ) and slower in multi-thread according to passmark:

Pentium G3250T : 2772 cpu marks ---> http://www.cpubenchmark.net/cpu.php?cpu=Intel+Pentium+G3250T+@+2.80GHz

Athlon 5350: 2604 cpu marks http://www.cpubenchmark.net/cpu.php?cpu=AMD+Athlon+5350+APU+with+Radeon+R3

So yeah, no wonder OEMs don't want to use Athlon 5350. It is just too expensive for what it offers.

They don't have to use HDL if they don't want to. Bristol Ridge is a desktop part that is (apparently) distinct from Carrizo. That is, it may not have the FCH on-die, though honestly we have no way of knowing whether or not that is the case. Also, keep in mind that, once Bristol Ridge comes out, the 28nm planar process should be more mature, as should the Excavator design. The engineers on that project will probably squeeze out some more performance here and there.

If performance pans out on that core (without HDL), then I think AMD should definitely make a hexcore then. Something that could even go up against an unlocked Core i3 if such a chip gets released. This in addition to other reasons I mentioned, including hexcore dies not yielding dual cores, etc.

Give it a dual channel DDR3/DDR4 memory controller so it works on both FM2+/FM3 boards.

Small iGPU, so the chip can actually be affordable.

Fix the desktop line-up so the performance of all parts can be pushed upward, not these big downbins like we see with 1M FM2/FM2+ (where a large amount of iGPU silicon is disabled on every chip) or Sempron 2650 (where the clocks of this dual core and its iGPU are pushed down even though it has a 25 watt rating)

 

[cbn]

Okay, are any of those 35W Intel CPUs in AiOs at the $400 price point?

About $450 with Pentium G3250T:

http://www.newegg.com/Product/Produc...82E16883266585

But notice it is a 21.5" 1920 x 1080p panel, not the cheaper 19.5" 1600 x 900 panel.

Assuming AMD could increase the performance of their cheap to make cat core chips (without having to be worried about product overlap and segmentation problems) then I think there is a good chance to see something a lot better at the $400 level with a 19.5" 1600 x 900 budget panel.

P.S. The ASUS 19.5" 1600 x 900 AIO with 35 watt Intel processor listed here ---> http://www.newegg.com/Product/Produc...82E16883220461 has a 35 watt Core i3 in it as well as the Pro versions of Windows 7 and Windows 8.1. It is obviously more expensive due to the price of the processor and software.

 

[ShintaiDK]

AMD's roadmaps clearly show an Excavator quad in Q3 2016 named Bristol Ridge with 512 shaders. It probably isn't Carrizo at all, but an Excavator APU with no on-die FCH on 28nm planar (no HDL).

Got a link to these roadmaps?

 

[DrMrLordX]

1.) I'm sure a hexcore SR could have a higher frequency than that considering AMD is able to clock the 65 watt quad core A10-7800 at 3.5 base/3.9 turbo.

Again, I invite you to go argue with The Stilt over it. He's been right about a frighteningly large number of things, and 3.5 ghz base clock is getting very, very close to 7850k speeds. Don't be so sure 3M Steamroller could clock that high at 95W or less.

2.) Hexcore SR could always overclocked to 4.5 Ghz+ with the proper board.

Sure, Crossblade Ranger or maybe A88x-Pro. Though I'm not sure it would hit 4.5 before maxing out the socket. FM2+ has issues with a shortage of power pins. I haven't found the upper power limit yet on the Pro, but then I don't even have a 512 shader chip so take that for what it's worth.

3.) FX-6300/FX-8310/FX-8320E, etc won't drop into a FM2+ board

So what? There's a comparatively-large installed base of AM3+ boards out there that can handle those chips.

Unfortunately Athlon 5350 is about the price of a 35 watt Pentium G3250T (http://ark.intel.com/products/83539/Intel-Pentium-Processor-G3250T-3M-Cache-2_80-GHz ) and slower in multi-thread according to passmark:

Doesn't matter if it's slower than an Intel chip or what have you. What matters is that it's faster than a Jaguar dual-core. AMD would be doing a lot better to have a chip like that in their big-box store OEM machines than an E1. Heck I'd rather have a 5350 than an E2-3800.

So yeah, no wonder OEMs don't want to use Athlon 5350. It is just too expensive for what it offers.

Ah, but there's the rub - it shouldn't be, at least not based on die size. It also shouldn't be too expensive to cool it in a bottom-tier AiO machine. For whatever reason, you just don't find chips like that in el-cheapo AMD AiOs.

If performance pans out on that core (without HDL), then I think AMD should definitely make a hexcore then.

Why, so it can compete with Zen/Summit Ridge? Good grief. It's almost crazy that they are even bothering with Excavator at that point, but they are for whatever reason. Are you seriously suggesting that they should expand their Excavator lineup just as Zen hits the market? There is no way in the Nine Hells I would waste my time with 3M Excavator when I could get 8M Zen instead. If Zen is bad enough that I need Excavator for something . . . better to jump ship to Intel by that point.

Something that could even go up against an unlocked Core i3 if such a chip gets released.

Yet another CPU we'll probably never see. But hey, who knows, Intel might get stupid and cannibalize their i5 sales.

This in addition to other reasons I mentioned, including hexcore dies not yielding dual cores, etc.

. . . that is not an advantage. Though, if history tells us anything, failed hexcore chips yield quads. Remember Zosma?

Give it a dual channel DDR3/DDR4 memory controller so it works on both FM2+/FM3 boards.

Likely impossible for very good reasons. FM3 will (hopefully) be pin-incompatible with FM2+ so that it can host more power pins, among other things.

Small iGPU, so the chip can actually be affordable.

. . .

Fix the desktop line-up so the performance of all parts can be pushed upward, not these big downbins like we see with 1M FM2/FM2+ (where a large amount of iGPU silicon is disabled on every chip) or Sempron 2650 (where the clocks of this dual core and its iGPU are pushed down even though it has a 25 watt rating)

If you really think that is AMD's problem on the desktop, then you are mistaken.

About $450 with Pentium G3250T:

That's a Lenovo machine at NewEgg. Lenovo has demonstrated the willingness to offer some nice processors near to the $400 price point, such as that $499 A6 machine with 8gb RAM I found on Best Buy's site. NewEgg is probably going to charge less than a big box brick n' mortar store. That represents an outlier.

Try Office Depot, Staples, Wal-mart . . . you know, places where you'd never even dream of buying a computer but where plenty of ill-informed consumers still do their shopping. HP, Dell, crap like that. Any Pentium G3250Ts at $450 there?

http://www.officedepot.com/a/browse/...Refine=391777/

There's a Lenovo G3240T that is on sale. Formerly $529, now $379. It's out of price range once the sale ends.

http://www.staples.com/Desktop-Compu...649|4224931103

Wow, nothin good there.

http://www.walmart.com/browse/electr...or_brand:Intel

There's another G3240T, this time for $480. Getting closer, but still out of reach of the bottom-dweller systems.

Best Buy omitted since their site is being slow at the time of this posting. Blah.

But notice it is a 21.5" 1920 x 1080p panel, not the cheaper 19.5" 1600 x 900 panel.

Lenovo seems to be curiously generous with features in their AiO offerings.

Assuming AMD could increase the performance of their cheap to make cat core chips (without having to be worried about product overlap and segmentation problems) then I think there is a good chance to see something a lot better at the $400 level with a 19.5" 1600 x 900 budget panel.

I don't. Are you noticing a pattern? You have to go to an e-tailer to consistently get something other than an E1 or a J1800 or some such garbage near the $400 price point. If you're lucky you can get something like that 3240T on sale. That is a huge hole in AMD's lineup, and OEMs refuse to plug it with 25-35W TDP chips. Intel chips in that TDP range are also scarce at that price point.

Got a link to these roadmaps?

Yes:

http://wccftech.com/amd-bristol-ridge-apu-2016/

which derives its content from:

http://translate.google.com/transla...ristol-ridge-samsas-om-sockel-fm3&prev=search

Interesting, I neglected to read about Promontory before. Promontory indicates that FM3 will have a Southbridge chip on the motherboard, ala A88x. That would seem to indicate that neither Zen (Summit Ridge) nor Bristol Ridge will have on-die FCH.

 

[mrmt]

AMD's roadmaps clearly show an Excavator quad in Q3 2016 named Bristol Ridge with 512 shaders. It probably isn't Carrizo at all, but an Excavator APU with no on-die FCH on 28nm planar (no HDL).

I would be wary of any roadmap published until today. AMD lost 15% of its R&D in the last quarter and there is a new boss on the company, ahe ought to new changes.

 

[ShintaiDK]

Yes:

http://wccftech.com/amd-bristol-ridge-apu-2016/

which derives its content from:

http://translate.google.com/transla...ristol-ridge-samsas-om-sockel-fm3&prev=search

Interesting, I neglected to read about Promontory before. Promontory indicates that FM3 will have a Southbridge chip on the motherboard, ala A88x. That would seem to indicate that neither Zen (Summit Ridge) nor Bristol Ridge will have on-die FCH.

So there isnt an official roadmap? Just an article going back to sweclockers that who knows the credibility of.

 

[Insert_Nickname]

Boo! I'm not asking for Anandtech's old review. I'm asking about your 6800k. You brought it up, so let's see how fast it goes. My 7700k is way out of spec and already faster than a stock 7850k.

Of course. It isn't overclocked at all because of severe thermal restraints. There should be an old thread around somewhere with details.

W/ 1600MHz 9-9-9-24 CR1

Aida64 5.00.3300:

CPU:
Queen - 21632
PhotoWorxx - 8419 MPixel/s
Zlib - 183.6 MB/s
AES - 9192 MB/s
Hash - 2186 MB/s

FPU:
VP8 - 3440
Julia - 7071
Mandel - 3615
SinJulia - 1481

Cinebench 11.5:

CPU - 3.52
OpenGL - 36.17FPS

 

[cbn]

Are you seriously suggesting that they should expand their Excavator lineup just as Zen hits the market?

Not expand, but rather replace Bristol Ridge with a hexcore/small iGPU processor that I think will have more appeal (for enthusiasts and OEMs) and that is cheaper to produce.

P.S. I don't think a hexcore (based on Excavator without HDL) will be able to compete against a 16 thread processor like Summit Ridge (Zen) in multi-thread.

 

[DrMrLordX]

I would be wary of any roadmap published until today. AMD lost 15% of its R&D in the last quarter and there is a new boss on the company, ahe ought to new changes.

So there isnt an official roadmap? Just an article going back to sweclockers that who knows the credibility of.

You know as well as I that AMD is playing with their cards close to their chest, especially with regard to anything desktop-related. They only recently admitted that Carrizo had been cancelled on FM2+. If the sole source were wccftech then yeah, maybe I'd wonder, but sweclockers is usually more reliable than that.

We'll find out in Q3 2016 whether or not Bristol Ridge is really going to launch at all.

Of course. It isn't overclocked at all because of severe thermal restraints. There should be an old thread around somewhere with details.

This thread? Okay, I can see why you are @ stock. A shame. I was hoping you had a 5 ghz 6800k. That's okay, I have an idea . . .

 

[mrmt]

You know as well as I that AMD is playing with their cards close to their chest, especially with regard to anything desktop-related. They only recently admitted that Carrizo had been cancelled on FM2+. If the sole source were wccftech then yeah, maybe I'd wonder, but sweclockers is usually more reliable than that.

What you call playing close to the chest I call not playing at all. AMD latest movies for the next couple of years have been cancelling or delaying products one after another. The simplest reason for AMD not to talk about Desktop Excavator is if it does not exist at all.

And think about it, the desktop market isn't really a tough market to serve. Nowadays the desktop market is comprised basically of trash silicon of server SKUs and trash silicon of mobile SKUs. If AMD had a competitive server part, performance wise, they could just launch it on the desktop and call it a day, the same for the mobile counterpart because in both cases whatever handicap they had in efficiency wouldn't matter too much for the desktop. If they are planning an Excavator SKU in 2016 then we can pretty much write Zen off already, because even if efficiency is ok, not significantly beating the performance levels of the unmitigated failure will be epic.

 

[DrMrLordX]

Not expand, but rather replace Bristol Ridge with a hexcore/small iGPU processor that I think will have more appeal (for enthusiasts and OEMs) and that is cheaper to produce.

So basically, what they are supposed to do is cancel yet another project, take a loss on all their R&D expenses incurred in production of Bristol Ridge, and start all over again with a design that would come out maybe ~4 months later than what would have been Bristol Ridge's launch date?

No. Just, no.

What you call playing close to the chest I call not playing at all. AMD latest movies for the next couple of years have been cancelling or delaying products one after another. The simplest reason for AMD not to talk about Desktop Excavator is if it does not exist at all.

It doesn't exist in any kind of meaningful timeframe. Q3 2016 is ages from now.

And think about it, the desktop market isn't really a tough market to serve.

I don't know that I really agree with that. Intel's top end consumer desktop processor is the 4790k which is an absolute monster, and that's just for the LGA1150 socket. Let's not talk about "enthusiast" LGA2011. AMD would have a very difficult time challenging the 4790k. If Intel felt threatened in the least, they could start pushing code initiatives (if they were on the ball) to encourage more widespread adoption of AVX2, putting AMD even further behind.

Yeah, that's right: Intel can extend their lead on the desktop in software. AMD TRIED to catch up in software (HSA) but they do not have the resources of Intel. One gets the impression that Intel has not pushed as hard for AVX2 adoption as they could have.

Nowadays the desktop market is comprised basically of trash silicon of server SKUs and trash silicon of mobile SKUs. If AMD had a competitive server part, performance wise, they could just launch it on the desktop and call it a day

Not if the competitive server part relies on mass parallelism to achieve performance. It's still somewhat difficult to make use of a consumer desktop processor's full power if it has the ability to handle more than four threads. DX12 could change that for gamers, but still, the point remains otherwise. Summit Ridge may wind up being an MT monster in a very nice power envelope, which does practically nothing for many end users. Don't get me wrong, I would like to have one when they come out, but I'm kind of a fanboy so don't mind me. I probably would have bought Bulldozer if I had been in the market for a chip back then . . .

If they are planning an Excavator SKU in 2016 then we can pretty much write Zen off already, because even if efficiency is ok, not significantly beating the performance levels of the unmitigated failure will be epic.

I wouldn't go that far. Zen is not going to be an APU, at least not to start. AMD apparently wants to keep producing APUs, so they need a successor to Kaveri, and Q3 2016 just happens to be the earliest date at which they can release one. It speaks more to their R&D failures outside of the Zen project than anything else.