AMD CPU speculation... and expert conjecture

[-Fran-]

I was actually considering one to replace my llano htpc, but not for $175.

I think we're better off waiting for stacked ram, DDR4 or embedded ram before making the upgrade/switch.

For all that I think Kaveri and Richland are better than Llano, I can still play stuff and do stuff in it with no suffering. I would say you're in the same boat. Plus, FM2 will be replaced sooner than later. I don't think the Excavator APUs will fit in FM2 if they sport DDR4 or other fancy type of external memory. Plus, RAMBUS opened up, so we might actually see new tech in non-closed RAM specs (JEDEC?).

Cheers!

 

[juanrga]

Carrizo APU competition details

http://wccftech.com/leak-intels-14-nm-broadwell-sku-details-gpu/

 

[juanrga]

Kaveri comes with 2400MHz support, just as someone predicted

http://www.hardwareschotte.de/preisvergleich/Amd-A10-7850K-Black-Edition-Apu-AD785KXBJABOX-p21826030#

 

[blackkstar]

For a company that just released Hawaii, which wasn't on any roadmaps, and Mantle, which wasn't on any roadmaps, some of you really take AMD roadmaps as written, unchangeable truths.

If anything, discussion should be around what is possible with AMD, not what they're saying they're going to do. It should be quite clear by now that whatever AMD says is going to happen is probably not actually going to happen.

 

[de5_Roy]

For a company that just released Hawaii, which wasn't on any roadmaps, and Mantle, which wasn't on any roadmaps, some of you really take AMD roadmaps as written, unchangeable truths.

If anything, discussion should be around what is possible with AMD, not what they're saying they're going to do. It should be quite clear by now that whatever AMD says is going to happen is probably not actually going to happen.

i think that hawaii was on a gpu roadmap, but it was never made public. because it had different specification from what is amd selling. my guess is that it had stacked ram in early iterations but got 'changed'. later, amd recycled the codename since it takes the hassle away from finding another codename.

 

[juggernautxtr]

Intel will cross licenses to AMD,because Intel needs AMD64, with out it Intel would be unable to do 64 bit processing.
AMD + arm is Intels worst nightmare coming to light, as they have already created server parts with both arm and X86 processing.
HP moonshot.

AMD will stay around, looking at the APU side of AMD, these are gaining more and more traction, they are now using these in servers too. and there they are doing actually really well and they gaining popularity over intel, they just need the production to supply demand.

amd is also trending APU before HPDT, BD was released on apu before it was sent to desktop, and i think they may have decided to forgo steamroller for the memory controller rework for high performance desktop, the splitting of the decoder and increasing the input lanes to each core is going to require faster speeds from memory, should be a really good increase. BD,vishera are starved for input. it didn't help that the windows 7 scheduler was/is bouncing core assignments like a little kid on a pogo stick.

marketing for as it started for the FX line should never have said 8 cores, the whole idea on these that they were supposed to have the extra core serving to run an extra thread. basically hyper threading, only with a real core doing work.
and this is what screwed AMD because 90% of the people can't get it through their thick skulls that this is not a "true" 8 core processor, respectfully only 2,3,4 core processors with extra integer core for replicating hyper-threading on an actual integer core., and it doesn't help matters when most bench testers for people just getting into higher end pc's to understand what it was meant for explaining what this architecture is about. automatically "oh it's 8 cores and it can't even keep up with 4!?"
AMD has had some really bad marketing when it comes to this processor. I think AMD is going to make sure it is right before they release a new processor this time, 2-3 more gens and AMD should be crawling up Intels rear end.

 

[anxiousinfusion]

For a company that just released Hawaii, which wasn't on any roadmaps, and Mantle, which wasn't on any roadmaps, some of you really take AMD roadmaps as written, unchangeable truths.

If anything, discussion should be around what is possible with AMD, not what they're saying they're going to do. It should be quite clear by now that whatever AMD says is going to happen is probably not actually going to happen.

i think that hawaii was on a gpu roadmap, but it was never made public. because it had different specification from what is amd selling. my guess is that it had stacked ram in early iterations but got 'changed'. later, amd recycled the codename since it takes the hassle away from finding another codename.

right around 36min the speaker was asked what cpu 8350 what motherboard Hawaii. Does amd have a hybrid 8 core cpu under there belt ? embedded aio ?

http://www.youtube.com/watch?v=QIWyf8Hyjbg

Everyone is misinterpreting this. He says FX-8350 and a Hawaii board which refers to the PCB board of the the Hawaii GPU. No it is not integrated, no it is not a system mainboard. I don't mean this offensively but perhaps it can sound confusing to those who don't speak American English as a first language. He is very clear.

 

[os2wiz]

I was actually considering one to replace my llano htpc, but not for $175.

You are not using your brain. That is a pumped-up pre-order price. once it is full production the price will be no more than $159 and likely less.

 

[juggernautxtr]

looks like mantle is going to scale across the cpu like a plague, i have a feeling that the "hawaii board" is next gen desktop board.
8350+ r290x
AMD is up to something, they have always been innovative in technology, AMD has never been this quiet.

 

[de5_Roy]

amd is also trending APU before HPDT, BD was released on apu before it was sent to desktop,

bd(zambezi) never made it to apu. too power hungry.

more gens and AMD should be crawling up Intels rear end.

quite possible. as long as they have the funding for it. before, i assumed calxeda would outrun amd and debut a 64bit soc before seattle.
right now, amd have too many high volume products that sell well, but don't make enough profit. except the high end desktop gpus and some firepro cards. high volume is good for showing people the increase in marketshare and probably pacifying investors.

 

[Master-flaw]

looks like mantle is going to scale across the cpu like a plague, i have a feeling that the "hawaii board" is next gen desktop board.
8350+ r290x
AMD is up to something, they have always been innovative in technology, AMD has never been this quiet.

I agree...I think they're already investing in what they have out..Seems like they have pretty big plans to bullhorn the market from the software side of things. They've already got the power...they just need developers to use it.
It's the only chance they've got against Intel line....and it may actually work, as single core performance is extremely restrictive and can easily hit a wall.

They seem to separate the two markets well...makes sense for why they came out with a new APU line and did nothing with their high end line.

 

[noob2222]

@os2wiz

Its not like amazon themselves to price gouge.

If it was a third party seller I could see that, but that one is sold by amazon.

 

[juanrga]

Intel will cross licenses to AMD,because Intel needs AMD64, with out it Intel would be unable to do 64 bit processing.
AMD + arm is Intels worst nightmare coming to light, as they have already created server parts with both arm and X86 processing.
HP moonshot.

AMD will stay around, looking at the APU side of AMD, these are gaining more and more traction, they are now using these in servers too. and there they are doing actually really well and they gaining popularity over intel, they just need the production to supply demand.

amd is also trending APU before HPDT, BD was released on apu before it was sent to desktop, and i think they may have decided to forgo steamroller for the memory controller rework for high performance desktop, the splitting of the decoder and increasing the input lanes to each core is going to require faster speeds from memory, should be a really good increase. BD,vishera are starved for input. it didn't help that the windows 7 scheduler was/is bouncing core assignments like a little kid on a pogo stick.

marketing for as it started for the FX line should never have said 8 cores, the whole idea on these that they were supposed to have the extra core serving to run an extra thread. basically hyper threading, only with a real core doing work.
and this is what screwed AMD because 90% of the people can't get it through their thick skulls that this is not a "true" 8 core processor, respectfully only 2,3,4 core processors with extra integer core for replicating hyper-threading on an actual integer core., and it doesn't help matters when most bench testers for people just getting into higher end pc's to understand what it was meant for explaining what this architecture is about. automatically "oh it's 8 cores and it can't even keep up with 4!?"
AMD has had some really bad marketing when it comes to this processor. I think AMD is going to make sure it is right before they release a new processor this time, 2-3 more gens and AMD should be crawling up Intels rear end.

Nobody doubts that Intel needs AMD64 license. What was speculated is if AMD needs x86-32 license, because some people is speculating that if AMD switch to ARM they will not need an Intel license anymore and can stop licensing AMD64 to Intel. Then Intel would be in big trouble!

I think AMD explained very well that its CMT architecture is in the middle between the CMP architecture and SMT architecture (e.g. Intel hyperthreading).

You are using a definition of core as integer+float, but this is not the only definition possible, neither the original. I have read many intel fanboys saying that FX has not true cores, but using their own definition, the 386 chip by Intel was not a true core, because it lacks floating point. They never make this last claim.

However, I agree with you that AMD marketing would be better if they had cataloged the 8000 series as 4M/8T instead of 8C/8T. and then explained that 1C+HT < 1 M < 2C, where C denotes CMP core and HT is hypertreading.

 

[juanrga]

Extreme OC of Kaveri

https://www.facebook.com/BiostarHQ/posts/10152048747775772

 

[-Fran-]

Juan, AMD needs "X86-32" even more than Intel needs X86-64. Simple reason being legacy support. There's still a LOT of 32 bit application and Windows 32bit installations out there.

And I agree with the Intel fanbois, hahaha. I've never agreed/accepted the fact that BD design is an "octocore" CPU. Like you said, it is silver lining, but we're all raging technical nerds, right?

Cheers!

EDIT: I think the pictures show a multi of 47 for the CPU.

 

[os2wiz]

looks like mantle is going to scale across the cpu like a plague, i have a feeling that the "hawaii board" is next gen desktop board.
8350+ r290x
AMD is up to something, they have always been innovative in technology, AMD has never been this quiet.

I think AMD is taking some steps to become competitive again. They are being careful not to make too many wild claims that will come back to hurt them. I think their marketing department has done an immensely better job than in the past. As far as having something up their sleeves, I have no idea, but it would be nice if sometime in 2015 they have some 8 core apus out there for enthusiasts who do work that requires excellent multi-threaded performance, not only better single-thread performance. HSA must take off in the next year or AMD will be very exposed in their lack of multi-threaded performance. Intel has announced some new xeon chips with 12 and 16 cores. AMD needs to get back into the general server market , not only microservers.

 

[etayorius]

When the hell does Kaveri gets released? exact date anyone?

 

[de5_Roy]

When the hell does Kaveri gets released? exact date anyone?

jan. 14 2014?

 

[anxiousinfusion]

When the hell does Kaveri gets released? exact date anyone?

January 14th http://www.anandtech.com/show/7507/amd-kaveri-apu-launch-details-desktop-january-14th

 

[etayorius]

Thanks!

 

[juanrga]

Juan, AMD needs "X86-32" even more than Intel needs X86-64. Simple reason being legacy support. There's still a LOT of 32 bit application and Windows 32bit installations out there.

And I agree with the Intel fanbois, hahaha. I've never agreed/accepted the fact that BD design is an "octocore" CPU. Like you said, it is silver lining, but we're all raging technical nerds, right?

Cheers!

EDIT: I think the pictures show a multi of 47 for the CPU.

No problem with people using different definitions of core, but with Intel fanboys that apply one definition to claim that 8350 is not a true octo-core but "a quad-core with extra integer units", and then apply another definition to claim that the i386 was a single-core chip.

47x ~ 6 GHz?

 

[palladin9479]

AMD looses across the board. If these numbers are real, AMD has a massive, massive problem performance wise.

When I see i3's (dual core) in the same bracket as i5's (quad core) I immediately disregard the benchmark. Whomever was running it was tinkering with settings / environment to force as much load on a single core as possible. This is obvious in the frostbite engine benchmarks as that engine is designed to scale dynamically with more stuff happening in the environment. Running a bench of you walking around an empty area looking at scenery produces radically different numbers then running through an actual battle field with dynamic environments and physics taking place. This is also why each site seems to have "different" benchmark numbers that show different things, editing different settings and exactly what your benchmarking have become more important then just running a 15m circle around an empty jungle area.

 

[palladin9479]

Juan, AMD needs "X86-32" even more than Intel needs X86-64. Simple reason being legacy support. There's still a LOT of 32 bit application and Windows 32bit installations out there.

And I agree with the Intel fanbois, hahaha. I've never agreed/accepted the fact that BD design is an "octocore" CPU. Like you said, it is silver lining, but we're all raging technical nerds, right?

Cheers!

EDIT: I think the pictures show a multi of 47 for the CPU.

x86-64 includes x64-32. AMD didn't design a new ISA, they designed a set of extensions of the x86 ISA to make it 64 bit capable similar to what SSE is. This is more akin to the 3dnow vs MMX vector extensions of old but with AMD being the one to make it to market first. AMD has a perpetual license to the x86 instruction set due to a court settlement a long time ago. Intel has access to the AMD 64-bit extensions as a result of another settlement where they agreed to give each other unrestricted access to any x86 related ISA that each makes, thus AMD got SSE3/4 and eventually AVX while Intel got x86-64 and anything HSA related (if AMD didn't make it an open standard, which they did).

Anyhow the 8xxx series is indeed an eight core CPU by the actual definition of a core. It has eight independently externally addressable general purpose processing units. The cores do not share ALU's which are the centrally addressable component of general purpose processors. The SIMD FPU's in x86 are technically co-processors with separate registers, stacks and pointer, they just happen to exist on the same die. L2 cache has been shared amongst cores before so that's nothing new. The only controversial part is them building one large front end decoder / scheduler for each module vs two independent decoders / schedulers. This isn't unheard of or new, but it's not something you typically do in a consumer orientated general purpose processor as getting it right requires the code to be tightly integrated with the metal. From what I've heard SR did quite a bit of work up front to widen up this part.

 

[FALC0N]

Anyhow the 8xxx series is indeed an eight core CPU by the actual definition of a core. It has eight independently externally addressable general purpose processing units. The cores do not share ALU's which are the centrally addressable component of general purpose processors. The SIMD FPU's in x86 are technically co-processors with separate registers, stacks and pointer, they just happen to exist on the same die. L2 cache has been shared amongst cores before so that's nothing new. The only controversial part is them building one large front end decoder / scheduler for each module vs two independent decoders / schedulers. This isn't unheard of or new, but it's not something you typically do in a consumer orientated general purpose processor as getting it right requires the code to be tightly integrated with the metal. From what I've heard SR did quite a bit of work up front to widen up this part.

Your a little off here.

There is no set definition for what constitutes a cpu "core". At least not that I am aware of. Thereby invalidating most of your argument.. I found the part about the front end and cache particularly questionable since the L2 and L3 cache are not a requirement for a functioning processor but the front end is. Yet you concede the entire front end is "controversial", while treating it as if its of equal or lesser importance than the cache.

Yes, the shared front end isn't new, your right about that. But you neglected one key point. On those server side parts that use multiple integer cores and shared front ends, the entire "module", to borrow AMD parlance, is considered the core, and not each individual integer execution engine. I would cite the IBM Power 7 as an example. Even AMD labeled their current bulldozer "modules" as cores in their original Bulldozer patent submission.

 

[FALC0N]

Then again, the lack of a definition means you could call anything a core I suppose. Maybe your right afterall Palladin.