• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD Launches Three Kaveri APU SKUs in February 2014 – Feature Set For A10 and A8 APUs Detailed

Why You Should Be Excited for AMD’s Next Line of APUs

Even though Advanced Micro Devices (NYSE: AMD ) is shifting away from the enthusiast market, investors should be excited after its CES 2014 press conference where the company officially unveiled its next-generation APU, the Kaveri. It's not only AMD's latest in their line of APUs, it's also sporting new technology that the company, as well as its backers, are hoping will cause a paradigm shift in computing.
Click to expand...

http://www.fool.com/investing/gener...ould-be-excited-for-amds-next-line-of-ap.aspx

qYRHVHY.png
 
Thats why I was letting down people gently very early on, to reduce the hype, martini.
I wanted to see 7750 gddr5 performance, based on the stagnation of richlands fail with rebranding vliw4.
Then people nameless to say were seeing 7750 shader leaks. Ignoring the differennce between a gddr5 7750 and a gddr3 version. And hyping over a leak 832 shader apu, ignoring the main issue shared system ddr3 memory bandwidth.
 
I dont think I will draw myself into the hype right now, I've owned both Trinity & Richland APU's. Very pleased with the former, but not with the latter, there was hardly any difference in day-to-day use & Richland ran a bit too toasty for my liking, not to mention the £20 price difference I paid. Both APU's are now in my nephews PC's.

I think I will buy a A88X board (I liked the A85X chipset boards) & pop an A8 5600K in there & wait for Kaveri's price to drop a bit. All I need to know is will faster ram benefit it like it did its predecessors.
 
d_brennen said:
Ain't that the truth. When they have a naff CPU product, it's hyped to the hills. When they have something genuinely compelling with soild USPs, they make about 5 of them and let OEMs put them in craptastic products at not quite cheap enough prices to redeem themselves
Click to expand...

To be fair to them they haven't really released a naff CPU, other than the Athlons they promised would take on the Core 2s.

The rest? failed to catch the imagination of lazy developers. If you're wondering what I mean? well Bulldozer wasn't really that bad. It just never got the software support it needed, and that was down to software.

AMD absolutely love releasing technology that isn't supported yet, then hoping that the software support arrives. Kaveri has HSA, not supported yet. However, the likelihood it will be is pretty much a given, because one of the consoles uses it.

AMD are spending more and more of their time lately, and their budget, on software technologies that will take advantage of their hardware. They'er pushing really, really hard to free us of DirectX and pushing 'metal to metal' about as hard as they can.

And like 64 bit they will eventually succeed. That wasn't an easy win either, given Intel's complete dismissive behavior.

I would concentrate less on the hardware AMD are releasing and look to the software.
 
ALXAndy said:
To be fair to them they haven't really released a naff CPU, other than the Athlons they promised would take on the Core 2s.

The rest? failed to catch the imagination of lazy developers. If you're wondering what I mean? well Bulldozer wasn't really that bad. It just never got the software support it needed, and that was down to software.

AMD absolutely love releasing technology that isn't supported yet, then hoping that the software support arrives. Kaveri has HSA, not supported yet. However, the likelihood it will be is pretty much a given, because one of the consoles uses it.

AMD are spending more and more of their time lately, and their budget, on software technologies that will take advantage of their hardware. They'er pushing really, really hard to free us of DirectX and pushing 'metal to metal' about as hard as they can.

And like 64 bit they will eventually succeed. That wasn't an easy win either, given Intel's complete dismissive behavior.

I would concentrate less on the hardware AMD are releasing and look to the software.
Click to expand...

Yeah, i remember that, now Intel can't live without it.

We may all laugh at and criticize AMD for their apparently lacklustre CPU's, but lest not forget even Intel, and Nvidia can't do without AMD's innovation and technology development.
 
humbug said:
Yeah, i remember that, now Intel can't live without it.

We may all laugh at and criticize AMD for their apparently lacklustre CPU's, but lest not forget even Intel, and Nvidia can't do without AMD's innovation and technology development.
Click to expand...

They're not really lacklustre though if they're used properly.

AMD should have known that no one would bother with Bulldozer. Supporting a CPU that isn't even supported by the operating system is just pointless. It's also pointless because to support Bulldozer in a game would have taken extra time and that means extra money. Thus, the world just continued to plod along in a quad core environment. I mean heck, even the quad cores barely got the support they needed.

AMD simply took for granted the support the Bulldozer needed. Their server CPUs had that support, so why not the desktop ones? well, reasons above. It's like coding a game to support SLI and Crossfire (and more, tri SLI and etc) natively. That takes time, time costs money. At times Nvidia or AMD side up with a game company and the support is there, but AMD simply couldn't get any one to bother with Bulldozer.

And it just continued on with Piledriver, right up to the point where they managed to get that technology into the consoles, thus, support would be there for games natively.

We're also seeing more of these benchmarks support these PD CPUs properly and the results are really rather good. A 8320 costing £109 can stand up there proudly just below the I7s that cost twice the price..

So what do AMD do? go and release another new technology that needs support :o

I don't know... Still, pushing for a better solution to game coding is always good I suppose. HSA is something else that could really stir things up, providing it's supported.
 
ALXAndy said:
They're not really lacklustre though if they're used properly.

AMD should have known that no one would bother with Bulldozer. Supporting a CPU that isn't even supported by the operating system is just pointless. It's also pointless because to support Bulldozer in a game would have taken extra time and that means extra money. Thus, the world just continued to plod along in a quad core environment. I mean heck, even the quad cores barely got the support they needed.

AMD simply took for granted the support the Bulldozer needed. Their server CPUs had that support, so why not the desktop ones? well, reasons above. It's like coding a game to support SLI and Crossfire (and more, tri SLI and etc) natively. That takes time, time costs money. At times Nvidia or AMD side up with a game company and the support is there, but AMD simply couldn't get any one to bother with Bulldozer.

And it just continued on with Piledriver, right up to the point where they managed to get that technology into the consoles, thus, support would be there for games natively.

We're also seeing more of these benchmarks support these PD CPUs properly and the results are really rather good. A 8320 costing £109 can stand up there proudly just below the I7s that cost twice the price..

So what do AMD do? go and release another new technology that needs support :o

I don't know... Still, pushing for a better solution to game coding is always good I suppose. HSA is something else that could really stir things up, providing it's supported.
Click to expand...

Intel live in the today, AMD live for tomorrow.

Intel have the money to change direction at the drop of a hat, AMD don't. So AMD do what they are good at, which is innovation and development, the idea being to turn the industry back to AMD's technologies.

Invest today for tomorrow, Intel just go with the flow, if technology takes a new direction they throw $30bn at redevelopment over night, and Pay AMD IP licensing if necessary, the money is pocket change to them.
 
Saw this on Anandtech forums(thanks Abwx):

http://translate.googleusercontent....a.html&usg=ALkJrhgsy-VVy3wsH_ZKaKFDiz-VUIBnBA

Tags: AMD , these 2014 ; Elpida ; GDDR5M ; JEDEC ; SK Hynix ;
Published on 11/01/2014 04:09 by Damien Triolet
Envoyer Imprimer
Share on facebookShare on twitter

This summer, we told you about the existence of a type of memory that had been more than discreet and who presented himself as particularly well suited to power APU: the GDDR5M . It should ultimately never see the day.

As a reminder, GDDR5M is a derivative of GDDR5 memory, the latter being primarily intended for GPU (but also operated on the PS4). If the basic technology is the same, the GDDR5M is optimized to allow a format "memory bar" without too explode costs, including through 4Gb chips interfaced 16-bit, 32-bit cons for classical GDDR5 . Enough to allow the establishment of memory modules SO-DIMM 2GB and 4GB


During CES, we had the opportunity to talk with Joe Macri which carries the dual role of Chairman of the committee in charge JEDEC DRAM and Chief Technology Officer in charge of public platforms for AMD. Of course we wanted to take the opportunity to find out what had happened to this GDDR5M.

Details of future A Series APU is still under embargo, and the eventual abandonment of the support of a technology originally intended which are unlikely to be formally discussed, we have of course not attempt to delve deeper by combining Kaveri the GDDR5M. We believe more than likely that AMD had planned to use for Kaveri but this remains speculation on our part.

What is not speculation by cons is that GDDR5M is well and truly dead and buried. Joe Macri has explained that he was a "wonderful memory" which presented extremely well at the start. Very promising, it would allow to save time compared to the arrival of the memory Wide I/O2 HBM or HMC, and two industrialists had decided to develop and produce.

This is the central point of the destiny of this GDDR5M as in the DRAM industry, it is necessary that at least two sources are confirmed for a technology can evolve into a commercial existence. Otherwise the risks are too high in availability or pricing. This is what killed in the bud development GDDR5M.

The two industries that were examined were Hynix and Elpida above. However, there are almost two years, Elpida had announced its bankruptcy and placed under protection by the authorities of Japanese financial regulation. They have ended all new projects which GDDR5M. Remained then was Hynix, and unfortunately no other DRAM manufacturer, Micron, for example, was willing to take over and start the adventure.

If our speculations are correct, you easily imagine that bankruptcy Elpida and could have important implications for AMD and Kaveri.
Click to expand...

That sucks.
 
CAT-THE-FIFTH said:
Saw this on Anandtech forums(thanks Abwx):

http://translate.googleusercontent....a.html&usg=ALkJrhgsy-VVy3wsH_ZKaKFDiz-VUIBnBA



That sucks.
Click to expand...

Indeed, while that is now dead AMD and Hynix are working on 3D Memory Stacking, so we have something to look-forward to.

With Elpida now backrupt there is an opening in the Memory market, i wonder if AMD could gather up some Elpida engenders, or even the brand along with them to get into that market themselves.

AMD have done surprising amount of work on Memory with huge success, not just the creation of IMC's and other Direct access architectures, but also GDDR5 and Random Access Memory Stacking.

Go on AMD, you know what your doing here, do it.
 
Last edited:
Memory stacking itself is pretty boring and relatively simple, transposers are the thing that is going to make memory stacking worthwhile and AMD(amongst many others) have been working on that.

the problem with stacked memory is, if you stack 4 chips, if 3 of them work and one doesn't, the stack is gone. So yields are.... not good. They are coming up and stacked memory will be viable when yields make it worth the cost basically.

The problem with stacked memory is how you connect it. If you build it as part of the die of a soc/chip, then every failed memory chip can fail an entire chip to that is a no go.

Stacked memory on dimms can increase capacity but not speed or bandwidth. The problem here is pin out under the chip, double bandwidth, double pin out(with same type of memory) and you literally can't fit more pins on these packages. The way you need to do it is on package, off die, but have connections small enough to actually route on package..... and that is where transposers come in. It's like a chip itself, it's basically a circuit board but rather than a normal circuit board like mobo with the traces, it's made in silicon so those traces are more than a magnitude smaller. So instead of struggling to route 128bit memory, you can pack traces to give you a 1024bit memory bus to on package memory with no problem.

Stacked memory, interposers are pretty simple tech, but can all destroy yields. Stacked memory, one chip in the stack fails, it's useless. As always chips can have failed yields, and tranposers won't have any difference there, some won't work off the wafer. But then you take a stack of memory chips which do work, a apu/soc that works, a transposer that works.... stick them together and some of those will fail as well. It's just taking time to get to the point where the pretty huge hit in yields becomes more cost effective, or when there is simply no other option to increase performance.

It's not quite why the Iris pro gpu is only on some Intel chips, but has the same reason, it's just far too expensive and decreases yields. it would decrease yields significantly further if it was part of the main die. I'm not actually sure how they've connected it and if they are using a transposer or a normal link, or a new link. Either way, it's all about getting things as close to the die as possible to accelerate performance but effecting yields as little as possible.

You also have the added advantage of being able to essentially mix and match. Premium part, add the stacked mem and transposer, budget part, package the chip as normal, meaning you can produce the one chip in large quantity and produce/package the rest however you want.

For Intel, if Iris pro chips stop selling, stop making cache, all the chips just get sold as normal Haswell's.
 
You must log in or register to reply here.
Back
Top Bottom