> It is unusual to see them do something on time and announce it reasonably close to its
> release date.

Analysts paint a different picture:

"Freescale plans to sample devices in 2H14, and we expect production shipments to start in 3Q15."
http://www.linleygroup.com/newsletters/newsletter_detail.php?num=5150

I didn't expect this:

AMD doing their own ARM core: WSJ

Interestingly it's pin compatible with an upcoming x86 chip and it's for client systems rather than servers.

> AMD doing their own ARM core: WSJ
> Interestingly it's pin compatible with an upcoming x86 chip

Sounds like AMD's SkyBridge is for ARM (Cortex-A57 and "K12") and x86 what Freescale's Layerscape is for ARM (Cortex-A7 and Cortex-A57) and PPC.

At first, I could not tell by the articles on this whether it was two pin compatible designs or one chip with both ISAs.
Now that I have realized it is the later, I am having trouble seeing the utility in this idea.
A totally ASMP design with two instruction sets.
Sounds like the complications inherent in using such a device would outweigh the benefits.

Quote:

PowerUP but faster? Strange plan.

Two separate pin compatible CPU's would be better IMO.

Quote:

That seems to be the plan, actually.

Quote:

It is the former, not the latter, as far as I can tell.

> At first, I could not tell by the articles on this whether it was two pin compatible
> designs or one chip with both ISAs. Now that I have realized it is the later

Is it? I understand it's the former.

> I am having trouble seeing the utility in this idea. A totally ASMP design with two
> instruction sets.

It can have utility for special purposes, for instance where there's a main ISA and another ISA for a security core. AMD did this before:

http://www.anandtech.com/show/6007/amd-2013-apus-to-include-arm-cortexa5-processor-for-trustzone-capabilities

And Applied Micro did this even earlier:

https://morph.zone/modules/newbb_plus/viewtopic.php?forum=3&topic_id=7675&start=268

> Sounds like the complications inherent in using such a device would outweigh
> the benefits.

If the two ISAs are supposed to be equipollent, yes. See:

http://en.wikipedia.org/wiki/PowerPC_600#PowerPC_615

Quote:

I believe they are two different chips but pin compatible with each other. One ARM, the other x86.

This means they only need to build one motherboard. I'd imagine most of the chip is the same as well (CPU cores are surprisingly small these days).

OK, two chips makes more sense, although most ARM Socs have functionality that X86 CPUs don't have.
An ARM chip with built-in AMD/ATI graphics would rock though.

BTW - Andreas, the 615 does sound like a bad idea.

http://www.amd.com/en-us/press-releases/Pages/ambidextrous-computing-2014may05.aspx


"Skybridge" (products available 2015)

A new family of 20 nanometer APUs and SoCs (ARM and x86) that are expected to be the world’s first pin-compatible ARM and x86 processors, full SoC integration, AMD Graphics Core Next technology, "HSA" (Heterogeneous System Architecture) and AMD Secure Technology via a dedicated Platform Security Processor (PSP).

- The 64-bit ARM variant of “Project SkyBridge” will be based on the ARM Cortex®-A57 core

- the x86 variant will feature next-generation “Puma+” CPU cores


"K12" (products available 2016)

A new 64-bit high-performance, low-power ARM-based core, designed by a core development team led by Chief CPU Architect​ Jim Keller.

The way I read this, is that the K12 will be a custom design ARM core, developed in-house by AMD. Apple did magic with their "Cyclone"/A7, which Apple promoted as a "Desktop Class" processor, a bold statement that actually has been confirmed as no exaggeration, as we have seen examples of in this thread. In a similar way, K12 seems to be AMD's take on going beyond the "standard" ARM Cortex-A57 core, and as a stopgap until it's done, they will be using the A57 in their "Skybridge" platform.


In that PR they also mention they have now publicly demonstrated for the first time their 64-bit ARM-based AMD Opteron™ A-Series processor, codenamed "Seattle", running a Linux environment derived from the Fedora Project.


Interesting times ahead for ARM indeed. As always!



(Edited: the CSS for the ul/li tags seems to be broken)

Yes, this one has my attention.
It will greatly enhance desktop availability and AMD's GPUs are top notch.
Smart move on AMD's part.

Quote:

Personally, I think the most interesting thing here is that unlike the AMD Opteron A-Series of processors (which targets server applications), the Skybridge seems to be where AMD is introducing ARM to client applications markets for the first time. And they are making it interchangeable with their next generation x86 Puma+ chips, which is interesting in more than one way IMHO.

And the "K12" is definitely a sign of determination and longevity in their ARM effort; to start developing their own 64-bit ARM core(s) that (presumably) will go way beyond the Cortex-A57, pretty much like Apple did. They will probably package the "K12" core(s) in both server and client CPU chips when its done...

Looking forward to see more info about this!



[ Edited by takemehomegrandma 08.05.2014 - 16:59 ]

Quote:

Quote:

He was one of the people responsible for the Alpha, AMD's K7 and Opteron, and most likely Apple's A7. Definitely worth looking at what the result of this will be.

...and he was senior at PASemi who did the PA6T processor in the X1000.

I've long held the opinion that there is something wrong in the setup of the processor in the X1000 that's holding it back - some of the early (non Amiga) benchmarks were very good.

I recently read a theory that they are using the wrong compiler. This would explain a lot.

> I recently read a theory that they are using the wrong compiler. This would explain a lot.

It is not just theory but fact that code compiled for OS4 is not tuned for the PA6T via compiler options. If it was, it would either run slower on the other OS4-supported CPUs or you'd need to provide several executables, one for each sub-arch. If you check OS4Depot, you will find that there's only one executable per archive with almost all packages (so either no specific compiles for G2/G3/G4 and PPC4x0, let alone PA6T, or tuned for one of them and thus running extra-slow on the others). So even when disregarding SIMD, compiling for OS4 is mostly a compromise where you target the lowest common denominator, which results in an executable that runs on all CPUs but not as fast as it could on any of them.
The same is true for MorphOS, btw (see the need for G5-specific executables for some applications), and the Sam460 port will only intensify this issue. Amigans know this whole shebang from the m68k days with specific compilation for 68000/20/30/40/60.

"AFAIK the old GCC version 4.2.4 in the current SDK may generate instructions not supported by the PA6T CPU which have to be emulated by an exception handler in the kernel, which is extremely slow."
http://www.amigans.net/modules/xforum/viewtopic.php?post_id=90747#forumpost90747

"there is no support for that processor target in the AmigaOS4 SDK yet."
http://euaejit.blogspot.com/2014/07/ppcjitbeta04b-hundreds-and-xthousands.html

[ Edited by Andreas_Wolf 28.07.2014 - 20:39 ]

Quote:

The CSS is fine but you need to use (minus all whitespace) [ ul ][ li ][ /li ][ /ul ] (unordered list) or [ ol ][ li ][ /li ][ /ol ] (ordered list).

Update:

>> in terms of 64-bit ARMv8 (AArch64) cores, there is announced so far:
>> - Cortex-A53 and Cortex-A57 from ARM Ltd.
>> - Denver from nVidia
>> - X-Gene from Applied Micro
>> - Thunder from Cavium
>> - Cyclone from Apple

> - Vulcan from Broadcom
> - K12 from AMD

More details on Cavium's ThunderX:

http://www.cavium.com/newsevents_Cavium_Introduces_ThunderX_A_2.5_GHz_48_Core_Family_of_Workload_Optimized_Processors_for_Next_Generation_Data_Center_and_Cloud_Applications.html
http://www.cavium.com/ThunderX_ARM_Processors.html

Quote:

I thought the AMD was impressive for an ARM... but this thing has 48 (!) 2.5GHz 64bit ARM cores. Half a Terabyte memory per chip and you can gang a pair together with a coherent bus. PCIe, SATA etc is all built in.

Mind = boggled.

I think it's safe to say the days of ARM being stuck in the low end are now over.

> I think Jen-Hsun Huang's words may hold ground and shouldn't be dismissed as
> "propaganda" quite so easily; he may actually know what he is talking about
> regarding his own product development; if he says they will release CPU's that
> will compete with x86 in PC, server and supercomputer markets, chances are
> kind of good that they will do just that.

Denver-based chips for servers (aka Project Boulder) cancelled:

http://www.pcworld.com/article/2366480/nvidia-abandons-64bit-denver-chip-for-servers.html

"while NVIDIA has never officially written off Denver’s server ambitions, it seems likely that POWER has supplanted Denver as NVIDIA’s server CPU of choice."
http://www.anandtech.com/show/8701/the-google-nexus-9-review/2


Edit: added Anandtech quote

[ Edited by Andreas_Wolf 08.02.2015 - 18:46 ]

ARM isn't a bad architecture, but it isn't my first choice either ( MIPS For LIFE / IRIX is forever! ) I have myself a G5 Quad though with an X1900/512MB inside which is much faster than any of my MIPS boxen. Hoping MorphOS will support it and have a 64-bit kernel eventually.

So long as it isn't on the horrid 8080-derived architecture that Intel and AMD push ( And I want one for convenience, ain't I a hypocrite? )

> the horrid 8080-derived architecture that [...] AMD push

They are in the process of establishing a second pillar:

https://morph.zone/modules/newbb_plus/viewtopic.php?forum=3&topic_id=7675&start=473
https://morph.zone/modules/newbb_plus/viewtopic.php?forum=3&topic_id=7675&start=551

Not that this makes the choice between x86 and ARM any clearer...

http://www.extremetech.com/computing/181867-amds-project-skybridge-new-arm-and-x86-chips-that-are-pin-compatible

btw As long as people are using high level programming languages, what does it matter if the underlying cpu architecture is a bit of a mess? What actual difference does it make to end users?

>As long as people are using high level programming languages, what does it matter if the underlying cpu architecture is a bit of a mess? What actual difference does it make to end users?

A bit of a mess is an understatement. Think about it. Does the engine in your car matter? No matter how you slice it, you're looking at problems if your engine is dreck. Look at the AMC Gremlin if you want an example. It sold well enough though and was a money maker for AMC.

To be clear I'm not a fan of PowerPC either due to lack of active desktop development.

I'm a proponent currently of SPARC64 and MIPS development as both architectures embody what I want in a computer: Low clock speed, legendary reliability, active desktop development and efficiency. Sun/Oracle hardware may be pricy, but its snappy and reliable for its age and I have never had one die on me, plus theres plenty of them. MIPS I like for the same reasons, I own SGI desktops, there are Chinese made MIPS desktop systems which have open documentation and are relatively inexpensive.

ARM has mostly market in cheap DIY hardware which may or may not have good documentation, has closed firmware or else is extremely expensive by comparison. Id say ARM has a long way to go before I consider it worthy to SPARC or MIPS

It would be nice for MorphOS to use MIPS especially if it has open documentation. But does MIPS laptop exist?

>TeamBlackFox wrote:
>... To be clear I'm not a fan of PowerPC either due to lack of active desktop development.
>I'm a proponent currently of SPARC64 and MIPS development as both architectures embody what I want in a computer: Low clock speed, legendary reliability, active desktop development and efficiency.

Could you show a link to MIPS desktops?
It's new to me.

Same for active SPARC64 desktops. Link?

I've been sure that PowerPC is the biggest desktop after x86/x64.

Other than that, IMO, MIPS is inferior to PPC.
(PPC is also far bigger in market share / bigger business http://www.brightsideofnews.com/2013/10/17/linley-processor-conference-arm-sweeps-next-gen-processor-designs/ ... unless there's cooking something that western world does not show)

btw. Does MIPS nowdays have SIMD? How about SPARC?
What's the state of RadeonHD or Nvidia support on those?

[ Edited by KimmoK 27.06.2014 - 06:22 ]