@WolfToTheMoon

"During the event, Freescale Technology Forum 2015 the company Freescale Semiconductor said its plans to transfer multicore QorIQ communications processors to produce 16-nanometer technology FinFET."

"Study samples first SoC Freescale, made on 16-nanometer technology FinFET, should be ready in about a year, in mid-2016."

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@OlafS25

Well, their is licensing to consider.
Funny that none of the FPGA enthusiasts has realized that Freescale already license 68000 processor designs.
And Coldfire designs...
So why not PPC designs?

@olegil

Quote:

, there's a LOT of differences between a T10xx and an LS10xxA, performance wise

Yes, the LS is designed for a different market segment.
It doesn't offer enough advantages.
But the T2080 would make a nice last act.

@KimmoK

Quote:

transfer multicore QorIQ communications processors to produce 16-nanometer technology FinFET.

Anyone now get why having a foundry makes sense?
Otherwise they would have to face all the competitors trying to elbow each other out at Global Foundries.

@iggy

I think Gunnar somewhen mentioned that IBM has such designs but they are too slow on FPGA and costly to license

@OlafS25

Anything from IBM is to costly.
I learned that from their Cell BE period.
"Got a few million? We MIGHT be interested. BUT we will have to QUALIFY your application first".

@iggy

Quote:

So why not PPC designs?

nobody gives a damn about PPC anymore. Applied Micro left PPC, there is no interest elsewhere, save for IBM's big cores.

_________________

@WolfToTheMoon

Quote:

WolfToTheMoon wrote: @cdimauro Freescale is on 28 nm currently... So 2-3 nodes away from the best of the industry.

Not that much old, I've to say. AMD and nVidia are still selling GPUs with such process.
Quote:

Don't know their future plans, however, going past planar 20-22 nm is going to be really expensive for their volume - NXP might help drive volume sufficiently and, who knows - maybe they get IBM's business (rumors that Mubadala wants to divest Global Foundries who now own ex IBM foundries)

New productive processes are always expensive, but they pay on the long run.


@olegil

Quote:

olegil wrote: @OlafS25 Comparison right now: LS2080A, $210, 8 A57 cores at 1.6GHz, dual channel 2100MT/s memory, SATA3, 1292 pins. T2080, $144, 4 dual-threaded e6600 cores at 1.8GHz, single channel 2100MT/s memory, SATA2, 896 pins. I have a hard time seeing how the LS2080A would in any way beat the socks off the T2080, but I guess the extra memory bandwidth and the bigger caches help. That's what you pay extra for, though.

You also have 8 cores, instead of 4 dual-threaded. A single A57 is capable of decoding/issuing/executing up to 3 instructions per clock, whereas an e6500 can do the same, but only for 2 threads (3 total, but maximum 2 per hardware thread).

The A57 has also the new 64-bit architecture (ARMv8), which is much faster than the old ARM 32-bit one (v7 was the last), and with a better & improved FPU and SIMD unit.


@KimmoK

Quote:

KimmoK wrote: @WolfToTheMoon "During the event, Freescale Technology Forum 2015 the company Freescale Semiconductor said its plans to transfer multicore QorIQ communications processors to produce 16-nanometer technology FinFET." "Study samples first SoC Freescale, made on 16-nanometer technology FinFET, should be ready in about a year, in mid-2016."

First study samples, so it has to be seen when the real production starts...

Quote:

by KimmoK on 9-Dec-2015 8:47:43 @Zylesea "That said I wouldn't bet all my money on ppc future, but it's not dead (yet)." +1 Odds are against PPC success, that's why one should bet a little on it. Currently, PPC is "good to use" if nice SoC is needed for custom hobby motherboards. UPDATE: https://community.freescale.com/docs/DOC-328531 (LayerScape roadmap update) (very little or no "Power", "Transition to ARM Processors" etc...) In page3 they show Power and ARM are handled in equal way. But after going through the material... marketing is pushing ARM 100%. Power is still more powerfull, but they seem to belittle it. To me it tells they will focus in ARM, they…

The update suggests a strategic shift towards ARM processors with less emphasis on power-centric technologies like those found in PPC architectures. While current applications still find value in the latter, industry trends indicate an increasing focus on ARM solutions for future developments. This could imply that while Power may not be phased out entirely due to existing customer bases and its robustness, it's likely being positioned as a secondary choice behind emerging ARM technologies moving forward by Freescale.