Subscribe to EDN

Power in the hands of ...

January 13, 2009

One intriguing takeaway in a recent FPGA users’ survey from Piper-Jaffray and EE Times was the fact that the PowerPC embedded core dominated all other RISC options by almost two to one (for RISC cores used within an FPGA SoC). Not only did it lead the next most popular “hard” processor core, ARM, by 64.5% to 31.6 %, but it beat the Microblaze and Nios soft processor cores by even bigger margins. MIPS, Tensilica, and other options were even further down the list.

Obviously, the fact that Power is the chosen core for FPGA market leader Xilinx plays a role here, as does the joint marketing of standard Power devices by Freescale, IBM, AMCC, and the Power.org coalition. But the fact that synthesizable cores from any FPGA vendor are still secondary contenders, and the low showing by alternative RISC cores outside ARM, means that mere market power can’t explain it all.

Here’s what seems to be happening. Power and ARM were designed from the start as streamlined, optimized cores for embedded use. Yes, Power was initially the base of the PowerPC, but the e series embedded options were offered from the late 1990s on. ARM always had a strong base in handheld, communications, and industrial applications. Developers of the MIPS and Tensilica instruction sets have made valiant efforts to establish embedded spins of their architectures, but particularly in the confines of an FPGA-based SoC, they must contend with the inertia of existing architectures.

Inertia also explains why Microblaze and Nios still must fight for share of mind. In theory, soft IP cores provide smaller footprints and lower overall costs. It was relevant in the survey that hard CPUs were beginning to plateau in users’ anticipation of future use, while soft cores still were on an upward path. The slow replacement of Power and ARM by synthesizable options can be seen as the inevitable user preference for the familiar. Until there is more comfort in using the design tools for fully configurable SoCs, Power architectures will no doubt reign supreme.

 

 

Posted by Loring Wirbel on January 13, 2009 | Comments (6)

April 16, 2010
In response to: Power in the hands of ...
Buy Cialis commented:

braille lucy personal accountant nonlocomotor diction measurable evening seedwiki oxford csir


January 19, 2009
In response to: Power in the hands of ...
Tim commented:

In my mind, part of the slow uptake of soft-procressors comes from the fact that they are only available as soft-processors... and so lack the versatility of application choice. I can get a single-chip ARM9 with flash, sram, and all peripherals for $10... a very low cost and small package. does anyone sell a similar NIOS2 or Microblaze chip for cases where the FPGA is not needed? No wonder software guys never use one. Decent ARM soft-cores are just becoming available, but cost can be an issue and they still lag performance from proprietary soft cores.


January 15, 2009
In response to: Power in the hands of ...
Loring commented:

Indeed, raising the issue of whether on-chip integration always is better, particularly as we get to multicore integer or DSP cores. Even with a single processor, there may be times when a low-power, off-chip ARM in conjunction with an FPGA for peripheral functions may be a better option than any on-chip processor core. One size does not fit all.


January 15, 2009
In response to: Power in the hands of ...
Ajay P commented:

With the cost and power being the major motivating factors behind the selection of ARM. Having it embedded in FPGAs which are inherently power hungry(leaving apart the exceptions)and which are mostly considered for applications with higher performance, the popularity of higher performance PPC among FPGA embedded processors seems justified.


January 13, 2009
In response to: Power in the hands of ...
Loring commented:

Yeah, the OS factor actually may override all others. Thanks for the input!


January 13, 2009
In response to: Power in the hands of ...
Andy T commented:

I think the mystery factor of FPGA core success had little to do with optimization for embedded use, and more with the inherent optimization of the architecture (primarily a hardware MMU) to support the OS of choice - Unix/Linux. Being real time in most apps, software emulation by the other processors didn''t cut the mustard. Momentum/intertia from there.

POST A COMMENT
Display Name
captcha

Before submitting this form, please type the characters displayed above. Note the letters are case sensitive:

Advertisement
Advertisement
Advertisement
About EDN   |   Site Map   |   Contact Us   |   Subscription   |   RSS
© 2012 UBM Electronics. All rights reserved.
Use of this Web site is subject to its Terms of Use | Privacy Policy

Please visit these other UBM Canon sites

UBM Canon | Design News | Test & Measurement World | Packaging Digest | EDN | Qmed | Pharmalive | Appliance Magazine | Plastics Today | Powder Bulk Solids | Canon Trade Shows