Interfacing FPGA to microcontrollers

An article going over high-level details of interfacing a microcontroller and FPGA: read here
[PLDesignLine]

FPGA-based hardware acceleration of C/C++ apps

link

Virtex 57 FPGAs break the PetaByte Clocking Barrier

In other words, it's April Fool's Day: Reach NEW Heights with Xilinx Virtex 57

99-cent FPGAs from Actel

Actel announced two new low-density FGPAs: IGLOO and ProASIC3 starting at only 99-cents! The target markets are communicatinos, consumer, medical and industrial applications (including smart phones, system controllers, portable medical devices, wireless sensors). With the example spec of 128 macrocells, 15k gates, 5 uW, and 1 Kb nonvolatile Flash ROM, it promises to beat the power consumption by 10x compared to the leading PLDs!

PLDesign Article

Using FPGAs for HDTV design

Tam Do's article from Altera on EDN. Read here.

Reconfigurable ECAs - new platform for SDR?

Article by Chen Zhang, Virginia Tech. Chen implemented a simple ZigBee using an ECA (Element CXI's Elemental Computing Array). ECAs provide "higher computational density, lower power consumption and higher structural robustness" [1] compared to GPPs, FPGAs and DSPs. It looks like a great new platform for Software Defined Radio (SDR) projects!

[1] Dynamically-reconfigurable ECAs - Part 4 (Student Project #2), Programmable Logic, Jan 2, 2008

Related:
ECA Architecture
ECA Programming Model
ECA Project (FIR Filter)
ECA Project (Image Processing Algorithm, ECA vs FPGA)

FPGA Optimized Soft Core Processors

Based on Programmable Logic Design Line article titled "Actel and ARM announce the Cortex-M1 soft processor core" (March 19, 2007)

ARM processors are widely used in the industry with applications ranging from automotive systems to wireless networking. The recent addition to their family, the 3-state pipeline Von Neumann Cortex-M1 processor, is developed with FPGA implementation in mind. It can execute Thumb code from previous generations of ARM processors (ARM7, ARM9 and ARM11), as well as compatible with the newer members of ARM's processor family - Cortex-M3, Cortex-A8, etc that support 32-bit Thumb-2 system instructions. Fast, and optimized for various FPGA devices (including Xilinx, Altera, Actel and QuickLogic), it is designed to take less FPGA fabric and produce less power.

Short history:
ARM7 - 1993, 3-stage pipeline (supports both, 32 and 16-bit ARM and Thumb instruction sets)
ARM9 - 1997, 5-stage Harvard memory architecture (32-bit RISC)
ARM11 - 2002, 8-stage Hardware memory architecture (32-bit RISC)
Cortex-M3 - 2004, 3-stage, high-performance, low cost (32-bit RISC)
Cortex-A8, 2005
Cortex-R4, 2006, 8-stage

Increasing Popularity of FPGAs

FPGAs are becoming more and more popular with every year of their existence. Previously slow and ignored by system designers working on speed-hungry applications, the newest FPGAs are reaching 600 MHz internal clock speeds. Altera announced it's Stratix III 65nm device at the end of 2006 (link). Xilinx Virtex 5, also 65nm, runs at around the same clock rates (link). Their ease of use, inexpensiveness and availability of read-on-your-own reference material on the internet makes them attractive not only to system designers, but also to students and DIY enthusiasts with a wide variety of applications.

From my experience in semiconductor industry I see that these devices create a serious competition for ASIC manufacturers. They caused at least one project to be canceled during the last year at my company. Companies choose to pay more for flexibility of re-programmable FPGA, rather than investing into a faster ASIC. With the increase of capacity and speed of FGPAs, ASICs will play a less important role then they did years ago.

As an interesting example of FPGA use, Pete Finnigan's security blog mentions that they could be used for faster password cracking. Read it here.

The low price, availability of development boards and easy to use integration software packages make FPGAs attractive to a wider range of users, interested in projects, for example, requiring custom hardware acceleration of simulations. One of my friends used FPGAs for neural net simulation during his Master's. FPGAs really speed up functional computations. A function that lasts hundreds of cycles in software can be computed in few clock cycles given that enough logic gates are used.

More and more FPGA fast prototyping software packages become available on the market. For example, LabView's FPGA kit allows a quick assembly of a system out of precoded optimized blocks ready to be wired together with a few mouse clicks. Even without prior HDL knowledge, one can add hardware accelerated functionality to their designs with ease, using these tools and FPGA development kits.

Numerous learning resources are easily accessible online for free. Xilinx has eLearning section and a quick Google search would reveal dozens of other sources of learning information. Check out www.fpgafromscrtach.com for a step-by-step guide of designing an embedded system on Virtex 4 FPGA.

Related links:
FPGA Design Tutorial from Scratch
Xilinx eLearning

FPGA vs. Microprocessor, Programming Comparison

Here is an interesting entry I found on Zeemz Blog about FPGA vs. Microprocessor programming comparison.

He's referencing Design and Reuse's article from March 2004.

FPGA Design Tutorial from Scratch

Check out this cool step-by-step tutorial that guides you through the process of designing embedded system on Virtex 4 FPGA. Steven Andersson from ZooCad consulting, with 15 years of ASIC design experience, is working on this project for a year on his own time and blogs his experience.

I'd love to put my hands on something like this in the near future.

Tutorial Link

Xilinx Free E-learning courses

Related articles:
Embedded.com
Programming Logic Design