Welcome Guest! To enable all features please Login.

Notification

Icon
Error

Why isn't FPGA use growing faster?
colin.holland
#1 Posted : Saturday, September 15, 2012 3:25:39 AM(UTC)
Rank: Newbie

Joined: 9/14/2012(UTC)
Posts: 2

The use of FPGAs in embedded designs is growing at a steady 3% a year but has still only reached 45% penetration. Over a third of those that dont use FPGAs at present say the trend towards FPGAs with built in multicore processors will change your mind. Do you agree or what do FPGAs have to provide to find a place in more designs?
Johan Dams
#2 Posted : Monday, September 17, 2012 9:01:37 AM(UTC)
Rank: Advanced Member

Joined: 9/16/2012(UTC)
Posts: 35

A lot of the embedded applications don't really need an FPGA, and FPGA's are not as accessible. Micro-controllers are plentiful, cheap and easy to get started with compared to FPGA's, so that's what people know and use. Tools are available free of cost and the amount of sample code available for embedded processors dwarfs FPGA code. Furthermore, the availability of embedded operating systems (and Linux in particular) makes it a much lower barrier to entry than FPGA. This means that people doing embedded as a hobby, or students, will more readily use micro-controllers and less FPGA based systems.

I think that the combination of CPU/FPGA in the same package will lower the barrier to entry for the casual user and hobbyist which hopefully spark a wide range of projects available to many (think Arduino, but with FPGA's). The combination of CPU and FPGA means people have access to something they are familiar with (CPU) while giving them the ability to learn something new (FPGA).

One thing that I personally believe should happen is for the tools to become more readily available, preferably as Open Source. They don't have to be as good as the commercial tools, or as feature-full, but should be easy enough to use to let people new to the field get familiar with it.


Johan.
David Bley
#3 Posted : Thursday, September 20, 2012 3:49:54 PM(UTC)
Rank: Newbie

Joined: 9/20/2012(UTC)
Posts: 1

I worked for a company that manufactured medical electronics. We had some products that contained FPGA's. Usually the lifetime of the design outlasted the lifetime of the FPGA as a part. When the FPGA's became obsolete, we designed then out and performed the process with a microcontroller.

They also did not fit into our engineering capabilities. We had hardware engineers and software engineers, but no FPGA engineers. We did not have time to get our software or hardware engineers up to speed due to agressive product development schedules.

The FPGA's also seem to be power hogs compared to 8 and 16 bit microcontrollers.

And last but not least, in order to use an FPGA in one of my designs, I need to tinker with a development board on my own first. This means that it needs to be something that I purchase on my own and run on my home computer and my budget for something that I am not sure that I can use.

There were just no wins using an FPGA in my particular product.
br12345
#4 Posted : Friday, September 21, 2012 10:42:54 AM(UTC)
Rank: Newbie

Joined: 9/21/2012(UTC)
Posts: 1

I have been using FPGAs for several years. The weakness in an FPGA is when I want a SPI interface I pay for the use of one or I develop my own. When I want an SDLC interface I pay for one or develop my own. When I want a SD card interface I pay for one or develop my own. I develop all of these on my own which is time consuming and error prone. I can buy a DSP or microcontroller with these built in. I just address the function. Secondly I have old code that was fine on older FPGAs that will not work on newer ones. All the issues are related to timing which you never contend with on a DSP or microcontroller. Tools for solving timing issues are expensive and require training or time to develop expertise in using them. I have developed my own soft processor on the FPGA in order to make the task easier. The FPGA development software needs to handle the timing issues more to put less strain on the developer.
Eddy_D
#5 Posted : Friday, September 21, 2012 10:46:13 AM(UTC)
Rank: Newbie

Joined: 9/21/2012(UTC)
Posts: 5

FPGAs are good for throwing a lot of tailored hardware at a processing problem, when you do not have sufficient processing resources to create a software model of that same hardware. This is true for any product that has to process a lot of data in real-time. I worked at a medical company that used 4-5 FPGAs per processing card to do digital filtering of signals. Multiple cards then aggregated their data onto a common bus to a data analysis PC. This was ~ 15 years ago and the real-time bandwidth was around 50 MBps (megabytes). Nowadays you could probably get away with using some hefty processors do do the same work, but that fact was these processors were just not there at that time.

So maybe FPGAs (for this use-case at least) would less likely be used now than say a decade ago.

- Eddy_D

I_O
#6 Posted : Tuesday, September 25, 2012 6:27:43 AM(UTC)
Rank: Newbie

Joined: 9/25/2012(UTC)
Posts: 1

Why don't I use FPGAs in my design? It is quite simple. They burn just too much power and cost too much money.

I cannot afford more than a couple of mA average operating current for the entire operational cycle of my satellite modem. This means that it must be possible to close down to a very low power state for any idle periods, and that any protocol and signal conditioning must be performed fast but with the minimum of processor power dissipation.

Oh, and the part must cost less than $3.

So far I have not seen an FPGA that can cut it.
jaybus
#7 Posted : Friday, September 28, 2012 4:27:22 AM(UTC)
Rank: Newbie

Joined: 9/28/2012(UTC)
Posts: 1

They are simply too expensive. Not just the cost of the FPGA itself, but all of the IP that must be purchased for a FPGA, when the same functionality (particularly interfaces) are included with nearly any SoC.
aurgathor
#8 Posted : Wednesday, October 03, 2012 8:27:51 PM(UTC)
Rank: Newbie

Joined: 10/3/2012(UTC)
Posts: 1

FPGAs are very nice for certain things, but not always cost effective, and as some people already mentioned, they have certain drawbacks.
I think they can be most useful when there are tasks that can and should be executed in parallel. Atmel used to make FPSLICs -- microcontrollers with an FPGA -- but I don't think they had much success with them.
Gavin.Li_#2
#9 Posted : Sunday, October 07, 2012 7:59:49 AM(UTC)
Rank: Newbie

Joined: 10/7/2012(UTC)
Posts: 1

I have been working with FPGAs for many years. The advantage is the parallel processing and high performance. It will just get the result in one clock cycle which compare to hundreds of thousands clock cycles in DSP or micro-controller software.
The drawback is it is totally different kind of programming practice, it has a long learning curve for programming, and lots of other skills need to learn to solve time issues.

For the power consumption that others pointed out. I couldn't agree with you too much. In my mind, low power always means low performance. You get high performance than you must pay more power. Just like Intel ATOM or ARM micro-processors, they are claimed to be low power, but they could never catch up the performance of Intel Dual core processors, am I right. It all depends how much performance do you need to finish your job, and then we can talk about the the power consumption. It is meaningless to talk about the power consumption without the condition to be able to get your job done.
vicky12
#10 Posted : Monday, April 15, 2013 11:25:00 PM(UTC)
Rank: Newbie

Joined: 4/15/2013(UTC)
Posts: 8

Principal / Senior / Engineers, Software Project Leader (Storage Media) (Software - IC)

Responsibilities
• Be part of the international software development team to develop embedded software for Radio, Radio-Navigation or Connectivity (Wifi, Bluetooth, GSM / 4G LTE features based) products to meet OEM customer requirements.
• Be involved through-out the software development life cycle (project planning, requirement definition and analysis, architecture analysis and definition, design, implementation, integration and validation), field-test and maintenance of the embedded software.
• Lead technical discussion, interface between internal application team and suppliers for drive modules integration into the system.
• Able to define SW Requirement, Interfaces and Testing methods to drive suppliers.
• Perform acceptance test on supplier deliverable base on define drive specification.
• Perform Design Reviews and recommend test cases to suppliers to achieve drive robustness.
• Experienced in supplier management to track SW deliveries from suppliers. • Involved in Drive Module Sourcing Activities.
• Principal and Senior Engineers will lead project team(s) in building competence for the software department.

Requirements:
• Masters / Degree in Electrical / Electronics / Computer Science or Computer Engineering with minimum 2 year's experience in embedded system development.
• Diploma holders with minimum 3 year’s experience in embedded system development are also encouraged to apply.
• Candidates with experienced in handling drive modules like CD/DVD/HDD/SSD and its SW interface.
• Self-motivated with a strong passion in embedded software development work.
• Willing to travel overseas frequently.
• Knowledge of Embedded Linux, Windows CE, uITron, mobile network protocols, shell scripts, and UML2.0 will be an added advantage.

Please send your updated profile in word format to recruit.intellect.minds@gmail.com
vicky12
#11 Posted : Monday, April 15, 2013 11:26:04 PM(UTC)
Rank: Newbie

Joined: 4/15/2013(UTC)
Posts: 8

Principal / Senior / Engineers, Software Project Leader (Storage Media) (Software - IC)

Responsibilities
• Be part of the international software development team to develop embedded software for Radio, Radio-Navigation or Connectivity (Wifi, Bluetooth, GSM / 4G LTE features based) products to meet OEM customer requirements.
• Be involved through-out the software development life cycle (project planning, requirement definition and analysis, architecture analysis and definition, design, implementation, integration and validation), field-test and maintenance of the embedded software.
• Lead technical discussion, interface between internal application team and suppliers for drive modules integration into the system.
• Able to define SW Requirement, Interfaces and Testing methods to drive suppliers.
• Perform acceptance test on supplier deliverable base on define drive specification.
• Perform Design Reviews and recommend test cases to suppliers to achieve drive robustness.
• Experienced in supplier management to track SW deliveries from suppliers. • Involved in Drive Module Sourcing Activities.
• Principal and Senior Engineers will lead project team(s) in building competence for the software department.

Requirements:
• Masters / Degree in Electrical / Electronics / Computer Science or Computer Engineering with minimum 2 year's experience in embedded system development.
• Diploma holders with minimum 3 year’s experience in embedded system development are also encouraged to apply.
• Candidates with experienced in handling drive modules like CD/DVD/HDD/SSD and its SW interface.
• Self-motivated with a strong passion in embedded software development work.
• Willing to travel overseas frequently.
• Knowledge of Embedded Linux, Windows CE, uITron, mobile network protocols, shell scripts, and UML2.0 will be an added advantage.

Please send your updated profile in word format to recruit.intellect.minds@gmail.com
openmcu123
#12 Posted : Tuesday, May 28, 2013 12:58:53 AM(UTC)
Rank: Newbie

Joined: 5/28/2013(UTC)
Posts: 1

colin.holland;149 wrote:
The use of FPGAs in embedded designs is growing at a steady 3% a year but has still only reached 45% penetration. Over a third of those that dont use FPGAs at present say the trend towards FPGAs with built in multicore processors will change your mind. Do you agree or what do FPGAs have to provide to find a place in more designs?



hi, maybe you can try this product.

http://cgi.ebay.com/ws/e...ssPageName=STRK:MESE:IT

I bought from this store and it works well. Also I think you can have a try!
Edgar Duarte
#13 Posted : Wednesday, June 05, 2013 11:20:53 AM(UTC)
Rank: Newbie

Joined: 6/5/2013(UTC)
Posts: 1

I dont't have much experience with both, but i have developed solutions in the two plataforms (MCU's and FPGA's) and I just think that it's not a fair comparison. Both have their advantages and disadvantages as every product. But my point is that, FPGA's are really good at tasks that MCU's don't even dream of being capable of. So it really depends in what you are trying to accomplish. If you are a hobbyst trying just a simple solution, or if you are developing something that doesn't require large amount of proccesing of course you are going to preffer a MCU. But if you are programming some big 32 bit MCU, or DSP, and you get out of juice, Maybe its better to invest more time and money and make a nice solution in the FPGA. FPGA's are complicated, and not very intuitive, even for really technical people. So maybe as the old architectures, of microprocessors, gave birth to all kinds of MCU's, maybe FPGA's are going to grow more in a few years, just think about image and video processing, or some AI solutions that require really huge amount of data processing, are going to extend FPGA's market in the future. Greetings.
Users browsing this topic
Guest
Forum Jump  
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.

Powered by YAF 1.9.5.5 | YAF © 2003-2011, Yet Another Forum.NET
This page was generated in 3.915 seconds.