Reusable Coding
Hi Dhananjayan,
Good topic. Looking forward for the next part.
Re: Good way to start
@JezmoSSL: I stand corrected -- given a choice, I woudl use both per review AND having a different person creating the text bench to the one creating the design :-)
JezmoSSL
7/30/2012 12:15:11 PM
User Rank Blogger
Re: Good way to start
Its not quite the onliest way to go, independant verification is fine if you have loads of money and time and resources and you are designing for high reliability, otherwise peer review is perfectly acceptable.
You shouldnt go giving people the idea that there is only one acceptabkle review process Max.
Re: Good way to start
@Adam: "...who writes the code does not write the test bench..."
YES!!! The problem is that if you are coding a design and you miss-read the spec / requirements, then you implement something incorrectly, and then you write the tests with the same missunderstanding in mind. Haveing someone else write the testbench really is the only way to go...
Re: Good way to start
@JezmoSSL: That's a really good point. When I started designing, we didn't have any automated tools, so all parts of the design went through a peer review -- even when the design was good I learned "stuff" from listening to the other engineers debate alternative implementation possibilities...
jacklsw86
7/29/2012 4:11:28 AM
User Rank Blogger
Re: Good way to start
@JezmoSSL, deja vu. At my side, looking at the firmware guy's Simulink design on a floating point division module part, it only registers the output and expects everything to be done within 1 clock cycle??? Thank god we rewrite the codes and use the floating point division from coregen (and set appropriate latencies). There's no way I can meet the timing closure if I was to use the guy's Simulink design as a blackbox module.
JezmoSSL
7/29/2012 2:12:16 AM
User Rank Blogger
Re: Good way to start
I recently had to work on some code produced by a software guy and he couldnt get it to work, I had a look at it and it didnt work because it didn't meet timing, it had asyncronous processes all over the place and so on so I had to rewrite it, and someone else might have sat there not knowing how to get it working.
Balkinization of coding styles at large companies
Have most of you noticed that large companies that have IP libraries that have acquired many smaller companies have a code base that contains multiple coding standards and thus is "Balkinized". This can mean significant re-work or adding interface layers or other work around code to the project.
Karl
7/28/2012 11:04:35 AM
User Rank Guru
Re: Good way to start
@JezmoSSL: I am a strong believer in analogies in various engineering fields and that the current approaches rely on developing a new language by computer science guys and that will never work.
As a hyardware systems guy who has learned a bit of programming in order to survive, I believe that programs and state machines are analogs and that makes me think that the control and data flow of a program can be parsed and implemenmted as a state machine.
What concerns me a little is that RTL/HDL(whatever you want to call it) still has to go through the FPGA tool chain and verification processes.
There may be need to implement only the critical algorithms in HDL.
JezmoSSL
7/28/2012 9:50:01 AM
User Rank Blogger
Re: Good way to start
This is the problem with letting software people loose on FPGAs, software vendors would like people to think that you don't need to have any FPGA experiance in order to port software to an FPGA, and people without any real idea of what they are doing sing the praises of SystemC et al.
The fact is it is never going to be possible to port any software to an FPGA with a single software tool, no matter what the sales guys would like you to think.
|
 |
Would you class these as adages, aphorisms, axioms, dictums, epigrams, maxims, precepts, saws, truisms, or... well, what?
Here we discover how to use the XADC (Xilinx Analog-to-Digital Convertor) in the Zynq All Programmable SoC to read the chip's internal temperature and voltage parameters and output them over an RS-232 link.
When extreme thermal cycling causes circuit boards and chip packages and the silicon die in the packages to expand and contract at different rates, problems may ensue.
In part 3 of this epic tale we consider how we might use tri-state buffers, leading up to the legendary bi-directional buffer.
Digital engineers are often confused among operational amplifiers, differential amplifiers, and instrumentation amplifiers; this is exacerbated by the fact that their circuit symbols can be similar.
|
Tweet This
