ESL tools like Catapult and C-to-Silicon

Hi,

I've heard about ESL tools like Mentor Catapult C and Cadence C-to-silicon.
As far as I read, they are very powerful and mature.

Also, Mentor Graphics website, they pretend that it has unmatched efficiency. Even Ericson developers couldn't make any closer to its results - tis is mentioned in the user testimony on mentor's website-. The same for Alcatel and many companies.

My question is: if so, why digital design companies still use veriolg or vhdl for their design? Why not using ESL tools directly? It's easier, shortening development time.

I know that these tools convert first to verilog or vhdl first.
My question in a modified way: why to start from the RTL level, not ESL ?

The problem with tools like this is that it requires the same engineering ability as you need for HDL. If you write badly designed C from a hardware perspective, you are going to get a bad design. This is NOT a tool that allows C developers to create hardware, this is a tool that allows hardware developers to write C and convert it to hardware.

Then, because you need hardware developers, who probably already know HDL, the only difference is the cost of the tool.
Why should we spend $1000s on a tool that we dont need?

---------- Post added at 09:44 ---------- Previous post was at 09:43 ----------

Also, afaik, Ericson are not in the silcone market - they make PCBs.

I understand your point. using these tools needs an HDL developer.

But what about a company using these tools? It can provide them with a better time to market. And in such a competitive market, this maybe a great advantage, especially if I'm going to get a satisfying and an optimized design. That's why companies may want to invest money in such tools.

Am I right or missing something ?

---------- Post added at 12:49 ---------- Previous post was at 12:49 ----------

I understand your point. using these tools needs an HDL developer.

But what about a company using these tools? It can provide them with a better time to market. And in such a competitive market, this maybe a great advantage, especially if I'm going to get a satisfying and an optimized design. That's why companies may want to invest money in such tools.

Am I right or missing something ?

You have to remember, it is just a tool, and that it all it is. They may sell it as a magic bullet, and claim better time to market, but as with all tools, on their own, they cannot help. A tool used badly will still have the same problems as if you didnt have the tool at all. I think if you see this tool as a solution to engineering problems, you are trying to solve your problems in the wrong way. This is not an engineering tool, its just a code converter. Without a good front end design and a large amount of verfication, it wont help.

Ive seen exactly the same sales pitch from mathworks and their HDL coder tool (generating HDL from simulink). It really only works in situations where you dont know what FPGA target you want and all you want to do is number crunching. When it comes to control logic and size/speed optimizations its very poor, and very hard to work with.