Cadence C-to-Silicon Compiler eliminates barriers to HLS adoption

Cadence Design Systems Inc. recently announced its C-to-Silicon Compiler, said to be the next-generation of HLS (high-level synthesis) technology.

The C-to-Silicon Compiler is said to eliminate historical barriers to HLS adoption to deliver the quality of results and net productivity gains engineers need. It also produces RTL (register transfer level) with quality at or above the 90th percentile of manual RTL design, while increasing the engineering productivity up to 10X. HLS incidentally, reduces the manual effort required to produce RTL, thereby enabling designers to avoid syntax errors common in traditional methodologies.

I was very fortunate enough to be able to speak directly with Steve Svoboda, marketing director for system level design products, Cadence, in the US, last evening, on the C-to-Silicon Compiler.

According to Svoboda, this tool can accurately predict timing estimates. Logic synthesis ability is embedded into the tool. Cadence logic sysnthesis has been embedded inside HLS. HLS transforms C and C++ into RTL.

What can this product actually do for the EDA industry? He says it can actually take EDA up to a new level in terms of delivering additional productivity to designers.

"When design compiler and logic synthesis came, it was during the golden era of the semiconductor industry. Productivity was increasing rapidly. But the problem is, since the early 1990s, there has been no real change in the RTL design methodology. The only productivity increase has come out in form of design re-use," he says.

"This (C-to-Silicon Compiler) could re-energize semiconductor and EDA industries by at least 10X times. About 20 years ago, there was 10X productivity increase. By having HLS, we can now close the gap and tackle the chips more effectively now."

So, first up, will C-to-Silicon Compiler compete with custom design projects? Svoboda it won't! Custom design projects typically utilize transistor-level design. C-to-Silicon is made to work within a standard ASIC design-flow.

Accelerate and improve verification
The C-to-Silicon Compiler will both accelerate and improve verification as well. The timing-approximate fast hardware models (FHMs) run 80-90 percent the speed of untimed C-models (or two-three orders of magnitude faster than RTL). This enables the hardware-software co-verification with greater timing accuracy.

The next question is: can people use third-party synthesis tools, along with the proprietary Cadence systhesis tool? Svoboda says that the C-to-Silicon Compiler outputs IEEE-standard Verilog RTL. Therefore, the output can go to any third-party synthesis tool. However, as the RTL output is generated using timing estimates from Cadence RTL Compiler, designers will get the best quality of results when using RTL Compiler for logic synthesis.

Will C-to-Silicon Compiler better predict performance and power? And if yes, has this cracked the low-power design issue? Svoboda adds that because of embedded logic synthesis, the C-to-Silicon Compiler can predict performance and (in principle) power better than other high-level synthesis tools.

He says: "Power estimation/optimization are key feature sets planned for upcoming releases of C-to-Silicon Compiler. We believe that those capabilities will enable the designers to create designs that are much better optimized for power, since design decisions with greatest power impact are made at the system-level."

Finally, how does C-to-Silicon compiler handle hardware allocation and scheduling operations? The answer is, C-to-Silicon Compiler handles hardware allocation and scheduling using various proprietary algorithms and heuristics. Many of these are based on previous research at Cadence Berkeley Labs.

Svoboda notes: "One should note that the better quality of results/performance of C-to-Silicon is due primarily to its inherent ability to generate more accurate timing-estimates than other HLS tools. The higher accuracy timing estimates result from the embedding of logic synthesis within the HLS tool/process, which enables gathering of full-context gate-level information to derive the timing estimates.

"Other HLS approaches rely on pre-characterization of technology libraries, which is not accurate enough, because those gate level estimates are only nominal values, and do not take into account the full-context of the design (fan-in, fan-out, buffers, etc.)"

Lastly, what happens to ESL (electronic system-level) tools? He believes that this tool will help the ESL market.

Svoboda says: "We now have a methodology to do design creation in C++ and SystemC. For example, they do virtual prototyping, hardware-software co-design, etc. In the past, when engineers created designs, they had to re-design in C++, etc. Our tool creates the RTL automatically for them. So, this could re-energize the ESL market very well."

It will be interesting to see what the other EDA firms such as Synopsys and Magma have in store!