( DAC'18 Item 2c ) ------------------------------------------------ [12/14/18]
Subject: Siemens/MENT InFact fails in the PSS Wars is #2c for Best of 2018
RAVI DROPPED THE BALL: Yes, Ravi Subramanian may have kicked some serious
ass this year with his Solido machine learning taking #1 Best of 2018 with
the users -- but the Ravi Subramanian, who's the VP & GM of All Digital IC
Validation at Siemens/MENT seriously messed up with PSS and InFact.
PSS Tool Word Counts in DeepChip "Best of" 2017 vs. 2018 EDA User Survey:
Cadence Perspec 2017: ############################### (3,097 words)
Cadence Perspec 2018: ##################################### (3,712 words)
Mentor InFact 2017: . (24 words) <-- damn it!!!
Mentor InFact 2018: #### (437 words) <-- double damn it!!!
Perspec & Breker have sucked up all the user mindshare air in the PSS Wars;
and those 3 users who talk about InFact are praising InFact for being a good
POINT TOOL -- plus they openly doubt InFact being PSS at all! D'oh!
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
QUESTION ASKED:
Q: "What were the 3 or 4 most INTERESTING specific EDA tools
you've seen this year? WHY did they interest you?"
---- ---- ---- ---- ---- ---- ----
At [name deleted] we had the opportunity to test out InFact from Mentor.
We found the results quite fascinating!
There were two separate problems we were interested in solving.
1. Fast Coverage Closure: We often come across the situation where
it is quite straight forward to cover 97-98% in Verilog simulation.
The last 2% takes a lot of effort and we end up merging many days
of coverage to get to it. This becomes especially cumbersome if
there is a late RTL bug that needs to be fixed and the coverage
re-closed.
We successfully used InFact to address this issue. InFact could
steer randomization in our Questa testbench with feedback from
Questa coverage data. This helped us reach 100% coverage in the
shortest time, consistently.
2. Vertical Reuse of Stimulus: we have separate teams for doing "unit"
level verification and "system" level verification. Most of our
units have a bunch of registers that can be configured differently.
Most of the registers have inter-dependency and legal limits. We
capture all those in UVM constraints. Later, when we move to
System Level Verification we have to re-do the same constraint
solver in C-code since the tests are written in C, to be run in
one of the CPU cores. This was quite inefficient.
Using InFact we could code our constraints in PSS -- and implement
our executable in C or UVM transactions. This eliminated the need
to write constraints twice.
Also whenever there was a modification in the spec, it was easy to
translate that to both the C and UVM testbenches.
For those two reasons, we're big InFact fans here.
---- ---- ---- ---- ---- ---- ----
The most exciting stuff in the EDA world as I see is the drive for PSS.
The concept is very intriguing and once adopted could save a lot of
cycles for SOC projects and reuse will be at a different level.
I've sat through some of the DAC presentations and later evaluated
Perspec vs. Breker vs. InFact. What I found:
- InFact meets the basic definition of Portable Stimulus in that it
lets you create reusable test scenarios on multiple platforms.
- InFact is more than just a Portable Stimulus tool. It accelerates
coverage closure by automatically generating coverage models from
your test specification and targets coverage in its test generation
to reach closure more quickly.
- After reaching coverage, InFact's Bug Hunting mode randomizes
everything to give you the benefits of constrained-random at the
system level.
- InFact's Constraint Explorer lets you make sure your constraints
are correct before you simulate and lets you do what-if exploration
of your constraints.
Questa InFact is what I would strongly suggest to folks out there.
---- ---- ---- ---- ---- ---- ----
We use InFact. It's not PSS. It's useful, but it's not PSS.
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
Related Articles
User buzz on Siemens/Solido machine learning is #1 for Best of 2018
CDNS Perspec is crushing it in the PSS Wars is #2 for Best of 2018
Breker TrekSoc is slipping in the PSS Wars is #2b for Best of 2018
Siemens/MENT InFact fails in the PSS Wars is #2c for Best of 2018
Again Spies hint Synopsys is making a PSS tool, but it's slideware
Join
Index
Next->Item
|
|
