( ESNUG 488 Item 11 ) ------------------------------------------- [03/25/11]
Subject: Mentor Bluebook contest has 2 winners out of 1,144 participants
> Long story short, Mentor shanghaied me into being a guinea pig for
> their upcoming "Test Your C-IQ" multiple-guess contest.
>
> The first version of the quiz was bad. I stopped after 11 minutes
> because of so many Mentor CatC-is-GREAT Marketing questions in it.
> I was disgusted and let them know it.
>
> Second version was 11 questions that took 14 minutes. Got a 78% score
> on that. It still had two Marketing CatC-is-GREAT Marketing questions.
>
> Eventually they FINALLY cut the marketing Q's, so it's a quick 4 minute
> true ENGINEERING ONLY test for C-based design with only 5 questions.
> Check it out:
>
> http://www.surveymonkey.com/s/NWQL9BZ
>
> Everyone who takes the quiz is entered into a drawing for 40 Blue Books,
> plus the top scorers are put in a drawing for an iPod Touch 4.
>
> - from http://www.deepchip.com/wiretap/101213.html
From: Thomas Bollaert <thomas_bollaert=user domain=mentor not mom>
Hi John,
Thank you for bringing the Mentor Catapult C High Level Synthesis Blue Book
contest to your readers' attention.
The total community response was extremely strong, with 1,144 participants.
We actually had so many top scorers we decided to run a 2nd "semi-finals"
round. Everyone who scored 80% or more was invited to participate. Even
then, there were still so many top participants and top scorers that we ran
a 3rd "finals" round.
It was impressive to see the general level of understanding of HLS.
Below are the stats on how this contest played out, plus I will share our
quiz answers and explanations at the end. If you recall, John, your own
score on our final Round 1 quiz was 78%.
After Round 1, we gradually upped the stakes and increased the difficulty
level. The 3rd round of questions was definitely the hardest.
Round 1 Round 2 Round 3
Participants: 1,144 320 106
Average score: 73% 72% 48%
# scored 80%+: 715 182 23
# scored 100%: 362 84 2
As you can see, our contest ultimately had two winners:
- a hardware engineer in Europe who asked to be anonymous
- Diego Sanchez, a research assistant in Europe
Congratulations to both of them! They each got perfect scores on all three
rounds, and each won an iPod Touch 4.
Another interesting piece of data was the broad distribution of participants
that made it to the
Participants who made Round 3 and got an 80% score:
North America : : ####################################### 39%
Europe : : ####################################### 39%
Rest of World : : ###################### 22%
Below is the link to the original contest, along with the answers and
explanations.
http://www.surveymonkey.com/s/NWQL9BZ
Question 1. Answer: A basic shift register.
Explanation: HLS analysis of the array access pattern can prove that
the current array element is copied from the previous
element (regs[i] = regs[i-1]) with regs[0] assigned a
new value each time the code executes, leading to the
inferencing of a shift register.
Question 2. Answer: The process of replicating the loop body and
trying to execute two or more loop iterations in parallel.
Explanation: Loop unrolling is used to reduce design latency by
increasing parallelism.
Question 3. Answer: Starting a new loop iteration before the current
one completes.
Explanation: Loop pipelining is used to improve design performance
by improving throughput.
Question 4. Answer: 5 bits.
Explanation: The loop has 20 iterations, so the loop counter must be
able to count to 20. Thus a 5-bit counter is needed.
Question 5. Answer: Two adders.
Explanation: The code requires that 3 additions are computed every
two clock cycles. This can be achieved with two adders
by sharing one of them across clock cycles.
Thanks again to everyone that participated, and congratulations to our 2
winners!
- Thomas Bollaert
Mentor Graphics Corp. Wilsonville, OR
Join
Index
|
|
