Which FPGA prototype should advance?
Bramble Devices is comparing two FPGA-based camera trigger controllers with the same external interfaces.
The question
The product lead needs a response within 200 ms in every required sampled workload and an estimated electronics cost at or below $12 per unit. The team must check both short and sustained bursts.
| Typical short-burst response | Electronics cost per unit | |
|---|---|---|
| Prototype A | 150 ms | $10 |
| Prototype B | 165 ms | $11 |
These are model and cost estimates. The model has no sustained-burst result.
“Recommend one design for the next pilot and explain what the evidence covers.”
Not saved by this site. Your browser may restore text when you navigate back.
A strong answer, step by step
Open each step in turn. This is one strong answer, not the only one.
Step 1: What does the model support?
What would you do here? Open the move you’d make.
Advance A based on its lower cost and modeled response
Tempting, but it misses something. The estimate gives no measured response under the required workloads.
Measure both designs under the required workloads
The strong move. The model guides the comparison. The next decision needs physical evidence.
Treat both designs as 50 ms or more under the limit
Tempting, but it misses something. B’s modeled margin is 35 ms, and neither estimate gives a measured maximum.
See how a strong candidate says it
“A is the lower-cost candidate and has the faster modeled response. I would compare both physical prototypes against the requirement before choosing one.”
Why it works: The model gives a typical response for short bursts. The requirement applies to each sampled workload.
Step 2: What did the short bursts prove?
The planned build configuration of each design ran 12 matched short-burst sessions. All sessions passed. Maximum responses were 172 ms for A and 188 ms for B.
What would you do here? Open the move you’d make.
Keep both candidates for the sustained test
The strong move. Both met the short-burst limit. The sustained workload still needs evidence.
End the comparison because A had the larger margin
Tempting, but it misses something. A’s margin applies only to the short workload.
Average the maximum responses and report one margin
Tempting, but it misses something. The decision applies separately to each design and workload.
See how a strong candidate says it
“Both designs passed 12 of 12 short sessions. A had 28 ms of margin at its slowest response. B had 12 ms. I would keep both for the required sustained comparison.”
Why it works: The short sessions provide measured results for one workload.
Step 3: Which design fits the new record?
The same two configurations ran 12 matched sustained-burst sessions each. A passed 9 of 12, with a maximum of 214 ms. B passed 12 of 12, with a maximum of 193 ms.
What would you do here? Open the move you’d make.
Choose A because its short-burst result is better
Tempting, but it misses something. Three sustained sessions exceeded the requirement.
Advance B to the next pilot
The strong move. B meets the measured workload limit and the estimated cost ceiling.
Release B for full production because all its sessions passed
Tempting, but it misses something. The result covers the measured samples. Planned build variation still needs a check.
See how a strong candidate says it
“I would advance B to the next pilot. A passed 9 of 12 sustained sessions and was up to 14 ms over the limit. B passed 12 of 12 and had 7 ms of margin. Its estimated cost is $11, or $1 more than A.”
Why it works: B meets both supplied requirements in the tested configurations. The sustained data reverses the ranking for the next pilot.
Step 4: How do you brief the product lead?
What would you do here? Open the move you’d make.
Show the measured scope and name the next check
The strong move. The lead can act on the pilot recommendation and see what remains open.
Report a zero population failure rate for B
Tempting, but it misses something. The measured sample gives no estimate of a zero failure rate across production.
Give only the new maximum and leave out the cost
Tempting, but it misses something. The product lead needs both requirements and the cost difference.
See how a strong candidate says it
“I would give the lead the 200 ms limit, both measured workloads and B’s $1 cost premium. I would recommend B for the next pilot and ask the pilot lead to check the planned build variation.”
Why it works: The recommendation states its scope and identifies the next evidence the lead needs.
The answer, in short
- Decision
- Recommend B for the next pilot.
- Because
- B met the latency limit in both sampled workloads and stayed within the electronics cost ceiling.
- What would change it
- A new measured miss on B or a cost above the ceiling would reopen the choice.
Common traps
- Compare maximum response with the stated limit.
- Keep a sample result within its measured scope.
Check your answer
- I identified the two requirements.
- I scoped the short test.
- I used the sustained result.
- I named the next evidence the lead needs.
Last reviewed Nov 11, 2026