← All services Service 05 / Design Rescue

For the problem your team has been stuck on for weeks.

Timing that will not close. A design that will not fit, or fits only on a part that costs too much. A board that works four times out of five and nobody can say why. Send us something we can reproduce and we will find the cause, then quote the fix.

Paid assessment, credited against the fix →
Timing closure Speed grade migration Size reduction CDC & metastability Intermittent failures Root cause report
rescue / engagementStages
Symptoms and reports reviewedIntake
Failure reproduced on our sideGate
Root cause identifiedAssessment
Fix quoted, scope agreedQuote
Fix implemented and verifiedDelivery
no reproduction, no engagement, no charge beyond intake
What we take on

Four kinds of stuck.

All four start the same way: an existing design, a problem your team has already spent real time on, and a target you have to hit.

Track 01

Timing Closure

The design will not meet its target frequency. We find the paths that actually dominate your slack, and fix them through retiming, pipelining, or restructuring rather than constraint tricks that hide the problem.

Critical paths Retiming Constraints
Track 02

Cost Reduction

The design fits and works, but the part is more expensive than it needs to be. Move it to a slower speed grade and still close timing. On volume production this is the highest-return work we do: the saving is per unit, every unit, for the life of the product.

Speed grade migration BOM reduction
Track 03

Size Reduction

The design does not fit. Over utilisation, a placer that gives up, or a router that never converges. We cut LUT, FF, BRAM, and DSP usage through resource sharing and restructuring until it fits the part you already committed to.

LUT & FF BRAM inference DSP sharing Routing congestion
Track 04

Stability & CDC

It works, mostly. It fails on one board, or at temperature, or once a day with no pattern. Usually a clock-domain crossing, a reset that is never properly released, or a path that passes static analysis and still loses the race.

CDC Metastability Reset & recovery Race conditions
The one hard requirement

We have to be able to reproduce it.

This is the whole engagement in one sentence. A failure nobody can trigger on demand cannot be debugged, only guessed at, and you have already paid people to guess. Before we quote anything, we establish that the problem happens in front of us.

There are two ways to get there, and either is fine.

  • A
    Send us the setup. The board, the test case, the stimulus, and the procedure that makes it fail. We reproduce it on our bench and work from there.
  • B
    Give us remote access. Your rig stays where it is. We connect to it, run the failing case ourselves, and instrument from our side.

If neither is possible yet, the first piece of work is making the failure reproducible. That is a real engagement on its own and often the hardest part.

What we need from you

Intake checklist.

  • 01
    Target part and speed grade. Exact device, package, and grade. For cost reduction, the grade you want to reach.
  • 02
    Current utilization. LUT, FF, BRAM, and DSP usage from your latest implementation run.
  • 03
    Timing numbers. Current WNS and TNS, your target Fmax, and the failing paths if you have them.
  • 04
    Failure symptoms. What goes wrong, how often, and under what conditions. Temperature, voltage, one board or all of them.
  • 05
    What you have already tried. Saves us repeating a week of your team's work, and tells us a lot about where the problem is not.
  • 06
    A project that builds. Sources, constraints, and toolchain version, building cleanly end to end on your side.
How an engagement runs

Diagnose first, quote second.

We do not quote a fix for a problem nobody has root-caused yet. Anyone who does is guessing at your expense.

01

Describe it

Symptoms, reports, and what your team has already ruled out. Free, and usually enough for us to say whether we can help.

02

Reproduce

You send the setup or open remote access. We confirm the failure happens in front of us before anything else.

03

Assessment

Fixed price, time-boxed. We root-cause the problem and write up what is actually wrong.

04

Fix quote

Now that the cause is known, a fixed price for the fix. The assessment fee comes off it.

05

Fix and verify

Implemented, verified against your environment, and handed back with the reasoning documented.

Pricing

Assessment first, then a fixed price.

There is no price list, because the price depends on what is actually wrong and nobody knows that until it has been root-caused. The first conversation is free. After that you pay a fixed, time-boxed assessment fee to find the cause, typically $2,500 to $4,000 for timing, cost and size work, and from $9,999 for stability and CDC where reproducing the failure is most of the effort. That fee comes off the fix if you go ahead.

Timing Closure
From $2,500
Design will not meet its target Fmax
  • Critical paths identified, not guessed
  • Retiming, pipelining, restructuring
  • Constraints reviewed and corrected
  • Assessment fee credited against the fix
Request a quote
Size Reduction
From $2,500
Design will not fit the target device
  • LUT, FF, BRAM, and DSP reduction
  • Resource sharing and restructuring
  • Fits and routes on the part you have
  • Assessment fee credited against the fix
Request a quote
Stability & CDC
From $9,999
Intermittent or unexplained failures
  • CDC, metastability, reset and recovery
  • Reproduction required before we quote
  • Written root cause, not a guess
  • Assessment fee credited against the fix
Request a quote
Why it matters

What being stuck actually costs.

Schedule, not hours

A design that will not close is not costing you engineer time. It is costing you a ship date, and everything downstream of it.

Per-unit savings compound

A speed grade migration saves money on every unit for the life of the product. On volume, it pays for itself many times over.

Intermittent means shipped

A failure that happens rarely gets through test and reaches customers. Finding it before that is far cheaper than a field recall.

Fresh eyes, no history

Your team knows the design too well to see past their own assumptions. That is not a criticism, it is why escalation works.

Pairs well with

Before and after a rescue.

Review finds problems early. Verification stops them coming back. Rescue is for when one already got through.

Tell us what is going wrong.

Symptoms, what you have tried, and the part you are on. The first conversation is free and we will tell you honestly whether we can help.

Contact us →