Semiconductors
Releases that wait on hand-run integration and test results
In machine software, integration speed and test reliability decide when you can deliver. We turn hand-run integration into a pipeline your team starts itself, with a check after every stage, in the tools you already run.
- Your tools stay
Release 6.1
Build 418, started by a developer at 08:14
What slows releases
Where the time goes between merge and machine
Integration is a checklist
Software from several teams comes together in one build. The integration steps repeat on every release, and an engineer still runs them by hand, in the right order.
Test time is scarce
Test rigs and machines are shared and booked ahead. When a run fails on its setup instead of on the software, the slot is gone and the feedback waits for the next one.
Ready is a judgment call
Results sit in different environments and tools. Before each release, someone gathers them by hand to answer one question: is it ready?
What changes
The same release, without the hand-run steps
A Self-Service Release Build automates the release flows that cost you the most, end to end. Plan on 4 to 8 weeks. The second half of the fee is due when the criteria you signed pass in your own environment.
-
TodayAn engineer runs the integration steps by hand, in order, for every release.
After a BuildHand-run steps and handoffs become one pipeline per release flow, started by the team.
-
TodaySomeone checks the results, environment by environment, before the release can go.
After a BuildEvery stage checks its own result, and rollback is tested before go-live.
-
TodayOnly a few people know the integration procedure end to end.
After a BuildWhat they know becomes a runbook, a trained team, and a pipeline your team has already changed on its own.
-
TodayRelease approvals happen in meetings and email, apart from the build.
After a BuildThe approvals your policy requires run inside the pipeline, and every production release leaves a change record your change board can check.
Results
What changed at ASML
ASML builds lithography systems, where integration speed and test reliability decide delivery. We built self-service CI for the Stage Position Measurement domain, and one reporting view to judge release readiness.
| ASML | Before | After |
|---|---|---|
| Integration steps | run by hand | one self-service CI pipeline |
| Release readiness | no single place to judge it | one Python and SQL reporting view of production machine test results |
| CI across teams and test environments | not standardized | standardized CI flows for multiple teams |
| Production machine tests | on ASML’s own test environments | offload of selected tests to Google Cloud Platform started |
| Engineering time on this work | 300+ hours a year back | |
“Mateusz created designs for automation of work processes, managed external dependencies and solved tooling prerequisites needed to implement the designs, build automations from the designs in Groovy (Jenkins pipeline scripts), Python and Shell script (Bash), tested the automations at different proprietary ASML test environments and worked at Linux systems in Git and ClearCase based development environments.
He gained experience about integration procedures and configuration management as applied in a large scale company. It was a pleasure for the team and me to work with Mateusz and I wouldn’t hesitate to hire him again.”
Reading
More on software delivery in semiconductors
- ArticleThe SDLC Reckoning in Semiconductor Engineering: Why Delivery Discipline Now Decides Who ShipsWhy AI assistance, toil, attacks on chip IP and the EU Cyber Resilience Act make delivery discipline the deciding factor for semiconductor teams.Read the article
- ArticleHow Software Development Drives Innovation in Semiconductor ManufacturingWhere software moves yield, uptime and design cycle time in a fab, and why the production system around a model matters more than the model.Read the article
- ArticleFive Ways Cloud Computing Benefits the Semiconductor IndustryElastic compute for design and verification, shared design data across sites, confidential computing for IP and pay-per-use costs, with the trade-offs named.Read the article
Where to start
Start with your last release
The free Release Check shows where your last release spent its time and what that costs a year, in under 5 minutes and without a call.
Contact
Talk to the engineer
Book 30 minutes and bring your last release, or send a message. You talk with Mateusz directly.
