Tales from the Frontlines: Startup War Stories
summary
A dozen startup war stories across 15–20 years show concrete wins and failures: building a 32‑register RISC JIT and hardware-enabled JVM that produces the lowest-latency GC on the planet, shipping boxes with 48–384 cores after silicon revisions, creating an exact distributed KV store that benchmarks as the fastest KV store, and delivering a parallel GBM/GLM that runs 7 TB in 90 seconds on 100 nodes. Alongside those wins are missed payrolls, fraud from rogue sales spending, product pivots, and team breakups.
AI-written summary · about
chapters (12)
- 0:00Startup war stories
- 1:27Why startups are thrilling and brutal
- 2:29Azul Systems hardware-software clash
- 4:53Debugging silicon bugs and latency
- 7:20Java memory model pitfalls
- 8:25Founding RAIN and distributed key-value store
- 9:47Pivoting to machine learning tools
- 12:37Growing pains of data startup
- 15:46Team dysfunction and leadership failures
- 17:31Catching fraud with faster data exploration
- 19:13Financial mismanagement and payroll disaster
- 20:55Startup lessons and next adventures
related talks
01h 16mA New Golden Age for Computer Architecture
24mKEYNOTE: RISC-V: Enabling a New Era of Open Data-Centric Computing Architectures
45mHackathon Presentations
01h 15mThe Evolution of Memory Technologies
01h 21mPanel+Closing
19mHow NOT To Program an Out-of-order Vector Processor
01h 20mExtending Moore’s Law with Low-Cost Memory Reliability
17mRISC-V is Ready for Powering the Era of Intelligent General Computing