CoRecursive: Coding Stories Tech Talk: Modern Systems Programming And Scala Native With Richard Whaling
21 snips
Feb 22, 2019 Richard Whaling, author of Modern Systems Programming with Scala Native, shares his unique insights on software development, focusing on low-level system programming. He discusses how to build faster software using glibc and modern techniques like manual memory management with raw pointers. The conversation highlights the importance of optimizing performance for data processing, the challenges of integrating Scala with C, and navigating software dependencies with a self-sufficient approach. Whaling also addresses modern systems programming challenges and the potential of emerging languages.
AI Snips
Chapters
Books
Transcript
Episode notes
Manual Memory Boosts Performance
- Native code avoids runtime GC overhead by manually managing bulk data memory.
- This can yield superlinear performance improvements for large data sorting compared to JVM.
Google Ngram Example in Scala Native
- Google Ngram data processing showed Scala Native excels at handling multi-gigabyte files.
- Reading lines into a single buffer avoids frequent GC and allocations unlike JVM.
Off-heap Memory for Distributed Systems
- JVM-based data systems like Spark struggle with GC pauses which can break cluster coordination.
- Scala Native offers approachable off-heap memory management that could improve data-intensive distributed apps.
