Fixing concurrency defects in multicore design
Paul Anderson, GrammaTech Inc.
EETimes (2/7/2012 2:11 PM EST)
Having reached the limits of performance gains that can be realized from miniaturization and integration, microprocessor manufacturers are more and more frequently choosing multicore processors as the best approach to increased performance in computationally intensive embedded systems.
However, realizing the potential performance gains with multicores has a cost. Software written to be single-threaded for a single core processor will realize little or no performance benefit when executed on a multicore processor. To take advantage of the cores it must be rewritten or adapted to use multithreading.
To read the full article, click here
Related Semiconductor IP
- RVA23, Multi-cluster, Hypervisor and Android
- 64 bit RISC-V Multicore Processor with 2048-bit VLEN and AMM
- NPU IP Core for Mobile
- RISC-V AI Acceleration Platform - Scalable, standards-aligned soft chiplet IP
- H.264 Decoder
Related White Papers
- MEMS -> Defects present hurdle to volume production
- Multi-core multi-threaded SoCs pose debugging hurdles
- Realising the Full Potential of Multi-core Designs
- Realising the Full Potential of Multi-core Designs
Latest White Papers
- QiMeng: Fully Automated Hardware and Software Design for Processor Chip
- RISC-V source class riscv_asm_program_gen, the brain behind assembly instruction generator
- Concealable physical unclonable functions using vertical NAND flash memory
- Ramping Up Open-Source RISC-V Cores: Assessing the Energy Efficiency of Superscalar, Out-of-Order Execution
- Transition Fixes in 3nm Multi-Voltage SoC Design