RISC-V Enhanced Trace Encoder
A complete trace solution for RISC-V processors
The Tessent™ Enhanced Trace Encoder is a fully‑featured RISC‑V trace solution that provides a mechanism to monitor the program execution of a CPU in real time. It encodes program execution (instruction trace) and optionally, the data from load and store instructions (data trace), outputting trace in a highly compressed format. External software can later take this data and use it to reconstruct the program execution flow. The Enhanced Trace Encoder can be finely tuned to balance the features and gate-count requirements of your system and includes a broad range of filters, giving complete control over what and when to trace. It complies with the latest standards produced by the RISC‑V International’s Efficient Trace for RISC‑V (E‑Trace) Working Group, supporting any standards‑compliant RISC‑V processor.
The need for RISC-V trace
Complex systems are prone to imperfect software behaviors.
These imperfections may be due to several factors, for
example, interactions with other processor cores or peripherals,
tight timing budgets, poor implementation, or a combination
of the above. The imperfections impact the real‑time behavior
of the system. Software that is not designed to minimize these
imperfections leads to avoidable costs, attributed to
underutilized CPUs, heating, power consumption, post-deployment
defects, crashes, and poor longevity of the
systems.
Understanding software behavior can be challenging but is key
to tackling the imperfections. Therefore, providing software
developers visibility of program execution is vital. Processor trace
capability enables the software engineer to view the behavior of
a program in detail, instruction‑by‑instruction without disrupting
the system.
Effficient trace for RISC-V specification
The Efficient Trace for RISC‑V specification specifies a minimum
level of trace functionality. The Enhanced Trace Encoder
satisfies the full requirements of the Efficient Trace
specification, while including many additional features not
typically found within implementations provided by CPU
vendors or in the open-source community. The following list of
features are few of the many features offered by the Tessent Enhanced
Trace Encoder.
Related resources
Machine learning-powered etch bias prediction for etch retargeting flow enhancement
The paper proposes a machine learning (ML) based approach to predict etch bias, which can replace the traditional rule-based etch bias tables
Guided random synthetic layout generation and machine-learning based defect prediction for leading edge technology node development
Combining synthetic layout generation and machine learning accelerates advanced semiconductor development. It enables early identification of process hotspots and defect prediction, improving efficiency and reducing costs.
A study on the improvement of machine learning (ML)-based defect prediction model reflecting process variations
This paper presents an ML-based risk pattern predictor that combines pattern segmentation, Greedy sampling, and unbiased statistics to find defects and improve yield.