Hardware‐assisted instruction profiling and latency detection

Sharma, Suchakrapani Datt; Dagenais, Michel R.

doi:10.1049/joe.2016.0127

Cited by 7 publications

(16 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…17 In our previous work, we observed Intel PT's low overhead as 2% to 3% for common workloads, as compared with other software-only program-flow trace approaches, which had an overhead of up to 63.5%. 17 Thalheim et al 18 used Intel PT for improving security and dependability in software by presenting a data parallel provenance algorithm. Analysis of virtual machines using host-only Intel PT hardware traces has also been explored recently.…”

Section: Literature Reviewmentioning

confidence: 81%

“…Intel PT generates this information in an encoded form, as multiple hardware packets such as taken‐not‐taken (TNT), target IP (TIP), and timestamp counter, which are then decoded postexecution to generate the program flow . In our previous work, we observed Intel PT's low overhead as 2% to 3% for common workloads, as compared with other software‐only program‐flow trace approaches, which had an overhead of up to 63.5% . Thalheim et al used Intel PT for improving security and dependability in software by presenting a data parallel provenance algorithm.…”

Section: Literature Reviewmentioning

confidence: 95%

“…Intel PT, however, records 1 bit per conditional branch, and the complete IP for indirect branches, thus reducing the trace bandwidth further. Intel PT generates this information in an encoded form, as multiple hardware packets such as taken‐not‐taken (TNT), target IP (TIP), and timestamp counter, which are then decoded postexecution to generate the program flow . In our previous work, we observed Intel PT's low overhead as 2% to 3% for common workloads, as compared with other software‐only program‐flow trace approaches, which had an overhead of up to 63.5% .…”

Section: Literature Reviewmentioning

confidence: 99%

“…Modern hardware tracing techniques, however, allow F ( P ) to be reconstructed postexecution, thus generating true execution profiles at very low overhead. We have discussed this in detail in our previous work . Therefore, for each branch encountered in CS p and CS r , the processor generates encoded trace packets representing the decision on a branch taken or not taken, along with the IP, if required.…”

Section: Background and Motivationmentioning

confidence: 99%

See 3 more Smart Citations

Hardware trace reconstruction of runtime compiled code

Sharma

Dagenais

2018

Softw Pract Exp

Self Cite

View full text Add to dashboard Cite

Summary Hardware tracing has emerged as a low‐cost technique to analyze systems at a very fine granularity, thus mitigating the need for software‐only trace approaches for performance analysis. State‐of‐the‐art trace hardware on modern Intel and ARM processors allows recording change‐of‐flow instructions in executable binaries, such as branches, for off‐line reconstruction. This conventional userspace–based trace reconstruction, however, is not robust enough in the common scenarios where runtime code is being generated, compiled, and executed. We therefore propose a novel kernel‐assisted mechanism called FlowJIT to reconstruct hardware traces with a low overhead of around 1.3 μs per code page modification event. We further show the efficacy or our technique with the help of 2 illustrative usecases that cover the JIT compiled code scenario and a same‐page instruction modification scenario. Our implementation has been open sourced as a patch for the Linux kernel.

show abstract

Section: Literature Reviewmentioning

confidence: 81%

Section: Literature Reviewmentioning

confidence: 95%

Section: Literature Reviewmentioning

confidence: 99%

Section: Background and Motivationmentioning

confidence: 99%

See 2 more Smart Citations

Hardware trace reconstruction of runtime compiled code

Sharma

Dagenais

2018

Softw Pract Exp

Self Cite

View full text Add to dashboard Cite

show abstract

“…Dedicated hardware blocks in the Intel architecture, such as Last Branch Record, Branch Trace Store [15], and more recently Intel Processor Trace (PT) follow this approach. A detailed study of hardwareassisted tracing and profiling with Intel PT has been recently presented in [12].…”

Section: Related Workmentioning

confidence: 99%