In

Capturing an Intel Processor Trace Data Stream by a PCIe-Attached FPGA Monitoring

Datum
21.09.2017
Zeit
09:00 - 10:00
Sprecher
Albert Schulz
Zugehörigkeit
Institut für Technische Informatik, Professur Prozessordesign
Sprache
en
Hauptthema
Informatik
Andere Themen
Informatik
Beschreibung
The development and operation of safety-critical processing systems requires broad capabilities with respect to system observability and online monitoring. This demand has been addressed by microprocessor manufacturers through chip area dedicated to a growing number of execution statistics and tracing units. The extremely valuable detail of the insight provided by instruction or even data traces, however, also implies a challenge in mastering the vast amounts of trace data produced already over small time frames. In consequence, an effective combination of online verification, event triggering and trace filtering is needed for satisfying long-running monitoring requirements. In this diploma thesis, the infrastructure to activate and capture the trace of an Intel x86 processor by a PCIe-attached FPGA monitoring engine has been developed.A decoder implementation is presented, that allows for the first layer of packet extraction, i.e. identification of packet types and their payload.Based on various measurements and statistics, the quality and completeness of the trace data has been examined. In addition, an outlook for the further processing is given.Many practical use cases require the reconstruction of the program flow.For that reason, expectations and requirements for such a processing stage are derived from the decoder implementation and the trace analysis. Diese Veranstaltung wird unterstützt von Professur Prozessordesign.

Letztmalig verändert: 21.09.2017, 10:03:08

Veranstaltungsort

TUD Georg-Schumann-Bau (Georg-Schumann-Str. 7A, 2. OG Raum 204)Münchner Platz301187Dresden
Homepage
https://navigator.tu-dresden.de/etplan/sch/00

Veranstalter

TUD InformatikNöthnitzer Straße4601069Dresden
Telefon
+49 (0) 351 463-38465
Fax
+49 (0) 351 463-38221
Homepage
http://www.inf.tu-dresden.de
Scannen Sie diesen Code mit Ihrem Smartphone and bekommen Sie die Veranstaltung direkt in Ihren Kalender. Sollten Sie Probleme beim Scannen haben, vergrößern Sie den Code durch Klicken darauf.
  • AuAusgründung/Transfer
  • BaBauing., Architektur
  • BiBiologie
  • ChChemie
  • ElElektro- u. Informationstechnik
  • Sfür Schüler:innen
  • GsGesellschaft, Philos., Erzieh.
  • InInformatik
  • JuJura
  • MwMaschinenwesen
  • MtMaterialien
  • MaMathematik
  • MeMedizin
  • PhPhysik
  • PsPsychologie
  • KuSprache, Literatur und Kultur
  • UmUmwelt
  • VeVerkehr
  • WeWeiterbildung
  • WlWillkommen
  • WiWirtschaft