Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Machine-Learning-Based Correction of Analytical NPU Performance Models

Gouveia, Joaquim LU and Johansson, Oliver LU (2026) EITM01 20261
Department of Electrical and Information Technology
Abstract
Modern Neural Processing Units (NPUs) must be evaluated long before silicon ex-
ists, yet early tools force a difficult compromise: fast analytical performance models
enable exploration, but may diverge from slower RTL simulation, which provides
a more accurate timing reference. This thesis investigates whether that gap can be
learned and corrected. Using early-available block-command descriptors, sequence
summaries, and analytical cycle estimates from Arm’s NPU modelling flow, super-
vised XGBoost regressors are trained as a correction layer rather than a replace-
ment for the existing model. The method is evaluated on anonymised single-block
and short-sequence workloads across three NPU units, comparing corrected pre-
dictions... (More)
Modern Neural Processing Units (NPUs) must be evaluated long before silicon ex-
ists, yet early tools force a difficult compromise: fast analytical performance models
enable exploration, but may diverge from slower RTL simulation, which provides
a more accurate timing reference. This thesis investigates whether that gap can be
learned and corrected. Using early-available block-command descriptors, sequence
summaries, and analytical cycle estimates from Arm’s NPU modelling flow, super-
vised XGBoost regressors are trained as a correction layer rather than a replace-
ment for the existing model. The method is evaluated on anonymised single-block
and short-sequence workloads across three NPU units, comparing corrected pre-
dictions against RTL cycle measurements with a 5% acceptance criterion. Results
show large improvements in difficult cases, raising several low baseline pass rates
above 90% while substantially reducing median and tail errors. SHAP-based ana-
lysis further identifies workload regimes linked to systematic mismatch, making
the correction model not only more accurate but also a practical diagnostic tool
for model improvement. (Less)
Please use this url to cite or link to this publication:
author
Gouveia, Joaquim LU and Johansson, Oliver LU
supervisor
organization
course
EITM01 20261
year
type
H2 - Master's Degree (Two Years)
subject
keywords
Neural Processing Unit, Performance modelling, RTL simulation, Machine learning correction, XGBoost, Model interpretability
report number
LU/LTH-EIT 2026-1143
language
English
id
9244213
date added to LUP
2026-06-26 09:19:05
date last changed
2026-06-26 09:19:05
@misc{9244213,
  abstract     = {{Modern Neural Processing Units (NPUs) must be evaluated long before silicon ex-
ists, yet early tools force a difficult compromise: fast analytical performance models
enable exploration, but may diverge from slower RTL simulation, which provides
a more accurate timing reference. This thesis investigates whether that gap can be
learned and corrected. Using early-available block-command descriptors, sequence
summaries, and analytical cycle estimates from Arm’s NPU modelling flow, super-
vised XGBoost regressors are trained as a correction layer rather than a replace-
ment for the existing model. The method is evaluated on anonymised single-block
and short-sequence workloads across three NPU units, comparing corrected pre-
dictions against RTL cycle measurements with a 5% acceptance criterion. Results
show large improvements in difficult cases, raising several low baseline pass rates
above 90% while substantially reducing median and tail errors. SHAP-based ana-
lysis further identifies workload regimes linked to systematic mismatch, making
the correction model not only more accurate but also a practical diagnostic tool
for model improvement.}},
  author       = {{Gouveia, Joaquim and Johansson, Oliver}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{Machine-Learning-Based Correction of Analytical NPU Performance Models}},
  year         = {{2026}},
}