The ITC'02 SOC Test Benchmarks is a set of benchmark circuits with a special focus on modular plug-and-play testing of core-based system chips. The purpose of these benchmark circuits is to stimulate research in new methods and tools for modular testing of core-based SOCs and to enable the objective comparison of such methods and tools with respect to effectiveness and efficiency. The set of benchmarks was officially presented at the IEEE International Test Conference (ITC'02) in October 2002 in Baltimore, MD, USA.

Table of Contents

Motivation and Background

With the ITC'02 SOC Test Benchmarks, we intend to combine the homogenity of the ISCAS'85 and ISCAS'89 benchmark circuits with the industrially-relevant sizes and characteristics of the ITC'99 benchmark circuits. On the one hand, we want to limit the amount of intellectual property that is released through the benchmarks, such that it will be easy for companies and other organizations to contribute one or more SOC designs to the benchmark set. On the other hand, we want to make sure that the information released contains sufficient data to solve relevant research problems in the SOC test automation domain, such as wrapper design, TAM design, and test scheduling. We believe we have found a format, that fulfills the requirements above. This format describes the core design hierarchy. For each core, the number of terminals, core-internal scan chains, length of these chains, number of tests and test patterns per test are provided. Optionally, the power dissipation during a test, as well as the layout coordinates of the cores might be provided. Note that the contents of the cores and SOC are not provided; also the test patterns themselves are not provided. Furthermore, the cores are numbered, and hence their original names are also not provided.

How to Contribute SOC(s)?

If you or your company or organization is interested in contributing to the set of ITC'02 SOC Test Benchmarks, you should code your SOC data according to the format described above. Send an e-mail to Erik Jan Marinissen (Erik . Jan . Marinissen @ nxp . com). Together with you, a name will be established for the SOC. The SOC data will then be made available through this Internet web page.

Your contribution can be recognized through the naming scheme of the SOCs; the first letter of the SOC name refers to the contributing company or organization. It is also possible to remain anonymous.

We recommend that (potential) contributors to the benchmark set subscribe to the e-mail reflector. Through this e-mail reflector you will be informed about extensions of the benchmarks set and other relevant news.

Please understand that the released SOC data will be available to anyone, and hence you should make sure that you do not release confidential information. However, the type of data we are asking for is expected not to pose confidentially issues for most companies.

How to Use the Benchmarks?

If you or your university or organization is interested in using the set of ITC'02 SOC Test Benchmarks, you can simply download the benchmark circuits from this web site. The benchmarks are freely available to anyone.

We recommend that (potential) users of the benchmarks subscribe to the e-mail reflector. Through this e-mail reflector you will be informed about extensions of the benchmarks set and other relevant news.

The set of benchmarks was first presented at ITC'02. However, the benchmark set, this web site, and the e-mail reflector continue to exist also after ITC'02. Hence, it is possible to continue to use the benchmark circuits for your research purposes also long after ITC'02, and publish the results in our conferences and/or journals.

We do encourage that you include a reference to the source of the benchmark circuits, either in the form of a reference to this web site, or, even better, to the ITC'02 paper presenting the benchmark SOCs:

E-Mail Reflector

The e-mail reflector related to the ITC'02 SOC Test Benchmarks is no longer working.

SOC Naming Scheme

Each SOC name consists of one letter and a number. The letter represents the origin of the benchmark. The number represents the test complexity of the SOC.

The details of the naming scheme used are described here.

Benchmark Format

The benchmark format contains the following information per SOC.

The details of the format used are described here.

Benchmark Files

The set of benchmarks currently contains the data of twelve SOCs.

SOC Number of Contributor
Modules Tests Levels å I/Os å SFFs
u226 10 9 2 376 1040 Universidade Federal do Rio Grande do Sul
d281 9 15 2 2931 882 Duke University
d695 11 10 2 1845 6384 Duke University
h953 9 8 2 929 4657 National Tsing Hua University
g1023 15 14 2 3707 1546 University of Stuttgart
f2126 5 4 2 1597 13996 Faraday Technologies
q12710 5 4 2 13167 12991 Hewlett-Packard
p22810 29 30 3 4283 24723 Philips Semiconductors
p34392 20 21 3 2057 20948 Philips Semiconductors
p93791 33 32 3 6943 89973 Philips Research
t512505 32 31 2 8663 68051 Texas Instruments
a586710 8 7 3 3755 37656 Analog Devices

The benchmarks are available as gzipped tar file of 4.8 Kbytes:
itc02benchm.tar.gz

Change History

Date Change Description
2004-05-10 Corrected #Modules and #Tests for t512505 in table above from (31,30) into (32,31).
Bug spotted by Sandeep Kumar Goel, Philips Research.
2002-08-28 "ScanUse 0" changed into "ScanUse 1" for SOC d281, Module 7, Test 1.
This module has internal scan chains and only one test. It was meant that this
test makes use of these scan chains. Bug spotted by Erika Cota, Universidade
Federal do Rio Grande do Sul.
2002-07-03 Added SOC d281 (contributed by Vikram Iyengar and Krish Chakrabarty,
Duke University) and SOC q12710 (contributed by Chul Young Lee, Hewlett-Packard
Company).
2002-06-06 Added SOC h953 (contributed by Cheng-Wen Wu, National Tsing Hua
University), SOC f2126 (contributed by Chuang Cheng, Faraday Technologies), and
SOC t512505 (contributed by Rubin Parekhji, Texas Instruments).
2002-06-03 Added SOC a586710, contributed by Luis Basto, Analog Devices.
2002-03-13 SOC g1096 was not included in web site; added.
Small mistakes in SOC benchmark files corrected: (1) "ScanChains 0" was
corrected by "ScanChains 0:"; (2) SOC names were mis-calculated. Sandeep Kumar
Goel (Philips) now implemented an automatic parser for the .soc format, that
also calculates the number in the SOC name.
2002-03-06 Adapted formula for determining the number in the benchmark name (1) to be
based on the number of tests (instead of on the number of modules), and (2) to
include 0/1 variables for "ScanUse" and "TamUse". Benchmark names are updated;
SOC u220 renamed to u251.
Added SOC g1096, contributed by Rainer Dorsch, University of Stuttgart.
2002-02-11 Introduced keywords "ScanUse" and TamUse" in benchmark format. Updated all SOC
files available.
Added SOC u220, contributed by Erika Cota, Universidade Federal do Rio Grande
do Sul.
2002-01-18 SOC d695 (contributed by Vikram Iyengar and Krish Chakrabarty, Duke
University), and SOCs p21241, p31108, and p93791 (contributed by Erik Jan
Marinissen, Philips) made available.

Papers on the Benchmarks

Papers ABOUT the Benchmarks

Papers USING the Benchmarks

A bibliography of papers in which the ITC'02 SOC Test Benchmarks are used for experimental research results can be found here.
Due to copyright issues, we cannot provide copies of these papers here; contact their respective authors!

This is a subset of the TECS Bibliography (which has a search facility!).

For inclusion of missing references, see here.

Acknowledgements

We thank the following people who have contributed to the ITC'02 SOC Test Benchmarks.

Furthermore, we thank Sandeep Kumar Goel (Philips Research) for making a parser for the .soc file format, and Bart Vermeulen (Philips Research) for providing us with the e-mail reflector facilities.

Interesting Links

Contacts

The ITC'02 SOC Test Benchmark Set is an initiative from
Erik Jan Marinissen, Vikram Iyengar, Krishnendu Chakrabarty - August 22, 2008