Introduction

This page summarizes some facts about the elaboration of the MCC'2013.

The Models

This section provides some data about the models tools had to process:

• “Known” models where known when the call for tool participation was issued and tool developers could choose the best technique to model these according to the capacity of their tool. Their analysis show the performance of the tool when it is used by people having a good knowledge of its internals;
• “Surprise” models where decided at the very last moment and come from various origins. To process these, tools had to use the provided PNML format and tool developers could not anticipate any tuning for their tool. Their analysis shows the performance of the tool when it is used by "newbies".

In total, each tool was confronted against 17 examinations applied on 255 models/instances, so a total of 4335 examinations. We performed semi-automatic analysis from the outputs provided by these executions.

The table below shows a summary of the characteristics of these models.

The Computers

Three computers were made available to operate the submitted tools. Since the computers have some differences in their characteristics, we processed all the examinations of all the instances of a model on the same machine to let us measure comparable data.

A short description of these machine is provided in the table below.

cluster1 (Univ P. & M. Curie)	ebro (Univ. Rostock)	quadhexa-2 (Univ. Nanterre)
CPU
total of 46 CPU	total of 64 CPU	total of 24 CPU
23 x Intel Xeon E5645 2.4 GHz, 6-Core, 6x 1536KB/12288KB L2/L3	4 x AMD Opteron™ 6200 Series (Interlagos) 2.7 GHz, 16-Core, 16x 1024KB/16MB L2/L3	4 x Intel Xeon X7460 2.66 GHz, 6-Core, 3 x 3MB/16NB L2/L3
Memory
23 x 8GB (2x4GB) DDR3 / PC1333	512GB (32x 16GB) DDR3 / PC1600	128BG (8x 16BG) DDR3 / PC1333
Disks
23 x 500GB SATA 7200 + 1TB SATA 7200	2 x 1TB SAS2-Server-RAID + 2 x 128GB SSD Samsung 830 SERIES SATA III MLC	4 x 400GB RAID 1 (mirror) Seagate SAS Cheetah
Linux Kernel
2.6.38.8-server-10.mga	2.6.32-358.11.1.el6.x86_64	3.8.1-server-1.mga3

In fact, we used:

• cluster1: 18 out of the 23 nodes (other experiments were running),
• ebro: 63 out of the 64 cores (one for the OS ;-),
• quadhexa-2: 23 out of the 24 cores (one for the OS as well).

The organizers thank the Universities of Paris-Ouest-Nanterre, Rostock and Pierre & Marie Curie for letting us using their computers.

The executions

This section provides some data about the executions of the examinations over the proposed benchmark, for both the “Known” and the “surprise” models. Please note that the number of execution does not includes a large number of preliminary tests in cooperation with the tool developers.