CPU Resource Table

This page is for the IT internal planning of CPU resources for the LHC experiments, non-LHC experiments and various other groups.

We have now 3 different areas of CPU resource allocations: openstack compute, openstack services and condor batch.

The unit for all CPU processing numbers is HS06.

I am assuming an error bar of about 5% for the presented numbers.

Experiment/Group status Sep 2020 pledge April 2021 pledge April 2022
  [KHS06] [KHS06] [KHS06]
ALICE 350 471 471
ATLAS 411 525 550
CMS 423 500 520
LHCb 98 175 268
         
TOTEM 10 10 12
ATLAS T3 CERN 8 8 10
CMS T3 CERN 8 8 10
LHCb T3 CERN 7 7 10
         
ATLAS T3 Wisconsin 7 4 4
ATLAS T3 Tokyo 9 9 9
         
COMPASS 79 100 120
AMS 205 205 240
Theory 70 85 100
NA62 50 60 75
NA61 10 15 20
TE 50 60 70
BE 27 35 45
Ship 10 10 10
NP02 7.5 10 12
NP04 7.5 10 12
DTEAM 10 10 10
GEANT 10 10 10
other groups/exp. 30 40 50
         
Sum 1897 2367 2638

other contains: IT,EN,EP,HSE,ILC,FCC,ESCAPE,RE37,NEXT,AWAKE,......

Resources for HPC (e.g. Theory QCD, CFD, BE parallel) are not included in the previous table. Also not included are service resources for IT, EN, GS and BE.

Resource details per experiment

The following table shows the split of the pledges resources across 6 different areas for the 4 LHC experiments. The openstack numbers are always the quotas and not the usage values. This is counting the cores and assigning about 10 HS06 per core. Actually 10 HS06/core is the right number when we assume that SMT on the Intel boxes is switched on. Due to the various Intel exploits (e.g. Meltdown), SMT was switched off on the openstack service nodes, which increases the HS06 values from 10 to 16 HS06/core. On the other side we have deployed some over-commitment on the services. Thus for the moment the value per core stays at 10 HS06.

The unit for all CPU processing numbers is [HS06],

ALICE

Resource type Apr 2020
  [KHS06]
openstack services+compute 14
condor batch 336
   
total sum 350

ATLAS

Resource type Apr 2020
  [K HS06]
condor dedicated 210
openstack services+compute 66
condor batch 220
   
total sum 496

CMS

Resource type Apr 2020
  [KHS06]
condor dedicated 318
openstack services+compute 72
condor 33
   
total sum 423

LHCb

Resource type Apr 2020
  [KHS06]
openstack services+compute 15
condor 83
   
total sum 98

Group codes and experiments

* group_code_mapping.txt: linux group code mappings

Policies

1. The pledged CPU resources for the LHC experiments can be split between the different areas (Voboxes, Shares, dedicated, openstack), but the total pledge is constant.

2. The openstack resources are calculated via the quotas, not the used resources.

History and Changes


Topic attachments
I Attachment History Action Size Date Who Comment
Texttxt group_code_mapping.txt r1 manage 16.3 K 2014-05-18 - 12:55 BerndPanzerSteindel linux group code mappings
Edit | Attach | Watch | Print version | History: r38 < r37 < r36 < r35 < r34 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r38 - 2020-09-15 - BerndPanzerSteindel
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Main All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2020 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback