{| class="wikitable" |- | Availability : '''March 31, 2025''' |- | Login node : '''tamia.alliancecan.ca''' |- | Globus collection : [https://app.globus.org/file-manager?origin_id=72c3bca0-9281-4742-b066-333ba0fdef72 TamIA's Globus v5 Server] |- | Data transfer node (rsync, scp, sftp,...) : '''tamia.alliancecan.ca''' |- | Portal : to be announced |} tamIA is a cluster dedicated to artificial intelligence for the Canadian scientific community. Located at [http://www.ulaval.ca/ Université Laval], tamIA is co-managed with [https://mila.quebec/ Mila] and [https://calculquebec.ca/ Calcul Québec]. The cluster is named for the [https://en.wikipedia.org/wiki/Tamias eastern chipmunk], a common species found in eastern North America. tamIA is part of [https://alliancecan.ca/en/services/advanced-research-computing/pan-canadian-ai-compute-environment-paice PAICE, the Pan-Canadian AI Compute Environment]. ==Site-specific policies== * By policy, tamIA's compute nodes cannot access the internet. If you need an exception to this rule, contact [[Technical_support|technical support]] explaining what you need and why. * crontab is not offered on tamIA. * Each job should be at least one hour long (at least five minutes for test jobs) and you can't have more than 1000 jobs (running and pending) at a time. * The maximum duration of a task is one day (24 hours). * Each task must use 4 GPUs, or 1 full node. ==Access== To access the cluster, each researcher must complete [https://ccdb.alliancecan.ca/me/access_services an access request in the CCDB]. Access to the cluster may take up to one hour after completing the access request. Eligible principal investigators are members of an AIP-type RAP (prefix aip-). The procedure for sponsoring other researchers is as follows: * In the '''[https://ccdb.alliancecan.ca/ CCDB home page]''', go to the ''Resource Allocation Projects'' table * Look for the RAPI of the aip- project and click on that to be redirected to the RAP management page * At the bottom of the RAP management page, click on '''Manage RAP memberships''' * To add a new member, go to ''Add Members'' and enter the CCRI of the user you want to add. ==Storage== {| class="wikitable sortable" |- | HOME
Lustre file system || * Location of home directories, each of which has a small fixed quota. * You should use the project space for larger storage needs. * Small per user [[Storage_and_file_management#Filesystem_quotas_and_policies|quota]]. * There is currently no backup of the home directories. (ETA Summer 2025) |- | SCRATCH
Lustre file system || * Large space for storing temporary files during computations. * No backup system in place. * Large [[Storage_and_file_management#Filesystem_quotas_and_policies|quota]] per user. * There is an [[Scratch_purging_policy|automated purge]] of older files in this space. |- | PROJECT
Lustre file system || * This space is designed for sharing data among the members of a research group and for storing large amounts of data. * Large and adjustable per group [[Storage and file management#Filesystem_quotas_and_policies|quota]]. * There is currently no backup of the home directories. (ETA Summer 2025) |} For transferring data via [[Globus]], you should use the endpoint specified at the top of this page, while for tools like [[Transferring_data#Rsync|rsync]] and [[Transferring_data#SCP|scp]] you can use a login node. ==High-performance interconnect== The [https://fr.wikipedia.org/wiki/Bus_InfiniBand InfiniBand] [https://www.nvidia.com/en-us/networking/quantum2/ NVIDIA NDR] network links together all of the nodes of the cluster. Each H100 GPU is connected to a single NDR200 port through an NVIDIA ConnectX-7 HCA. Eeach GPU server has 4 NDR200 ports connected to the InfiniBand fabric. The InfiniBand network is non-blocking for compute servers and is composed of two levels of switches in a fat-tree topology. Storage and management nodes are connected via four 400Gb/s connections to the network core. ==Node characteristics== {| class="wikitable sortable" ! nodes !! cores !! available memory !! CPU !! storage !! GPU |- | 42 || 48 || 512GB || 2 x [https://www.intel.com/content/www/us/en/products/sku/232380/intel-xeon-gold-6442y-processor-60m-cache-2-60-ghz/specifications.html Intel Xeon Gold 6442Y 2,6 GHz, 24C] || 1 x SSD de 7.68TB || 4 x NVIDIA HGX H100 SXM 80GB HBM3 700W, connected via NVLink |- | 4 || 64 || 512GB || 2 x [https://www.intel.com/content/www/us/en/products/sku/232398/intel-xeon-gold-6438m-processor-60m-cache-2-20-ghz/specifications.html Intel Xeon Gold 6438M 2.2G, 32C/64T] || 1 x SSD de 7.68TB || none |} ===Software environments=== [[Standard software environments/fr|StdEnv/2023]] is the standard environment on tamIA. ==Monitoring jobs== {{note|type=reminder|The portal is not yet available.}} From the [https://portail.tamia.calculquebec.ca/ tamIA portal], you can monitor your jobs using CPUs and GPUs in real time or examine jobs that have run in the past. This can help you to optimize resource usage and shorten wait time in the queue. You can monitor your usage of * compute nodes, * memory, * GPU. It is important that you use the allocated resources and to correct your requests when compute resources are less used or not used at all. For example, if you request 4 cores (CPUs) but use only one, you should adjust the script file accordingly.