Overview¶
The Metrix portal is a website for Alliance users. It leverages information collected from compute nodes and management servers to interactively generate data, allowing users to monitor their real-time resource usage (CPU, GPU, memory, file system).
| Cluster | Portal URL |
|---|---|
| Rorqual | https://metrix.rorqual.alliancecan.ca |
| Narval | https://portail.narval.calculquebec.ca |
| Nibi | https://portal.nibi.sharcnet.ca |
File System Performance¶
This section presents bandwidth and metadata operation graphs, with the following visualization options: last week, last day, and last hour.
Login Nodes¶
CPU and memory usage statistics, system load, and network are presented in this tab, with the following visualization options: last week, last day, and last hour.
Scheduling¶
This tab presents statistics on allocated cluster cores and GPUs, with the following visualization options: last week, last day, and last hour.
Scientific Software¶
The most used software with CPU cores and GPUs are presented as graphs.
Data Transfer Nodes¶
Bandwidth statistics for data transfer nodes are presented in this tab.
User Summary¶
Under the user summary tab, you will find your quotas for different file systems, followed by your 10 most recent jobs. You can select a job by its number to access its detailed page. Additionally, by clicking on 'More details', you will be redirected directly to the Job Statistics tab, where you can find all your jobs.
Job Statistics¶
The first block displays your current usage (CPU Core, memory, and GPUs). These statistics represent the average resources used by all currently running jobs. You can easily compare the resources allocated to you with those you actually use.
Next, you have access to an average over the last few days, presented as a graph.
On the left, the graph shows the number of disk write operations you have performed. (input/output operations per second (IOPS)) On the right, you see the amount of data transferred to the servers over a given period. (Bandwidth)
The following section presents all the jobs you have already launched, which are currently running or pending. In the top left, you can filter jobs by status (OOM, completed, running, etc.). In the top right, you can search by job number (Job ID) or by name. Finally, in the bottom right, an option allows you to quickly navigate between pages by making multiple jumps.
CPU Job Page¶
At the top, you have the job name, its number, your username, and the status. Details of your submission script are displayed by clicking 'View Job Script'.
Interactive Jobs
If the job was launched in interactive mode, the submission script will not be available.
The working directory and submission command are accessible by clicking 'View Submission Command'.
The next section is dedicated to scheduler information. You can access your CPU account tracking page by clicking on your account number.
In the Resources section, you can get an initial overview of your job's resource usage by comparing the Allocated and Used columns for the various listed parameters.
The CPU graph allows you to visualize the CPU cores you requested over time. On the right, you can select/deselect different cores as needed.
Short Jobs
Note that for very short jobs, this graph is not available.
The Memory graph allows you to visualize the usage of the memory you requested over time.
The Process and threads graph allows you to observe various parameters related to processes and threads. Ideally, for a multithreaded job, the sum of the Running threads and Sleeping threads parameters should not exceed twice the number of requested cores. However, it is quite normal to have some processes in sleeping (Sleeping threads) mode for certain types of programs (Java, Matlab, commercial software, or complex programs). You also have the program applications executed over time as a parameter.
The following graphs represent file system usage for the current job, not the entire node. On the left, a representation of the number of input/output operations per second (IOPS) is displayed. On the right, the graph illustrates the data transfer rate between the job and the file system over time. This graph helps identify periods of intense activity or low file system utilization.
Node Resource Statistics
For statistics on the entire node's resources, please note that they may be imprecise if the node is shared among multiple users.
The graph on the left illustrates the evolution of bandwidth used by the job over time, in relation to software, licenses, etc. The graph on the right represents the evolution of network bandwidth used by a job or a set of jobs via the Infiniband network over time. This can show periods of massive data transfer (e.g., read/write on a Lustre file system, MPI communication between nodes).
The graph on the left illustrates the evolution of the number of input/output operations per second (IOPS) performed on the local disk over time. The one on the right shows the evolution of bandwidth used on the local disk over time, i.e., the amount of data read or written per second.
Graphical representation of local disk space usage.
Graphical representation of power usage.
CPU Job Page (Job Array)¶
A CPU job page within a job array is identical to that of a regular CPU job, except for the Other jobs in the array section. This table lists other job numbers belonging to the same job array, along with information about their status, name, start time, and end time.
GPU Job Page¶
At the top of the page, you have the job name, its number, your username, and the status. Details of your submission script are displayed by clicking 'View Job Script'.
Interactive Jobs
If you launched an interactive job, the submission script is not available.
The working directory and submission command are accessible by clicking 'View Submission Command'.
The following section is reserved for scheduler information. You can access your GPU account page by clicking on your account number.
In the Resources section, you can get an initial overview of your job's resource usage by comparing the Allocated and Used columns for the various listed parameters.
The CPU graph allows you to visualize the usage of requested CPU cores over time. On the right, you can select/deselect different cores as needed.
Short Jobs
Note that for very short jobs, this graph is not available.
The Memory graph allows you to visualize the usage over time of the memory you requested for the CPUs.
The Process and threads graph allows you to observe various parameters related to processes and threads.
The following graphs represent file system usage for the current job, not the entire node. On the left, a representation of the number of input/output operations per second (IOPS) is displayed. On the right, the graph illustrates the data transfer rate between the job and the file system over time. This graph helps identify periods of intense activity or low file system utilization.
The GPU graph represents your GPU usage. The Streaming Multiprocessors (SM) active parameter indicates the percentage of time the GPU executes a warp (a group of consecutive threads) in the last sampling window. This value should ideally be around 80%. For SM occupancy (defined as the ratio between the number of warps assigned to an SM and the maximum number of warps an SM can handle), a value around 50% is generally expected. Regarding the Tensor parameter, the value should be as high as possible. Ideally, your code should leverage this part of the GPU, optimized for multiplications and convolutions of multi-dimensional matrices. Finally, for Floating Point operations (FP64, FP32, and FP16), you should observe significant activity on only one of these types, depending on the precision used by your code.
On the left, you have a graph indicating GPU memory usage. On the right, a graph of GPU memory access cycles, representing the percentage of cycles during which the device's memory interface is active for sending or receiving data.
The GPU power graph shows the evolution of GPU power consumption (in watts) over time.
On the left, GPU bandwidth on the PCIe bus (or PCI Express, for Peripheral Component Interconnect Express). On the right, GPU bandwidth on the NVLink bus. The NVLink bus is a technology developed by NVIDIA to enable ultra-fast communication between multiple GPUs.
Node Resource Statistics
For statistics on the entire node's resources, please note that they may be imprecise if the node is shared among multiple users.
The graph on the left illustrates the evolution of bandwidth used by the job over time, in relation to software, licenses, etc. The graph on the right represents the evolution of network bandwidth used by a job or a set of jobs via the Infiniband network over time. This can show periods of massive data transfer (e.g., read/write on a Lustre file system, MPI communication between nodes).
The graph on the left illustrates the evolution of the number of input/output operations per second (IOPS) performed on the local disk over time. The one on the right shows the evolution of bandwidth used on the local disk over time, i.e., the amount of data read or written per second.
Graphical representation of local disk space usage.
Graphical representation of power usage.
Account Statistics¶
The Account Statistics section groups your group's usage into two subsections: CPU and GPU.
CPU Account Statistics¶
Here you will find the sum of your group's requests for CPU cores, as well as their corresponding usage over the last few months. You can also track the evolution of your priority, which varies based on your usage.
This graph shows the most commonly used applications.
Here you can view the resource usage for each user in your group.
This graph shows the evolution over time of wasted CPU cores per user in the group.
Here you can view memory usage for each user in your group.
This graph represents memory wasted per user.
Next, you have a representation of your activity on the file systems. On the left, the graph shows the number of disk write commands you have performed. (input/output operations per second (IOPS)) On the right, you see the amount of data transferred to the servers over a given period. (Bandwidth)
You have a list of the latest jobs that have been performed for the entire group.
GPU Account Statistics¶
Here you will find the sum of your group's GPU requests, as well as their corresponding usage over the last few months. You can also track the evolution of your priority, which varies based on your usage.
This graph represents the most commonly used applications.
Here you can view the resource usage for each user in your group.
The following graph represents, over time, the amount of wasted GPU per user.
Next, you have the CPU cores allocated and used in your GPU jobs.
This figure illustrates the wastage of CPUs within your GPU jobs.
Here you can visualize memory usage for each user in your group.
This graph illustrates memory wasted per user.
Next, you have a representation of your activity on the file systems. On the left, the graph shows the number of disk write commands you have performed. (input/output operations per second (IOPS)) On the right, you see the amount of data transferred to the servers over a given period. (Bandwidth)
Here is the list of the latest jobs performed at your group level.
Cloud Statistics¶
The first table, 'Your Instances,' presents all virtual machines associated with an account. The 'Flavor' column refers to the virtual machine type. The 'UUID' column corresponds to a unique identifier assigned to each virtual machine.
Subsequently, each virtual machine has its own usage statistics (CPU Cores, Memory, Disk Bandwidth, Disk IOPS, and Network Bandwidth) viewable for the last month, last week, last day, or last hour.