ICDS Roar

• For data that will be worked with and retrieved often
• Per TiB/month

All storage is sold in 1TiB blocks. Please consult ICS for recommended storage strategies.

• For data that will be worked with and retrieved often
• Per TiB/month

All storage is sold in 1TiB blocks. Please consult ICDS for recommended storage strategies.

A basic core allocation give priority (immediate) access to Roar Collab basic compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A basic core allocation give priority (immediate) access to Roar Collab basic compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A general unit of compute on our collaborative system for uncontrolled data, to "spend" as needed on computational resources.  Each compute token represents one core-month (720 core hours) of compute on a basic core system.

A general unit of compute on our collaborative system for uncontrolled data, to "spend" as needed on computational resources.  Each compute token represents one core-month (720 core hours) of compute on a basic core system.

A full A100 GPU card allocation gives priority (immediate) access to a single full Roar Collab A100 GPU card for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A full A100 GPU card allocation gives priority (immediate) access to a single full Roar Collab A100 GPU card for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A half A100 GPU card allocation gives priority (immediate) access to a single half Roar Collab A100 GPU card for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A half A100 GPU card allocation gives priority (immediate) access to a single half Roar Collab A100 GPU card for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A high memory core allocation give priority (immediate) access to Roar Collab high memory compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A high memory core allocation give priority (immediate) access to Roar Collab high memory compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A single A100 GPU instance (1/8) card allocation gives priority (immediate) access to a single Roar Collab A100 GPU instance (1/8 card) for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A single A100 GPU instance (1/8) card allocation gives priority (immediate) access to a single Roar Collab A100 GPU instance (1/8 card) for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

Aa A40 GPU allocation gives priority (immediate) access to a single Roar Collab A40 GPU for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

Aa A40 GPU allocation gives priority (immediate) access to a single Roar Collab A40 GPU for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A P100 GPU allocation gives priority (immediate) access to a single Roar Collab P100 GPU for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens. This includes the 28 standard cores needed to run the GPU.

A P100 GPU allocation gives priority (immediate) access to a single Roar Collab P100 GPU for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens. This includes the 28 standard cores needed to run the GPU.

A standard core allocation give priority (immediate) access to Roar Collab standard compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A standard core allocation give priority (immediate) access to Roar Collab standard compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

• For data that will be worked with and retrieved often
• Per TiB/month

All storage is sold in 1TiB blocks. Please consult ICS for recommended storage strategies.

• For data that will be worked with and retrieved often
• Per TiB/month

All storage is sold in 1TiB blocks. Please consult ICS for recommended storage strategies.

A basic core allocation give priority (immediate) access to Roar Restricted basic compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A basic core allocation give priority (immediate) access to Roar Restricted basic compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A general unit of compute on our restricted system for controlled data, to "spend" as needed on computational resources.  Each compute token represents one core-month (720 core hours) of compute on a basic core system.

A general unit of compute on our restricted system for controlled data, to "spend" as needed on computational resources.  Each compute token represents one core-month (720 core hours) of compute on a basic core system.

A standard core allocation give priority (immediate) access to Roar Restricted standard compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A standard core allocation give priority (immediate) access to Roar Restricted standard compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A general unit of compute on our collaborative system for uncontrolled data, to "spend" as needed on computational resources.  Each compute token represents one core-month (720 core hours) of compute on a basic core system.

A basic core allocation give priority (immediate) access to Roar Collab basic compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A standard core allocation give priority (immediate) access to Roar Collab standard compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A high memory core allocation give priority (immediate) access to Roar Collab high memory compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A P100 GPU allocation gives priority (immediate) access to a single Roar Collab P100 GPU for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens. This includes the 28 standard cores needed to run the GPU.

A full A100 GPU card allocation gives priority (immediate) access to a single full Roar Collab A100 GPU card for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A half A100 GPU card allocation gives priority (immediate) access to a single half Roar Collab A100 GPU card for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A single A100 GPU instance (1/8) card allocation gives priority (immediate) access to a single Roar Collab A100 GPU instance (1/8 card) for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

Aa A40 GPU allocation gives priority (immediate) access to a single Roar Collab A40 GPU for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A general unit of compute on our restricted system for controlled data, to "spend" as needed on computational resources.  Each compute token represents one core-month (720 core hours) of compute on a basic core system.

A basic core allocation give priority (immediate) access to Roar Restricted basic compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A standard core allocation give priority (immediate) access to Roar Restricted standard compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A general unit of compute on our collaborative system for uncontrolled data, to "spend" as needed on computational resources.  Each compute token represents one core-month (720 core hours) of compute on a basic core system.

A basic core allocation give priority (immediate) access to Roar Collab basic compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A standard core allocation give priority (immediate) access to Roar Collab standard compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A high memory core allocation give priority (immediate) access to Roar Collab high memory compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A P100 GPU allocation gives priority (immediate) access to a single Roar Collab P100 GPU for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens. This includes the 28 standard cores needed to run the GPU.

A full A100 GPU card allocation gives priority (immediate) access to a single full Roar Collab A100 GPU card for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A half A100 GPU card allocation gives priority (immediate) access to a single half Roar Collab A100 GPU card for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A single A100 GPU instance (1/8) card allocation gives priority (immediate) access to a single Roar Collab A100 GPU instance (1/8 card) for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

Aa A40 GPU allocation gives priority (immediate) access to a single Roar Collab A40 GPU for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A general unit of compute on our restricted system for controlled data, to "spend" as needed on computational resources.  Each compute token represents one core-month (720 core hours) of compute on a basic core system.

A basic core allocation give priority (immediate) access to Roar Restricted basic compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

A standard core allocation give priority (immediate) access to Roar Restricted standard compute cores for Slurm jobs.  This functionality is similar to the previous GReAT model without the attached bursting capability.  Bursting can still be achieved via compute tokens.

• For data that will be worked with and retrieved often
• Per TiB/month

All storage is sold in 1TiB blocks. Please consult ICS for recommended storage strategies.

• For data that will be worked with and retrieved often
• Per TiB/month

All storage is sold in 1TiB blocks. Please consult ICS for recommended storage strategies.

• For data that will be worked with and retrieved often
• Per TiB/month

All storage is sold in 1TiB blocks. Please consult ICDS for recommended storage strategies.

• For data that will be worked with and retrieved often
• Per TiB/month

All storage is sold in 1TiB blocks. Please consult ICS for recommended storage strategies.