Cores and RAM Fair Usage Policy

This page defines the fair usage limits for CPU cores and system RAM, clarifies the distinction between system memory and GPU VRAM, and outlines that GPU memory can supplement but never replace required system RAM.

Definitions

RAM (Random Access Memory)

RAM is a type of high-speed, volatile memory that temporarily stores data and instructions a processor needs while performing tasks. It allows for rapid read and write access, enabling efficient execution of programs.

CPU RAM (System memory)

Location and role

CPU RAM, often just called "system memory," is attached to the computer’s mainboard and directly accessible by the central processing unit (CPU).

Purpose

It holds the operating system, active applications, and data that the CPU is currently working on.

Characteristics

• General-purpose: optimized for diverse workloads.

• Typically, larger in capacity (e.g., 8–128 GB in most systems).

• Higher latency than GPU RAM, but more flexible in usage.

GPU RAM (Video RAM or VRAM)

Location and role

GPU RAM resides on the graphics card and is directly accessible by the graphics processing unit (GPU).

Purpose

It stores textures, framebuffers, shaders, and parallel-processing data needed for rendering images or running GPU-accelerated computations (AI, simulations, etc.).

Characteristics

• Specialized: optimized for massively parallel workloads.

• Typically, smaller in capacity than CPU RAM (e.g., 4–48 GB), but with much higher bandwidth.

• Designed for rapid transfer of large blocks of data (e.g., pixel data, tensor arrays).

Key distinction

• CPU RAM: General-purpose, flexible memory for the entire system’s computations.

• GPU RAM: Dedicated, high-bandwidth memory specialized for graphics and parallel data processing.

Fair usage

This Fair Usage Policy (“Policy”) governs use of the Licensed Software. Its purpose is to ensure fair access and system stability by enforcing CPU core-to-system RAM limits, and by clarifying how GPU VRAM may be used in addition to (but not in place of) system RAM.

Resource boundaries

• Each deployment should remain within the guidance CPU core-to-system RAM limits in Schedule A – Guidance Configurations.

• GPU VRAM is permitted as an addition to system RAM where indicated in Schedule A – Guidance Configurations. GPU VRAM does not reduce, offset, or substitute for the required system RAM.

• Both physical and virtual cores count toward your licensed total.

Prohibited actions

• Use of utilities such as taskset (Linux man page) or similar methods to hide, pin, or mask CPU cores is prohibited.

• Artificially capping, throttling, or misreporting RAM or GPU VRAM allocations to appear compliant is not allowed.

• Virtualization, container, or OS-level tricks to bypass these limits are expressly forbidden.

Fairness principle

This Policy ensures no user gains unfair advantage by misreporting or bypassing system limits. The goal is fair access and system stability.

Clarifications on GPU VRAM

• Where Schedule A indicates “+ GPU VRAM”, GPU-attached memory may be used in addition to system RAM.

• GPU VRAM must never be counted toward minimum system RAM or used to exceed maximum system RAM caps.

• Compliance is measured on system RAM independently of GPU VRAM.

Schedule A – Guidance configurations

CPU to System RAM Limits (GPU additions where allowed).

vCPUs (Cores)	Permitted System RAM (Non-GPU)	Permitted System RAM (GPU Configs)	GPU VRAM Use Allowed?
1	0.5 GB – 4 GB	4 GB – 16 GB	Yes – VRAM is additional
4	3.6 GB – 16 GB	4 GB – 16 GB	Yes – VRAM is additional
8	7.2 GB – 32 GB	61 GB – 224 GB	Yes – VRAM is additional
16	14 GB – 104 GB	64 GB – 224 GB	Yes – VRAM is additional
32	28 GB – 128 GB	128 GB – 440 GB	Yes – VRAM is additional
64	56 GB – 512 GB	512 GB – 680 GB	Yes – VRAM is additional
96	192 GB – 768 GB	1.1 TB – 1.9 TB	Yes – VRAM is additional
128+	256 GB – 15 TB	3.6 TB – 3.97 TB	Yes – VRAM is additional