docs(server): clarify that --ctx-size is total context divided among parallel slots

[kitaekatt]

Summary

When using --parallel N, the --ctx-size value is the total context divided among all slots, not the per-slot context. This is a common source of confusion (see #11681, #5732).

Changes

Added clarification to two flags in tools/server/README.md:

--ctx-size: Added note explaining that when using --parallel N, this is the total context divided among all slots. Each slot gets ctx-size / parallel tokens. To allocate X tokens per slot with N parallel slots, set --ctx-size to X * N.

--parallel: Added note that the total context is divided equally among these slots.

Example

• --ctx-size 4096 --parallel 4 → each slot gets 1024 tokens
• To get 4096 tokens per slot with 4 parallel slots, use --ctx-size 16384 --parallel 4

Related Issues

• Fixes Misc. bug: llama-server --ctx-size is divided by --parallel and cannot be increased? #11681 (ctx-size divided by parallel confusion)
• Related to Context length documentation confusion #5732 (context length documentation confusion)

[ngxson]

this documentation is auto-generated, modify its source in arg.cpp instead

[taronaeo]

This is a common source of confusion (see #11681, #5732).

#17671 as well