# Intra-domain point-to-point transfer (`p2p_intra`) | ServingStudio Kernel Library

Send one payload between two GPUs in the same NVLink domain.

The simulator prices each MoE dispatch and combine transfer between GPUs in one NVLink domain by looking up its bytes on this curve; an fp8 payload is fewer bytes on the same curve. One GPU sends message_size_bytes to one peer, measured at bf16.

Category: Communication. Interactive page: https://servingstudio.cs.washington.edu/kernels.html?kind=p2p_intra. All kernel kinds: https://servingstudio.cs.washington.edu/kernels.md.

## Formula

- `actual bytes = max(1, ⌊message_size_bytes / bytes per element⌋) · bytes per element`
- `algbw = actual bytes / time`
- `busbw = algbw`

## How it is measured

CUDA-event time around 100 repeated transfers after 50 warm-up transfers and a barrier. Rank 0's mean per-transfer time is kept. NCCL times dist.send to rank 1; NVSHMEM times put followed by quiet on the sender's stream.

## Caveats

- The runner rounds the requested byte count down to whole dtype elements, with at least one element transferred.
- The fabric label keys the result but does not select links; the reserved GPUs determine the physical connection.

## Arguments

| Argument | Type | Unit | Meaning |
| --- | --- | --- | --- |
| `message_size_bytes` | number | bytes | Payload bytes sent from one GPU to its peer. |
| `dtype` | dtype |  | Element type of the payload; bf16 is used for the sweep. |
| `fabric` | label |  | Interconnect label used to identify the measurement. |

## Metrics

- Time (`time_ms`, ms)
- Algorithm bandwidth (`algbw_gbps`, GB/s)
- Bus bandwidth (`busbw_gbps`, GB/s)
- Energy (`energy_j`, J)

## Backends

- `nccl`: torch.distributed send and recv over NCCL, rank 0 to rank 1. any CUDA GPU.
- `nvshmem`: nvshmem4py put from rank 0 into rank 1's symmetric memory, then quiet to wait for completion. any CUDA GPU.

## Measured rows

57 rows in all.

| GPU | Backend | Precision | Rows |
| --- | --- | --- | --- |
| NVIDIA B200 | `nccl` |  | 19 |
| NVIDIA H200 | `nccl` |  | 19 |
| NVIDIA H200 | `nvshmem` |  | 19 |

## Used by these deployments

- [Qwen3 235B-A22B FP8: FFN side of an attention–FFN split](https://servingstudio.cs.washington.edu/models/Qwen3-235B-A22B-FP8/qwen3_fp8_ffn_moe.md) (`Qwen3-235B-A22B-FP8/qwen3_fp8_ffn_moe`)
- [Qwen3 235B-A22B FP8: DP attention, EP MoE, DeepGEMM experts](https://servingstudio.cs.washington.edu/models/Qwen3-235B-A22B-FP8/qwen3_moe_fp8_dp_attn_ep_ffn.md) (`Qwen3-235B-A22B-FP8/qwen3_moe_fp8_dp_attn_ep_ffn`)
- [Qwen3 235B-A22B FP8: vLLM, DP attention, EP MoE](https://servingstudio.cs.washington.edu/models/Qwen3-235B-A22B-FP8/qwen3_vllm_moe_dp_attn_ep_ffn.md) (`Qwen3-235B-A22B-FP8/qwen3_vllm_moe_dp_attn_ep_ffn`)
- [Qwen3 235B-A22B Thinking 2507 FP8: DP attention, EP MoE, DeepGEMM experts](https://servingstudio.cs.washington.edu/models/Qwen3-235B-A22B-Thinking-2507-FP8/qwen3_moe_fp8_dp_attn_ep_ffn.md) (`Qwen3-235B-A22B-Thinking-2507-FP8/qwen3_moe_fp8_dp_attn_ep_ffn`)
- [Qwen3 235B-A22B Thinking 2507 FP8: vLLM, DP attention, EP MoE](https://servingstudio.cs.washington.edu/models/Qwen3-235B-A22B-Thinking-2507-FP8/qwen3_vllm_moe_dp_attn_ep_ffn.md) (`Qwen3-235B-A22B-Thinking-2507-FP8/qwen3_vllm_moe_dp_attn_ep_ffn`)
- [Qwen3 235B-A22B: FFN side of an attention–FFN split](https://servingstudio.cs.washington.edu/models/Qwen3-235B-A22B/qwen3_ffn_moe.md) (`Qwen3-235B-A22B/qwen3_ffn_moe`)
- [Qwen3 235B-A22B: DP attention, EP MoE](https://servingstudio.cs.washington.edu/models/Qwen3-235B-A22B/qwen3_moe_dp_attn_ep_ffn.md) (`Qwen3-235B-A22B/qwen3_moe_dp_attn_ep_ffn`)
- [Qwen3 Coder 480B-A35B: FFN side of an attention–FFN split](https://servingstudio.cs.washington.edu/models/Qwen3-Coder-480B-A35B-Instruct/qwen3_fp8_ffn_moe.md) (`Qwen3-Coder-480B-A35B-Instruct/qwen3_fp8_ffn_moe`)

## API

- This kind: https://servingstudio.cs.washington.edu/api/public/v1/kernels/p2p_intra
- Measured rows: https://servingstudio.cs.washington.edu/api/public/v1/kernels/p2p_intra/rows (filter by any column, such as `?gpu=NVIDIA%20H200`; `format=csv` for CSV)
- Configs the public deployments ask of it: https://servingstudio.cs.washington.edu/api/public/v1/kernels/p2p_intra/configs

---

This page as Markdown: https://servingstudio.cs.washington.edu/kernels/p2p_intra.md. A guide to the whole site for AI agents: [llms.txt](https://servingstudio.cs.washington.edu/llms.txt).
