forked from vllm-project/vllm
-
Notifications
You must be signed in to change notification settings - Fork 15
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
[Doc] Compatibility matrix for mutual exclusive features
Signed-off-by: Wallas Santos <wallashss@ibm.com>
- Loading branch information
Showing
2 changed files
with
192 additions
and
0 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,191 @@ | ||
| .. _compatibility_matrix: | ||
|
|
||
| Compatibility Matrix | ||
| ==================== | ||
|
|
||
| The table below shows mutually exclusive features along with support for some device types. | ||
|
|
||
| .. list-table:: | ||
| :header-rows: 1 | ||
| :widths: 20 8 8 8 8 8 8 8 8 8 8 8 | ||
|
|
||
| * - Feature | ||
| - Chunked Prefill | ||
| - APC | ||
| - LoRa | ||
| - Prompt Adapter | ||
| - Speculative decoding | ||
| - CUDA Graphs | ||
| - Encoder/Decoder | ||
| - Logprobs | ||
| - Prompt Logprobs | ||
| - Async Output | ||
| - Multi-step | ||
| * - APC | ||
| - ✅ | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| * - LoRa | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/config.py#L1558>`__ | ||
| - ✅ | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| * - Prompt Adapter | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| * - Speculative decoding | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/config.py#L1200>`__ `[T] <https://github.com/vllm-project/vllm/issues/5016>`__ | ||
| - ✅ | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/spec_decode/spec_decode_worker.py#L86-L87>`__ | ||
| - ✅ | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| * - CUDA Graphs | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| * - Encoder/Decoder | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/worker/utils.py#L25>`__ | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/worker/utils.py#L17>`__ `[T] <https://github.com/vllm-project/vllm/issues/7366>`__ | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/worker/utils.py#L35>`__ | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/worker/utils.py#L55>`__ | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/worker/utils.py#L47>`__ `[T] <https://github.com/vllm-project/vllm/issues/7366>`__ | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/worker/utils.py#L51>`__ `[T] <https://github.com/vllm-project/vllm/issues/7447>`__ | ||
| - | ||
| - | ||
| - | ||
| - | ||
| - | ||
| * - Logprobs | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - | ||
| - | ||
| - | ||
| - | ||
| * - Prompt Logprobs | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/engine/output_processor/multi_step.py#L65>`__ `[T] <https://github.com/vllm-project/vllm/pull/8199>`__ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - | ||
| - | ||
| - | ||
| * - Async Output | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/config.py#L396>`__ | ||
| - ✅ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/config.py#L383>`__ | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/config.py#L200>`__ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/worker/utils.py#L51>`__ | ||
| - ✅ | ||
| - ✅ | ||
| - | ||
| - | ||
| * - Multi-step | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/7de49aa86c7f169eb0962b6db29ad53fff519ffb/vllm/engine/arg_utils.py#L944>`__ | ||
| - ✅ | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/engine/output_processor/multi_step.py#L130>`__ | ||
| - ✅ | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/engine/arg_utils.py#L951>`__ | ||
| - ✅ | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/worker/utils.py#L47>`__ | ||
| - ✅ | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/engine/output_processor/multi_step.py#L65>`__ `[T] <https://github.com/vllm-project/vllm/issues/8198>`__ | ||
| - ✅ | ||
| - | ||
| * - NVIDIA | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| * - CPU | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/engine/arg_utils.py#L954>`__ | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/executor/cpu_executor.py#L345>`__ | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/executor/cpu_executor.py#L31>`__ `[T] <https://github.com/vllm-project/vllm/pull/4830>`__ | ||
| - ✗ `[T] <https://github.com/vllm-project/vllm/issues/8475>`__ | ||
| - ✅ | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/executor/cpu_executor.py#L327>`__ | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a84e598e2125960d3b4f716b78863f24ac562947/vllm/worker/cpu_model_runner.py#L125>`__ | ||
| - ✅ | ||
| - ✅ | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/config.py#L370>`__ | ||
| - ✗ `[T] <https://github.com/vllm-project/vllm/issues/8477>`__ | ||
| * - AMD | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✗ `[C] <https://github.com/vllm-project/vllm/blob/a2469127db6144eedb38d0b505287c0044e4ce06/vllm/attention/backends/rocm_flash_attn.py#L343>`__ | ||
| - ✅ | ||
| - ✅ | ||
| - ✅ | ||
| - ✗ `[T] <https://github.com/vllm-project/vllm/issues/8472>`__ | ||
|
|
||
| Note: | ||
|
|
||
| - [C] stands for code checks, that is, there is a checking on running that verify if the combinations is valid and raises and error or log a warning disabling the feature. | ||
| - [T] stands for tracking issues or pull requests on vLLM Repo. | ||
| - APC stands for Automatic Prefix Caching. | ||
| - Async output processing needs CUDA Graphs activated to work, there is a code check in the table to inform that. It is the only ✅ with a [C]. | ||
| - Encoder/decoder currently does not work with CUDA Graphs, therefore it is not compatible with Async output processing as well. | ||
|
|
||
|
|
||
| .. | ||
| TODO: Add support for remaining devices. |