A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 15, 17:15 UTC
[Doc] Update V1 user guide for fp8 kv cache support (#15585)
Signed-off-by: weizeng <weizeng@roblox.com>
Wei Zeng committed
Mar 26, 2025 at 19:39 UTC
7f301dd8ef1d91c8f356c21ec9ee118a44553d5a
1 file changed
+1
-3
docs/source/getting_started/v1_user_guide.md
+1
-3
@@ -47,9 +47,9 @@ This living user guide outlines a few known **important changes and limitations*
47
| **Logprobs Calculation** | <nobr>🟢 Functional</nobr> |
48
| **LoRA** | <nobr>🟢 Functional ([PR #13096](https://github.com/vllm-project/vllm/pull/13096))</nobr>|
49
| **Multimodal Models** | <nobr>🟢 Functional</nobr> |
50
+| **FP8 KV Cache** | <nobr>🟢 Functional on Hopper devices ([PR #15191](https://github.com/vllm-project/vllm/pull/15191))</nobr>|
51
| **Spec Decode** | <nobr>🚧 WIP ([PR #13933](https://github.com/vllm-project/vllm/pull/13933))</nobr>|
52
| **Prompt Logprobs with Prefix Caching** | <nobr>🟡 Planned ([RFC #13414](https://github.com/vllm-project/vllm/issues/13414))</nobr>|
52
-| **FP8 KV Cache** | <nobr>🟡 Planned</nobr> |
53
| **Structured Output Alternative Backends** | <nobr>🟡 Planned</nobr> |
54
| **Embedding Models** | <nobr>🟡 Planned ([RFC #12249](https://github.com/vllm-project/vllm/issues/12249))</nobr> |
55
| **Mamba Models** | <nobr>🟡 Planned</nobr> |
@@ -134,8 +134,6 @@ in progress.
134
135
#### Features to Be Supported
136
137
-- **FP8 KV Cache**: While vLLM V1 introduces new FP8 kernels for model weight quantization, support for an FP8 key–value cache is not yet available. Users must continue using FP16 (or other supported precisions) for the KV cache.
138
-
137
- **Structured Output Alternative Backends**: Structured output alternative backends (outlines, guidance) support is planned. V1 currently
138
supports only the `xgrammar:no_fallback` mode, meaning that it will error out if the output schema is unsupported by xgrammar.
139
Details about the structured outputs can be found