Skip to content

Refine README content for clarity and conciseness #263

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
FujitsuPolycom wants to merge 1 commit into from
Closed

Refine README content for clarity and conciseness #263

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
22 changes: 6 additions & 16 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,11 +26,6 @@ reproducible benchmarks, and [test results](performance/).

## Profiles

KV is approximate total token capacity. The shared TP4 figure is for DCP1;
the TP2 figure is a reference estimate. `—` means no capacity
is recorded. Startup reports the actual capacity, which is separate from the
per-request Context limit.

### Four Sparks

| Model / predictor | Serving stack | Transport | Layout | Context | Sequences | KV (tokens) | Guide |
Expand All @@ -45,14 +40,9 @@ per-request Context limit.

The Vision-Exp [artifact contract](runtime/deepseek-vision-exp/profile.json)
identifies the Anemll image, MiaAI-Lab recipe, and SparkRing transport separately.
Contributor-reported results are linked from the guide; independent reproduction
of the selected artifacts is not claimed.

The shared-image GLM profiles enable continuation coalescing and token-sharded
mHC. Ring profiles use hardware-forwarded mesh paths and NCCL across both host
PCIe domains. Their source integration is **implemented**; full shared-image
serving qualification remains **research-only**. Switched deployments are
provided as-is and have not been validated on switched hardware.
Contributor-reported results are linked from the guide.

Switched deployments are provided as-is and have not been validated on switched hardware.
Shared-image GLM ring deployment requires the [managed-mesh setup](runtime/glm53-spark-mtp3-mesh/MANAGED_MESH.md).

### Two Sparks
Expand All @@ -64,9 +54,7 @@ Shared-image GLM ring deployment requires the [managed-mesh setup](runtime/glm53
| Qwen3.8-27B EXL3 K5/K6 | [SparkRing vLLM/ExLlamaV3 build](runtime/qwen38/README.md) | [Patched NCCL](spark_transport/nccl/README.md) | TP2/DCP1 | 1M | 32 | — | [Quickstart](docs/QWEN38_27B_EXL3_K5K6_PAIR_QUICKSTART.md) |

The NVFP4-Spark pair uses 8.75 GiB KV per rank, one DAC, both host PCIe
domains, coalescing and mHC. SparkCache is unsupported for this TP2 composition. Shared-image GPU
qualification remains **research-only**. The NVFP4-Spark reference deployment
reported 1,050,118 KV tokens; that is capacity, not a tested maximum workload.
domains, coalescing and mHC.

See the [profile index](docs/profiles/README.md) for evidence scopes and
[SparkCache compositions](recipes/sparkcache/README.md) for persistent-cache
Expand Down Expand Up @@ -106,6 +94,8 @@ Retired profiles retain their image references in their linked guides.
The [shared-image publication record](runtime/sparkring/source_image/publication.json)
contains the download digest and profile verification scope.

Image work is in flux and will reach a cleaner, condensed state in the future.

## Benchmark results

Each row links the exact measured configuration. These records include retired
Expand Down
Loading