Skip to main content
Glama
alphaparkinc

genpark-paged-attention-kv-cache-budget-calculator-skill

by alphaparkinc
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Remote

Related Servers

Alternatives to genpark-paged-attention-kv-cache-budget-calculator-skill

No user-submitted related servers found.

    Related Servers

    • genpark-kv-cache-paged-attention-allocator-skill by alphaparkinc

      genpark-kv-cache-pagedofficial

      alphaparkinc
      A
      license
      Not graded
      quality
      B
      maintenance
      Enables MCP clients to manage LLM inference memory through virtual paged-attention KV-cache block mapping, non-contiguous physical page allocation, zero-copy fragmentation tracking, radix-trie prefix caching, INT8 quantized compute, and speculative decoding verification. It also exposes prefill and decode latency telemetry so edge deployments can be benchmarked and tuned without external dependencies.
      7
      MIT

    • genpark-speculative-decoding

      Alpha-Park
      A
      license
      Not graded
      quality
      B
      maintenance
      Enables verification of draft-target speculative decoding with rejection sampling, acceptance-rate telemetry, and dynamic speedup estimation, alongside edge inference primitives such as INT8 quantization, paged KV allocation, and radix prefix caching.
      7
      MIT

    • genpark-dynamic-prompt-prefix

      Alpha-Park
      A
      license
      Not graded
      quality
      B
      maintenance
      Enables inference engines to match longest common prompt token prefixes via a radix trie, so cached KV-cache blocks are reused instead of re-prefilled. This eliminates redundant prefill computation and lowers time-to-first-token, alongside paged attention allocation, speculative decoding verification, INT8 quantization, and latency telemetry.
      7
      MIT

    This server cannot be deployed

    One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

    Maintenance

    ActivityMaintained
    ResponsivenessNo issues

    Resources