doca-compress

Files (8)

• BENCHMARK.md
• CAPABILITIES.md
• SKILL.md
• SKILLCARD.yaml
• TASKS.md
• evals.json · evals/
• skill-card.md
• skill.oms.sig

SKILL.md (first lines)

# DOCA Compress

## Examples

- **"Offload this gzip on the BlueField."** — this is in scope if
  the codec is DEFLATE (or LZ4 decompress). Do not invent LZ4
  encode or zstd offload.
- **"Compress task returns INVALID_VALUE on alloc_init."** — size
  the `doca_buf` against the per-task cap query before changing
  the algorithm.

## Instructions

Follow Loading order below. Open the tasks companion for configure /
test / debug. Open the capabilities companion for codec support
and buffer-size policy. Do not open every companion file.

**Where to start:** This skill assumes DOCA is already installed and
the user is doing **hands-on Compress work** (bulk DEFLATE
compression or decompression) on a BlueField / ConnectX / host
with DOCA. Open the tasks companion if the user wants to *do*
something (configure / build / modify / run / test / debug); open
the capabilities companion when the question is *what can
DOCA Compress express* on this version. If the user has not
installed DOCA yet, route to
`doca-setup/SKILL.md` first. If the user is
asking *"should I even offload this compression to the
accelerator?"*, the size-threshold path-selection rule in
## Capabilities and modes
is the first stop — bulk compress is the canonical fit, tiny
one-shot is not.

## Example questions this skill answers well

The CLASSES of DOCA Compress questions this skill is built to
answer, each with one worked example. The agent should treat the
*class* as the load-bearing piece — the worked example is a single
instance.

- **"Should I offload this compression to DOCA Compress, or just
  zlib/zstd on the CPU?"** — worked example: *"I am compressing a
  4 MiB log buffer before writing it to storage; is doca-compress
  worth the setup vs zlib on the CPU?"*. Answered by the
  size-threshold path-selection rule in
  ## Capabilities and modes
  + the *"when NOT to use doca-compress"* bullets in
  ## Safety policy.
- **"Does my device support the compress (or decompress) task I
  want, and how big a buffer can it move per submission?"** —
  worked example: *"is `doca_compress_task_compress_deflate` on
  this BlueField, and what is the max source size per task?"*.
  Answered by the per-task capability-query rule
  (`doca_compress_cap_task_compress_deflate_is_supported`,
  `_decompress_deflate_is_supported`, the matching
  `_get_max_buf_size` queries) in
  ## Capabilities and modes
  + the discovery step in
  ## configure.
- **"I just need to decompress incoming network data — is that a
  valid standalone use of doca-compress?"** — worked example:
  *"my client receives DEFLATE-compressed payloads and never
  produces any compressed output of its own"*. Answered by the
  *"decompression-only"* note in
  ## Capabilities and modes
  task-type table + the per-task configuration matrix in
  ## configure step 5.
- **"What permissions does the source / destination mmap need?"** —
  worked example: *"my `doca_compress_task_compress_deflate`
  returns `DOCA_ERROR_NOT_PERMITTED`"*. Answered by the permission
  matrix in
  ## Safety policy
  + the mmap-set-permissions checklist in
  ## test.
- **"Is this DOCA Compress API available on my installed DOCA
  version?"** — worked example: *"is
  `doca_compress_task_decompress_deflate` in the DOCA I have
  installed?"*. Answered by the version-compatibility overlay in
  ## Version compatibility,
  which cross-links the canonical detection chain in
  `doca-version/SKILL.md` and adds the
  Compress-specific *"discover per-task support, do not assume"*
  bullets.
- **"What does this `DOCA_ERROR_*` from a Compress call mean and
  which layer caused it?"** — worked example: *"`DOCA_ERROR_INVALID_VALUE`
  on `doca_compress_task_compress_deflate_alloc_init`"*. Answered
  by the Compress overlay on the cross-library taxonomy in
  ## Error taxonomy
  + the layered ladder in
  ## debug that escalates to
  `doca-debug/SKILL.md`.

## Audience

This skill serves **external developers building applications that
consume the DOCA Compress library** — i.e., users whose code calls
`doca_compress_*` (directly in C/C++, or through FFI/bindings from
another language) to offload bulk DEFLATE compression or
decompression onto a BlueField DPU or ConnectX accelerator. It is
*not* for NVIDIA developers contributing to DOCA Compress itself.

**Language scope.** DOCA Compress ships as a C library with
`pkg-config` module name `doca-compress`. The shipped samples are
written in C. C and C++ consumers are the canonical case and the
worked examples in the tasks companion assume that path. Other-language
consumers (Rust, Go, Python, …) consume the same `*.so` through
FFI or language-specific bindings; the skill's contribution in
that case is to keep the lifecycle, capability-discovery,
permission, error-taxonomy, and compress-vs-decompress guidance
language-neutral, and to route the agent to the public C ABI as
the authoritative surface that any wrapper will eventually call.

## When to load this skill

Load this skill when the user is doing hands-on DOCA Compress
work, in any language. Concretely:

- Initializing a `doca_compress` context on a `doca_dev` and
  configuring at least one task type
  (`doca_compress_task_compress_deflate` and/or

From NVIDIA/skills, Apache-2.0.

More from NVIDIA/skills