{
  "slug": "pfpc",
  "term": "pFPC",
  "aliases": [
    "Parallel FPC"
  ],
  "definition": "pFPC is a parallel lossless binary64 compressor that assigns chunks to separate FPC-style predictor histories.",
  "note": "Partitioning: Chunks go to threads in round-robin order. Changing thread count or chunk size changes which values share a prediction history, so parallelization can change compressed size as well as execution time. [paper]\n\nMemory budget: With n threads and level x, predictor tables alone total n * 2^(x+4) bytes. Four threads at x=20 use 64 MiB, versus 16 MiB for one thread. Holding the total at 16 MiB instead requires x=18 for four threads; buffers and other overhead remain additional. [project] [ledger]\n\nEvidence: The 2009 evaluation used nine datasets, a subset of the earlier thirteen, and three processor systems. Its ratio-versus-thread-count experiments kept predictor memory per thread fixed, increasing total predictor memory. Those results are not equal-memory comparisons. [paper]\n\nIntegration: The distributed example performs sequential input/output around its parallel routines. Its command-line elapsed time must not be equated with the paper's isolated parallel-routine throughput. Small chunks can also incur cache and TLB costs. [project] [paper]",
  "status": "shipping",
  "advantages": [],
  "limitations": [],
  "related": [
    "fpc"
  ],
  "supersedes": [],
  "supersededBy": [],
  "firstSeen": null,
  "lastSeen": null,
  "updated": "2026-11-17",
  "citations": [
    {
      "id": "paper",
      "title": "Burtscher and Ratanaworabhan: pFPC, Data Compression Conference (2009), sections 2, 4 and 5",
      "url": "https://userweb.cs.txstate.edu/~burtscher/papers/dcc09.pdf",
      "accessed": "2026-11-17"
    },
    {
      "id": "project",
      "title": "pFPC 1.0: per-thread memory, chunk assignment and sequential-I/O example warning",
      "url": "https://userweb.cs.txstate.edu/~burtscher/research/pFPC/",
      "accessed": "2026-11-17"
    },
    {
      "id": "ledger",
      "title": "Hesela: exact FPC residual costs and pFPC predictor-memory accounting",
      "url": "https://hesela.dev/research/fpc-residuals/README.md",
      "accessed": "2026-11-17"
    }
  ],
  "provenance": [
    "Source synthesis and integer accounting, with no measured throughput, representative compression ratio or original-paper replication.",
    "The paper reports an unsuccessful pipelined parallelization attempt; its conclusions depend on the tested hardware and memory traffic. [paper]"
  ]
}
