Solvers

RetiredCoder

Took four keys of the puzzle transaction with a GPU kangaroo of their own making. Then quit the race in public and left 140 to everyone else.
IDretired-coder05 / 13person · 4 solves · 2023-02-27 → 2026-07-28
Subject

RetiredCoder

bitcoin

Author of RCKangaroo, an open source Pollard kangaroo solver for Nvidia GPUs, who took four of the puzzle transaction's keys and then retired from the race.

Solves
4
Prize recorded
40.200034 BTC
solved 4

Solves

Channels
githubgithub.com/RetiredC
bitcointalkbitcointalk.org/index.php
Addresses
3Emiwzxme7Mrj4d89uqohXNncnRM15YESs
1Prestige1zSYorBdz94KA2UbJW3hYLTn4

Log

 

Registered the bitcointalk account on October 5, 2024.

bitcointalk.org/index.php
 

The GitHub profile says: I've solved 134-bit range point on Secp256k1 (Bitcoin challenges #120, #125, #130, #135). Probably, it's a world record.

github.com/RetiredC
 

Publishes RCKangaroo, a CUDA implementation of Pollard's kangaroo method with symmetry that solves the discrete log in a known range.

github.com/RetiredC/RCKangaroo
 

Announced #135 on bitcointalk: about 5 months on 200 GPUs, and I quit, officially.

bitcointalk.org/index.php
 

Released RCKangaroo v4.0 with the ASM turbo kernels the same day, calling its main loop without inversions the Triple Montgomery trick.

bitcointalk.org/index.php
 

Said the run paid, but never again: it was too boring to pay bills for five months.

bitcointalk.org/index.php
local dataset / no network · data 3c17 73e5 a65a

Who this is

The author of RCKangaroo, an open source Pollard kangaroo for Nvidia GPUs, and the reason four of the big b1000 addresses are empty. 120, 125, 130 and 135 all fell to the same method: the public key is on chain, the range is known, so you send out herds of kangaroos and wait for two of them to land on the same point. Waiting is the hard part. For 135 it was about five months on 200 GPUs.

The funny thing is the order. The bitcointalk account is from October 2024, twelve days after 130 moved, and 120 and 125 were already gone in 2023 without a name on them. So the handle showed up after most of the work was done. The GitHub bio then claims all four and adds "Probably, it's a world record". Hard to argue with a 134 bit range.

How it ended

On July 28, 2026 the 135 prize moved to an address that starts with 1Prestige1. Twenty minutes later came the post: five months, 200 GPUs, "I quit, officially. Happy solving #140 without me." Same day, RCKangaroo v4.0 went up with the ASM kernels and a main loop without inversions, which the author calls the Triple Montgomery trick. And when someone asked if it paid off: yes, but never again, "it was too boring to pay bills for five months". Honest ending. Most people who retire from something at least pretend it was fun until the last day.

What the record says

The four solves each carry the same key, so the page joins them. The profile and the two addresses sit on all four records. The facts sit where they happened: the bitcointalk registration on 130, and on 135 the bio, the repository and three posts from one day. Nothing here says who is behind the handle, and nothing needs to. The keys are the proof.

The record

tssrc/collections/b1000/135.ts
solver: party("RetiredCoder", {
  key: "retired-coder",
  kind: PartyKind.Person,
  about: "Author of RCKangaroo, an open source Pollard kangaroo solver for Nvidia GPUs, who took four of the puzzle transaction's keys and then retired from the race.",
  addresses: ["3Emiwzxme7Mrj4d89uqohXNncnRM15YESs", "1Prestige1zSYorBdz94KA2UbJW3hYLTn4"],
  profiles: [
    profile("github", "https://github.com/RetiredC"),
    profile("bitcointalk", "https://bitcointalk.org/index.php?action=profile;u=3657819"),
  ],
  facts: [
    fact("The GitHub profile says: I've solved 134-bit range point on Secp256k1 …", "https://github.com/RetiredC"),
    fact("Announced #135 on bitcointalk: about 5 months on 200 GPUs, and I quit, officially.", "https://bitcointalk.org/index.php?topic=5517607.msg66986617#msg66986617", { date: "2026-07-28" }),
    // …
  ],
}),