How to Mine CapStash (CAP)

GPU mining on the Whirlpool-512 algorithm: 60-second blocks, 1 CAP rewards, and a chain design where even a single card gets a genuine shot. Windows, Linux, and HiveOS covered.

Published 2026-07-28 · Updated 2026-07-28 · Written by the CapsPool operator

CapStash (CAP) is a proof-of-work UTXO chain built around the Whirlpool-512 hashing algorithm (Whirlpool XOR Folded) rather than SHA-256. That single choice defines the mining landscape: no Bitcoin ASIC can touch it, so CAP is mined with ordinary GPUs. Blocks arrive every 60 seconds carrying a 1 CAP reward — 1,440 blocks a day network-wide — and difficulty retargets every single block via Dark Gravity Wave, so the chain reacts to hashrate changes within minutes, not weeks. The chain also includes a deliberate anti-concentration mechanism — periodic difficulty-1 "lottery" rounds — that gives small miners real chances no matter how much hashrate the big players bring. CapsPool runs the native-difficulty CAP pools with correct Whirlpool hashrate reporting.

What you need

• A GPU. NVIDIA GTX 1060 through RTX 5090 (driver 525 or newer), or AMD RX 470/480/500-series, Vega, and RX 5000–9000 series via OpenCL. Intel Arc works experimentally. Hybrid NVIDIA + AMD rigs run fine in one miner instance. There are no CAP ASICs.
• A Whirlpool miner. DankMiner is the reference miner — use the latest release (v1.5.3 at the time of writing; anything from v1.2.9 up works on this pool). Barracuda is an alternative. Both speak standard stratum.
• A CAP wallet address — next step.

Step 1 — Create a CAP wallet

1. Download CapStash-Core from the official GitHub releases and install it.
2. Let the node sync — with 60-second blocks the chain moves fast, but it is compact.
3. Create a receiving address under File → Receiving Addresses. Bech32 addresses start with cap1q; legacy addresses start with C. Both work on the pool — every address you mine with is validated against a full CapStash node before shares are accepted.
4. Back up the wallet file offline.

Step 2 — Choose Solo or PPLNS

Native Solo: you hunt whole blocks. When you find one, the coinbase transaction itself pays your address directly — instant, trustless, nothing held by anyone. With 1,440 blocks a day network-wide and lottery rounds in the mix, solo on CAP is far less lonely than solo on a 10-minute chain; small rigs genuinely land blocks here.

Native PPLNS: team mining with a 1% fee. Every block the pool finds is split across the last 10,000 shares in proportion to your contribution, paid out once the block reaches 100 confirmations (minimum payout 0.001 CAP). Steadier income, same infrastructure.

Step 3 — Connection details

Username is your CAP address, optionally with a rig name: cap1qyouraddress.rig1. Password is x. Always use the full stratum URL:

Both regions run independent full deployments of the pool and the CapStash node — connect to whichever is closer, and use the other as a failover entry in your miner config.

Step 4a — Mining on Windows or Linux

Download the build for your OS from the DankMiner releases page, extract it, and launch with your address and the full pool stratum:

PPLNS (steady income):

dankminer -a capstash -w cap1qyouraddress -p stratum+tcp://caps.capspool.io:8333 -W rig1

Solo (whole blocks):

dankminer -a capstash -w cap1qyouraddress -p stratum+tcp://caps.capspool.io:7333 -W rig1

AMD cards are auto-detected via OpenCL; if detection misses, add --force-opencl. While mining, open http://localhost:4068 in a browser for DankMiner's live dashboard — per-GPU hashrate, shares, temperature, fan, and power draw.

Step 4b — Mining on HiveOS

DankMiner runs as a custom miner on HiveOS — no manual SSH work needed, the flight sheet installs it for you. Create a flight sheet for your CAP wallet, choose Miner → "Custom", then open Setup Miner Config and fill it exactly like this:

Apply the flight sheet and the rig downloads, installs, and starts the miner automatically. The single tarball covers every supported GPU generation including RTX 50-series — no separate builds. One prerequisite for newer NVIDIA cards: make sure the rig's NVIDIA driver is 525 or newer (nvidia-driver-update from the Hive shell if not).

GPU tuning for Whirlpool

Whirlpool-512 is core-heavy, not memory-heavy — the opposite of Ethash-era habits. Raise the core clock, drop the memory clock, and you'll gain hashrate while cutting power. If hashrate is unexpectedly low, check temperatures, riser cables, and power delivery first; a starved riser shows up as a slow card long before it shows up as errors.

What to expect: blocks, odds, and time

Expected time to find a solo block is difficulty × 2³² / your_hashrate seconds. Because CAP difficulty retargets every block, your real-world average follows your share of network hashrate: if the network is at 300 GH/s and you bring 3 GH/s, you're expected to find about 1% of 1,440 daily blocks — roughly 14 blocks a day on average, plus whatever the lottery rounds hand you. That's the honest math; individual days will swing hard in both directions, because that's what solo variance is. You can watch the live network hashrate and difficulty on the pool dashboard.

Why "native difficulty" matters

Whirlpool-512 hashes are not distributed like SHA-256d hashes — CapStash's difficulty-1 target is roughly twice Bitcoin's — and pool software that recycles Bitcoin constants without accounting for that misreports your hashrate by about 2x. CapsPool's CAP pools run a native-difficulty implementation (CAP-native truediffone) so the hashrate you see on the dashboard matches what your card is actually doing. If you've mined CAP elsewhere and seen numbers that made no sense, this is why.

Lottery blocks — why some blocks show no effort %

By design, the CapStash chain periodically drops network difficulty to 1. Whoever submits first wins that block essentially instantly — a built-in lottery that redistributes chances toward small miners and keeps any single farm from locking up the chain. Roughly 6–7% of all CAP blocks are lottery blocks, and the chain's difficulty algorithm deliberately excludes them from retarget math so they never distort regular mining. On our dashboard those blocks are labeled Lottery instead of showing an effort percentage, because measuring "effort" against a difficulty of 1 is meaningless. It's one of several places where we choose honest statistics over pretty ones — see the FAQ for the full explanation, including what the ~ estimated-effort marker means.

Monitoring

No account, no registration. Enter your CAP address on the pool dashboard under My Stats for per-worker hashrate, shares, and blocks found; the URL is bookmarkable. PPLNS miners can also see their pending and paid amounts per block. Everything is available as JSON via the public API, and every found block links to the CapStash explorer so you can verify the coinbase payout on-chain.

Troubleshooting

• Authorization rejected ("Invalid CAP address"): check for uppercase characters in a cap1 address, typos, or a wrong-coin address. The pool validates every username against a live CapStash node.
• "GLIBC not found" on a rig: you're running the Linux desktop build on HiveOS — switch to the HiveOS tarball from the flight sheet setup above.
• "no kernel image available": NVIDIA driver older than the build needs — update to 525+.
• "CUDA not available — trying OpenCL": normal on AMD-only rigs, no action needed.
• Old DankMiner won't stay connected or shows zero shares: upgrade — versions before v1.2.9 mishandle difficulty messages during reconnects.
• Low hashrate: Whirlpool is core-bound — raise core clock, drop memory clock; then check temps, risers, and power delivery.
• Found a block but the reward isn't spendable yet: coinbase rewards mature at 100 confirmations (~100 minutes on CAP). It's yours; the chain just makes you wait.

Why mine CAP on CapsPool?

Independent infrastructure, native Whirlpool difficulty for correct stats, solo payouts written into the coinbase so the pool can't hold your coins, published fee addresses verifiable on-chain, US + EU servers, and statistics honest enough to label their own limitations. The operator runs the pool hands-on — and the same HiveOS fleet documented above — and answers questions directly on Discord.