Parano1d ① Proof-native Layer 1 ordered by PoW

[Grumo]

GPU miner for NOID — measured, and it works

The maintainer asked in discussion #12 whether a GPU could help with the nonce search, and said only a real implementation and a benchmark could answer it. I built one. Here is the answer.

Why a GPU can't mine Parano1d alone — and why that stops mattering behind a pool

Every block needs a fresh HistoryStep proof before there is anything to search, and that proof is CPU work. A GPU cannot build it. On its own, a GPU would sit idle between blocks waiting for a CPU, and if a new block landed first the prepared work would be thrown away.

A pool removes that: its node builds and proves the template once, and every GPU attached to it searches nonces on that same template in parallel. The GPU never touches the proof. It only does the part it is good at.

The arithmetic question, answered with numbers

Poseidon2b over GF(2^128) needs carryless multiplication. CPUs have it in silicon (VPCLMUL, PMULL); GPUs have to emulate it, which is expensive per operation. The open question was whether parallelism pays for that emulation. It does, and not marginally.

Measured per card, from the pool's own share accounting — the rate implied by shares actually delivered, not what the miner claims about itself:

Code:

   RTX 5090      62 MH/s per card
  RTX 4090      42 MH/s per card
  RTX 5080      32 MH/s per card
  RTX 5070 Ti   26 MH/s per card

One process drives every card in a machine. A four-card rig has been running at ~130 MH/s for hours against a live pool, with rejects in the low single digits per thousand.

Download

Windows, Linux, and a HiveOS custom-miner package : github.com/lucasan123/noid-gpu-miner

Code:

  noid-gpu --gpu --coinbase o1YOURADDRESS --worker rig1

Tried and supported pools:
Ariapool: https://pool.ariabrain.com/noid.html
My pool: https://parano1d-pool.fun/

Your NOID address is your login and your payout address; there is no registration. Generate it on your own machine and give out only the public address — never a seed phrase or a wallet file. Mining does not need them, and anyone asking you for them is stealing from you.

Needs an NVIDIA driver 580 or newer. On HiveOS, a jammy or noble image. Verify what you downloaded against SHA256SUMS.txt in the release.

The dev fee, stated plainly

10% of the shares are submitted under the dev fee address instead of yours.


That address is used for nothing else, so anyone can see on chain exactly what this software earns. The fee never changes the coinbase of a block, and never switches mid-search — only between jobs. It is printed in the header on every single run, with the address in full: you do not have to take my word for it, just read the screen. And if the fee cannot be delivered, the miner stops instead of mining, rather than run while showing you something different from what it does.

Yes, 10% is higher than the 1–3% that is usual. It is disclosed, it is verifiable, and nobody has to use it.

Source

Only binaries are published. If you would rather run something you can read, the official parano1d-miner is right there and that is a completely fair choice — I am not going to argue against it.

No warranty of any kind. Mining software moves money; run it only from a source you decided to trust, and decide that for yourself.

[alt_x]

GPU dominance determines the mining cost.

At 60 MH/s, 4,000 MH/s network hashrate, 20-second blocks and a 45 NOID reward, the estimated electricity cost is ~€0.00037 per NOID at €0.10/kWh.

As GPU miners dominate, this low production cost becomes the key benchmark for NOID's price.

I honestly wonder why the dev created a GPU-dominated coin when the original announcement presented it as a CPU-minable coin.
That seems like a serious mismatch between the original vision and the actual mining economics.

The epic fail.

[minerminer3456]

GPU dominance determines the mining cost.

At 60 MH/s, 4,000 MH/s network hashrate, 20-second blocks and a 45 NOID reward, the estimated electricity cost is ~€0.00037 per NOID at €0.10/kWh.

As GPU miners dominate, this low production cost becomes the key benchmark for NOID's price.

I honestly wonder why the dev created a GPU-dominated coin when the original announcement presented it as a CPU-minable coin.
That seems like a serious mismatch between the original vision and the actual mining economics.

The epic fail.

most coins start as a cpu minable and goes with gpu fpga asic. isn it. son it is not wrong to say "cpu minable" imo.

[Ima_o_lambao]

GPU dominance determines the mining cost.

At 60 MH/s, 4,000 MH/s network hashrate, 20-second blocks and a 45 NOID reward, the estimated electricity cost is ~€0.00037 per NOID at €0.10/kWh.

As GPU miners dominate, this low production cost becomes the key benchmark for NOID's price.

I honestly wonder why the dev created a GPU-dominated coin when the original announcement presented it as a CPU-minable coin.
That seems like a serious mismatch between the original vision and the actual mining economics.

The epic fail.

most coins start as a cpu minable and goes with gpu fpga asic. isn it. son it is not wrong to say "cpu minable" imo.

I dont know the tech behind and this chain might be good and have its surprises, but telling its cpu minable and gpus cant mine it then get cracked on day one is def a lmao

[Tongpu]

Lmao. The two noobs above just discovered pool mining in 2026 and called it a crack and an epic fail 😄 The GPU miner dev wrote it in bold - the pool CPU builds HistoryStep, while GPUs only search nonces. Ignotus explained exactly the same thing before this miner even existed. The pool simply moved that CPU mining part from each miner to one shared server. That is what makes GPU workers possible.Both Paranoid pools are doing impressive work.

https://parano1d-pool.fun/ looks fire and released GPU miner for nonce part.
https://pool.ariabrain.com/noid.html is also top. They released AriaMiner-NOID v1.2 and claim around 2–2.5x over v1.1 on Zen 2/3 and EPYC without AVX-512. I didnt test it yet, but definitely will.

Thinking about launching my own pool and have a proposal for pool owners:
In a usual PoW pool, the server does not spend power building a cryptographic proof before workers can hash. Here the pool server is a real part of mining and does significant work.
Maybe we should not even call it a normal pool fee. It is a split between the HistoryStep proof producer and nonce workers. I think 30-50% for the proof side can be fair, depending on prover speed and server costs. Yes, this sounds crazy for classic blockchain, but Paranoid mining is not normal mining lol. What you guys think?

[IgnotusNemo]

GPU miner for NOID — measured, and it works

The maintainer asked in discussion #12 whether a GPU could help with the nonce search, and said only a real implementation and a benchmark could answer it. I built one. Here is the answer.

Why a GPU can't mine Parano1d alone — and why that stops mattering behind a pool

Every block needs a fresh HistoryStep proof before there is anything to search, and that proof is CPU work. A GPU cannot build it. On its own, a GPU would sit idle between blocks waiting for a CPU, and if a new block landed first the prepared work would be thrown away.

A pool removes that: its node builds and proves the template once, and every GPU attached to it searches nonces on that same template in parallel. The GPU never touches the proof. It only does the part it is good at.

The arithmetic question, answered with numbers

Poseidon2b over GF(2^128) needs carryless multiplication. CPUs have it in silicon (VPCLMUL, PMULL); GPUs have to emulate it, which is expensive per operation. The open question was whether parallelism pays for that emulation. It does, and not marginally.

Measured per card, from the pool's own share accounting — the rate implied by shares actually delivered, not what the miner claims about itself:

Code:

   RTX 5090      62 MH/s per card
  RTX 4090      42 MH/s per card
  RTX 5080      32 MH/s per card
  RTX 5070 Ti   26 MH/s per card

One process drives every card in a machine. A four-card rig has been running at ~130 MH/s for hours against a live pool, with rejects in the low single digits per thousand.

..

Thank you for building this and publishing the measurements. A working implementation is a much better answer than speculation.

[GofyGofy]

GREETINGS, CRYPTO PARANOIDS.

I have watched blockchains drift toward the very thing they were meant to destroy.
History grows without end. Verification becomes heavier. People stop determining truth for themselves and borrow it from APIs, explorers, exchanges, foundations and data centers.
This is not decentralization. It is rented verification.
The post-quantum transition will make this worse. Larger signatures and heavier cryptography, appended forever, will turn today's debt into tomorrow's dependency.
You can build post-quantum signatures and other primitives. But they alone do not make a system post-quantum.
Security targets and conjectures are not proofs. Marketing is not soundness.
I wanted a system that ordinary hardware could fully verify today and decades from now. Post-quantum from the first block. Not promised for later. Proven end to end.

That became Parano1d.
Proofs establish validity.
Proof of work establishes canonical order.
State is validated from genesis in O(1).
End-to-end post-quantum soundness meets the NIST PQC Category 1 threshold.

I searched for a blockchain where the present could prove the past without replaying it. I found none.
So I built one ①.

---

Parano1d mainnet is live from August 21, 16:00 UTC.

code: https://github.com/ignotusnemo/parano1d

official discussion space on github: https://github.com/ignotusnemo/parano1d/discussions


Links

https://parano1d.org

https://docs.parano1d.org

https://lab.parano1d.org

---

The Present Must Prove the Past

Can a blockchain validate its current state without replaying the entire execution history from genesis?

https://lab.parano1d.org/research/the-present-must-prove-the-past/

---

This is where it started:

Specter was the first idea. It proved signatureless authorization could work, but the design had major problems and was archived. Parano1d was built from scratch.

A Signatureless Transaction Model and Recursive Proof-Based PQ Blockchain

https://bitcointalk.org/index.php?topic=5578193.0

Hey, are you considering setting up an official Discord or Telegram community for Parano1d?

I honestly think it’s becoming a must-have for a project like this nowadays. A dedicated community would make it much easier for miners, developers, node operators and users to communicate, report bugs, share ideas and coordinate development in real time.

It would also help build a stronger community around the project instead of having everything happen through GitHub issues. Discord in particular could be useful for development discussions, testing, mining support and quick feedback.

Even a simple Discord with a few channels to start with would be a big step toward building the Parano1d ecosystem and making the project feel more alive and accessible.

Just wondering if you have plans for this in the future.

[Seikar]

I honestly think it’s becoming a must-have for a project like this nowadays. A dedicated community would make it much easier for miners, developers, node operators and users to communicate, report bugs, share ideas and coordinate development in real time.

It would also help build a stronger community around the project instead of having everything happen through GitHub issues. Discord in particular could be useful for development discussions, testing, mining support and quick feedback.

Technically there is no intrinsic difference in response frequency between checking Github discussions or this thread X times a day, versus doing so X times a day on Discord.  If participants are online simultaneously, a post can be written and responded to in discussion instantly at any of those locations.

https://github.com/ignotusnemo/parano1d/discussions

Yet maybe unstated expectations / specific reasons for liking Discord are any of:

1. The developer/participants are assumed to respond faster by leaving the app running and getting background notifications, more like carrying a pager or being on call?
2. People feel they can not write as quick feedback on an old-style forum because it encourages a more formal writing style?
3. People often prefer to type in a phone rather than using a laptop or desktop?
4. Subscribing to a channel for update notifications is desired?
5. People already have Discord / Telegram as regularly visited sites or installed apps for many purposes, which is less so with Bitcointalk or Github Discussions?

Personally I don't like Discord or any site which requires a login to read anything, but the popularity is undeniable after maybe a half-dozen posts have asked for such. There is the possibility an unofficial channel could be community-created.

There is another potential reason, which might be the strongest argument:
6. The exchanges open to listing new coins often require Discord/Telegram, up to even a specific minimum subscriber count, for standard eligibility for listing.

[rugrumx (not an imposter)]

Definitely need a discord guys, telegram optional.
Can start OTC in a couple of spots and take a look with safetrade.
Will check out the miners shortly

[IgnotusNemo]

Parano1d v1.0.1 mainnet update is published.

Large pool payout batches exposed a P2P issue. After a new block or reorg, some transactions became stale or conflicting but continued circulating through the network, adding unnecessary load to live nodes. v1.0.1 validates transactions before relaying them, so these transactions are dropped locally instead of being amplified across the mesh. I recommend everyone update, especially pools and public nodes. v1.0.0 remains compatible. No data reset or configuration change is required.

UPD
Regular outbound seed and discovered-peer connections now use an ephemeral TCP source port instead of reusing the P2P listener port. This avoids VPN, NAT and stateful-firewall paths that drop symmetric 9600 -> 9600 flows. DCUtR keeps port reuse for hole punching.

Discussion https://github.com/ignotusnemo/parano1d/discussions/15
Download https://github.com/ignotusnemo/parano1d/releases/tag/v1.0.1About Discord, Telegram, etc.

I dont build for exchanges. I dont care about subscriber counts. Those who understand why Parano1d matters will support it and list it themselves. The code is open. Discord and Telegram mean spam, marketing proposals and constant moderation. I would rather write code. Anyone can run an unofficial channel.

GitHub Discussions already gives us the useful parts of a forum and Discord in one place. It is public, searchable and tied directly to the code.

[GofyGofy]

About Discord, Telegram, etc.

I dont build for exchanges. I dont care about subscriber counts. Those who understand why Parano1d matters will support it and list it themselves. The code is open. Discord and Telegram mean spam, marketing proposals and constant moderation. I would rather write code. Anyone can run an unofficial channel.

GitHub Discussions already gives us the useful parts of a forum and Discord in one place. It is public, searchable and tied directly to the code.

I have to say, I was a bit surprised by your answer about Discord and Telegram, but I totally understand your point. I actually like the fact that you prefer to spend your time building instead of chasing exchanges, followers and hype.

I really like the GUI wallet too, especially considering how new Parano1d is. There’s already so much built around it that I honestly started wondering — are you really doing all of this as a solo developer, or is there a small team behind Parano1d? 😄

Anyway, I respect the approach. Keep building, I think that’s the best thing you can do right now.

[weesdr]

About Discord, Telegram, etc.

I dont build for exchanges. I dont care about subscriber counts. Those who understand why Parano1d matters will support it and list it themselves. The code is open. Discord and Telegram mean spam, marketing proposals and constant moderation. I would rather write code. Anyone can run an unofficial channel.

GitHub Discussions already gives us the useful parts of a forum and Discord in one place. It is public, searchable and tied directly to the code.

Maybe looks at this way, Discord or Telegram are there to build the community for the project. Newbie who needs help setting up the nodes, wallet, mining can go there to get help with. Github discussion is mostly for technical people and it isn't noob friendly like me.

[IgnotusNemo]

Parano1d changes the blockchain architecture, but architecture alone cannot keep a network decentralized.
Users must behave like participants, not customers.

1. Verify for yourself
2. Keep the shared State compact
3. Preserve your own proofs
4. Back up your wallet key
5. Verify claims, not marketing

ONETHICS ① is a simple code for doing that - https://github.com/ignotusnemo/parano1d/discussions/19

[ocean1223]

If anyone is looking for a Parano1d community Discord, I’ve set up an unofficial server here: https://discord.gg/WeBqbzYX
Please note that this is an unofficial community server and is not affiliated with the Parano1d team.
I’m also a developer working on mining infrastructure, and I find this project very interesting. I’d be happy to contribute and offer technical assistance as the community continues exploring and developing the Parano1d ecosystem

[max2000irc]

GPU miner for NOID — measured, and it works

The maintainer asked in discussion #12 whether a GPU could help with the nonce search, and said only a real implementation and a benchmark could answer it. I built one. Here is the answer.

Why a GPU can't mine Parano1d alone — and why that stops mattering behind a pool

Every block needs a fresh HistoryStep proof before there is anything to search, and that proof is CPU work. A GPU cannot build it. On its own, a GPU would sit idle between blocks waiting for a CPU, and if a new block landed first the prepared work would be thrown away.

A pool removes that: its node builds and proves the template once, and every GPU attached to it searches nonces on that same template in parallel. The GPU never touches the proof. It only does the part it is good at.

The arithmetic question, answered with numbers

Poseidon2b over GF(2^128) needs carryless multiplication. CPUs have it in silicon (VPCLMUL, PMULL); GPUs have to emulate it, which is expensive per operation. The open question was whether parallelism pays for that emulation. It does, and not marginally.

Measured per card, from the pool's own share accounting — the rate implied by shares actually delivered, not what the miner claims about itself:

Code:

   RTX 5090      62 MH/s per card
  RTX 4090      42 MH/s per card
  RTX 5080      32 MH/s per card
  RTX 5070 Ti   26 MH/s per card

One process drives every card in a machine. A four-card rig has been running at ~130 MH/s for hours against a live pool, with rejects in the low single digits per thousand.

Download

Windows, Linux, and a HiveOS custom-miner package : github.com/lucasan123/noid-gpu-miner

Code:

  noid-gpu --gpu --coinbase o1YOURADDRESS --worker rig1

Tried and supported pools:
Ariapool: https://pool.ariabrain.com/noid.html
My pool: https://parano1d-pool.fun/

Your NOID address is your login and your payout address; there is no registration. Generate it on your own machine and give out only the public address — never a seed phrase or a wallet file. Mining does not need them, and anyone asking you for them is stealing from you.

Needs an NVIDIA driver 580 or newer. On HiveOS, a jammy or noble image. Verify what you downloaded against SHA256SUMS.txt in the release.

The dev fee, stated plainly

10% of the shares are submitted under the dev fee address instead of yours.


That address is used for nothing else, so anyone can see on chain exactly what this software earns. The fee never changes the coinbase of a block, and never switches mid-search — only between jobs. It is printed in the header on every single run, with the address in full: you do not have to take my word for it, just read the screen. And if the fee cannot be delivered, the miner stops instead of mining, rather than run while showing you something different from what it does.

Yes, 10% is higher than the 1–3% that is usual. It is disclosed, it is verifiable, and nobody has to use it.

Source

Only binaries are published. If you would rather run something you can read, the official parano1d-miner is right there and that is a completely fair choice — I am not going to argue against it.

No warranty of any kind. Mining software moves money; run it only from a source you decided to trust, and decide that for yourself.

Great job !  One question , is it possible to make your pool a little bit faster , it seems is submiting the blocks too slow , so other pool or miner submit it fast , basically your pool have like 20-30% of the network but it's finding 10-15% of the blocks. Thanks in advance !

[weesdr]

Also, any plans to bring down pool fee? 10% is crazy high on top of miner dev fee: https://parano1d-pool.fun/

[leogeeko]

Man, you need a Discord. So we can discuss about the coin.

Also, the HiveOS miner is not working correctly.

[Grumo]

Also, any plans to bring down pool fee? 10% is crazy high on top of miner dev fee: https://parano1d-pool.fun/

Great job !  One question , is it possible to make your pool a little bit faster , it seems is submiting the blocks too slow , so other pool or miner submit it fast , basically your pool have like 20-30% of the network but it's finding 10-15% of the blocks. Thanks in advance !

Thanks — both points are fair.


Pool improvements

We have confirmed that the pool is not yet converting its full hashrate into the number of blocks it should find. This is not only bad luck: there are periods when miners do not receive new work quickly enough, especially when the network changes to a new block.

We are improving the entire process: delivering new jobs faster, immediately removing outdated work and submitting new blocks to the network with less delay. We are also adding better monitoring so we can follow every block from the moment it is found until it is confirmed.

The first node upgrade is already live and has reduced an important part of the processing time by approximately 18%. More pool-side improvements are now being implemented and tested.

We are also developing an integrated blockchain explorer for the pool website. Miners will be able to check blocks, confirmations, rewards and pool activity directly from one place. This will make the pool easier to use and its real performance much more transparent.

After the updates have been running long enough, I will publish a simple before-and-after comparison based on real results.

Faster GPU miner

The optimized miner is already showing an improvement of around 19% in our first RTX 5070 Ti tests. Linux/HiveOS, RTX 4090 and RTX 5090 testing will follow before the public release.

Our goal is simple: better performance, an easier mining experience and clear, verifiable results.

Pool fee

I understand that the current pool fee looks expensive, but there is an important reason behind it.

The pool server costs approximately €600 every month. This is not a basic hosting server: it uses powerful hardware to efficiently handle the CPU-intensive part of the network’s proof of work.

Using a cheaper server would reduce our costs, but it would also reduce the pool’s performance and most likely result in fewer blocks being found. That would save money for the pool while costing miners more in lost rewards — and I do not believe that is the right compromise.

I still agree that combining the pool fee with the miner development fee is too expensive. The plan is to reduce both fees from 10% to 7.5% as the optimized miner and pool improvements are released.

The reduction is not live yet, so the currently published fees remain in effect until the official announcement. Every change will be clearly displayed on the pool website, in the miner and in the release notes. There will be no hidden fees or silent changes.

A faster pool, a more powerful miner, lower fees and an integrated explorer are all part of the same goal: building a better and more transparent mining platform.

Thanks for the honest feedback and for supporting the project.

[GofyGofy]

I just bought some ① NOID.

But honestly, I don't feel like I simply bought a coin. I feel like I bought a small piece of the technology behind it.

The more I look into Parano1d, the more I like the idea behind it. The proof-native architecture, the way the State is handled, the focus on verification and the fact that the developer seems much more interested in building than marketing — that's what got my attention.

I have no idea where the market will take NOID, and I'm not buying it because I expect some easy pump. I bought it because I find the technology genuinely interesting and I want to have some skin in the game while it's still this early.

Maybe I'm completely wrong. Maybe I'm not.

But I'd rather own a small piece of something I find technically interesting than buy another coin just because someone on Twitter says it's going to 100x.

①

[Ima_o_lambao]

I just bought some ① NOID.

But honestly, I don't feel like I simply bought a coin. I feel like I bought a small piece of the technology behind it.

The more I look into Parano1d, the more I like the idea behind it. The proof-native architecture, the way the State is handled, the focus on verification and the fact that the developer seems much more interested in building than marketing — that's what got my attention.

I have no idea where the market will take NOID, and I'm not buying it because I expect some easy pump. I bought it because I find the technology genuinely interesting and I want to have some skin in the game while it's still this early.

Maybe I'm completely wrong. Maybe I'm not.

But I'd rather own a small piece of something I find technically interesting than buy another coin just because someone on Twitter says it's going to 100x.

①

Where tf you bough it from? i thought there were no exchange listing yet