More people are moving away from Windows to Linux, with all the new improvements for gamers in Linux and the EU dumping down Windows. We need to bring the download version for Linux back. I know it’s easy to build once you’ve got all the dependencies, but we need to make it easy for people to start using the TU wallet.
One of my wishlist hopes for the community shift in this project is for TU to be across all major GNU Linux distros. It of course will need to be well-supported, and where graphics drivers are concerned for mining that’s not easy… there can be a lot of proprietary tinkering ![]()
agree.
it’s unreasonable of removing the support of linux os in the certain vesion of wallet, becaue it’s built on cross-platform frameworks, and supported linux since it’s born.
linux are widley used in early users, please consider this advice.
We removed Linux just because of the sheer number of unique runtimes there are; it becomes a support nightmare. This is just gotten worse. Remember linux is not an OS, its a kernel. Supporting every OS becomes super hard if you don’t have a user that championions it and supports the install stack.
And every other project out there has a Linux version without a problem; if it’s a specific bug, just remove what’s causing the bug in TU, but bring back the Linux version. I want to be able to use the bridge too.
I used the Linux Tari Universe build quite a bit before its support was discontinued and it worked mostly fine for me. However, I have an idea what would have made it especially complicated to support: Mining, and especially GPU support for mining.
I think that it’s reasonable to draw a fence around what is going to be directly supported in the Linux build, and what is going to be ‘best effort’, ‘community supported’, or ‘recommend other solution’.
For instance, it remains a pain in the ass to install and use the correct Nvidia drivers for many systems, especially systems which have multiple GPUs where Linux may be configured to favor one over the other.
However, this difficulty shouldn’t prevent us from making available other wallet features, such as sending/receiving money and the bridge. Linux’s marketshare is growing tremendously of late due to Microsoft’s continual shooting itself in the foot.
We may also be in a better position to address the edge cases now that there’s more public contribution.
Though I agree here, it’s possible to target the most popular distros.
Ubuntu LTS and Debian Stable would go a long way before even looking at RHEL + Fedora, Arch, etc.
I was going to comment about the difficulties with CUDA etc. but those are largely inherited from the xmr ecosystem and are just a general issue with GPU mining as you note. Start with a build that removes or disables these features, and then if there is enough interest in it, people will build ways to do it in their own community repos.
If those 2 lanes, GPU and ASIC, cause so much trouble, why not remove them? Would that solve the majority of the problem? They even talk about the problems with those 2 lanes. Update Tari's Difficulty Algorithm - #3 by kinkajou and An observation on hashrate concentration on the Tari RandomX lane - #2 by Memloch
My thread on the difficulty algorithm is an issue that effects all lanes, not just GPU/ASIC, and the other thread you linked to is about the Tari CPU lane.
Either way, we shouldn’t ever be removing mining lanes - that’s our network security. It isn’t practical to reduce network security just so a few users have access to a GUI wallet without building it themselves.
@profdiggity and @Fox have the right idea. Simply remove the GPU miner. It’s just using lolminer anyway. It’s trivial for miners to go install that themselves if they want. Is that what you meant by removing the lanes? The Sha3x miner was removed a while ago so it’s just c29/randomx (CPU/GPU) in TU now.
I’m not sure on this point-- Monero is operating with solid network security using only CPU mining, and GPUs/ASICs are becoming harder to come by for everyone. However, I feel it’s a separate question, and should be separately considered, from this problem.
Nearly every single person that plays video games on their computer has a GPU. I can buy a GPU cheaper than I can get a set of DDR5 for my CPU right now. Many of the AI hobbyists/developers we’re apparently targeting now have their own GPU, or many of them. And there’s plenty of ASICs available for sale. Only the first batch went out of stock.
Monero also isn’t operating with solid network security using only CPU mining, Qubic just caused a 30 block reorg with their malicious mining attack and could have done more. Monero had to implement some emergency “block checkpoints” fix to prevent total disaster and exchange delistings/indefinite suspensions.
Tari also isn’t operating on solid security with our CPU lane - we just linked to a thread on this very forum showing nearly the entire lane is a single miner/botnet! ![]()
Generally speaking, GPU mining requires proprietary drivers so it can’t be done with a fully free software stack. Let’s start with a stripped down build and proprietary GPU support can be added by the people who want it (though I won’t be doing it).
At the moment, XTM needs a boost of users and the simplest way to get a stable Tari Universe in the hands of more GNU/Linux users is to start with an app that can be built and maintained as simply as the Windows and MacOS builds (maybe simpler). Otherwise, we’ll end up where we are now, with no build at all for any free distros.
So, let’s start from “first principles” so to speak, and then the folks who want to worry about GPU mining etc. can do that.