how to speed up releases

[smitty-codes]

There is quite the lag between when golem main releases and when this library is updated. How can we help to expedite this fork being updated to latest main release?

Thanks

[MarijnStevens]

I left the scene a while ago, but in short, why does the Golem people only talk x86, that was my motivation to try the same goal on arm, or in the future risk or whatever, a compute is a compute, we can argue that one arch take n ms, or another takes m ns, but in the end, a mainframe multi-arch should support *almost any arch possible, so thats why I went for it. The core problem I got stuck on is the rust filtering allocation of jobs, there is no reason to schedule arm jobs to arm processors, but the infrastructure (im talking 2 years ago) didnt support it. | Nevermind other reasons like you want to run your own written powerpc assembly on the Golem network, so im not saying only ARM, im talking ALL qumu architectures. |

Im old skool I guess, if there are multiple programs running in an ancient programming language, because of banks for example, why not provide the environment to run them, not on IBM machines, nor Cray Machine's, but on any of them. The performance issue is weird to me, because companies run them on 2 decade old hardware in weeks, where you phone would do the same job in seconds.

[smitty-codes]

@MarijnStevens I also lost interest in Golem for these reasons:

• Looking at the network stats, less than 50 requestors (i.e. customers) make up the bulk of the requests, so there is a minuscule customer base.
• Combine that with the silly pricing providers are using in the network its more like a charity of compute power than a serious way to make some extra $ with your idle hardware. If you try to price your compute/ram/storage to something comparable to what you'd pay on Azure/AWS/GCP and THEN drop that price 75% off, its still NOT competitive on the network with other providers. Then you have to factor in electricity costs, which apparently no providers care about. Most providers are essentially providing their compute for free. Its just not attractive from a provider stand point IMO.
• Regarding ARM support - yeah, it was a cool idea (and you did good work) but Golem overall is just a niche crypto hype project so I think you were right not investing anymore energy into it. I wish it was more after this many years but its not.