Meanwhile everybody has seemingly forgotten how groundbreaking the Ryzen platform was/is. Everybody thought Intel and Xeon was unstoppable until Ryzen came out and Epyc ate Intel’s lunch to the point that they’re nearly irrelevant. I literally have a desktop motherboard running an AMD Pro APU with ECC for a NAS that uses like 50W of power at load - it can’t really be overstated how unthinkable that was a decade ago or how wildly expensive/power hungry it would have been to do that on an Intel mobo (if you could even run it at a desktop form factor and not on a server blade). Also the fact that each new generation has largely been backwards compatible and doesn’t force new mobo or hardware upgrades to leverage improvements is extremely nice.
Nvidia has a lot bigger head start on AMD in a lot of ways because they only really started shaking the cobwebs off like 10-12 years ago. But they also are developing GPU capabilities pretty steadily and ROCm is picking up steam and also AMD consistently develops OSS drivers/tooling and also supports Linux natively to the point that amdgpu on Linux oftentimes performs better than the Radeon drivers on Windows. People want so badly to pile on hate but medium/long-term stuff Lisa Su has done has really been to the benefit of both enterprise and hobbyist consumers.
Not immediately off-hand. I’ve been eyeing a few of the newer ones that seem promising like Fluxer, Stoat, etc. but haven’t deployed any of them. Ultimately I mostly just still use my matrix.org account and think I’m largely settling on hosting xmpp for friends because ejabberd is trivial to deploy and Movim (https://movim.eu/) seems like a pretty mature client atp and supports a lot of the features that someone who wants to migrate off Discord would want.
I imagine the landscape might continue to develop but ultimately self-hosting a service comparable to Discord isn’t easy and people genuinely want to flock onto the thing that just works or has the lowest barrier to entry.
Matrix is not really much of a viable alternative to Discord in the way that you present it. For e.g. one big barrier is a lack of RBAC in the spec (https://spec.matrix.org/v1.19/client-server-api/#mroompower_levels) having power_level mirror classic IRC style permissions isn’t really flexible enough to provide the same behavior you get from a Discord server. It would be awesome if you had the ability to have a
m.spaceand provide different permissions to child objects or assign roles so that you can opt into seeing particular text channels similar to how larger/well-organized Discord servers work.The fragmentation across clients is also still a pretty big sticking point. Some of the more popular clients still don’t support spaces (like you mention) and threads which introduces a lot of friction. Also the entire way that e2ee works when you want to use >1 device is pretty much a nonstarter for any casual user which kind of defeats a lot of the points of using Matrix to begin with.
There’s also historically (not sure if it’s still an issue because I’ve largely detached myself from Matrix development after Element went all-in on supporting government sectors as their primary demographic) been problems of Synapse implementing experimental features before the MSCs are fully ratified which created churn across homeservers and oftentime left you with limited choice but to use Synapse which carries its own set of annoyances because of how heavyweight and sometimes difficult it is to administer.
I’d love for Matrix to succeed and I’ve been rooting for and using it for a decade, but at this point I’m not so sure it’s actually going to happen and have largely resigned to the “maybe all the IRC v3 and XMPP people have been right all along” camp.
Introducing a revolutionary new cube where each cell is a dynamic LCD. Pair the cube to our app and for only $9.99/mo you can have access to unlimited cubing puzzle challenges.
Does the announcement or article specify if this something that will be able to be used in-lieu of prometheus node_exporter? It’d be nice to have a built-in way to ship logs and metrics without having to deploy additional tooling on every server.
E: further digging says not yet but its intended to be laying the base layer for that functionality
1 points on AMD’s Twitter brag that it ‘loves gamers’ met with savage Community Note in !pcgaming@lemmy.ca ·
I don’t think Intel’s mismanagement is solely what caused AMD to produce a genuinely better platform that has compatibility across server/desktop. Their mismanagement enabled AMD to take over those spaces very quickly but it was genuinely good engineering that let the timing of Intel being complacent working out in their favor.
The GPU space is a completely different competitive landscape. My only point was that up to this point AMD under Lisa Su’s leadership has been a much better company overall. I think they are genuinely working on improving their graphics stacks to be competitive with Nvidia but there’s a much larger gap to close, but now that they have largely captured the CPU server space they can focus on building out GPU capabilities but it’s going to take time. I’m merely trying to dispel the GPs point that it’s some kind of conspiratorial plot by AMD to deliberately be bad or something by pointing out there’s a decade long track record to the contrary (at least for now).