Kadah Coba

CTO. IT cat. Developer for the shitty Firestorm Viewer for Second Life. Skeptic. Secular humanist. Independent. Stubborn. Massive introvert. Feline privateer.

[Kadah Coba @KadahCoba]

@LodestoneRock @MOSS_Global @ArtificialAnlys Or the open weight Krea 2 models. Feels a bit cherry picked.

[Kadah Coba @KadahCoba]

@LodestoneRock That was quick.

[Kadah Coba @KadahCoba]

@discord @saltydkdan You misspelled "disrespectful"

[Kadah Coba @KadahCoba]

@discord Its fixed for myself. Have some users still reporting issues after restarting.

I still have no trust in Discord about any of the renewed claims about not storing PII after the last time. :\

[Kadah Coba @KadahCoba]

@discord Over half of users across all of my servers that can report do have 18+ status but still restricted. I suspect far more effected due being fully locked out or are just part of the large inactive population.

Got a lot of admins that are locked out.

I am locked out on mobile.

[Kadah Coba @KadahCoba]

@LodestoneRock

[Kadah Coba @KadahCoba]

@LodestoneRock One concern for me is that MJ is know for being a generative imagery company. Generative and medial imaging are kinda on opposite end of an several axises. We're talking about an industry that is concerned about the possibility of single dead pixels and uncalibrated displays.

[Kadah Coba @KadahCoba]

@LodestoneRock Hard to tell from their promo images, but it kinda looks like resolution is less than a most of the random CTs I've seen. It would be nice to see a set of comparative DICOMs between CT, MRI, and MS (or whatever acronym they end up with).

[Kadah Coba @KadahCoba]

@PetesQtrLifeCrs @ID_AA_Carmack I'm/In/s

Autocorrect being overly helpful.

[Kadah Coba @KadahCoba]

@PetesQtrLifeCrs @ID_AA_Carmack I'm many contexts, having 2/4/8 fixed values is more information than 1 fixed value that has 2/4/8x greater maximum range.

[Kadah Coba @KadahCoba]

@taker_of_whizz If that model only needs 96GB of VRAM, I will be impressed. The full weights of current open models need closer to 40GB for around 20 second generation. So at for a bit more than 2x the size with that length and consistency, it seems a little unlikely.

[Kadah Coba @KadahCoba]

@discord I haven't seen a faster self-destruct move in quite a long time. Seems like every random server for every varied hobbies are creating new servers on alternate services and planning their migrations.

I hope all these other chat services can handle the sudden scale up and out.

[Kadah Coba @KadahCoba]

@MSFT365Status Zero improvment. Nearly all EXO services are still down.

Have you tried just turning off Copilot and reverting the agentic changes it made instead of continuing to ask it to fix the issue?

Who am I kidding... Copilot is the one writing and posting these status updates. :V

[Kadah Coba @KadahCoba]

@MSFT365Status We're still fully down over here. I'm just going to send everybody home early and hope Microslop can sort out their today-issues overnight.

[Kadah Coba @KadahCoba]

@MSFT365Status O365 services were up about 20-30 minutes ago with intermittent before that, but now most aspects of Outlook/Exchange Online are almost all completely down.

MS should at least credit everybody for an entire day of business interruption.

[Kadah Coba @KadahCoba]

@jiggiefun Cannot read properties of undefined (reading '0')

:V

[Kadah Coba @KadahCoba]

@__tinygrad__ Able to do full extension? Pulling out far enough to lift the cover to access the ram/cpu and not have to derack to service is always preferable. xD

I ask since the rail mount look like Supermico 4-5u and those don't at this depth. Its pain on deep servers with rear mobos.

[Kadah Coba @KadahCoba]

@__tinygrad__ What power supplies does it use. I'm guessing 5x 1.2kw or higher CRPS form factor PSUs with a custom power backplane.

[Kadah Coba @KadahCoba]

@planetdeimos @__tinygrad__ It would be mostly blank space and block airflow. The mobo doesn't even had PCIe slots.

[Kadah Coba @KadahCoba]

@__tinygrad__ @latentspacehack What was the ambient cold side temp? Noise levels?

Our current self-built 8x4090 4u server runs at similar temps with 25C ambient using 6 3196-FN-GALE-01-ND fans at low to mid speed. 2-slot GPUs make things easier, but trying to get those in volume would invite nv's wrath.