James Ostheimer
5.3K posts

James Ostheimer
@jostheim
physics, astronomy and data
Charlottesville, VA Katılım Haziran 2007
413 Takip Edilen412 Takipçiler

.@Frigidaire my oven racks no longer fit my oven, the oven warped which seems like an enormous quality control issue
English

who is doing #stablediffusion of how to videos? So i could prompt “how to replace the carburetor on a Poulan chainsaw?” and i get a customized how to video. Plenty of training data on youtube.
English
James Ostheimer retweetledi

@jamesdensmore followed by my presentation “neither snowflake or duckdb will fix your orgs incentive misalignment” 😀
English

@BrianRoemmele so they just calculate Pi non-stop or something? what a wasteful design. my background is in physics so i’ll just accede to you given your background, I am astonished that ARMs take full power all the time
English

@jostheim James, just an aside. I have a background in chip design.
English

A quick look at nvidea-smi says this is simply not true, my 1080 ti uses 70W at 0% and 205 W at 100%.
Brian Roemmele@BrianRoemmele
Reality check 98.771% of your CPU/GPU cycles do almost nothing—most of your multi-cores are idling yet are STILL consuming energy. If mining took the trillions of idle CPU/GPU cycles YOU would get paid and very little new net energy used. This could have been ETH proof-of-work.
English

@BrianRoemmele Same thing with the cpu, if you are doing processing you are using energy… I am not sure exactly what an idle processor does, but I do know a lot of progress has been made in lowering energy usage when inactive. I don’t think there is a there there.
CHARLOTTESVILLE 🇺🇸 English

@jostheim James, indeed. Yet many of the SETI @ Home type of calculations will not require just the GPU but the cores idling in the ARM CPU.
English

@BrianRoemmele If you ask your GPU to do calculation it will use more power, that is simply thermodynamics.
English

@jostheim James, indeed IF we use the same type of mining code. I am not proposing using a system that always favors larger compute power. In fact it could benefit by using nominal power.
English
James Ostheimer retweetledi

causality is hard… there are so many confounders here my guess is Scrum was never the real problem
Gergely Orosz@GergelyOrosz
Skype went all-in on Scrum 2011-12 right as WhatsApp overtook it in market share. I asked the WhatsApp founder Jan Koum if they used it at all. His response: “To be honest i have no idea what this Scrum garbage is and we never uttered the word Scrum when i was at WhatsApp.”
English

@UHaul_Cares your upacks arrived 4 hours after their window and caused me to spend $600 on movers to stand around only to then have me have to move everything… i want my $600 and time back
English

Mission Driven, hope everyone remembers that blog post…
.@WallStCynic
2.9%…?! $COIN
MILWAUKEE 🇺🇸 English


@titusandtate this was a special episode… the him, one of dem ones, and who got the dog in em’ is a timeless debate
English

