Twitxy Saved

Warren Togami

Verified account This account is verified. Learn more

@wtogami

LC since 2022. Founder @fedora Linux. ex-Red Hat.
Austin, TX Joined May 2012
2.1K Following 14K Followers 23K Posts

Latest posts 5

Warren Togami Verified account @wtogami
1/ RTX 6000 bottlenecks are mostly power/clock/thermal. Max-Q prefill performance 40.8% slower. Decode is only 9% slower. Not saturated: DDR4 or PCIE 4.0 x16.

DeepSeek v4.1 Flash 1017
Threadripper 5995WX DDR4 PCIE 4.0 4xMax-Q 325W
vs
Threadripper 9995WX DDR5 PCIE 5.0 4xWS 600W
Warren Togami Verified account @wtogami
This is a key question for Motorola's new phone that officially ships with GrapheneOS.
Warren Togami Verified account @wtogami 3d
@GrapheneOS Now that GrapheneOS is officially supported by Motorola does that mean it would continue to be blocked by the Play Integrity API while EOL vendor phones lacking years of security updates are considered "secure" and are not blocked?
View quoted post on X
1 replies 1 reposts 10 likes 844 views View on X
Warren Togami Verified account @wtogami
WTF is this?
It lets me write reviews for other products.
I wanted to write that a product is awesome but has a potential safety hazard because it violates spec. But it is awesome specifically because it violates spec.
Warren Togami Verified account @wtogami
A few years ago we discovered a rare crash of a network-facing golang service. I was totally unable to debug or reproduce it locally but it kept happening in production. It turned out to be limited to Alpine and musl libc. DeepSeek v4.1 Flash found it is a musl specific crash.