Nomecks

Since they renamed M362 to Copilot the usage numbers have been off the chart!

It’s the “Magnificent 7”: Alphabet, MS, Amazon, NVIDIA, Tesla, Apple and Meta

We know what you doing

“EVs lack comparable torque to ICE” - guy in my rearview mirror

Not surprising, the card was filled with… UNCOMPRESSIBLE DATA!

New form of key compression unlocked!

This describes pretty much every large company.

It’s not just the chips that keeps NVIDIA ahead, it’s the 20ish years of optimizing their software across every industry.

I remember when Gentoo came out and people said it was the simple one

• Walk away and find a not incompetent lawyer.

• Send him docx files and tell him if he can’t work with modern files he needs to fix it. It’s like $30/mo for an office subscription. There’s no excuse for him being so cheap.

Please see my original comment.

The difference is you can use inline compression and dedupe in a high performance environment. HDDs suck at random IO.

Major update? 1 hour. Minor update? 1 hour.

Spinning platter capacity can’t keep up with SSDs. HDDs are just starting to break the 30TB mark and SSDs are shipping 50+. The cost delta per TB is closing fast. You can also have always on compression and dedupe in most cases with flash, so you get better utilization.

|sudo bash

So you would be using CoW in-memory in this case?

Is there a benefit to doing CoW with Pandas vs. offloading it to the storage? Practically all modern storage systems support CoW snaps. The pattern I’m used to (Infra, not big data) is to leverage storage APIs to offload storage operations from client systems.