706 karma · joined July 29, 2008
Its eye-roll inducing.
AWS charges $32/hr for an 8xA100s (p4d.24xlarge) which comes out to $4/hour/gpu. Yes you can get lower pricing with a 3 year reservation but thats not what this question is asking.
You also need 256 nodes to be colocated on the same fabric -- which AWS will do for you but only if you reserve for years.
https://fortune.com/2023/03/11/silicon-valley-bank-svb-ceo-g...
https://techcrunch.com/2023/03/01/addressing-criticism-opena...
The US, perhaps acting on intelligence preceding the Viasat attack, provided Zelenskywith an Iridium 9575A.
https://www.cnn.com/europe/live-news/ukraine-russia-putin-ne...
This has nothing to do with the M1 silicon nor is it a design flaw.
Learning concepts like branch prediction, out-of-order execution, pipelining, register planning, caches, memory management is far far more important than Lexing/Parsing/Intermediate Representations and SSA.
What it is: they made a poison which will kill a cell when two-parts of the poison touch each other in a certain way (think bubble gum in mission impossible). They've designed the parts of the poison so it should only trigger when a virus is present.
Here's why no big pharma compnay will touch it: the criterion for "virus present, KILL CELL" is potentially highly highly error prone, could easily go haywire killing all the cells.
Contrast to Immunotherapy -- the state of the art in clinically approved cancer treatments which uses the body's own immune system to selectively kill cancer cells (which also faced significant financial barriers to its development), even if the signal goes haywire, the body's immune system still has fail-safes to not killing all of its own cells.
disclaimer: i am not a biologist, but i took some basic college bio courses
From wikipedia: "Differentiation between infected and healthy cells is made primarily via the length and type of RNA transcription helices present within the cell. Most viruses produce long dsRNA helices during transcription and replication. In contrast, uninfected mammalian cells generally produce dsRNA helices of fewer than 24 base pairs during transcription. Cell death is effected via one of the last steps in the apoptosis pathway in which complexes containing intracellular apoptosis signaling molecules simultaneously bind multiple procaspases. The procaspases transactivate via cleavage, activate additional caspases in the cascade, and cleave a variety of cellular proteins, thereby killing the cell."
From the DRACO paper (http://journals.plos.org/plosone/article/asset?id=10.1371%2F...) :
"In its simplest form, a DRACO is a chimeric protein with one domain that binds to viral dsRNA and a second domain (e.g., a procaspase-binding domain or a procaspase) that induces apoptosis when two or more DRACOs crosslink on the same dsRNA. If viral dsRNA is present inside a cell, DRACOs will bind to the dsRNA and induce apoptosis of that cell. If viral dsRNA is not present inside the cell, DRACOs will not crosslink and apoptosis will not occur."
What was the story here? I'm not familiar with this and I can't seem to find anything on Google about it :)
https://www.mail-archive.com/golang-nuts@googlegroups.com/ms...