Still, i don't think it'll need to get much more complex to be very useful for AI workloads.
People have been discovering that more, and simpler, calculations seem to work better? the trend in AI workloads seems to have gone from FP32 -> FP16 -> INT16 -> INT8 and possibly even INT4?
Seems like just having lots of simple calculations is more efficient/effective than more complex stuff.
He did not play a meaningful role in the beginning of the company and is not responsible for its success. Money was responsible
You say that, but applies just as well to the first 2 founders.
the two founders' expertise was responsible
What expertise? Seriously, tell me what they actually brought to the table aside from pitching their idea for a company and attracting venture capitalist money. They registered the name of a company and had ideas. Not expertise. They hired the expertise, with Musk's money.
Speaking of missing the forest for the trees, tell me this: Is an automotive company "founded" as soon as someone registers the name, or when they begin actual engineering efforts towards building an automobile?
I will respond to this by asking "is registering the name of a company the only thing that counts when founding a company?"
Because that's what the original founders did. They registered the name. No patents, no designs, no engineering, no staff. They registered the name, then went searching for VC money.
Yes. Like, they literally corrected everything that went wrong in the first test. And it only took 7 months.
launch pad blown to shreds -> fully intact water suppression system
Engines exploding on takeoff -> all engines on both the booster and ship operational on first ignition
stage separation failed -> HOT staging successful
Self-destruct system didn't destruct fast enough -> self destruct happened immediately
The next launch will probably focus on the fail points of this launch. That is, re-lighting the engines on the booster after turnaround. And whatever caused the starship to go off course (?) and activate the self-destruct.
meanwhile Boeing discovers some valves were stuck, takes half a year to fix it only to discover they're still stuck, gonna need another half a year... oh wait, we took too long trying to fix it, we gotta completely replace them, that'll be another year...
cyberpunk is actually great for that. Sometimes if I don't wanna do anything in particular I'll just... fight the police, lol. Or just randomly do some open world stuff, like drive around the badlands. Then if I'm feeling like doing something more "meaningful" I'll jump into an actual story quest
Ok, it sounds like you're trying real hard to split hairs.
Not just the company itself and Wikipedia say so, but legally, he is a founder. That was the outcome of the lawsuit.
It's true that the first 2 founders legally registered the corporate entity known as "Tesla Motors". Then for the next year, they didn't do jack shit involving anything automotive... they were just going around looking for investors.
Musk was basically their first, and biggest, investor. They didn't actually hire any engineers or, you know, actually start doing anything until Musk's money came into play.
because Google maps DOES list hiking trails. And for the popular trails it's accurate enough. People who have used Google maps to guide their first 5 trails will be led to believe that their 6th one will be just fine and dandy.
Still, i don't think it'll need to get much more complex to be very useful for AI workloads.
People have been discovering that more, and simpler, calculations seem to work better? the trend in AI workloads seems to have gone from FP32 -> FP16 -> INT16 -> INT8 and possibly even INT4?
Seems like just having lots of simple calculations is more efficient/effective than more complex stuff.