I am building a free poker training web app, specifically aimed my friend JG who wants to place sufficiently high enough in a WSOP or WPT satellite tournament to qualify for the main tournaments: https://holdempuzzles.com/
With Ollama i got the 20B model running on 8 TitanX cards (2015). Ollama distributed the model so that the 15GB of vram required was split evenly accross the 8 cards. The tok/s were faster than reading speed.
Also, when I am not driving my car I park it, it is not adding to congestion. Waymo Taxis roam the streets in SF adding to congestion. I cant see how congestion can be reduced...the worst congestion happens during rush hour...replacing personal cars with a Waymo will have no effect.
You can only fit 3 cars across on most side streets near me. You have more space, but does it really create better flow?
Many of these side streets and neighbourhoods are designed and modified NOT to flow, because then people would use them to get to work. Would this change that?
While I admit I see these more in the ‘burbs, this is becoming more widespread and is no longer exclusive to them.
It does increase throughput. And it’s not relevant to a weighing of reducing ridesharing and increasing lanes—both would induce demand.
And as another person said, a single streetside-parked car has made that entire street more dangerous for bikers. Take out streetside parking and you can add lanes for cars, light rail, busses, bikes or even just widen sidewalks and add more trees.
Parked cars add to congestion far more uselessly than taxis, ride sharing vehicles or AVs.
Streets with parked cars are residential and don’t need better throughput. They need safety rails so little kids can exist without being run over. They need to be exempt from gps routing so people can quietly live their lives.
In some places throughput isn’t needed… safety is.
You could widen sidewalks, place bike lanes, or plant trees in those parking spaces for the equivalent effect. Also the residential dichotomy isn’t true for lots of areas of the city. I live next to an office building, for example.
Unless every passenger pickup is at the location of the last drop off and the timing is perfect, some portion of robotaxi driving must be empty as the taxi drives to the next pickup or at least goes somewhere to park. And I'm honestly not sure Waymos are smart enough to know where it is and isn't legal to park for some unknown period of time while waiting for the next trip, so that might not even be an option.
Contrast this with personal vehicles, which are always transporting at least one person to their destination. It seems essentially impossible that robotaxis could result in less congestion unless people share robotaxis significantly more frequently than they do personal cars or regular taxis.
I'm immediately suspicious of any chain that links "denser environments" to "less congestion", since everywhere I've ever been, the densest environments have the most congestion.
It's a bit like adding extra lanes, to some degree, demand expands to meet capacity (But I mantain that in this case the net effect is possitive)
There would be less space to be congested by fewer vehicles, but in this context, less congestion also means fewer people experience the congestion directly (because it also works to disincentivize car usage), but those affected have it the same or worse.
I wouldn't take my car to a large city center if I can at all avoid it, which seems to be the common reaction. These people are "transparent" to the congestion—they don't add to it and (mostly) aren't affected by it.
If it’s parked on the street, that is taking up a lane that could be carrying traffic. (And in some cities it’s common for parallel street parking to turn into a lane from 4-8 PM or similar)
Robo-taxis can pack significantly more densely into a dedicated parking space than regular cars. Snout to butt and no room side to side, as there's no need for maneuvering. A robo-taxi parking lot just becomes a dense FIFO queue.
Other folks can use it too.