Recent content by portdirect

Not fast and loose at all - mfrs can either use a 0.7 multiplier to go from the dyno results to ‘real world’ or use a series of factors to produce a potentially more accurate result. Rivian do the latter for all the measurements they have done to date - which actually shows a minuscule...

using the following: https://dis.epa.gov/otaqpub/display_file.jsp?docid=65095&flag=1 and fitting it to the numbers published at energyefficiency.org: Vehicle EPA doc raw CD range EPA doc raw highway CD range Raw 55/45 blend Published EPA range Inferred EPA factor Effective city range...

Sorry - I meant R1 G1 and G2

Re ADAS compute: Gen1 had no nvidia. Gen2 peak TOPs is around 508 (2x Nvidia Orin AGX at 256 each). Rivian's marketing has muddied the waters here a bit - mostly referring to the capability of a single orin, contrasting that to a pair of RAP1s. Re infotainment: Rivian has never used an...

Also whats up with the paint on the middle one? Doesn't look 'marketing photo ready' by a long shot?

eww. 🤣 (and with that I realize I'm old - thanks I guess @captainjp)

It’s pretty truthful… …if you know how to cut through their marketing. OP looks to be seeing around EPA rated range for their config - 20” AT‘s on a quad gen1 large pack. Rivian get funny by blending drive modes for their ‘headline’ figures - eg the 371 mile ‘rating’ for a gen2 tri with 22”...

I would separate the model from the harness here. Rivian absolutely should and does own the harness: vehicle state, command execution, safety boundaries, permissions, context, memory/chat retention, and the handoff between local and cloud paths. That is where Rivian-specific differentiation is...

That actually makes the governance question more concrete. If Mind is already part of Rivian’s production environment, then this is not just a theoretical future concern. The relevant questions are whether Rivian is paying Mind, whether those terms are arm’s length, whether they were approved...

No - and I dont think there should be any expectation - as you are the one driving the R1. Full context - I asked for similar in my R1T - as at one point of the 6k miles on it (at the time). ~3K had been done by Rivian whist in for warranty work. I didn't expect to get a positive outcome, but...

There is another angle here that is worth separating from the broader automation debate. The issue is not whether industrial robotics is useful. It clearly could be. The governance question is whether Rivian shareholders are fully protected if Mind Robotics becomes deeply tied to Rivian’s...

I mostly park our R1’s outside here in STL - as other have said it’s the cold starts that really take the toll on range. As a result for very short trips (ie ~5 miles or so) can see a massive impact, but by the time we are actually going somewhere I’d put it at around 15-25% impact during Jan/Feb.

I think you’re conflating two separate Rivian claims. Rivian’s R2 media material places the 200 sparse TOPS figure under the R2 infotainment compute module. RAP1 is described separately as part of the Gen 3 Autonomy Computer, which Rivian rates at 1600 sparse INT8 TOPS (using a pair of RAP1...

where did you find that? I’ve only seen some stuff for RAP1 - not the updated infotainment system.

“Local LLM” does not mean much beyond marketing unless the implementation details are clear. “200 sparse TOPS” raises more questions than it answers. At what precision? BF16, INT8, INT4? What sparsity format? Does the runtime actually use it? Most normal LLMs are dense, and I am not talking...