Sure but it doesn't launch or land humans on Earth, which is the hardest part. And that's fine for now. Once it's launched and landed a few hundred times we can think about launching people on it.
SLS might be less safe than some hypothetical future version of starship, but in reality the current starship's most recent flight had a 15 minute window where a no-go for orbit was called and the mission plan would have resulted in a loss of crew. Safe crewed launches and earth landings are not a priority for them and likely won't be for a long time. Even then, SLS+Orion have the advantage of an LES and the option for unpowered landings essentially anywhere, both of which are fundamentally unworkable with starship's design.
Liquid fueled rockets are intrinsically safer than solid fuel rockets. There is a floor on just how low you can get the risk level with solid fuel, you can't test the rockets you fly, and the failure mode is catastrophic blow-up à la Challenger.
It's like saying flying is safer than driving per passenger-mile. It is. By orders of magnitude. I don't care how safe your new Volvo or electric vehicle is, flying in a helicopter is still safer, and a passenger jet vastly more so.
"But Starship blew up 5 times!" Yes, and flying a test plane is dangerous too. Apples to oranges. Starship and SLS have vastly different design certification strategies. SLS spends 100x as much to deliver something that works as reliably as it ever will by the second or third launch. SpaceX gets Starship flying quickly and discovers bugs in prod. So you see more failures. But even now, Starship is probably safer than SLS ever will be. If not now, then very, very soon.
SLS has an abort mechanism. That alone mitigates most risks of the first stage in a way starship never can, meaning the latter has to fight an uphill battle to deliver the same safety through flight numbers and experience.
> But even now, Starship is probably safer than SLS ever will be. If not now, then very, very soon.
Any failure to orbit is essentially a guaranteed loss for Starship. Any booster explosion or total engine failure on starship would also be essentially a guaranteed loss. Even the space shuttle had better backup options than the current version of starship.
Idk about you, but if I'm the astronaut I'd take 1-(P(SLS failure)*P(LES/parachute failure)) over P(superheavy success)*P(Starship to orbit)*P(Raptor relight) every time.
You're telling me starship is safe for humans when they failed to land the booster?
How about we let them succeed first before pressuring them to perform and inevitably embarrass themselves because the silly SpaceX fans were too smug on the internet.
I like the rocket less, imho solids have no place in human spaceflight, but the Orion has a LES, which elevates it to being more safe than any spacecraft without one.
Which makes the SLS safer than the Shuttle but airliners also don’t have one. They are considered safe because of how much they have flown and that is how Starship will be determined safe enough.
Fuel transfer seems like a trivial problem next to the reusable heat shield. And even if for some weird reason it never works, the program could likely pay for itself just launching Starlink/Starshield with full reuse.
Or use SuperHeavy to launch non-reusable second stage and fueled Moon module of different design. After all SuperHeavy doesn’t have to be used exclusively for Starship.
Yeah, I'm eagerly awaiting results from inspection of the heat shield (which hopefully they will disclose). Reflight without refurbishment is the last goal they need to hit before the system can be considered a success. After that, optimizing for increased payload and nailing reliability will be the things to work on. And of course orbital refueling for HLS.
They already recovered a ship from that one time they accidentally landed one in the ocean. SpaceX got to have a lot of heatshield endurance data by now.
They don’t need to fly without refurbishing as long as they have minimal refurbishing needed. They are able to get cheaper than F9 per kg by reducing Raptor costs.
It’s probably better to think of it on the spectrum of “atmosphere-rated” vessels. Ships landing in zero-g don’t need feet, they need clamping surfaces. Atmosphere rated might need both, depending on usage. The is just an almost atmosphere rated craft. If the moon and mars weren’t covered with horrible dust we could have it land on feet. But it is instead some of the worst dust in the solar system so we would have to catch it even there wasn’t a big weight pensly for carrying it around.
The method seems sound and the findings correct. But I think many are missing that (as the research itself says) this doesn't prove any illegal behavior. It's perfectly legal for Chinese companies to train models on GPUs that they rent outside of China. It's also legal for companies to buy GPUs and install them in Malaysia with the intention of renting them to Chinese companies.
There isn't even an allegation that the Chinese companies lied about this. They simply didn't disclose which chips were used. In all cases where they did disclose the training chips, this research found the disclosures to be accurate.
The small print in TFA states the many limitations of their method, i.e. that it actually provides no evidence about which GPUs were really used for training, and much less about who was their owner and where they were located.
The method just strongly suggests that the initial values of the weights, before starting training, were generated with the NVIDIA cuRAND library on a Blackwell GPU (or on an AMD GPU, where the NVIDIA library is emulated for compatibility).
The training might have been done with thousands of GPUs of other types. It is plausible that the GPUs used for training or at least a part of them may have been of the same type with the one used for initializing the model, but it is not proven in any way.
A scenario that matches what is reported in the article, but in which there is only a single Blackwell GPU involved, is if the LLM was intialized on the workstation of some lead developer, who had the privilege to use the best GPU they had, and then it was transferred to a server cluster for training, where other kinds of GPUs were used.
Now that this has been published, it would be very simple in the future for any Chinese AI company to avoid this "fingerprinting", by just using a different PRNG, instead of cuRAND.
Yes it's simple to avoid this fingerprinting now that people know about it, but nobody knew about it before. I don't believe people are casually transferring snapshots between chip architectures without a really good reason, either. It's hard enough to coordinate these training runs without involving multiple chip architectures in a single run.
'Legal' and 'illegal' apply to Nvidia or other US companies facilitating these activities. Perfectly legal for Nvidia to supply GPUs to be rented by Chinese companies, or for Oracle to build datacenters for Chinese companies to rent, as long as they are installed outside of China.
You'd think so but the USA is also the country that's most advanced at getting its legal system to apply outside of it. E.g. they shut down the International Criminal Court.
Indeed, but the USA has been flexing it's muscles against small countries all over the world for decades. However, go back and study the Korean war and Vietnam war, where China and even Russia intervened, and the US lost their arses and was sent packing both times. My eldest uncle was an officer/pilot in the USAF involved in the Vietnam war, flew many bombing missions. What do we have to show, for both those conflicts ?
Not nostalgia for me, I play Quake 3 about 15 times a week with a group of people, it's still awesome fun. We're all so good at it, it's genuinely a difficult competition every time. After 15 years we still get in ridiculous/interesting situations in the same maps all the time.
Same here. My BFF and I have played Q3 on and off since it was released. But we've clocked more hours in the past couple years than in the previous ten.
Even after all this time we have a blast and sometimes situations erupt into huge bouts of laughter. I'll never not think that an errant rocket splash knocking him off the map is hilarious.
Starlink Direct-to-Cell fills that use case better than Starlink Mini. It works with existing cell phones. Unlike the iPhone's built in satellite feature, it doesn't require pointing your phone at the satellite, and it supports WhatsApp, Google Maps, and several other apps in addition to RCS and SMS.
Removable battery is cool. Does "quick-swap" mean you can switch the battery while it's operating? I always thought it would be a good idea for devices like this to have a tiny secondary battery or large capacitor in the main unit that would let it keep going for a few seconds while you swap the main battery.
My first laptop back in... 2001? had this, it was very handy especially because batteries didn't last as long back then and hibernate was either non-existent or at least less reliable. You could also get a wall charger for the battery so you could charge them up outside of the laptop.
Also curious about this; I'm not much of a tinkerer but I did see their battery has a USB-C port. Is there some efficiency loss in using a spare battery as a power bank rather than actually replacing the battery where it lives?
I wonder about the potential for iron nitride magnets. They're actually theoretically stronger than neodymium, but demagnetize more easily. Would it be possible to design competitive servos around their limitations?
reply