eldenring

Whats your definition for reasonable temp? my envelope math tells me at 82 celsius (right before h100s start to throttle) you'd need about 1.5x the surface area for radiators. Not exactly back to back, but even 3x surface area is reasonable.

Also this assumes a flat surface on both sides. Another commenter in this thread brought up a pyramid shape which could work.

Finally, these gpus are design for earth data centers where power is limited and heat sinks are abundant. In the case of space data centers you can imagine we get better radiators or silicon that runs hotter. Crypto miners often run asics very hot.

I just don't understand why every time this topic is brought up, everyone on HN wants to die on the hill that cooling is not possible. It is?? the primary issue if you do the math is clearly the cost of launch.

Interesting.

According to this other source https://www.satellitetoday.com/connectivity/2026/02/02/space...

the filing mentions this

> these satellites would operate between 500 km and 2,000 km altitude and 30 degrees and Sun-Synchronous Orbit inclinations (SSO)

Interesting, That could surely simplify things.

Solar in space is a very different energy source in terms of required infrastructure. You don't need batteries, the efficiency is much higher, cooling scales with surface area (radiative cooling doesn't work as well through an atmosphere vs. vacuum), no weather/day cycles. Its a very elegant idea if someone can get it working.

Yeah the main benefits are:

1. solar is very efficient at generating energy, no moving parts, simple physics etc.

2. in space you don't deal with weather or daylight cycle, you can just point your panels at the sun and generate very stable energy, no batteries required

3. environmental factors are simpler, no earthquakes, security, weather. Main problem here is radiation

In theory its a very elegant way to convert energy to compute.

The radiators would be lighter compared to the solar panels, and slightly smaller surface area so you can line them back to back

I don't think dissipating heat would be an issue at all. The cost of launch I think is the main bottleneck, but cooling would just be a small overhead on the cost of energy. Not a fundamental problem.

You can line the solar panels and radiators facing away from each other, and the radiators would take up less surface area. I think maybe the tricky part would be the weight of water + pipes to move heat from the compute to the radiators.

Google tested the radiation tolerance of tpus which include hbm and they performed fine. https://research.google/blog/exploring-a-space-based-scalabl...

the radiators would be lighter compared to the solar panels, and slightly smaller surface area so you can line them back to back

these same comments pop up every time someone brings up satellite data-centers where people just assume the only way of dissipating heat is through convection with the environment.

There aren't too many base models out there to compare against.

I think this is more of a result of big US tech being extremely productive (with their main competency)

I think Claude is genuinely much smarter, and more lucid.

Google did a study with their TPU v6

> For ML accelerators to be effective in space, they must withstand the environment of low-Earth orbit. We tested Trillium, Google’s v6e Cloud TPU, in a 67MeV proton beam to test for impact from total ionizing dose (TID) and single event effects (SEEs). > > The results were promising. While the High Bandwidth Memory (HBM) subsystems were the most sensitive component, they only began showing irregularities after a cumulative dose of 2 krad(Si) — nearly three times the expected (shielded) five year mission dose of 750 rad(Si). No hard failures were attributable to TID up to the maximum tested dose of 15 krad(Si) on a single chip, indicating that Trillium TPUs are surprisingly radiation-hard for space applications.

You should read the linked article, they talk about it there. You radiate the heat into space which takes less surface area than the solar panels and you can just have them back to back.

In general I don't understand this line of thinking. This would be such a basic problem to miss, so my first instinct would be to just look up what solution other people propose. It is very easy to find this online.

I think the main draw is its elegance. You have very efficient power from the sun, put that directly into your compute, radiate it out. Energy is ~free, no heavy infrastructure required, just a closed circuit for computing.

I'm guessing they were waiting to figure out more efficient serving before a release, and have decided to eat the inference cost temporarily to stay at the frontier.

the only real benefit is privacy which 99.9% of people dont get about. Almost all serving metrics (cost, throughput, ttft) are better with large gpu clusters. Latency is usually hidden by prefill cost.

Depending on how hot/dense/clocked you run your compute, the Radiators take _less_ surface area than the solar panels, so you can have them back to back and the radiators will take less space.

Obviously there are some unanswered questions but there is clearly a path forward.

The comment you were replying to mentioned this. Yes you cant remove heat via convection, but you can use radiators to emit heat as radiation into space.

There's a great way to express the utility something has to you to get it. You can spend more money.

Its the other way around too, HLE questions were selected adversarially to reduce the scores. I'd guess even if the questions were never released, and new training data was introduced, the scores would improve.

I think something with more uniform training and inference setups, and otherwise equally hardware friendly, just as easily trainable, and equally expressive could replace transformers.

You can just turn off GC for the duration of the match, or during rounds.

Buy phase is also used for navigating the map on your "side" before you can shoot at the other guys. Also for planning.

I find a lot of Rust libraries "seem" dead, based on github activity, but looking into it, they are actively used in many projects. I think Rust projects just tend to have less open issues, and don't need to be maintained as often. This is also the case internally at my company.

the 2nd token is generated without knowing what token was chosen for the 1st token

Ah interesting, good point. So I guess expert-choice routing leaks across the batch. Now I'm not sure.

Very impressive! I guess this still wouldn't affect their original example

> For example, you might observe that asking ChatGPT the same question multiple times provides different results.

even with 0.0 temperature due to MOE models routing at a batch level, and you're very unlikely to get a deterministic batch.

> Not because we’re somehow leaking information across batches — instead, it’s because our forward pass lacks “batch invariance”, causing our request’s output to depend on the batch size of our forward pass.

The router also leaks batch-level information across sequences.

So what do you do if you have a better product and a "name brand" disadvantage? Advertising commodifies information flow instead of letting it pool with the people who already have access to it. Think of all the products that got big nowadays because they could convince VCs to fund ad spend, and saw a return for it.

I think advertising has a huge, positive, 2nd order effect on the world.