This is a really fun problem! I suggest anyone who likes optimization in a very broad sense to try their hand at it. Might be the most fun I've had while interviewing. I had to spend a week-worth of evenings on it to fully scratch the itch, and I managed to get 1112 cycles. But that was mostly manual, before the current crop of agentic models (clopus 4.5, gpt5.2). I wonder how far you can RalphWiggum it!
What a garbage way to gate visibility to the sport. You are already looking at a niche audience who would be interested in the idea, and you hope to collect some emails for marketing as well?
> MTG-S1 is the first geostationary meteorological sounder satellite to fly over Europe
I was confused for a minute on how it's both _geostationary_ and _over Europe_ -- you can't be geostationary if your orbit is not over the equator!
Turns out[1] the MTG-S1 satellite is in fact geostationary and parked at exactly 0°00'00"N 0°00'00"E (off the coast of Ghana), 42164 km up from the center of Earth, it's just pointing at Europe at an angle.
I had doubts about the "parked at exactly 0°00'00"N 0°00'00"E", thinking it was over Null Island just because the data wasn't updated yet and it was showing uninitialized values.
So there could reasonably be dozens [0] of satellites "parked at exactly 0°00'00"N 0°00'00"E". Definitely an unnecessary level of precision.
[0] A few sites give 10km as a standard minimum separation for geostationary satellites. That theoretically allows a thousand of them in the 100km cube, but I am guessing a lattice of them every 10 km in all 3 dimensions would not be manageable.
I don’t think that is a management problem but the mechanics will work against you and you would squander too much of the precious fuel if you stack’em in three dimensions.
But some geostationary satellites are close enough so that there can be failover without adjusting receiving antennas on the ground.
So you can of course keep them dense around the equator. Probably very close down to hundreds of meters (if not less) if you coordinate the station keeping. After all the forces that push or pull the satellites out of orbit (tidal forces and particle streams) should be very similar for close neighbours. Problem is that you have to share the bandwidth of the up- and downlink then because the dishes of the groundstations cannot focus so sharply.
Given that, and redundancy put aside, one bigger satellite with more payloads would usually be cheaper than two smaller ones without any disadvantages.
NOAA/NASA (USA), EUMETSAT (European organization), JMA (Japan), KMA (Korea), and CMA (China) all have a geostationary satellite (one or more actually). So, northern hemisphere countries, but the coverage is global thanks to the fact that you need to be, as you say, above the equator.
I am surprised they would pick 0 for the latitude, it seems that most of Europe, whether it's the land or the people is east of that. Maybe some important weather systems develop over the Atlantic and they want to track that?
It’s exactly that. In fact, information propagates along with the winds. If you don’t observe upstream, you instead propagate an information hole. Each new model run incorporates the output of the previous run to preserve sparse weather information. It’s not that there are few observations, it’s that Earth is really big.
How the heck is 0,0 still available! Was nobody interested in this position before for any purpose?
Is there a way to list what's all in geostationary orbit (either stationary at the equator, or at which longitudes they commonly cross through the equator)? Edit: found https://en.wikipedia.org/wiki/List_of_satellites_in_geosynch... (geosynchronous is a superset of geostationary). The closest is H2Sat at 0.5°. Article notes: "Some of these satellites are separated from each other by as little as 0.1° longitude [or] approximately 73 km". Trickier than keeping them apart is apparently getting a narrow enough communications beam width. /edit.
How long until we can see this ring above the equator from the ground? Although I guess the thickness would rival Saturn's rings and we would probably not be able to make it out even if the sats were shoulder to shoulder. We do see satellites from the ground when the sun hits them right, but those are typically around 1000x closer
Tbh it still puzzles me why gameplay degradation specifically was chosen as a way to try to discourage piracy. I imagine many more people hit the degradations, thought the game was just buggy and abandoned it, compared to people who were motivated by bad gameplay to give the developers money.
The mindfuck angle is pretty effective though. This article wouldn't have been written otherwise.
I have a vague memory of a "game-dev studio tycoon" sim game which, if you played the pirated version, would have your sales taper off super hard and you'd go bust because pirates. There was, however, an explicit nod to this happening and it was at least clear that the failure was making a point
1. It is harder to see if your crack was successfully completed, especially if the degradation happens late in game.
2. If the game is fun until it goes wonk and the person learns it is pirated, they may decide to buy the real deal.
3. The potential damage, if you didn't have a noticeable false-positive rate, is limited and for those unwittingly hit and find out their software is pirated, they're likely to not buy/get from the downloaded source again.
4. From the standpoint of a developer, it is creative fun.
> Tbh it still puzzles me why gameplay degradation specifically was chosen as a way to try to discourage piracy.
Yeah, I get that maybe it made the developer and/or publisher feel some kind of 'justice' was done, but it's ultimately bad for your brand to have any game out in the world that has subtly degraded performance. The players of pirated versions probably just assumed the company makes games that are buggy or with really bad difficulty scaling. Reducing piracy by making players not want your games doesn't seem like a winning long-term strategy.
Some publishers instead layered their piracy checks later in the game play or delayed stopping pirated play until some number of game events after detection. If they were concerned crackers would find an explicit error message, another option is to change game play in some other way like this game ended the luge race before the player finished. A legit game behavior at the wrong time is still harder to find than an error box or specific text.
This is the question and for people that are older the scales steadily tip towards just taking the damn drug because in 20 years they're likely to be dead anyways.
That's where I am too right now for personal projects, and I ended up reimplementing parts of Dokuploy for that, but I don't feel much of a need to move from "fun little docker compose" for some reason
That's a great point! I'd agree that just the extra emotional motivation from having your own thing is worth a ton. I get some distance down that way by having a large RAM no GPU box, so that things are slow but at least possible for random small one offs.
I was thinking of doing something similar, but I am a bit sceptical about how the economics on this works out. On vast.ai renting a 3x3090 rig is $0.6/hour. The electricity price of operating this in e.g. Germany is somewhere about $0.05/hour. If the OP paid 1700 EUR for the cards, the breakeven point would be around (haha) 3090 hours in, or ~128 days, assuming non-stop usage. It's probably cool to do that if you have a specific goal in mind, but to tinker around with LLMs and for unfocused exploration I'd advise folks to just rent.
> On vast.ai renting a 3x3090 rig is $0.6/hour. The electricity price of operating this in e.g. Germany is somewhere about $0.05/hour.
Are you factoring in the varying power usage in that electricity price?
The electricity cost of operating locally will vary depending on the actual system usage. When idle, it should be much cheaper. Whereas in cloud hosts you pay the same price whether the system is in use or not.
Plus with cloud hosts reliability is not guaranteed. Especially with vast.ai, where you're renting other people's home infrastructure. You might get good bandwidth and availability on one host, but when that host disappears, you should hope that you did a backup, which vast.ai charges for separately, and if so, you need to spend time restoring the backup to another, hopefully equally reliable host, which can take hours depending on the amount of data and bandwidth.
I recently built an AI rig and went with 2x3090s, and am very happy with the setup. I evaluated vast.ai beforehand, and my local experience is much better, while my electricity bill is not much higher (also in EU).
Sure, but unless you're using them for training, the power usage for inference will vary a lot. And it's cumbersome to shutdown the instance while you're working on something else, and have to start it back up when you need to use it again. During that time, the vast.ai host could disappear.
Most people don't think of storage costs and network bandwidth. I have about 2tb of local models. What's the cost of storing this in the cloud? If I decide not to store them in the cloud, I have to transfer them in anytime I want to run experiments. Build your own rig so you can run experiments daily. This is a budget rig and you can even build cheaper.
Let me add that moving data in and out of vast.ai is extremely painful. I might be overprivileged with a 1000 MBit line but these vast.ai instances have highly variable bandwidth in my experience; plus even when advertising good speeds I'm sometimes doing transfers in the 10-100 KiB/s range.
> Well if you are not using a rented machine during a period of time, you should release it.
If you're using them for inference, your usage pattern is unpredictable. I could spend hours between having to use it, or minutes. If you shut it down and release it, the host might be gone the next time you want to use it.
> what do you use a 2x3090 rig for? Bulk not time-sensitive inference on down quanted models?
Yeah. I can run 7B models unquantized, ~13-33B at q8, and ~70B at q4, at fairly acceptable speeds (>10tk/s).
if you are just using it for inference, i think an appropriate comparison would just be like a together.ai endpoint or something - which allows you to scale up pretty immediately and likely is more economical as well.
Perhaps, but self-hosting is non-negotiable for me. It's much more flexible, gives me control of my data and privacy, and allows me to experiment and learn about how these systems work. Plus, like others mentioned, I can always use the GPUs for other purposes.
to each their own. if you are having really high-sensitive conversations with your GAI that someone would bother snooping in your docker container, figuring out how you are doing inference, and then capturing it real-time - you have a different risk tolerance than me.
i do think that cloud GPUs can cover most of this experimentation/learning need.
with runpod/vast, you can request a set amount of time - generally if I request from Western EU or North America the availability is fine on the week-to-month timescale.
fwiw I find runpod's vast clone significantly better than vast and there isn't really a price premium.
With the current more-or-less dependency on CUDA and thus Nvidia hardware it's about making sure you actually have the hardware available consistently.
I've had VERY hit-miss results with Vast.ai and I'm convinced people are cheating their evaluation stuff because when the rubber meets the road it's very clear performance isn't what it's claimed to be. Then you still need to be able to actually get them...
Unless you are training, you never hit peak watts. When inferring, the watt is still minimal.
I'm running inference now and using 20%. GPU 0 is using more because I have it as main GPU. Idle watt sits at about 5%.
Device 0 [NVIDIA GeForce RTX 3060] PCIe GEN 3@16x RX: 0.000 KiB/s TX: 55.66 MiB/s
GPU 1837MHz MEM 7300MHz TEMP 43°C FAN 0% POW 43 / 170 W
GPU[|| 5%] MEM[|||||||||||||||||||9.769Gi/12.000Gi]
Device 1 [Tesla P40] PCIe GEN 3@16x RX: 977.5 MiB/s TX: 52.73 MiB/s
GPU 1303MHz MEM 3615MHz TEMP 22°C FAN N/A% POW 50 / 250 W
GPU[||| 9%] MEM[||||||||||||||||||18.888Gi/24.000Gi]
Device 2 [Tesla P40] PCIe GEN 3@16x RX: 164.1 MiB/s TX: 310.5 MiB/s
GPU 1303MHz MEM 3615MHz TEMP 32°C FAN N/A% POW 48 / 250 W
GPU[|||| 11%] MEM[||||||||||||||||||18.966Gi/24.000Gi]
When you compute the break even point did you factor in that you still own the cards and you can resell them? I bought my 3090s for 1000$ and after 1 year I think they go for more in the open market if I resell them now.
I just made a clone of diskprices.com for GPUs specifically for AI training, and it has a power and depreciation calculator: https://gpuprices.us
You can expect a GPU to last 5 years. So for 128 days break even you are only looking at 6.67% utilization. If you are doing training runs, I think you are going to beat it easily.
P.S. coincidentally or not, but shortly after it got mentioned on Hacker News, Best Buy run out of both RTX 4090s and RTX 4080s. They used to top the chart. Turns out at descent utilization they win due to the electricity costs.
the current economics is a low ball to get costumers. it's absolutely not going to be the market price once commercial interests have locked in their products.
but if you're just goofing around and not planning to create anything production worthy, it's a great deal.
Well, almost. GPUs have not be depreciating. The cost of 3090's and 4090's have gone up. Folks are selling it for what they paid for or even more. With the recent 40's SUPER series from Nvidia, I'm not expecting any new releases in a year. AMD & Intel still have ways to go before major adoption. Startups are buying up consumer cards. So I sadly expect prices to stay more or less the same.
He can use these cards for 128days non stop and re-sell, claiming back the purchase price almost fully since OP bought them cheap. Buying doesn't mean you use the GPUs to a point where they end up costing 0, yes there is risk with GPUs going but but c'mon.... Renting is money you will never see again.
The third effort is referred to sometimes as AI not-kill-everyone-ism, a tacky and unwieldy term that is unlikely to be co-opted or lead to the unproductive discussion like around the OP article.
It is pretty sad to read people bash together the efforts to understand and control the technology better and the companies doing their usual profit maximization.
More effort spent on early commercialization like keeping ChatGPT running might mean less effort on cutting edge capabilities. Altman was never an AI safety person, so my personal hope is that Anthropic avoids this by having higher quality leadership.
