if they start eating rat, does that technically define them as cannibals? given how much of their ilk become diet target...
froztbyte
joined 2 years ago
haha yeah I got very "we just need $29.95 billion bro just $29.95b trust me bro it'll be so intelligent bro just watch" impression from that as well
reasonable point, but still leaves the ethical parts of it (e.g. it gives more ammo to the kinds of creeps who make non-consensual imagery)
that list undercounts far more than I expected it to
when digging around I happened to find this thread which has some benchmarks for a diff model
it's apples to square fenceposts, of course, since one llm is not another. but it gives something to presume from. if g4dn.2xl gave them 214 tok/s, and if we make the extremely generous presumption that tok==word (which, well, no; cf. strawberry), then any Use Deserving Of o3 (let's say 5~15k words) would mean you need a tok-rate of 1000~3000 tok/s for a "reasonable" response latency ("5-ish seconds")
so you'd need something like 5x g4dn.2xl just to shit out 5000 words with dolphin-llama3 in "quick" time. which, again, isn't even whatever the fuck people are doing with openai's garbage.
utter, complete, comprehensive clownery. era-redefining clownery.
but some dumb motherfucker in a bar will keep telling me it's the future. and I get to not boop 'em on the nose. le sigh.
following on from this comment, it is possible to get it turned off for a Workspace Suite Account
- contact support (
?button from admin view) - ask the first person to connect you to
Workspace Support(otherwise you'll get some made-up bullshit from a person trying to buy time or Case Success or whatever, simply because they don't have the privileges to do what you're asking) - tell the referred-to person that you want to enable controls for "Gemini for Google Workspace" (optionally adding that you have already disabled "Gemini App")
hopefully you spend less time on this than the 40-something minutes I had to (a lot of which was spent watching some poor support bastard start-stop typing for minutes at a time because they didn't know how to respond to my request)
also, my inbox earlier:
24661 N + Jan 21 Apple Developer ( 42K) Explore the possibilities of Apple Intelligence.
so, for an extremely unscientific demonstration, here (warning: AWS may try hard to get you to engage with Explainer[0]) is an instance of an aws pricing estimate for big handwave "some gpu compute"
and when I say "extremely unscientific", I mean "I largely pulled the numbers out of my ass". even so, they're not entirely baseless, nor just picking absolute maxvals and laughing
~~parameters~~ assumptions made:
- "somewhat beefy" gpu instances (g4dn.4xlarge, selected through the tried and tested "squint until it looks right" method)
- 6-day traffic pattern, excluding sunday[1]
- daily "4h peak" total peak load profile[2]
- 50 instances mininum, 150 maximum (let's pretend we're not openai but are instead some random fuckwit flybynight modelfuckery startup)
- us west coast
- spot instances, convertible spot reserves, 3y full prepay commit (yeah I know full vs partial is a big diff; once again, snore)
(and before we get any fucking ruleslawyering dumb motherfuckers rolling in here about accuracy or whatever: get fucked kthx. this is just a very loosely demonstrative example)
so you'd have a variable buffer of 50..150 instances, featuring 3.2..9.6TiB of RAM for working set size, 800..2400 vCPU, 50..150 nvidia t4 cores, and 800..2400GiB gpu vram
let's presume a perfectly spherical ops team of uniform capability[3] and imagine that we have some lovely and capable active instance prewarming and correct host caching and whatnot. y'know, things to reduce user latency. let's pretend we're fully dynamic[4]
so, by the numbers, then
1y times 4h daily gives us 1460h (in seconds, that's 5256000). this extremely inaccurate full-of-presumptions number gives us "service-capable life time". the times your concierge is at the desk, the times you can get pizza delivered.
x3 to get to lifetime matching our spot commit, x50..x150 to get to "total possible instance hours". which is the top end of our sunshine and rainbows pretend compute budget. which, of course, we still have exactly no idea how to spend. because we don't know the real cost of servicing a query!
but let's work backwards from some made-up shit, using numbers The Poor Public gets (vs numbers Free Microsoft Credits will imbue unto you), and see where we end up!
so that means our baseline:
- upfront cost: $4,527,400.00
- monthly: $1460.00 (x3 x12 = $52560)
- whatever the hell else is incurred (s3, bandwidth, ....)
>=200k/yper ops/whatever person we have
3y of 4h-daily at 50 instances = 788400000 seconds. at 150 instances, 2365200000 seconds.
so we can say that, for our deeply Whiffs Ever So Slightly values, a second's compute on the low instance-count end is $0.01722755 and $0.00574252 at the higher instance-count end! which gives us a bit of a handle!
this, of course, entirely ignores parallelism, n-instance job/load/whatever distribution, database lookups, network traffic, allllllll kinds of shit. which we can't really have good information on without some insider infrastructure leaks anyway. if we pretend to look at the compute alone.
so what does $1000/query mean, in the sense of our very ridiculous and fantastical numbers? since the units are now The Same, we can simply divide things!
at the 50 instance mark, we'd need to hypothetically spend 174139.68 instance-seconds. that's 2.0154 days of linear compute!
at the 150 instance mark, 522419.05 instance-seconds! 6.070 days of linear compute!
so! what have we learned? well, we've learned that we couldn't deliver responses to prompts in Reasonable Time at these hardware presumptions! which, again, are linear presumptions. and there's gonna be a fair chunk of parallelism and other parts involved here. but even so, turns out it'd be a bit of a sizable chunk of compute allocated. to even a single prompt response.
[0] - a product/service whose very existence I find hilarious; the entire suite of aws products is designed to extract as much money from every possible function whatsoever, leading to complexity, which they then respond to by..... producing a chatbot to "guide users"
[1] - yes yes I know, the world is not uniform and the fucking promptfans come from everywhere. I'm presuming amerocentric design thinking (which imo is probably not wrong)
[2] - let's pretend that the calculators' presumption of 4h persistent peak load and our presumption of short-duration load approaching 4h cumulative are the same
[3] - oh, who am I kidding, you know it's gonna be some dumb motherfuckers with ansible and k8s and terraform and chucklefuckery
view more: next ›
tfw when you recognize it's Quality Rationalist Content