Hacker News new | past | comments | ask | show | jobs | submit | deadalus's comments login

I wish AMD did well in the Stable Diffusion front because AMD is never greedy on VRAM. The 4060Ti 16GB(minimum required for Stable Diffusion in 2024) starts at $450.

AMD with ROCm is decent on Linux but pretty bad on Windows.


I run A1111, ComfyUI and kohya-ss on an AMD (6900XT which has 16GB, the minimum required for Stable Diffusion in 2024 ;)), though on Linux. Is it a Windows specific Issue for you?

Edit to add: Though apparently I still don't run ollama on AMD since it seems to disagree with my setup.


You don't need 16gb, literally majority of people don't have that and use 8gb and up. especially with forge


They bump up VRAM because they can't compete on raw compute.


Or rather Nvidia is purposefully restricting VRAM to avoid gaming cards canibalizing their supremely profitable professional/server cards. AMD has no relevant server cards, so they have no reason to hold back on VRAM in consumer cards


Nvidia released consumer RTX 3090 with 24GB VRAM in Sep 2020, AMDs flagship release in that same month was 6900 XT with 16GB VRAM. Who is being restrictive here exactly?


it doesn't matter how much compute you have if you don't have enough vram to run the model.


Exactly. My friend was telling me that I was making a mistake for getting a 7900 XTX to run language models, when the fact of the matter is the cheapest NVIDIA card with 24 GB of VRAM is over 50% more expensive than the 7900 XTX. Running a high quality model at like 80 tps is way more important to me than running a way lower quality model at like 120 tps.


I cannot parse this. The Radeon RX 7900 XTX also has 24GB of vram, so how does it help you run higher quality models? I would understand if it had more ram.


Only the RX 7900 XTX has 24 GB of VRAM at its price point. If I went with an NVIDIA card, I would either have to spend over 50% more on the card, or use much worse models to fit on their 16 GB cards.


They lag on software a lot more than the lag on silicon.


Astronauts age slower and grow taller in space.



Due you mean in the Einstein "twin paradox" sense or are they actually aging significantly slower?


According to Wikipedia astronauts that spent about 800 days in space have aged about 20 milliseconds less. But I'm not sure if that's the correct phrasing and usage of word "age".


If you would consider someone who time traveled ten years into the future with magic to not be ten years older, then aged is correct.


What about, "has experienced ..." or "passed through 20 ms fewer"?


like hobbits drinking ent water


I will always remember the automatic removal of the Christchurch shooting video from users' personal Google Drive and Dropbox accounts, with no prior warnings or explanations provided.


I use Bing for porn. For reverse image search, I use Yandex. And for researching psychedelics and conspiracy(censored) topics - I use Brave Search.

I use Google only for vanilla search that shows I am a good citizen(navigation, online shopping, general research on movies and sports...etc). I don't want my search Identity to be fully profiled by Google. Police(and other bad actors like a hostile government) are more likely to access my data on Google than on other lesser know search engine.


How do you know all search engines aren’t compromised?

(OK, reasonable chance Yandex isn’t compromised by your government, but the others?)


semi-related: I do get a jolly deep belly laugh when "a threat to national security" is loosely tossed out as doublespeak for "an augment to individual privacy".

Without a VPN or tor on at least some of those browsers, though, I imagine the metadata strewn about could get stitched back together on the data-broker (ad-tech) end of all those transactions.

this is why RMS ssh's into a pine server to read email, i reckon.


Yandex is absurdly compromised


By whom?


Why use Google at all, instead of something like DuckDuckGo?


DDG is based in the US, what assurances do you have they would not comply to this kind of data request?


> ... what assurances do you have they would not comply ...

While compliance is certainly not optional, the value of your compliance is highly dependent on your technology. (The authors mention this in the article by writing that Apple does not collect the same kind of data as Google.)

I think it is like choosing VPN providers. You never know if they were involved in compromising their customers until they are caught compromising their customers.

So if you have a choice between two similar VPN providers, and one of them is known to be involved in compromising their customers, you use the other one.

I do not see how this would be any different with search.


I quit using DDG because when it ran out of answers based on what I searched for, it would spam results of 'random word in search query + city/area based on IP address'. Which was not only completely useless, but creepy as fuck.


Bing over yandex to avoid kompromat?


Da, comrade


I have never seen a more cute deer.

I wish all ancient drawings had such a version so we can easily interpret what they are talking about in the article. Early cave drawings are very hard to decipher.


It's an auroch, not a deer (actually the correct term seems to be "aurochs", but it's not what the article uses): https://en.wikipedia.org/wiki/Aurochs


I saw a few of examples in biology. Here I got a random image from Google https://bmcecolevol.biomedcentral.com/articles/10.1186/1471-... It shows a fossil of an insect in a stone, A and B are photos, C and D are hand drawings that show the important features. I'm not sure if it's common to superimpose the drawings, but in my case I thought it was better.


clearly a cow, yep. You missed the posterior ear


Cloudflare banned 8chan without a legal requirement to do so. They banned it based on moral reasons.


Wow, so the sun can spin! I never knew that.

https://www.space.com/does-the-sun-rotate


Even black holes "rotate".

Since black holes aren't really normal objects, what we mean by that is that the sum of all the matter that created black hole had an angular momentum. This also affects their behavior.

See https://en.wikipedia.org/wiki/Rotating_black_hole for more.


Almost nothing in the universe can not spin. Everything is rotating all the time.


This stuns me for some reason. I was not explicitly aware of this. Why? Just left over momentum from something? Is there a bias in direction (overall or among subsets)?


Lots of reasons, stuff like "an unevenly spread cloud of something that collapses under it's own gravity" will spin.

http://burro.case.edu/Academics/Astr221/SolarSys/Formation/s...


Link suggests that is because the cloud is already rotating due to "gravitational shearing in the galaxy's disk", not because it is uneven.


tl;dr parent and parent's link are not wrong, just very simple. Unfortunately it does not make explicit that "at least a little" does not mean the entirety of the rotation, or even its dominant component.

In the context above gravitational shear is the first of the empirical https://en.wikipedia.org/wiki/Oort_constants (differential rotation shears the non-solid disc). Differential rotation means gas in the direction of the galactic centre drags a bit compared to gas further from the centre, effectively acting as a small torque on a large gas cloud. If the gas cloud has no net rotation at all, this torque will develop it. That's what I take the sentence you focus on to be trying to say.

Unevenness is related to the Jeans instability, which describes the failure of internal gas pressure to prevent gravitational collapse. Chandrasekhar developed a more complicated treatment useful for when the angular momentum of the bulk gas cloud is non-negligible.

Roughly, collisions within the gas cloud convert kinetic energy to light which carries energy out of the collapsing cloud. The cloud consequently cools and contracts gravitationally.

However, still roughly, superimposing a net rotation on random motions within the gas prevents collapse in the direction perpendicular to the net rotation's spin axis. So contraction is mainly along the spin axis. Result: a disc, which is essentially what parent's link says.

The coupling of the galaxy's spin to the solar system's spin is at best weak; the solar system's spin axis is tilted about 60 degrees from the galactic plane; the plane is perpendicular to the galaxy's spin axis, in line with the previous paragraph.

Here's a diagram <https://i.stack.imgur.com/VIipS.jpg>.

The sun's path around the centre of the galaxy is a bit messy compared to a kinematically hot star (which will feel perturbations like the bar or spiral arms less). The sun tends to bob up and down relative to the midplane of the galactic disc. The Oort constants are local and position-dependent, so the solar system's migration -- and any migration of its precursor -- means wandering into regions with different gravitational shear.

Questions that afaik are still open (but I would be happy for a galaxy or solar system dynamics person to correct me, particularly if the correction includes magnetics and chemodynamics/metallicity-dependent pressure!): has the solar system's spin axis tilt to the galactic disc evolved since formation? Accepting the Coatlicue hypothesis, was the remnant of the explosion of the heavy star that was the precursor of the solar system (and others) aligned with the star's rotation? Was that precursor star's rotational axis aligned with the galactic disc? The remnant was almost certainly not of uniform density. Alternatively accepting the Wolf-Rayet bubble Giant Molecular Cloud (GMC) nebular hypothesis, did the GMC fragmentize under galactic shear, and if so was "our" fragment's spin dominated by that initially, with spin perpendicular to the disc but evolving to the present tilt, or was our fragment's spin axis initially tilted close to the present approx. 60 degrees to the galactic disc where it has since remained?

Finally, also afaik, the spins (more broadly the angular momentum vectors) of stars in the local bubble are essentially random. So repeat the questions in the preceding paragraph for each of those...


See that's what I love about HN :).

I didn't study physics past A-Level and none of this stuff really came up, what little understanding I have is as an interested bystander (I loved physics at school but wasn't the career for me), if I had multiple life times I'd have loved to have studied it at university though, so much stuff to learn in a short life time.


I like this video from minutephysics to get an intuitive sense for it: https://www.youtube.com/watch?v=tmNXKqeUtJM

It addresses a different question (Why is the Solar System Flat?) but it touches on the question of why things spin when they clump together from gravity.

An answer through the lens of my own understanding: it's just more difficult NOT to have angular momentum.

If you have a lot of particles falling toward each other due to gravity, imagine how difficult it would be to set it up such that all of them fall straight into their collective center of mass. They'd have to be in a precise, orderly configuration (e.g. equally spaced apart on a unit sphere). Note that each particle affects each other particle -- if any one particle gets too close to another, their gravitational interaction will cause them to move toward each other, and add an angular component to their motion with respect to the center of mass.

There are so much more disorderly configurations that will result in the particles moving with at least SOME angular momentum about their center of mass. Vector sum them all together, and it'd again be difficult for that sum -- the total angular momentum -- to be zero; they'd have to cancel each other out exactly, and there's just way more configurations where that isn't the case.

That's why it's simply much more likely for anything made out of particles in space to be spinning than not.

Additionally, as they fall toward the center of mass, the radius lowers, too, which means to conserve angular momentum you'll see their angular velocities increase. Helps make it more subjectively noticeable to us that everything is spinning.


That was excellent work by minutephysics.

Your intuition about Boltzmann entropy (disguised as the probability of finding a cloud with no net rotation in its bulk vs one with some) interested me. I think it highlights a "past hypothesis" problem in the wake of your link to the minutephysics video.

Your description is pretty much entirely gravitational. When you add matter-matter collisions, especially inelastic ones, you get changes in internal energy of the grains, and radiation outwards, and that's what drives collapse of gas clouds in timescales shorter than the age of the universe.

Changes in internal energy (e.g. internal rotational degrees of freedom in large dust grains, large being at least the size of a hydrogen molecule) alters the "bounce" from a collision making it easier for collisional grains to clump together. See <https://news.ycombinator.com/item?id=36418364> for a neat example involving partially filled water bottles being dropped to the floor. When the water is swirled, the bounce is reduced. Analogously, one can swirl or flex an isolated molecule around some axis.

For much smaller grains (e.g. atomic hydrogen) that feel electromagnetism, the light emitted by collisions ("scattering") is even more important, as it carries away internal pressure from the cloud, rather than just shifting it around internally to the cloud. In other words, for each in a pair of collided grains, subsequent collisions will be at lower energies, and so have smaller recoils.

Looking at this through the lens of General Relativity (it being a gravitational problem, after all), we want to consider a covariant angular momentum quantity. In suitable coordinates (~ Cartesian, with the origin at the centre of momentum of the cloud), this angular momentum is best associated with the off-diagonal spatial shear components of the stress-energy tensor. That's the upper blue triangle in this diagram:

https://en.wikipedia.org/wiki/Stress%E2%80%93energy_tensor#/...

In the diagram the 0-3 are the four Lorentzian spacetime dimensions, and 0 is the timelike one. Roughly, for T^{mn}, m is the (signed) "goesinto" direction and n is the (signed) "goesoutof" one, or if you prefer, the flux of m-momentum in the n-direction.

Essentially the goal in the early collapse of a cloud is to shuffle the nonzeros out of the pressure diagonal (in green) into the energy (T^00 in red, which dominates a tensor contraction to a Newton-like mass) and the components above the pressure diagonal, which one can think of as heat, angular momentum, and radiation.

Matter-matter interactions enable this shuffling much faster than for matter that only interacts gravitationally. A cloud of cold collisionless, non-radiating, non-interacting (except via gravity) dust with few internal degrees of freedom has trouble collapsing gravitationally. "Their gravitational interaction will cause them to move towards each other" is true, but if they don't interact non-gravitationally they'll mostly just slide right past each other. We see this in galaxy-galaxy collisions (like in the famous Bullet cluster) where stars are spaced far enough apart that they're essentially as non-colliding as dark matter; it's the less-compacted interstellar gas and dust clouds which smack into each other and throw off lots of X-ray radiation, which helps the gas swirl around near the site of the collision and collapse into star forming regions.

> [many] disorderly configurations ... with at least some angular momentum

Sure, but the thrust of the minutephysics video (if not exactly answering the question your comment's parent asked) is evolving from a system with relatively little angular momentum to a system with a lot, and contracting from a blob to an arrangement with a clear axis and eventually towards a thin disc. I would think that initial conditions of condensed objects moving on roughly circular Keplerian orbits roughly constraned to a plane is less typical than initial conditions of a cloud with random internal motion. From a Boltzmann perspective, there are a lot more microstates which can describe the cloud-blob macrostate than the star system macrostate. But the "past hypothesis" problem is that if the latter evolves from the former, then surely the primordial gas-cloud must be of lower entropy than the nice orderly-looking star system, especially if one finds cats, cars, computers, and teapots in it?

So the question from my second paragraph becomes: does a cloud with random internal motions and no bulk rotation have more entropy than a cloud that has a clear rotational axis? As short-duration snapshots, the answer appears to be yes, and for reasons very similar to the ones you gave in your comment. But because of the evolution of the former to the latter, which is evidently common (we see stars and galaxies everywhere in our sky, and have good models for star formation and ok ones for galaxy formation), the answer appears to be no.

https://en.wikipedia.org/wiki/Past_hypothesis (2nd last paragraph) is a terse starting point if you want more. There are bound to be youtube videos about it too, hopefully at least as good as the minutephysics video at the top.

PS: for experts who want to think about angular momentum differently, e.g. the J parameter in a Kerr black hole or some other vacuum spacetime, are directed to §5.11 of Misner Thorne & Wheeler vs e.g. the problems at the end of Wald's chapter on asymptotic flatness.


There’s a debate over if the whole universe is rotating, called rotational anisotropy.

At every scale below that you’re definitely rotating: cosmic filaments, clusters, and galaxies are all observed to rotate.

As the theory goes, we’re seeing the small perturbations/rotations from when the universe was tiny inflated to cosmic scale.


> debate over if the whole universe is rotating

Wait, debate ? Or am I failing to understand what you mean ?

\Lambda-CDM at the linear level requires the decay of vector-mode perturbations associated with vorticity; at the very least, the small scalar perturbations must totally dominate. This motivated a couple decades of searches for vector (and tensor) perturbations encoded in the CMB. Post-WMAP/Planck polarization data, aren't vector modes dead as a doornail in the linear theory ? (e.g. <https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.11...> | <https://link.aps.org/accepted/10.1103/PhysRevLett.117.131302> aka astro-ph/1605.07178 which kills off early-time vorticity imho, conflicting with your final paragraph.)

I also know of some work on late-time vorticity, proposing studies of CMB lensing curl, kinetic SZ and moving-lens tomography, galaxy rotation planes and so forth, but isn't practically all of this work explicitly generating null-tests of \Lambda-CDM, rather than a cosmology where "the whole universe is rotating", Bianchi or otherwise ?

Finally, I also know of some work trying to use large N-body simulations to look at scales where predictions from the linear theory become unreliable (~ Mpc). However, I don't think this is what you mean.

If there is a published review of "a debate over if the whole universe is rotating" that you could direct me to, I would be grateful. I got nowhere with "rotational anisotropy" as a search term, and seem to lack the capacity to imagine possible synonyms other than as above. Don't spare me, I'm prepared to admit and embrace the consequences of my ignorance.


I searched for “rotation anisotropy universe” and got this as the first result:

https://physics.aps.org/articles/v9/s103

And the second:

https://www.scirp.org/html/12-7501435_36098.htm

And the fourth:

https://bigthink.com/starts-with-a-bang/new-anomaly-universe...

This one is about the preferred rotation of galaxies:

https://news.umich.edu/the-universe-may-have-been-born-spinn...


I'm sorry to say that none of these is close to the review I asked for, and afaics your first result explicitly quotes from and your fourth is preceded by an image taken from the PRL letter (Sadeeh et al) I supplied in my comment's second paragraph. The former is a summary of Sadeeh et al's results: the constraints on vorticity are unforgivingly strong and vanishingly small. The latter does not even discuss vorticity or rotation or spin outside the caption of said image. Search-fu failure. That the bigthink article is simply pop-sci (Ethan Siegel is very well known) and not a literature review is something I think you should have noticed.

As to your last link, the conjecture about a preferred handedness of spiral galaxies went nowhere (you found a link from more than a decade ago; the preprint (and a visit to google scholar and citeseerx) does not match a peer-reviewed publication <https://arxiv.org/abs/0812.3437>), and additionally says nothing about galaxy clusters, ellipticals, and so forth. Even more importantly it's only for z < 0.085 which is in the region where the linear theory of cosmology is not expected by anyone to apply, c.f. my comment's penultimate paragraph.

Finally, your second result to a 2013 paper is different and interesting, and at last has evidence of engagement by other authors <https://scholar.google.co.uk/scholar?cites=10702549710128965...> however considerably more than half of those are by U V Satya Seshavathram et al talking about their very non-standard hypotheses in many fields <https://www.researchgate.net/scientific-contributions/U-V-S-...> (note the cold fusion stuff, and the at least four very different rotating cosmologies, quantum and not, dark foam and not). Also, the first (by GS ordering at the link above) citation of the 2011 paper you found, amusingly, is a self-citation (Chechin, Astronomy Reports 2016). Another Chechin self-cite, in a 2014 paper, also looks interesting (although it's only cited twice).

So it's pretty clear on your supplied evidence that there is no debate about the rotation of the universe. That's not too surprising, even BOOMERaNG data fails to support (and even somewhat undermines) early-time (pre-CMB) relevance of anything but scalar perturbations.

Thanks anyway. I'll enjoy reading these two wild Kazakh ideas I didn't see when they were fresh last decade, so your efforts weren't wasted.

(ETA: to be clear, Chechin and his coauthors are doing physics. Their math is sensible, their argument based on physical principles. They demonstrate professional familiarity with the standard cosmology. They advance a hypothesis (well a couple hypotheses, they too have multiple spinning universe models; their Generalized Jeans one is the most interesting) which is probably in conflict with observations available not long after publication date. They aren't pretending to be physicists, and they're not cranks. But the idea that there is an angular velocity proportional to the square root of the (dark) energy density is ... pretty wild! As in nobody's put it in a zoo of models.)


Yes, it is left over momentum from the big bang. It's crazy to think, but objects are still in the process of "settling down" 13.5 billion years later. They will probably continue to do it since there is no other force acting upon them.


What's a known exception?


A Pulsar, that's not really spinning but oscillating back and forth, becoming a very highly precise instrument of time measurement as well:

https://physicsworld.com/a/pulsar-timekeepers-measure-up-to-...


Pulsars are dense compact spinning objects.

From your link: "Pulsars are neutron stars that rotate at very high speeds and appear to emit radio pulses at extremely regular intervals. The pulses are actually all we see of a radio beam that is focused by the star’s magnetic field and swept around like a lighthouse beacon."


The Higgs Boson has spin 0. See https://en.wikipedia.org/wiki/Scalar_boson


Is using quantum mechanical spin a fair comparison? Isn't it moving the goalposts a bit?

> As the name suggests, spin was originally conceived as the rotation of a particle around some axis. Historically orbital angular momentum related to particle orbits. While the names based on mechanical models have survived, the physical explanation has not. Quantization fundamentally alters the character of both spin and orbital angular momentum.

> The classical analog for quantum spin is a circulation of energy or momentum-density in the particle wave field: "spin is essential a wave property".

https://en.wikipedia.org/wiki/Spin_(physics)


I thought quantum mechanical spin is completely unrelated to Newtonian angular momentum, but IANAP.


Its not completely unrelated, but it is not clearly related. That is, spin contributes to overall angular momentum, but that contribution is much too large to be explained as the actual rotation of subatomic particles. So, like most quantum properties, our Newtonian intuitions are poor guides.


Any tidally locked satellite such as the moon.


A tidally locked satellite rotates: one rotation per orbit.


unless the body it rotates around rotates at the exact same speed around another body in the opposite direction


Then it still rotates.


It can only rotate relative to an arbitrary point, not in some objective absolute sense. Here it does not rotate relative to chosen point.


Something about privileged reference frames...


But what's the alternative, right?


It is also the reason why the planets revolve around it


Spinning is so much cooler than not spinning!


warmer*


In the 15-year period of 2005 through 2019, canines killed 521 Americans. Pit bulls contributed to 66% (346) of these deaths. Combined, pit bulls and rottweilers contributed to 76% of the total recorded deaths [1]

[1] https://www.dogsbite.org/


The pit bull is bred so that it keeps biting no matter what.

Their bite force is about 235 psi and some can pull over 3000 lbs.

It's the most abandoned breed and the one with most safety incidents.

So:

- keeping a pit bull on leash doesn't make a difference

- once it's aggravated you cannot control it anymore

- nobody actually needs a pit bull

- it can disfigure or kill a person or animal at any moment: you, your family, your neighbors, their pets, dog sitter, etc.


And literally the only ways to get a pitbull off a person are to a) kill it b) choke it unconscious.

No other internet trick works on them, they've specifically been bred to bite and bite hard.


When I looked for a dog several years ago the local animal shelters were mostly filled with pit bulls. Very few other breeds (except maybe chihuahuas). I certainly wouldn't want one in my home, especially not knowing its background. Most of them are probably fine but you never know. Not worth the risk.



That's a perfect example of transparency from Wikipedia.


its not transparent when they don't tell you the individual pages and people whose pages are constantly edited with lies


Overall I agree that Wikipedia is largely neutral, and I'm sure great effort is put into it, but we are all human and everyone has biases. It's just whether we're aware of them or not, and even if you're aware of a bias, it can be a trap because we think "oh I'm aware of it, I wouldn't do that" and then trick ourselves into doing it anyway, we just got more skilled at deceitfully rationalising it to ourselves. All this to say, we're all flawed, and as a consequence so is Wikipedia.

A really good example of where this happened is the Toxic Masculinity article, which evolved over time to be a bit more... "Contemporarily woke"[1] instead of purely psychological (last time I checked).

Similar thing happened to a lot of trans articles that tend to advertise benefits of transitioning (and don't get me wrong, there are many life changing benefits for people with gender dysmorphia, I'm not disputing them), but diminish criticism of lack of process and due diligence in clinics prior to green lighting transition, and citations being removed (under pretense of transphobia and the source being biased), etc. Trans representation in related history articles is also being embellished through subtle linguistic edits; they're not wrong per se, but are still deceitful word games (which I think just makes everyone more distrusting and sceptical as is evident from that thread).

I do think the woke neurosis will settle down eventually, and we'll have increased our "decency" baseline in society thanks to it, since all historic movement were riddled with bad actors, but it's still difficult to watch an otherwise good movement soil itself like this, and as an extension influence Wikipedia.

So in certain "niche" categories where things are largely a matter of opinion, it's a bit of a wild west unfortunately. I waged a small edit war in the past, only to be burnt out. I do care about the topics outlined here, but I'm not pathologically online (alas, but apparently I am enough to post this comment) to dedicate so much of my free time on playing edit pong with an on-line asshole (I'm sure their intentions are for the net good since they're technically arguing for things I want in society myself, but I don't like that it comes at the cost of candor and increased overall social outrage and distrust)

[1]: I think being woke in the traditional, pre social-media era was a good thing. Now it's just performative due to communal narcissists infiltrating, abusing and feeding of socially vulnerable demographics. Ie, being woke is good, but performative wokeness is a shield people now use online to be bullies, unfortunately.


What is your point?


India Remains Internet Shutdown Capital of the World for Fifth Year Running : https://thewire.in/tech/india-remains-internet-shutdown-capi...


A few years ago, Algeria started to shut down the internet every year during the baccalaureate exams [1].

[1] https://www.theguardian.com/world/2018/jun/21/algeria-shuts-...


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: