-
Posts
3490 -
Joined
-
Last visited
-
Days Won
20
Content Type
Profiles
Forums
Blogs
Everything posted by Zoraptor
-
Weird, random, interesting - now with 100% less diacriticals
Zoraptor replied to Amentep's topic in Way Off-Topic
Quoting myself, but unsurprisingly it turns out Microsoft itself was directly hacked via Solarwinds as well. About as unsurprising as them releasing the news on... New Year's Eve, where it would obviously garner a lot of attention and dominate the news cycle. At the moment they're claiming that the hackers could 'only' do read only stuff like, er, download source code etc, so no biggie. As the NPR article points out, there's an awful lot of nefarious stuff you can do without being able to write to a system. -
Weird, random, interesting - now with 100% less diacriticals
Zoraptor replied to Amentep's topic in Way Off-Topic
NSA checking pngs for encoded stolen nuclear blueprints, but no, it's some neckbeard's favourite Linux distro and their collection of Moria clones hidden in those cute cat pics. -
Faces melting off, starvation, diseased and dying siblings; six was about the age I first saw 'Watership Down'.
-
Expanse S5E5
-
-
4000 series will definitely be hot and power hungry judging by the leaks, possibly even more so than Ampere is. I'd personally suspect that Hopper will slip further and it will be more than a year between them. I would have thought a 2080Ti would be fine at very least until late next year. Lovelace almost certainly wasn't planned until quite recently, so it's likely to be a bit of a stopgap. Doesn't necessarily mean it won't be a good value stopgap though.
-
It seems unlikely they ever went for 'genuine' herd immunity, but more 'managed infection'. The theory was that you'd have a sort of ping pong situation with lock downs where you'd lock down, the rate would drop, and then rebound as soon as the lock down ended resulting in a perpetual up/down cycle- andor perpetual unsustainable restrictions- so they wanted to smooth out the cycle while exposing only those at least risk. That would result in similar overall infection numbers, increased immunity in the overall population and less economic damage with no spikes in infection numbers that would overwhelm the health system and cause a lot of excess deaths. They got aspects right and had some unusual circumstantial situations (very rapid vaccine development) that made them look more stupid than deserved, but overall you can't really argue that the approach was a success when the death rate is considerably higher than their neighbours, and they got no economic benefit either.
-
kopite7kimi does seem to think that Lovelace (AD102) will be 4000 series rather than 3x50 or supers, and he's usually accurate. Probably Q1 2022 is the realistic estimate, though it would depend on how much if any actual improvement it will have over Ampere and no doubt they'd like to have them available for xmas/ NH winter as that's big season. Presumably Hopper would then be 5000 series, perhaps a year after that. Samsung '5nm' is a bit of a punt though. It's an iteration/ rebranding of their 7nm process much as '8nm' is an iteration/ rebranding of their 10nm, done because neither of the parent nodes are very good at all. Last I checked there was literally one customer for Samsung's 7nm due to teething problems, and it was Samsung themselves. If ~19000 CUDA cores, ie +80% over Ampere, is accurate that's going to be an absolutely massive chip even if they've done it by moving from doubling on Ampere to tripling up on per core resources/ dependencies, and on a hot and unreliable node. But it can't be TSMC 5nm as Apple has that booked out, and AMD is buying every available wafer on 7nm. Slightly OT, but the US government would have been far better off paying Global Foundries to make a 7nm plant in Arizona than getting TSMC to make one. Top end fabrication is now a near monopoly, and GloFo's (well, really IBM's) 7nm process was actually better than TSMC's, they just couldn't get financing for it.
-
Probably want a settlement. That's the problem with caving to Sapkowski's sellers' remorse when they didn't have to, shows you're an easy mark.
-
Yes, that's one of the things which is simplistic in the explanation. Epidemiological matters tend to be rather less exact/ precise than you'd like from scientific stuff. 'Herd immunity' itself being a prominent example, since its seldom actual immunity that is being talked about but rather protection against uncontrolled spread. The weak immune response I mentioned for respiratory viruses in general is because it's fundamentally difficult for the immune system to get at a respiratory tract infection as compared to a more general infection- epithelial cells have the 'outside' on one side, after all. So you can have a beautiful immune response to a vaccine but still get a mild infection because the immune system has difficulty reaching the infection site (and for some people it will simply be ineffective for whatever reason). There's also the dichotomy between SARS-CoV-2 as a virus, and covid19 as the effect of the virus. The most simple example would be a linear 95% reduction giving a death number around half that of annual flu, without taking into account any reduced spread due to reduced symptoms or outright immunity given. The virus itself would be present a lot more than that 95% reduction suggests, but you'd have reduced covid19 (the viruses effects) by a theoretical 95%. Practically it doesn't quite work that way of course, but the way it 'really' works is too complicated for an easy explanation- and would be well beyond my expertise level.
-
A vaccine usually gives better immunity over natural exposure especially for respiratory tract diseases as the lungs/ airways give poor natural immune responses. In terms of antigens, if it's an attenuated (or dead) virus the antigen is the same as natural, in the case for SC2 all the approved or nearly so vaccines use encoded pieces of the spike protein either via mRNA (Pfizer/ Moderna) or it being stuck into a different viral vector (A-ZOxford/ Sputnik) which is deliberately designed and won't (necessarily) be the specific antigen segment(s) targeted naturally. SC2 is not a slow evolver, it's a fast one. It's a ssRNA virus, they all evolve rapidly because ssRNA is inherently the most unstable genetic option- single stranded gives no redundancy, RNA is less stable than DNA. It isn't as fast evolving as its relative ssRNA virus influenza, but flu is an extreme outlier there due to the way it is assembled in segments. Spike had no known mutations for months, but the new UK variant has (iirc) 8 substitutions from the original spike, which probably explains its extra infectiousness. Since all the vaccines target spike due to it previously being stable mutations there are a big deal as they potentially reduce the induced immunity. 'Natural' herd immunity is not a viable strategy, but not because it's a slow infector. It isn't as fast as measles but again, that's a big outlier in terms of infectivity. Bit simplistic but, herd immunity works when the virus cannot find enough non immune people to continue spreading, so the R number drops, eventually below 1. That's ~65 to 70% for standard SC2, and probably a bit more for the new strain. IIRC the immunity for common cold coronaviruses lasts a few years, so using that as an estimate if something like 25% of the population was infected per year you'd never actually reach herd immunity levels because you'd not reach 75% immunity- those infected in the first year would be losing immunity as those infected in the 3rd were gaining it. That makes sense, because most viruses don't 'die out' naturally due to herd immunity, even those with 'permanent' immunity like chicken pox or mumps. They percolate in the background, occasionally flaring up when immunity levels drop enough. Vaccine based immunity should last a bit longer than natural since the immune response is stronger, but how long and how consistently long is a very open question; but with vaccines you can at least theoretically get to (effectively) 100% immunity (in this case the highest possible seems to be 95%) per smallpox and nearly polio.
-
It could be a red herring, though Discovery is certainly not much of a 'red herring' type show. They've occasionally done a good job of obfuscating things but once revealed they've been almost painfully linear.
-
So that wasn't the season finale of Discovery last week? Fricking Netflix making me look stupid. Looking forward to finding out what caused the burn, if only to see how dumb the explanation is.
-
No way to know with any confidence. Hopper is still a fair way out and would be expected to slip somewhat if it really is a MCM design since that's likely to have teething problems. 2 years seems to be a reasonable estimate, I'd suspect maybe a bit longer. Sub generational there's likely to be an Ampere refresh at some point maybe a year after Ampere's initial release, ie around 9 months away. That's assuming AMD is releasing RDNA3 as scheduled.
-
In the non EE version of Icewind Dale the key to the Yxunomei fight was choke points and using Messenger of Sseth (? or whatever that speed bow was called) well. But really, single player game, play on whatever difficulty you want and don't feel bad about it. Especially if it's the first playthrough.
-
Most of the international stuff on that list is neolibs/cons neolib/ conning. Dalai Lama reincarnation is to annoy China who wants to appoint its own Dalai Lama when the current one dies instead of the one anointed by the Tibetans. Sudan was bribed to recognise Israel- not at all popular, internally- and that's part of the bribe, Nepal has a border dispute with China, Ukraine has a border dispute with Russia, Cambodia is seen as a bit of an influence battleground with China (plus Pakistan, Burma, and even Tunisia). There's a certain amount of irony in things like the Sudan situation which is obviously not America first in any real respect, but Israel first, but such things are in general the cost of running an empire with the alternative being surrendering influence to others. Which may be OK with the Guard Dogs of the US, but isn't with the vast majority of the ruling classes who write the bills. Look forward to a lot more such things from Biden and Harris, since their foreign policy will be failed neolib policy after failed neolib policy, because if at first you don't succeed keep going perpetually because anything else is failure.
-
Expanse S5E4 (minor spoilers)
-
Lucky here that things will be much the same as any other year: roast turkey, xmas pudding etc in 25 degrees celsius heat and 90+% humidity... Only covid related issue is that my sister and family were going to be here from Ockerland for the first time in 5 years and that has been cancelled. Compared to elsewhere that's utterly trivial.
-
Expanse S5E4
-
Yeah, I suspect the Brits are a victim of doing a lot of viral sequencing such that they've been able to detect why the spread has been more severe. It's certainly not just the SE of England that has had a far worse time of it in the 2nd wave than the 1st, and amelioration efforts that worked previous are a lot less effective in multiple places too. As a confirmed cynic I'd also suspect the state of the UK/ EU Brexit negotiations is a factor too, with France wanting to illustrate what a no deal brexit might look like.
-
Mutation is on the spike protein, which is the target of all (?) the vaccines. It being a functional mutation makes it more likely the shape of the protein has changed significantly which could reduce vaccine effect. OTOH we are being assured it won't, and they should be able to do decent modelling of what effect the mutation has on 3d structures. Turns out that it isn't actually a UK mutation either, the first known detection of it was Brazil in in the first half of the year, and it has popped up elsewhere (including Australia) prior to the UK.
-
Yep. Credit where due, in a series that is often sloppy that reveal was near perfectly handled and actually made a lot of sense in retrospect, and it earned a certain amount of vagueness of description.
-
Yeah, 7770 equivalent for the baseline xbone, more/ less on specs/ frequency. Fat fingers gave it a bit of an upgrade. PS4 original was roughly 7850 spec wise, iirc. Perform a bit better than base specs suggest due to the direct hardware access etc, but still pretty antiquated either way.
-
So, that was the season finale of Discovery. Randomly. Guess it was mostly OK as an episode, as always the alt universe bits were a lot more fun than the real universe ones, and the 'cameo' might be a bit polarising, to some. Otherwise, that was a really random stop point for the main plot- presumably due to covid, but it does just stop abruptly. I presume Georgiou is off to a spin off show now but Discovery will miss her being snarky a fair bit given how tepid and inoffensive the rest of the characters are. The one positive I'll unreservedly give to this season is that it has done more much needed fleshing out of minor characters. The other big positive is that it's been a lot more self aware about its flaws and weaknesses; not an unreserved positive because it would be better if it actually addressed those flaws a bit better rather than just acknowledging that they're there. But, they're doing a decent job of leaving it ambiguous whether Admiral Beard/ Starfleet already knows what the story is with the Kiev/ Burn or not, and had some of the Significant Looks in the alt universe actually end up being Insignificant Looks instead which is more nuanced than it has been. OTOH, the episode plots are still bullet points far too often strung together incoherently and inconsistently and they are still awful at conveying any sort of real emotion, though at least thsi episode's dose of over long schmaltz was for a major character. Somewhat relevant to that, I did rather like that they brought back Aeryom (?) in the alt universe (minus cyborg suit), only to near immediately and unceremoniously kill her off again... Overall- not bad enough for me to drop it, but certainly not enough improvement there to get any sort of recommendation.
-
Don't really know what CDPR was trying for there, it must have been patently obvious there were going to be massive problems with the ps4/xb1 for ages. Fundamentally, if someone tried playing the PC version on a Bulldozer CPU- not exactly inspiring, even when new- underclocked to ~2 Ghz and a 7870 GPU I doubt they'd be thrilled either; yet that is what base last gen console hardware was. Probably should have gone with an AMD partnership instead of nVidia, at least.