They already torpedoed bootcamp so Apple's notch is no loss. Are there even native full-screen apps for MacOS?
:devious:
The Battlefront Miscellaneous Thread
- Thread starter Horatio
- Start date
- Status
:devious:
I was mostly
Full screen presentations and the like would qualify but I'm not sure how many people use the built in monitor for those situations
I thinking they are going for consistent competence. Maybe not actually achieving it consistently but doing far better than most manufacturers (and most other Google ventures lately)
A simplified lineup with decent quality and decent support is what the Android ecosystem needs *badly*, not yet another expensive phone with everything turned up to 11
Android 12 is... different I guess?
I don't hate it yet but I'm kinda wondering what the point was
I don't hate it yet but I'm kinda wondering what the point was
MS makes more money selling the game than the console
They probably prefer people on gamepass but completely locking out Sony is counterproductive
I like cheap Fire tablets for kids but all they do is play PBS games and kids shows. Amazon makes the money on content though, or something like that
If people wanted other Android tablets with actual margins for hardware OEMs there would be a market for them. There isn't
If people wanted other Android tablets with actual margins for hardware OEMs there would be a market for them. There isn't
This is a major issue for them
Moto has a minimal product support window on most of their products. Quite a few of them listed for sale will be out of support practically (or literally) the day you buy them
Controversial?
Investors want to get paid and they have voting stock.
Presumably the lenders would want to get paid too.
It's definitely not the only priority but bias between short and long term gains are ultimately determined by the shareholders
Apple has been hoarding cash. Someone needs to put it to use
More than a million people bought Windows Phone. They can definitely snatch defeat from the jaws of victory
He said it was the Daily Beast in his post. I post links from my news feed like that though MSN hosts their copy of the article at least
You'd get more memory per addressable unit. That would also keep the read/write lines in the cache a bit more simple. Tradeoff is the ineffeciency of reading/writing more than you might need
IIRC, TLB doesn't come into play into you start accessing main memory (system RAM). Everything on the processor is physical, everything off of it is virtual and it has no freaking clue what fixed storage is. That's the kernel guys' job
No, larger pages means fewer addresses. The cache can hold more data with the same number of entries
I'm a bit rusty on all this but that depends on your cache hierarchy
Copyright isn't bikes
The news agency or whoever has configured their server to perform a copy of the article and send it to you. The copy is authorized.
That you choose to modify how your browser displays that data afterwards is no business of theirs.
Plenty of sites have figured out how to only show the first paragraph or two to anonymous users; its not rocket science
They are still GPUs even if optimized for other use cases. They dont need an onboard video port; streaming works just fine for their intended use case. The H100 Hopper architecture isn't optimized for gaming (your frames/watt and frames/$ would be awful) but it almost certainly would work; I'd be very surprised if it didn't support the full instruction sets from previous generations. A100 is based on Ampere, the same GPU as the RTX 4000 series. Its literally a datacenter version of the 4090 with no functionality restrictions.
CUDA isn't "faking 3d screens", I promise
The "A" in A100 stands for Ampere. I was wrong in my previous post. Ampere is the RTX 3000 series and not the 4000, but the point remains. Its literally the same architecture as the cheaper desktop parts.
Hopper is specialized but it can run games (badly). Its even been benchmarked apparently
https://www.techspot.com/news/99148...-gaming-performance-surpassed-integrated.html
They are not general purpose processors, true. They usually require a host CPU to dispatch instructions and collect results.
But your description of how they process instructions is completely and utterly wrong. The GPU doesn't need to read/write from a framebuffer (neither for PhysX nor for CUDA) and it can directly exchange data from its memory to that of the host CPU (and vice versa)
GPUs are just specialized coprocessors
GPUs support graphical APIs (amongst others) and are specialized in matrix operations and other highly parallel operations
Among the debris recovered from the Alaska 737 Max 9 jet which had an emergency hatch blown out was an iPhone that still works after being sucked out of the plane and dropping 16k feet. A boy had his clothes ripped off as the pressure escaped the cabin.
Did it have a case? Otherwise, great Apple build quality?
Probably luck, landed on some plush vegetation or soil.
I‘ve never cracked an iPhone screen or nicked a corner. Case all the time. I did crack an iPad screen which was in a case but it landed screen first onto a tile floor from a height of about a foot and a half — coffee table.
Randall, if you are reading this, there's a new XKCD to be made about the likelihood of a phone breaking being inversely proportional to how ridiculous the circumstances are.
A super cycle for Windows 10 EoL? Sure
For average Joe/Contoso to run AI on client systems? I doubt it
For average Joe/Contoso to run AI on client systems? I doubt it
Maybe eventually but such a distributed AI ecosystem doesn't exist yet.
Is this the next AR hype train?
Is this the next AR hype train?
The local system wouldn't be able to store enough of most datasets to really do much, I'd expect
I could see that for visual/image searches that local processing might at least offload some early transcoding/compression and save some bandwidth but that's not necessarily a massive use case
I could see that for visual/image searches that local processing might at least offload some early transcoding/compression and save some bandwidth but that's not necessarily a massive use case
I wouldn't call any of that AI and certainly none of it requires new hardwareThat's right but it would be great for automation tasks. "Start word"., Enter the time in notepad", "Move all files from this folder to dmore"
Most of that stuff you can do now via various automation tools
They do that now. Tag someone's face and the software is smart enough to be bale to sort by that face. Extend that to a general image attributes. Some android launchers have been able to sort icons by major color (i.e sort by blue icons and it shows everything that mostly blue)) and have been for quite some time
Google's Picassa, before the jerks canceled it, could identify unique individuals in images 15 years ago
My comparison to AR wasn't so much a direct technical one; its the tendency of tech execs to try and push something well before its ready for mass adaptation
It definitely still generated white people but it seems like it was programmed to never make an image of a group of people with only white members.
This got awkward fast when people started generating examples of minorities in WWII Nazi uniforms and appearing at the drafting of the Declaration of Independence, IIRC.
Well, unfortunately you can't trust LLM's to be consistent either especially if you dont ask the exact same question. I phrased two questions (Houthi, cannibalism) just like from the X screen shots and got reasonable answers similar to what papadage had for the former and a good argument against for the latter
I also tried asking the foie gras recipe exactly like papadage did yet I got a similar denial to what Curry's X post showed:
LLM responses can vary based on context. At the least questions you have previously asked in you current session are known to influence it but there are definitely other factors.
These tiresome accusations of lies and bad faith are a bit uncalled for and this makes an excellent demonstration for yet another pitfall of relying on a non-deterministic language interpolator that's being constantly (and opaquely) reprogrammed.
The SolarWinds attack was carried out via a compromise of the software vendor of the same name
Possibly SolarWinds, the company, might have been compromised by an ADFS vulnerability. Regardless I'm not sure why fixing an ADFS vulnerability would have endangered a contract negotiation. Its not exactly a secret that software has flaws and MS patches major ones every month
Lets look at the flaw itself which is mainly explained via frustrating slideshow: if you make a copy of the private key from a SAML server you can forge access tokens. Like yeah, duh. That's how SAML works and this isn't unique to MS. The only way to get that key is via obtaining highly privileged access to the authentication service which would be very bad on any platform. Or finding the key laying around if an admin is very sloppy. Neither is directly a flaw of the technology.
Thats, uh, not how the internet works. A forged token can elevate your access but it doesn't hide your IP. There are other ways to do that of course. (and WTH is with the scare quotes?)
Later it goes on to talking about how smartcard SSO was part of the problem but dont explain how that anything to do whatsoever with a compromised SAML key
Further the link to a CyberArk article that's hilariously bad
https://www.cyberark.com/resources/...technique-forges-authentication-to-cloud-apps
Either the author of the article badly mangled the description of the flaw or they got played. Either way its a bad look for them
Possibly SolarWinds, the company, might have been compromised by an ADFS vulnerability. Regardless I'm not sure why fixing an ADFS vulnerability would have endangered a contract negotiation. Its not exactly a secret that software has flaws and MS patches major ones every month
Lets look at the flaw itself which is mainly explained via frustrating slideshow: if you make a copy of the private key from a SAML server you can forge access tokens. Like yeah, duh. That's how SAML works and this isn't unique to MS. The only way to get that key is via obtaining highly privileged access to the authentication service which would be very bad on any platform. Or finding the key laying around if an admin is very sloppy. Neither is directly a flaw of the technology.
Thats, uh, not how the internet works. A forged token can elevate your access but it doesn't hide your IP. There are other ways to do that of course. (and WTH is with the scare quotes?)
Later it goes on to talking about how smartcard SSO was part of the problem but dont explain how that anything to do whatsoever with a compromised SAML key
Further the link to a CyberArk article that's hilariously bad
https://www.cyberark.com/resources/...technique-forges-authentication-to-cloud-apps
No shit. If your root admins get popped and you do a bad job of cleaning up everything they could have compromised, you are still vulnerable. This has been true since the first physical lock and key were invented.
Either the author of the article badly mangled the description of the flaw or they got played. Either way its a bad look for them
There's nothing to patch. If a credential like a token signing cert is compromised you must replace it. If you dont realize you've already been compromised then you're just boned. There's nothing unique about ADFS in that regard. It would be just the same for literally any other bit of federated infrastructure ever. There will necessarily be a root of trust that can be compromised to affect multiple systems. The "Golden SAML" text above neglects to mention that someone whose gained such access would already be able to trivially intercept user passwords or even create their own accounts with any level of access they wish.
The solution to this is just to limit the extent of trust between systems. That's an architectural call made by the customer. Its not like segmentation was an unknown practice prior to these reports
A security group. They have a lot of them. I'd guess because he seems to be a bit of an obsessive crackpot.
Glancing at the article again it even they state Solar Winds wasn't hacked by this. They were just sloppy on multiple fronts and didn't clean house after the original breach
Its really tough for internet news agencies to make a dollar these days. You cant really blame a newcomer like Apple News for exploring every option to raise enough money to support their thankless mission of aggregating everyone else's hard work. Its not like they belong to some bottomless source of funding like WaPo
Bing rewards is my retirement grease piggy bank for when my willpower breaks down and I need to add a new Xbox controller to the fleet
- Status