I get why JPEG isn't perfect but the big issue here is that for the vast and overwhelming use cases, JPEG is "good enough".
It's going to be hard to come up with an alternative that's not ridiculously patent encumbered.
I don't think patents are the most serious issue. It's a handy whipping boy, but the legacy installed base is the big one.
There have been numerous "replacements" to JPEG, as well as other image formats. It took decades for the vastly suprior PNG to overtake GIF (even with patent issues in GIF), but even then, GIF is very much alive and well.
The same can be said about older video codecs (Many Blu-rays are still encoded using MPEG-II, in spite of the availability of better codecs). Nobody complains about the 1992-era Dolby AC-3 audio on DVD's, and 1993's MP3 shows no few of slowing down, in spite of considerably better codecs being developed & deployed in the past two decades.
The bottom line is that while higher quality is possible, the "better" level isn't enough for most people to even care.
Glare on your device's screen can make a bigger difference.
The car was already stopped, so "doing nothing" wouldn't surprise me. Since they're still developing autonomous driving, it's better for "the company" to have the car stay put and let the Police sort it out. Better to pay a fine for blocking traffic than make the news for hurting somebody.
I seriously doubt any autonomous driving system is sophisticated enough to decide whether fleeing from (and possibly running over) attacking pedestrian(s) is the correct course of action.
There's also a good chance that the pedestrian committed felony assault, and in some nations/jurisdictions, failing to report a felony is a crime as well, as is driving away.
I'm pretty sure that's a thicket that developers would like to put in the backlog, for after the simple problems like driving in the snow are taken care of.
It never ceases to amaze me how often people are surprised when Apple blocks applications the first time it goes through the review, especially as the apps are nigh invariably approved after an appeal.
Spoiler: That's exactly what happened here, if you read the update in TFA.
there is absolutely no reason why Apple should withhold this from the store
TFA has since been updated: Apple has since decided to approve it.
It's common for Apple to deny apps their first go-round, and just as common for the app to be approved after additional review. It shouldn't be news to anybody, as it's been observed by developers for nearly a decade at this point.
I wouldn't be surprised if the practice is similar to SMTP servers requiring the originating server to re-send the message -- spam bots don't bother, and earnest developers will try again.
That or the reviewer was exerting his/her political views...
Or maybe one of Apples App reviewers is just a die hard Trumpkin and an enthusiastic Ajit Pai fan?
Eh... you can dislike regulations and not be a fan of either Trump or Pai.
Regardless, the TFA has since been updated, and Apple is going to allow the app:
Apple asked Choffnes to provide a technical description of how his app is able to detect if wireless telecom providers throttle certain types of data, and 18 hours after he did, the app was approved.
Honestly, App store denials for the first version of a new program are fairly common -- hardly a surprise given the number of developers shitposting into the App store.
Update: After this article was published, Apple told Dave Choffnes that his iPhone app, designed to detect net neutrality violations, will be allowed in the iTunes App Store. According to Choffnes, Apple contacted him and explained that the company has to deal with many apps that don't do the things they claim to do. Apple asked Choffnes to provide a technical description of how his app is able to detect if wireless telecom providers throttle certain types of data, and 18 hours after he did, the app was approved.
"The conversation was very pleasant, but did not provide any insight into the review process [that] led the app to be rejected in the first place," Choffnes told us in an email.
the problem is this 'renewable' is not even a net zero return.
In many ways, that's not even the point. If we can create enough energy through other methods (Fission, Solar, Wind, Hydro -- even Fusion if it works out), then net zero is not as important as being able to convert the energy to a useful form. (The same applies for Hydrogen, for that matter).
Petroleum isn't going to last forever, and it's increasingly looking like we'll have supply problems by the middle of the century. That is a problem, and using biodiesel helps lengthen timeframe to develop something better.
Why do we care? Aerospace and shipping.
While there is some work for short-range electric jets, long flights at altitude are at a disadvantage compared to fuel-burning turbines (which are extremely efficient at operational altitude & speed). Jet fuel is very, very similar to diesel, and biodiesel is a possible replacement. Many of the large new ships are diesel powered - either piston or gas turbine. As you can imagine, militaries (especially America's) love the stuff because it's domestic, and it could power most of our Navy, our tanks, and our aircraft.
The primary fuel of booster rockets today is kerosene. Methane may work well in the future (and with a lower ISP).
Or have you become so repressed that you pretend it does not happen and that there's never a reason to talk about it?
I'd mod you up if I could.
Fecal humor exists in every human culture for a reason: We can't escape our biology.
Humanity is united in knowing that even Emperors and Kings can't escape it. It doesn't take much of a Google search to see the POTUS doing it. (I don't think it matters which POTUS...)
There's nothing dignified about cutting a turd loose. There's nothing elegant about removing shit stuck to your body.
We all know that from the smallest baby to the most gorgeous supermodel: all must go through the humiliation of giving birth to a politician.
If there's one thing that unites all of humanity, it's the knowledge that we all must check our egos at the door, and rendezvous with last night's supper.
On top of that, GPUs don't run kernel code (so cannot leak it),
It's entirely different from saying it's a problem with the GPU, but: There is the much-maligned NVIDIA kernel module which does interact with the GPU.
It's not far-fetched to say that there may be Meltdown/Spectre related patches -- though it's far more reasonable to say that any fixes would be a mitigation between the kernel module/driver and the CPU.
Anybody actually paying attention knew well before The Register printed anything.
The flaw was spelled out reasonably well by LWN as far back as November 15th, and it was noted that it was highly unusual for the patchset to be fast-tracked as it was. LWN also mentioned the initial KPTI patchset (then called KAISER) about a week earlier than that (Nov 10th). A month later, LWN followed up (including notes that ARM64 was affected) - more than a week before The Resister printed anything.
It was clear that something monumental was on the horizon, and that it was related to memory protection.
It was even clear that there was an information embargo in place, because comments were scrubbed from the associated patches.
It's been reasonably public for close to two months now.
The unknowns were more along the lines of "How deep is this pool of excrement," and "Which animal made it." Major OS patches were a fargone conclusion.
Why not simply take the device away from the child after x time elapses?
The old "take the toy away, give them a textbook to do homework" model doesn't work with computers.
Unlike a toy, computers have many, many abilities. By taking away the device, you may take away the unapproved toy, but also approved things you want your child to be doing -- reading books, news, maybe even textbooks, educational programming, or using tools to create art and music, and so on.
So, parental controls can be used to to limit functionality for reading, instead of gaming, which seems perfectly valid to me.
And I can walk over and look at what my child is doing on the device, but it only takes a fraction of a second to close the offending app and appear compliant. They can hide the evidence while I'm walking over.
Honestly... "My toddler's too quiet" is a meme for a reason: even the youngest toddlers know when they're sneaking behind their parent's back. Even the best parents have an extremely limited capacity to pay attention, and aren't capable of paying attention all the time. That little child-safety latch over the cupboard with drain cleaner may be a "technological gimmick", but I have yet to meet anybody who considers it bad parenting.
So parental controls can also be part of a "defense in depth" kind of thing -- yes, you monitor what your kids are doing. Eventually, you're going to be human and slip up, and that's where ever-watchful automation can help.
AMD original response is that there is a "near zero" risk of exploit for variant 2 and a "zero" risk for variant 3
Claims about "near zero" or "zero" risk of exploit are traditional CYA effluent, after all...
Even AMD has claimed (for a different CPU errata) that problems "would not occur during normal desktop usage and we've never encountered it during our testing." The claim was funny, as all you had to do was fire up a VM... or the right benchmark... It didn't speak well to AMD's testing at the time.
Having run across a nearly four-year old CPU having PCID (Haswell Xeon E3-1276), I'm less inclined to say that "newer" is entirely accurate; more that it was less common...
A few years back, John Carmack tweeted that he could ping transatlantic faster than putting a pixel on screen. He then explained more fully at superuser.
Since I run smokeping from my home server, I can definitely say that most online services are reachable from my home internet connection in ~20 ms round-trip - or 10 ms one-way.
For contrast, one frame at 30 FPS is 33 ms. (Kind of slow by modern standards, but I'm old enough for 30 FPS to be the threshold of being acceptable).
While it certainly matters to the experience, 2-3 frames of latency between my input and rendered pixels doesn't detract from the experience -- especially for games played with a gamepad (instead of a mouse).
It might not be suitable if you're playing an FPS game professionally, but it's fine for RPG and strategy games.
One less wire for my Toddler (or pet) to shove in his mouth is good. Charging pads (short-field charging) don't get me there.
There are, however, the mid-field and long-field wireless charging options in the pipeline that seem more compelling - Energous just received FCC approval for their version which lets you charge anywhere within three feet, with a more powerful version in development that can hit 15 feet. Those are more interesting.
Mobile news is markedly worse with AMP; I've read several accounts where publishers lament that it's slower than their native version.
My first sign that AMP is horribly broken is that every single AMP page I've ever visited all point to "google.com", and the URL bar shows "google.com" regardless of the site I'm actually visiting. Phishers can (and have) conceal pretty much anything behind AMP, and few users would have a clue because they see the lock with "google.com" at the top of their browser.
The next problem with AMP is that you can't turn it off: Google feeds you AMP pages if you use a mobile browser, and you have to load the AMP page first and then click to additional times to get to the non-AMP page.
The alternative is to use a different search engine, which is what a lot of us are doing.
AMP is a bigger problem than anything it was trying to replace.
no amount of wishing will put the AES-256 toothpaste back in the tube
Or the fact that pen & paper cryptography with the Vigenère_cipher and a sufficiently long key of random characters is still considered unbreakable. (Key sharing is a pain, but AES has the same problem)
It's all especially ironic as the "Advanced Encryption Standard" was a US Federal Government program where the world's cryptographers competed & collaborated to come up with unbreakable encryption. Or that other governments have done the same with NESSIEECRYPT, and CRYPTEC.
And that US Federal Government is sponsoring workshops & standardization on post-quantum cryptography, ostenably so we'll have something secure & standardized before we can crack RSA, DSA, El Gamal, Elliptic Curves, etc. with quantum computers.
It's almost as if law enforcement doesn't want to go back to the heady days of 2006 when they did stake outs.
Hardware isn't software; it takes a lot more than changing code and a recompile to fix.
Logic problems are only part of the story -- after the logic is done, there's physics to deal with (voltage breakdown, heat, crosstalk, etc). Then there's the "rat's nest" of conductive paths for the signals to follow.
Coders may complain about following spaghetti, but they've got nothing compared to what hardware guys deal with. Even minor changes can require considerable amounts of re-routing.
After that is complete, they still have to build the chips, which will take months more to get the fab machinery set up.
Expect to wait into 2019 and possibly later. Nobody has a processor that does out-of-order speculation that is immune...
Especially in the case of Microsoft, Hanlon's razor should be applied.
I don't have a problem with the idea that Intel is trying to actively sabotage AMD (they've done it in the past), but that it seems a lot more likely that Microsoft made a stupid mistake.
I get the feeling that the virtualization instructions (VT-x, AMD-V, and whatever the equivalents are on s390, ARM, and POWER) aren't really involved with either Spectre or Meltdown.
I wonder about SPARC, but given Oracle & Fujitsu appear to be killing SPARC... it probably doesn't matter.
To play devil's advocate: I don't know of a single chip that doesn't have errata published by the manufacturer. The first "real" lesson I learned in my EE career is "you can't take the manufacturer's documentation at face value."
The important thing is how problems are handled, and I don't know of any chip maker who handles bugs like this well.
When their GeForce 8600M had issues with overheating and dying, NVIDIA first blamed notebook manufacturers, and later the foundry. It was later found to be a problem with NVIDIA's silicon layout. Apple had to replace millions of MacBooks whose NVIDIA card died after less than a year (Ever wonder why Apple uses AMD graphics these days?)
Even AMD has issues when it goes into CYA mode: Back in the day, AMD had major TLB issues with their Phenom processor. At the time AMD claimed the "... TLB erratum is a highly random event that would not occur during normal desktop usage and we've never encountered it during our testing of Phenom." "Normal desktop usage" was defined to exclude a Xen hypervisor running Windows XP, or running the SPEC 2006 benchmark -- not exactly common, but they're clearly doing the PR spin.
Then there's Intel... Some serve as an example of others to follow. Others serve as a warning to others.
In the past two months alone, from serious bugs in the Intel Management Engine to Meltdown, they've done a great job serving as a warning.
Slashdot Mirror
I get why JPEG isn't perfect but the big issue here is that for the vast and overwhelming use cases, JPEG is "good enough".
It's going to be hard to come up with an alternative that's not ridiculously patent encumbered.
I don't think patents are the most serious issue. It's a handy whipping boy, but the legacy installed base is the big one.
There have been numerous "replacements" to JPEG, as well as other image formats. It took decades for the vastly suprior PNG to overtake GIF (even with patent issues in GIF), but even then, GIF is very much alive and well.
The same can be said about older video codecs (Many Blu-rays are still encoded using MPEG-II, in spite of the availability of better codecs). Nobody complains about the 1992-era Dolby AC-3 audio on DVD's, and 1993's MP3 shows no few of slowing down, in spite of considerably better codecs being developed & deployed in the past two decades.
The bottom line is that while higher quality is possible, the "better" level isn't enough for most people to even care.
Glare on your device's screen can make a bigger difference.
The car was already stopped, so "doing nothing" wouldn't surprise me. Since they're still developing autonomous driving, it's better for "the company" to have the car stay put and let the Police sort it out. Better to pay a fine for blocking traffic than make the news for hurting somebody.
I seriously doubt any autonomous driving system is sophisticated enough to decide whether fleeing from (and possibly running over) attacking pedestrian(s) is the correct course of action.
There's also a good chance that the pedestrian committed felony assault, and in some nations/jurisdictions, failing to report a felony is a crime as well, as is driving away.
I'm pretty sure that's a thicket that developers would like to put in the backlog, for after the simple problems like driving in the snow are taken care of.
It never ceases to amaze me how often people are surprised when Apple blocks applications the first time it goes through the review, especially as the apps are nigh invariably approved after an appeal.
Spoiler: That's exactly what happened here, if you read the update in TFA.
there is absolutely no reason why Apple should withhold this from the store
TFA has since been updated: Apple has since decided to approve it.
It's common for Apple to deny apps their first go-round, and just as common for the app to be approved after additional review. It shouldn't be news to anybody, as it's been observed by developers for nearly a decade at this point.
I wouldn't be surprised if the practice is similar to SMTP servers requiring the originating server to re-send the message -- spam bots don't bother, and earnest developers will try again.
That or the reviewer was exerting his/her political views...
Or maybe one of Apples App reviewers is just a die hard Trumpkin and an enthusiastic Ajit Pai fan?
Eh... you can dislike regulations and not be a fan of either Trump or Pai.
Regardless, the TFA has since been updated, and Apple is going to allow the app:
Apple asked Choffnes to provide a technical description of how his app is able to detect if wireless telecom providers throttle certain types of data, and 18 hours after he did, the app was approved.
Honestly, App store denials for the first version of a new program are fairly common -- hardly a surprise given the number of developers shitposting into the App store.
After all that's been said, there's an update:
Update: After this article was published, Apple told Dave Choffnes that his iPhone app, designed to detect net neutrality violations, will be allowed in the iTunes App Store. According to Choffnes, Apple contacted him and explained that the company has to deal with many apps that don't do the things they claim to do. Apple asked Choffnes to provide a technical description of how his app is able to detect if wireless telecom providers throttle certain types of data, and 18 hours after he did, the app was approved.
"The conversation was very pleasant, but did not provide any insight into the review process [that] led the app to be rejected in the first place," Choffnes told us in an email.
the problem is this 'renewable' is not even a net zero return.
In many ways, that's not even the point. If we can create enough energy through other methods (Fission, Solar, Wind, Hydro -- even Fusion if it works out), then net zero is not as important as being able to convert the energy to a useful form. (The same applies for Hydrogen, for that matter).
Petroleum isn't going to last forever, and it's increasingly looking like we'll have supply problems by the middle of the century. That is a problem, and using biodiesel helps lengthen timeframe to develop something better.
Why do we care? Aerospace and shipping.
While there is some work for short-range electric jets, long flights at altitude are at a disadvantage compared to fuel-burning turbines (which are extremely efficient at operational altitude & speed). Jet fuel is very, very similar to diesel, and biodiesel is a possible replacement. Many of the large new ships are diesel powered - either piston or gas turbine. As you can imagine, militaries (especially America's) love the stuff because it's domestic, and it could power most of our Navy, our tanks, and our aircraft.
The primary fuel of booster rockets today is kerosene. Methane may work well in the future (and with a lower ISP).
I'm covered in the dust of the leader. He favors me!
I am even dustier -- dustier than thou!
It certainly qualifies: Makes you laugh at first, but then you think.
It's not about farts so much as modeling the gut.
I've got a niece who can't eat without debilitating pain once the food hits her stomach, so having a better idea why may be very useful.
Or have you become so repressed that you pretend it does not happen and that there's never a reason to talk about it?
I'd mod you up if I could.
Fecal humor exists in every human culture for a reason: We can't escape our biology.
Humanity is united in knowing that even Emperors and Kings can't escape it. It doesn't take much of a Google search to see the POTUS doing it. (I don't think it matters which POTUS...)
There's nothing dignified about cutting a turd loose. There's nothing elegant about removing shit stuck to your body.
We all know that from the smallest baby to the most gorgeous supermodel: all must go through the humiliation of giving birth to a politician.
If there's one thing that unites all of humanity, it's the knowledge that we all must check our egos at the door, and rendezvous with last night's supper.
On top of that, GPUs don't run kernel code (so cannot leak it),
It's entirely different from saying it's a problem with the GPU, but: There is the much-maligned NVIDIA kernel module which does interact with the GPU.
It's not far-fetched to say that there may be Meltdown/Spectre related patches -- though it's far more reasonable to say that any fixes would be a mitigation between the kernel module/driver and the CPU.
Anybody actually paying attention knew well before The Register printed anything.
The flaw was spelled out reasonably well by LWN as far back as November 15th, and it was noted that it was highly unusual for the patchset to be fast-tracked as it was. LWN also mentioned the initial KPTI patchset (then called KAISER) about a week earlier than that (Nov 10th). A month later, LWN followed up (including notes that ARM64 was affected) - more than a week before The Resister printed anything.
It was clear that something monumental was on the horizon, and that it was related to memory protection.
It was even clear that there was an information embargo in place, because comments were scrubbed from the associated patches.
It's been reasonably public for close to two months now.
The unknowns were more along the lines of "How deep is this pool of excrement," and "Which animal made it." Major OS patches were a fargone conclusion.
Why not simply take the device away from the child after x time elapses?
The old "take the toy away, give them a textbook to do homework" model doesn't work with computers.
Unlike a toy, computers have many, many abilities. By taking away the device, you may take away the unapproved toy, but also approved things you want your child to be doing -- reading books, news, maybe even textbooks, educational programming, or using tools to create art and music, and so on.
So, parental controls can be used to to limit functionality for reading, instead of gaming, which seems perfectly valid to me.
And I can walk over and look at what my child is doing on the device, but it only takes a fraction of a second to close the offending app and appear compliant. They can hide the evidence while I'm walking over.
Honestly... "My toddler's too quiet" is a meme for a reason: even the youngest toddlers know when they're sneaking behind their parent's back. Even the best parents have an extremely limited capacity to pay attention, and aren't capable of paying attention all the time. That little child-safety latch over the cupboard with drain cleaner may be a "technological gimmick", but I have yet to meet anybody who considers it bad parenting.
So parental controls can also be part of a "defense in depth" kind of thing -- yes, you monitor what your kids are doing. Eventually, you're going to be human and slip up, and that's where ever-watchful automation can help.
AMD original response is that there is a "near zero" risk of exploit for variant 2 and a "zero" risk for variant 3
Claims about "near zero" or "zero" risk of exploit are traditional CYA effluent, after all...
Even AMD has claimed (for a different CPU errata) that problems "would not occur during normal desktop usage and we've never encountered it during our testing." The claim was funny, as all you had to do was fire up a VM... or the right benchmark... It didn't speak well to AMD's testing at the time.
PCID is only supported on newer cpus
Having run across a nearly four-year old CPU having PCID (Haswell Xeon E3-1276), I'm less inclined to say that "newer" is entirely accurate; more that it was less common...
A few years back, John Carmack tweeted that he could ping transatlantic faster than putting a pixel on screen. He then explained more fully at superuser.
Since I run smokeping from my home server, I can definitely say that most online services are reachable from my home internet connection in ~20 ms round-trip - or 10 ms one-way.
For contrast, one frame at 30 FPS is 33 ms. (Kind of slow by modern standards, but I'm old enough for 30 FPS to be the threshold of being acceptable).
While it certainly matters to the experience, 2-3 frames of latency between my input and rendered pixels doesn't detract from the experience -- especially for games played with a gamepad (instead of a mouse).
It might not be suitable if you're playing an FPS game professionally, but it's fine for RPG and strategy games.
One less wire for my Toddler (or pet) to shove in his mouth is good. Charging pads (short-field charging) don't get me there.
There are, however, the mid-field and long-field wireless charging options in the pipeline that seem more compelling - Energous just received FCC approval for their version which lets you charge anywhere within three feet, with a more powerful version in development that can hit 15 feet. Those are more interesting.
AMP changed that and forced Facebook to follow suit with the lightning articles, and the mobile news reading experience is *infinitely* better now.
Bullshit on all accounts. Facebook's Instant Articles predated Google's AMP by several months, and Google is the one who was following.
Mobile news is markedly worse with AMP; I've read several accounts where publishers lament that it's slower than their native version.
My first sign that AMP is horribly broken is that every single AMP page I've ever visited all point to "google.com", and the URL bar shows "google.com" regardless of the site I'm actually visiting. Phishers can (and have) conceal pretty much anything behind AMP, and few users would have a clue because they see the lock with "google.com" at the top of their browser.
The next problem with AMP is that you can't turn it off: Google feeds you AMP pages if you use a mobile browser, and you have to load the AMP page first and then click to additional times to get to the non-AMP page.
The alternative is to use a different search engine, which is what a lot of us are doing.
AMP is a bigger problem than anything it was trying to replace.
no amount of wishing will put the AES-256 toothpaste back in the tube
Or the fact that pen & paper cryptography with the Vigenère_cipher and a sufficiently long key of random characters is still considered unbreakable. (Key sharing is a pain, but AES has the same problem)
It's all especially ironic as the "Advanced Encryption Standard" was a US Federal Government program where the world's cryptographers competed & collaborated to come up with unbreakable encryption. Or that other governments have done the same with NESSIE ECRYPT, and CRYPTEC.
And that US Federal Government is sponsoring workshops & standardization on post-quantum cryptography, ostenably so we'll have something secure & standardized before we can crack RSA, DSA, El Gamal, Elliptic Curves, etc. with quantum computers.
It's almost as if law enforcement doesn't want to go back to the heady days of 2006 when they did stake outs.
Screw what Gwyneth Paltro does with Orange Goop.
I want to see what Covetton House does with it.
Hardware isn't software; it takes a lot more than changing code and a recompile to fix.
Logic problems are only part of the story -- after the logic is done, there's physics to deal with (voltage breakdown, heat, crosstalk, etc). Then there's the "rat's nest" of conductive paths for the signals to follow.
Coders may complain about following spaghetti, but they've got nothing compared to what hardware guys deal with. Even minor changes can require considerable amounts of re-routing.
After that is complete, they still have to build the chips, which will take months more to get the fab machinery set up.
Expect to wait into 2019 and possibly later. Nobody has a processor that does out-of-order speculation that is immune...
Especially in the case of Microsoft, Hanlon's razor should be applied.
I don't have a problem with the idea that Intel is trying to actively sabotage AMD (they've done it in the past), but that it seems a lot more likely that Microsoft made a stupid mistake.
I'm not quite sure what you mean by "VT" -- do you mean Intel's VT-x?
It's probably best to let the Xen folks explain far better than I could hope to.
Quemu has their own explanation (and I think KVM is included in the explanation)
I get the feeling that the virtualization instructions (VT-x, AMD-V, and whatever the equivalents are on s390, ARM, and POWER) aren't really involved with either Spectre or Meltdown.
I wonder about SPARC, but given Oracle & Fujitsu appear to be killing SPARC... it probably doesn't matter.
To play devil's advocate: I don't know of a single chip that doesn't have errata published by the manufacturer. The first "real" lesson I learned in my EE career is "you can't take the manufacturer's documentation at face value."
The important thing is how problems are handled, and I don't know of any chip maker who handles bugs like this well.
When their GeForce 8600M had issues with overheating and dying, NVIDIA first blamed notebook manufacturers, and later the foundry. It was later found to be a problem with NVIDIA's silicon layout. Apple had to replace millions of MacBooks whose NVIDIA card died after less than a year (Ever wonder why Apple uses AMD graphics these days?)
Even AMD has issues when it goes into CYA mode: Back in the day, AMD had major TLB issues with their Phenom processor. At the time AMD claimed the "... TLB erratum is a highly random event that would not occur during normal desktop usage and we've never encountered it during our testing of Phenom." "Normal desktop usage" was defined to exclude a Xen hypervisor running Windows XP, or running the SPEC 2006 benchmark -- not exactly common, but they're clearly doing the PR spin.
Then there's Intel... Some serve as an example of others to follow. Others serve as a warning to others.
In the past two months alone, from serious bugs in the Intel Management Engine to Meltdown, they've done a great job serving as a warning.