|
Unknown for NT, but they have been working on this for months.
|
#
?
Jan 3, 2018 18:39
|
|
|
|
| # ? May 15, 2024 12:59 |
|
|
Rastor posted:That may go some ways to explain why the Linux kernel developers set the pti=on to be the default for all x86 processors in their patch rather than having an exception for AMD. The really funny thing is that 50% number actually comes from running the patch on Epyc processors, not Intel (where the pathological case is 30%). In other words, a rough guess might be that this patch hits AMD twice as hard as it hits Intel. If they switch it on by default, Intel could actually gain relative performance  10 years from now we'll still be hearing about how Bernie Would Have Won if only Intel wasn't cheating, release the nopti benchmarks shintel! It's really not the world's worst idea to just do it by default if the performance impact is that minimal, though.
|
|
#
?
Jan 3, 2018 18:53
|
|
|
Cygni posted:so the current Linux fix has the performance hit regardless of CPU manufacturer, do we know if thats the case with the NT kernel fix yet? Paul MaudDib posted:The really funny thing is that 50% number actually comes from running the patch on Epyc processors, not Intel (where the pathological case is 30%). In other words, a rough guess might be that this patch hits AMD twice as hard as it hits Intel. If they switch it on by default, Intel could actually gain relative performance Latest news is that AMD's request to turn off the PTI on their chips *will* be accepted into the Linux kernel, though it may go into 4.16 instead of 4.15 https://www.phoronix.com/scan.php?page=news_item&px=Linux-Tip-Git-Disable-x86-PTI
|
|
#
?
Jan 3, 2018 19:01
|
|
|
Paul MaudDib posted:So yeah, to be frank this one's a nothingburger as far as end users are concerned. That's not spin, that's a realistic assessment of the situation. Performance is one thing, security is the other really big problem here. Sure it'll be patched out, it still doesn't change the hand waving you've done  .
|
|
#
?
Jan 3, 2018 21:01
|
|
|
repiv posted:I don't envy the people who write GPU drivers https://twitter.com/FioraAeterna/status/948481524013268992 security through mindfuck
|
|
#
?
Jan 3, 2018 21:26
|
|
|
oh no https://twitter.com/FioraAeterna/status/948473228686647296
|
|
#
?
Jan 3, 2018 21:37
|
|
|
Generic Monk posted:i mean the only potential downside to treating this issue as the serious thing it is, is if you own any intel stock. which patently seems to not have been affected by the issue anyway. forcing them to own their mistakes, and at the very least getting them to abide by the nominal rules of the free market hellhole in which we currently reside, should be a moral imperative Apple patched for this as of 10.13.2
|
|
#
?
Jan 3, 2018 21:39
|
|
|
Paul MaudDib posted:The really funny thing is that 50% number actually comes from running the patch on Epyc processors, not Intel (where the pathological case is 30%). In other words, a rough guess might be that this patch hits AMD twice as hard as it hits Intel. If they switch it on by default, Intel could actually gain relative performance It seems like you're bipolar or something when it comes to Intel or AMD and you need to address that or at least learn to tone down the hyperbole 20 or so notches when it comes to anything anti AMD or pro Intel you might say.
|
|
#
?
Jan 3, 2018 21:39
|
|
|
https://twitter.com/brainsmoke/status/948561799875502080 Lawl
|
|
#
?
Jan 3, 2018 21:42
|
|
|
Over 100 million? 1060 or 970? Betting on 970, with its weird as gently caress memory config 
|
|
#
?
Jan 3, 2018 21:43
|
|
|
Truga posted:Over 100 million? 1060 or 970? It's probably something mobile, I think Fiora works for one of the SoC manufacturers. No idea which one though.
|
|
#
?
Jan 3, 2018 21:46
|
|
|
yeah, I googled and it doesn't appear any single nvidia SKU sold >100 million. Has to be either something mobile or an intel iGPU in laptops.
|
|
#
?
Jan 3, 2018 21:48
|
|
|
I'm a little skeptical because the GPU market is about 15M units/year. The discrete desktop GPU market anyway.
|
|
#
?
Jan 3, 2018 21:48
|
|
|
Truga posted:yeah, I googled and it doesn't appear any single nvidia SKU sold >100 million. Has to be either something mobile or an intel iGPU in laptops. Yeah I'm betting Intel GMA or some other turd like that edit: repiv posted:It's probably something mobile, I think Fiora works for one of the SoC manufacturers. No idea which one though. Oh, yeah, could be VideoCore or Adreno then.
|
|
#
?
Jan 3, 2018 21:50
|
|
|
Mediatek or Rockchip SoC would be my bet.
|
|
#
?
Jan 3, 2018 21:51
|
|
|
Yeah, I enjoyed that one too Xae posted:I'm a little skeptical because the GPU market is about 15M units/year. To be fair, it could be the core architecture of a very popular GPU as opposed to a specific SKU
|
|
#
?
Jan 3, 2018 22:04
|
|
|
https://www.fool.com/investing/2017/12/19/intels-ceo-just-sold-a-lot-of-stock.aspx From the 19th of last year, could be a case of insider trading.
|
|
#
?
Jan 3, 2018 22:17
|
|
|
SwissArmyDruid posted:https://www.fool.com/investing/2017/12/19/intels-ceo-just-sold-a-lot-of-stock.aspx The SEC makes you schedule your sales two quarters in advance, it's kinda doubtful that BK knew the exact week that it was going to blow up. The general approach has been known for a while, but the attempts at the time were not successful. I think it's more likely that he didn't want to hold more shares than necessary just based on a negative general outlook... a resurgent AMD, a 10nm process that is permanently stuck in the prototype phase, the urgent need for a serious uarch revision, etc etc (all of which are publicly known). Intel still has a lot of inertia in the market, but they are not in a great strategic position as far as the next year or two is concerned and it's likely that inertia will continue to deteriorate. Paul MaudDib fucked around with this message at 22:34 on Jan 3, 2018 |
|
#
?
Jan 3, 2018 22:26
|
|
|
It's not biased to say it affecting epyc more then Intel is funny as hell, because it really is in a "everything is actually poo poo" way. I was on the fence about ryzen refresh or an 8700k sooner rather then later, but I'm getting AMD regardless of anything else now. Still funny.
|
|
#
?
Jan 3, 2018 22:46
|
|
|
GRINDCORE MEGGIDO posted:It's not biased to say it affecting epyc more then Intel is funny as hell Since the fix won't apparently be needed for AMD's chips its not a issue.
|
|
#
?
Jan 3, 2018 23:06
|
|
|
PC LOAD LETTER posted:it wasn't the bug effecting Epyc in that test, it was the fix for Intel's chips that was causing the poor performance. I was referring to Paul's post about the patch. Really we should all hug. GRINDCORE MEGGIDO fucked around with this message at 23:29 on Jan 3, 2018 |
|
#
?
Jan 3, 2018 23:21
|
|
|
https://security.googleblog.com/2018/01/todays-cpu-vulnerability-what-you-need.htmlGoogle Security posted:These vulnerabilities affect many CPUs, including those from AMD, ARM, and Intel, as well as the devices and operating systems running them. AMD? Is there a second disclosure coming that affects them too?
|
|
#
?
Jan 3, 2018 23:37
|
|
|
No, and there's no patch for it. Read bottom to top for some reason:
|
|
#
?
Jan 3, 2018 23:39
|
|
|
Paul MaudDib posted:The SEC makes you schedule your sales two quarters in advance, it's kinda doubtful that BK knew the exact week that it was going to blow up. Intel stock was up like 30%, taking all that money off the table isn't that unusual of a scenario
|
|
#
?
Jan 4, 2018 00:23
|
|
|
https://videocardz.com/74580/amd-arm-google-intel-and-microsoft-issue-official-statements-on-discovered-security-flawsAMD PR posted:There is a lot of speculation today regarding a potential security issue related to modern microprocessors and speculative execution. As we typically do when a potential security issue is identified, AMD has been working across our ecosystem to evaluate and respond to the speculative execution attack identified by a security research team to ensure our users are protected. https://twitter.com/ryanshrout/status/948683677244018689
|
|
#
?
Jan 4, 2018 02:29
|
|
|
Has anyone seen any benchmarks of the impact on heavily parallel loads on post-update Windows on Zen? I know we've seen gaming loads but I haven't seen anything for server and workstation type loads.
|
|
#
?
Jan 4, 2018 03:48
|
|
|
Hey Paul what’s worse, segfaulting chips or leaking data through JavaScript and web browsers  poo poo is out of control
|
|
#
?
Jan 4, 2018 04:24
|
|
|
Lol at Paul trying to spin this as a loss for AMD
|
|
#
?
Jan 4, 2018 04:25
|
|
|
Rastor posted:Latest news is that AMD's request to turn off the PTI on their chips *will* be accepted into the Linux kernel, though it may go into 4.16 instead of 4.15 It's in Linus' 4.15 tree now: code:The_Franz fucked around with this message at 04:29 on Jan 4, 2018 |
|
#
?
Jan 4, 2018 04:26
|
|
|
https://blog.mozilla.org/security/ Lmao. If this happened to AMD the company would go bankrupt
|
|
#
?
Jan 4, 2018 04:27
|
|
|
Maybe Intel can pay everyone in
|
|
#
?
Jan 4, 2018 04:35
|
|
|
Risky Bisquick posted:Hey Paul what’s worse, segfaulting chips or leaking data through JavaScript and web browsers Obviously hard locking due to segfaults, duh.
|
|
#
?
Jan 4, 2018 04:40
|
|
|
SourKraut posted:Maybe Intel can pay everyone in I certainly didn't get any rebates for RAM when rowhammer was invented.
|
|
#
?
Jan 4, 2018 04:47
|
|
|
Segfaults, since there's no mitigations available. A 1% performance impact from the mitigation really isn't a big deal. Also, let's remember that AMD is leaking data through Javascript right now too - the paper stated the researchers have made the Spectre attack work on AMD too, they just won't admit it at the moment. Parsing what they've said very carefully, they only said they aren't vulnerable to Meltdown, which is correct, but they've only stated that there is only a "low risk" from Specter, which is kinda correct since it's a more difficult and more limited category of exploit, but equally true of Intel. And unlike Meltdown, there is no broad-spectrum software mitigation that's possible, you have to go program by program and patch bugs - and it's not really clear to me if there's even a cookie-cutter fix that will work. Meltdown is a lot of hot air but in 6 months everyone will have sucked it up and patched. Specter is going to be the one that sucks to fix, unfortunately, and I really don't see a good fix except for re-implementing speculative execution and being a lot more careful the next time around. Paul MaudDib fucked around with this message at 04:55 on Jan 4, 2018 |
|
#
?
Jan 4, 2018 04:48
|
|
|
fishmech posted:I certainly didn't get any rebates for RAM when rowhammer was invented. Think AMD wrote any checks when Phenom had the TLB bug and performance actually dropped by 20% across the board?  In the grand scheme of mitigations this really isn't that bad. If disabling opcache or some other fix made Ryzen work properly, I wouldn't give it as much poo poo as I do.
|
|
#
?
Jan 4, 2018 04:55
|
|
|
https://twitter.com/BryanLunduke/status/948430797266042880 E this stings even worse. My amd stock going to the moon https://mobile.twitter.com/attritionorg/status/948759303153856512 Risky Bisquick fucked around with this message at 05:04 on Jan 4, 2018 |
|
#
?
Jan 4, 2018 04:59
|
|
|
That's the Meltdown patch though, it doesn't have anything to do with Specter. AMD is still vulnerable, as is every CPU on the market that uses speculative execution. edit: from the Spectre paper, the authors tested the attack on Ryzen and it's vulnerable:  The only processors that aren't affected are low-end ARM, Atom, and OG Pentiums (and older). There is no software mitigation for this one either. The low-hanging fruit is stuff like lowering the timer resolution in browsers... but to really fix it you either need to patch every single application that does a bounds check on an array, or you need to replace your processor with one that implements speculative execution securely or not at all. Paul MaudDib fucked around with this message at 05:14 on Jan 4, 2018 |
|
#
?
Jan 4, 2018 05:04
|
|
|
SwissArmyDruid posted:https://videocardz.com/74580/amd-arm-google-intel-and-microsoft-issue-official-statements-on-discovered-security-flaws And also to be clear, Google didn't put together that chart (blog posts here and here), and that's a statement from AMD PR there, not Google. It's not clear where that image came from and it contradicts the papers themselves in several areas. Spectre cannot be patched in software and it was verified in Ryzen according to the paper. And issue 3 is the one that's driving the PTI software patches, issue 1 and 2 collectively make up the category of Specter exploits. The "software fix" for issue 1, if you can call it that, is patching thousands of individual applications. Paul MaudDib fucked around with this message at 05:25 on Jan 4, 2018 |
|
#
?
Jan 4, 2018 05:22
|
|
|
Verified in a construction core APU*
|
|
#
?
Jan 4, 2018 05:23
|
|
|
|
| # ? May 15, 2024 12:59 |
|
|
Risky Bisquick posted:Verified in a construction core APU* That's not what the paper stated. Maybe you should read it? Or excerpt I previously linked for you?  
Paul MaudDib fucked around with this message at 05:29 on Jan 4, 2018 |
|
#
?
Jan 4, 2018 05:26
|
|