[Closed] Nevermind the iPhone - everyone's computer is going to slow down a lot

That's a big cock up....

yep was told about this earlier, not sure if the article mentioned it but potentially 2-30% performance hit, though I believe the 30% is only for (very) specialist tasks in linux.

EDIT: Article suggests 17 to 23%,hmmmm

z1ppy - Member
yep was told about this earlier, not sure if the article mentioned it but potentially 2-30% performance hit, though I believe the 30% is only for (very) specialist tasks in linux.

Anything that does a significant amount of system calls (i.e. mostly everything) will be affected. 30% seems unlikely except for very odd workloads though.

Must say I'm not overly impressed with Intel of late, after this and learning of the IME exploits.

ETA: if i'm reading the postgres example correctly, what they're timing is how fast their DB can connect and essentially do nothing (well, return a 1 anyway), and it's over a loop back, so no network overhead, minimal work to parse the sql and no i/o. In other words it's probably not too relevant to most database applications.

Grounds for a refund for my cpu?

Blah blah blah performance hit blah blah malware blah blah exploit blah vulnerability blah speculation.... nothing we haven't seen before.
Maybe this is fake news put out by Intel to get people to buy new processors...

From what I can gather, its only data centers, people who run VM's...

We don't know how performance will be with further optimizations, only the initial patch based on the phoronix benches, which seem to show it probably won't affect 99% of what regular users do.

The CEO of Intel sold a large amount of his stock in the company yesterday.

Coincidentally.

https://www.fool.com/investing/2017/12/19/intels-ceo-just-sold-a-lot-of-stock.aspx

EDIT: Me and wwaswas playing snap!

It was back in November, by the looks of it, probably when he had a good idea of what was coming down the track, but the markets didn't.

A class action of some kind is inevitable if people start seeing 30% drops in CPU performance.

Grounds for a refund for my cpu?

Well this is the interesting bit. *If* Intel have to refund/replace all CPU's made in the last 10 years it'll ruin them.

Be interesting to keep an eye on and see what comes of all of this.

nothing we haven't seen before.
Maybe this is fake news put out by Intel to get people to buy new processors...

No, this is a very significant security flaw and the only processors people will be buying as a result of it are AMD ones.

The reason for the performance hit is because the only way to deal with the issue to prevent the CPU from undertaking certain tasks that it would normally do that speed everything up.

This is right down at microcode level - deep deep within the CPU. This is not a fudge or fake news. Trust me - it *could* send Intel under.

No windows benchmarks yet, but bad news for servers, SQL etc: [url= https://www.phoronix.com/scan.php?page=article&item=linux-415-x86pti&num=2 ]https://www.phoronix.com/scan.php?page=article&item=linux-415-x86pti&num=2[/url]

Gonna be a lot of angry data centers with angry customers.

And realtime cloud computing at Google etc.

How about if you're running a 150,000 core Xeon 16C machine, do you think that might be a problem?

Gonna be a lot of angry data centers with angry customers.

And mountain biking forums.

Likely hood of the CEO knowing about this is high, reported that OS programmers knew about it and was trying to fix it back end of November early December. Selling his stock seems very fishy now. They brought forward all their new cpu's to 2017 because AMD's Ryzen chip was better than expected on all accounts, if that has anything to do with it. Faulty rushed coding?

I think the Champers will be flowing in the AMD boardroom today 😯

The CEO of Intel sold a large amount of his stock in the company yesterday.

Coincidentally.

I lol'd. Can that be legal? Even if he holds the minimum required of a CEO, surely its some kind of insider dealing.

This going to break the internet quicker than Kardashians with Kittens. Or sell AMD 17-30% more CPUs ...

I lol'd. Can that be legal? Even if he holds the minimum required of a CEO, surely its some kind of insider dealing.

It says a lot about his competencies as a CEO that he clearly thought he'd get away with it.

Live footage from AMD HQ

[img] [/img]

mattyfez - Member
Live footage from AMD HQ

[code]- /* Assume for now that ALL x86 CPUs are insecure */
- setup_force_cpu_bug(X86_BUG_CPU_INSECURE);
+ if (c->x86_vendor != X86_VENDOR_AMD)
+ setup_force_cpu_bug(X86_BUG_CPU_INSECURE);[/code]

https://lkml.org/lkml/2017/12/27/2

[img] [/img]

Normal users don't seem to be hit, but the server market is thier bread and butter:

[url= https://www.hardwareluxx.de/index.php/news/hardware/prozessoren/45319-intel-kaempft-mit-schwerer-sicherheitsluecke-im-prozessor-design.html ]https://www.hardwareluxx.de/index.php/news/hardware/prozessoren/45319-intel-kaempft-mit-schwerer-sicherheitsluecke-im-prozessor-design.html[/url]

Does that mean this place will get even more slowerer?

is that possible?

How about if you're running a 150,000 core Xeon 16C machine, do you think that might be a problem?

No, this is a very significant security flaw and the only processors people will be buying as a result of it are AMD ones.

Most people just buy a laptop online and neither know nor care what processor it uses...

Are you sure Singletrack didn't post this as fake news to explain why the site has been slow lately? 😉

what intel processors does it affect?

seosamh77 - Member
what intel processors does it affect?

For now, all intel CPUs.

what intel processors does it affect?

Everything from Clarkdale/Westmere (2010) onward, so yes effectivley all of them unless youve got a really old PC/server

Hmm… ARM sell off now looks like a daft move…

The sooner we have open hardware, the better.

scaredypants - Member
I reckon your framerate might drop a bit on FIFA

Dammit - Rpeak (TFlop/s) reduced to 3,750 - but WoW should be ok, no?

Everything from Clarkdale/Westmere (2010) onward, so yes effectivley all of them unless youve got a really old PC/server

Looks like I'll be fine, then. (-: Mine's... Penryn I think, off the top of my head.

I wonder, will they be patching all OSes, or just the ones with affected CPUs? Kind of a bit shit being forced to patch a vulnerability I don't have.

I wonder, will they be patching all OSes, or just the ones with affected CPUs? Kind of a bit shit being forced to patch a vulnerability I don't have.

The Linux kernel option can be disabled on the kernel boot line, and there's a patch floating around to disable it for AMD.

It also affects Aarch64.

Intel shares down 4.2% since this morning, that's a shed load of money.

AMD shares up 7.2%.

[i]The fix is to separate the kernel's memory completely from user processes using what's called Kernel Page Table Isolation, or KPTI. At one point, Forcefully Unmap Complete Kernel With Interrupt Trampolines, aka ****WIT, was mulled by the Linux kernel team, giving you an idea of how annoying this has been for the developers.[/i]

I was about to return my laptop as the speakers have gone all crackly...got another 18 months on the warranty so I might hold fire incase they end up replacing it with an updated processor!

ARM sell off now looks like a daft move…

Arm processes may be affected also - see article for further details.

Still got a XT 286 so I win!!

Unless it’s still broken from that Millennium bug

Intel shares down 4.2% since this morning, that's a shed load of money.

AMD shares up 7.2%.

Correction:
Intel now at -4.4%

But AMD still climbing at +8.2%

I wonder if the Intel CEO bought AMD shares with the money he made on his stock 🙂

I'll stick to my Cyrix 6x86MX for now until they get things sorted.

Looks like the share prices for AMD and Intel are settling down now.. That's not much consolation if you've kitted out a billion pound data centre based on Intel chips in the last few years though.

I imagine intel reps are going to have some 'interesting' meetings with customers in the near future.

Haha, my QX6850 is too old to care! Still going to get nerfed though presumably.

Kinda glad I never got that i3 8100 package now. Odds on 8th gen supply issues being closely related due to a pre-launch redesign?

I'll stick to my Cyrix 6x86MX for now until they get things sorted.

If that's still going it must be liquid cooled.

If that's still going it must be liquid cooled.

My PCs have been liquid cooled since the late 90s. I mainly just use an old laptop now though.

Gaming sites have said the fix won't affect gaming performance, which is where most users push their machines.

Gaming sites have said the fix won't affect gaming performance, which is where most users push their machines.

Yeah the average home user, including gamers won't notice any difference realistically.

There's still a huge problem for data centers and cloud hosts etc.

House full of AMD processors here! Would feel smug, but my electicity bill shows that I use AMD as they are hungry wee things!

Two I7's (gaming desktop and gaming laptop), two I5's and a Zeon (older workstation).

Massive issue for data centres though.

Yeah, Intel here!.... 🙁

This must impact Macs too I'm guessing?

I don't use the pc an awful lot these days. Most of my stuff is done on my smartphone.

Still, INTEL, not looking great for you guys?

I used to run AMDs, as they were great value and I liked their underdog status for some reason.

It doesn't really affect home users, you might take a one or two FPS hit for your favourite game.

What it does affect is big business as that performance degradation could cost a lot of money in terms of broken SLA's and OLA's.

You don't spend wad of cash with a remote host and accept a 30% degradation on your Web, SQL servers etc.

This must impact Macs too I'm guessing?

Of course it affects macs, they all run Intel CPU's. But you probably won't be able to tell unless your using your mac as an industrial server, which nobody ever did ever.

Pfui! Just use Linux on POWER8 chips. You know it makes sense!

Or I suppose we could migrate everything back to the mainframe.

I guess this finally puts the kybosh on running Windows Server 2016 on a dl980 then .... 😛

Wrongly reported. Not just intel

https://security.googleblog.com/2018/01/todays-cpu-vulnerability-what-you-need.html?m=1

[url= https://newsroom.intel.com/news/intel-responds-to-security-research-findings/ ]Intel statement[/url]

"...many types of computing devices — with many different vendors’ processors and operating systems — are susceptible to these exploits."

"Contrary to some reports, any performance impacts are workload-dependent, and, for the average computer user, should not be significant and will be mitigated over time."

Yeah the average home user, including gamers won't notice any difference realistically.

Unless you play online. What do you think servers run with? I mean, besides the STW tuber based data farm.

You know what I'm looking forward to, multiple topics on the first page full of people trying to figure out how to un**** Windows Update so they can patch this.

Intel is committed to the industry best practice of responsible disclosure of potential security issues, which is why Intel and other vendors had planned to disclose this issue next week when more software and firmware updates will be available. However, Intel is making this statement today because of the current inaccurate media reports.

And their stock value!

If you have an up to date Mac, it appears this was fixed with the last update. Can't say I've seen anything noticeable in day to day use of my MBP.

https://www.macrumors.com/2018/01/03/intel-design-flaw-fixed-macos-10-13-2/

Cert seems quite clear what the solution is;

[img] [/img]

..many types of computing devices — with many different vendors’ processors and operating systems — are susceptible to these exploits."

"Contrary to some reports, any performance impacts are workload-dependent, and, for the average computer user, should not be significant and will be mitigated over time."

Note this is Intel's statement and is contradicted by pretty much every IT security specialist who has looked at the problem.

darrenspink - Member

Wrongly reported. Not just intel

R4 saying AMD as well - sell those shares quick?

[i]Note this is Intel's statement [/i]

Google's statement as well:

"These vulnerabilities affect many CPUs, including those from AMD, ARM, and Intel, as well as the devices and operating systems running on them."

There are (at least) two related issues, maybe AMD/ARM is only affected by one of them.

https://meltdownattack.com/

It has to be remembered that this vulnerability is not to execute malicious code, but to potentially read kernel memory.

So for cloud hosts, and esx servers etc (ie the systems likely to be running at high capacity) the threat is actually quite minimal.

I am going to call some meetings this morning to see how we will deal with the situation, but there is absolutely no need to panic, I am quite happy to allow a mid to long term strategy to implement a fix.

[i]there is absolutely no need to panic[/i]

*assumes it must be really bad*

[img] [/img]

MSP - being able to read the host kernel from a VM on a cloud host is *really* bad. Like about as bad as you can imagine. Basically, any other guest on that cloud host can potentially read your data. Think about it for a minute...

Rachel

[i]So for cloud hosts, and esx servers etc (ie the systems likely to be running at high capacity) the threat is actually quite minimal.[/i]

That's what I was thinking - blocked externally?
So the threat is to normal users, who won't be affected by the patch... and datacentre servers?

It allows cross-VM data copying, how is that a minimal threat?

You don’t know who is buying space on the same host as your super-important service. They can upload code without your knowledge that simply harvest data from ring-0 and send it anywhere.

MSP - being able to read the host kernel from a VM on a cloud host is *really* bad. Like about as bad as you can imagine. Basically, any other guest on that cloud host can potentially read your data. Think about it for a minute...

There is no indication that the virtual system will have "pass through" capability to read physical kernel memory that I am aware of. But it is still a developing situation I expect to be getting more information thought the next few days.

There is no indication that the virtual system will have "pass through" capability to read

Yes, it appears there is: https://www.geekwire.com/2018/intel-cloud-vendors-shed-light-two-major-chip-security-issues-emerged-week/

Rachel

There is no indication that the virtual system will have "pass through" capability to read physical kernel memory that I am aware of. But it is still a developing situation I expect to be getting more information thought the next few days.

I acknowledge It is a developing situation, but It would appear thats exactly what this flaw will allow people to do, once the exploit has been developed 'fully'.

I'd say its the public cloud providers and people running workloads on public cloud providers that are at the biggest risk.

At present it appears no one has developed this exploit yet, but you can expect all the bad'uns to be beavering away at it now.

All my Azure subs were patched last night with very little warning at around 2am.

Apparently performance impact is minimal...

https://azure.microsoft.com/en-us/blog/securing-azure-customers-from-cpu-vulnerability/

There is no indication that the virtual system will have "pass through" capability to read physical kernel memory that I am aware of. But it is still a developing situation I expect to be getting more information thought the next few days.

Some further analysis from El Reg, particularly of Intel's statement...

https://www.theregister.co.uk/2018/01/04/intel_meltdown_spectre_bugs_the_registers_annotations/

Wow, that doesnt sound biased in any way. Especially when everone else seems to be acknowledging that it isnt confined to Intel, omething they seem to have glossed right over.

Might the increased overhead in patching this be problematic for people who rely on battery life of laptops a lot? A few percent might not impinge on processor performance per se if you're not running at 100%, but it will presumably chew battery?

Someone asked that question in the reg comments. No answer.

As suggested I imagine it will have an impact even if not immediately apparant. More activity means more power means more heat. And that's just at consumer level.

In other news, once we reach the year 2000, all the computers in the world will stop working with cataclysmic consequences. Next to that, taking an extra 20 seconds to process an image in Photoshop will seem like a walk in the park.

Apple are fairly clear it's ARM processors too, and by implication AMD.
[i]
“Security researchers have recently uncovered security issues known by two names, Meltdown and Spectre,” it added. “These issues apply to all modern processors and affect nearly all computing devices and operating systems. All Mac systems and iOS devices are affected, but there are no known exploits impacting customers at this time.[/i]

[url= https://www.theguardian.com/technology/2018/jan/05/apple-mac-spectre-meltdown-iphone-ipad-hackers?CMP=twt_a-technology_b-gdntech ]https://www.theguardian.com/technology/2018/jan/05/apple-mac-spectre-meltdown-iphone-ipad-hackers?CMP=twt_a-technology_b-gdntech[/url]

There still seems confusion/lack of information about which CPUs are vulnerable to which of the attacks. I'm hoping things get clarified quickly!

At least as far as I understand:

Meltdown does seem primarily Intel related (although the Apple position implies could also affect ARM), from what I've read the whole "Intel CPUs since 2010" was derived from what the Google zero day team tested on, they have actually said it could be an issue since branch prediction microcode was introduced in the 90's (they just haven't tested CPUs that old to confirm).

Spectre is much more widespread in what CPUs are affected but is more difficult to exploit