Spectre Patch Causes 'Significant Slowdown' On Older OS, Chips

[Lucian Armasu]

Microsoft and Intel show "significant slowdown" in performance after applying Meltdown and Spectre patches to Windows PCs.

Spectre Patch Causes 'Significant Slowdown' On Older OS, Chips : Read more

 

[nostrocrompt]

*reads article*
*looks at beautiful, over-achieving 5820k*
*cries*

 

[Kennyy Evony]

I ran a test on my haswell system it said it wasn't vulnerable to the defects? win7 4790k cpu

 

[Math Geek]

i certainly noticed on my older machine. that was once i got it booting again after the patch torched the boot record and rendered it unable to boot to 1 of 2 OS's.

fun fun fun

 

[takeshi7]

So has Intel given any hints to when they'll release a CPU that has this issue fixed?

 

[danwat1234]

Yep, so I won't be allowing the patch(es) on my Win7/8 computers at all. Risk is low. Performance and efficiency is very important to me.

Mass public will tho... planned obsolescence conspiracy . sub 10% performance gains in CPUs year over year average, oh but now your old computer sucks, gotta upgrade to 2017+ hardware..

 

[TadashiTG]

Typical Intel and Microsoft, they want people to spend money on their new products.

 

[Giroro]

"Microsoft and Intel’s message today seems clear: it's time to upgrade."

To what, exactly? Neither company offers a product that fixes the security issue without degrading performance.

 

[InvalidError]

Typical Intel and Microsoft, they want people to spend money on their new products.

Microsoft isn't making much money with Windows 10... it wanted to pass the billionth install mark in the first year with free upgrades to help it along and it only got half-way there over the first two years if I remember correctly. Probably the second slowest new Windows version adoption after Vista, perhaps third after ME.

 

[tpi2007]

Tom's Hardware, please don't inflate the problem before having done benchmarks and don't put words in Microsoft's mouth. Specifically:

"According to the company, machines running Windows 7 and 8, as well as computers based on Haswell chips or older, will see “significant slowdowns” from the update."

That is not what they said. This is:

"With Windows 8 and Windows 7 on older silicon (2015-era PCs with Haswell or older CPU), we expect most users to notice a decrease in system performance."

https://cloudblogs.microsoft.com/microsoftsecure/2018/0...


"significant" is not there, so you can't put quotes on that.


Similarly here:

"No matter which version of Windows, machines that use 2015-era Haswell CPUs or older will experience “significant slowdowns,” according to Microsoft."

That's also not what they said and they differentiated between OS versions. To "some" users it will only be in "some benchmarks". And we already know how some benchmarks exaggerate the issue.

 

[CerianK]

Intel’s findings: “six-cores and 12 threats”

Seems odd that Intel would admit it in that way. ??

Anyway, I have 6/12 ‘threats’. Good thing I’ve got my Xeon W3690 OC locked to x29 multiplier already... maybe it won’t suck quite so bad if Win 10 forces nasty patches down my throat. Just to be clear: I run all of my cores all of the time doing research work... I will notice if a 10 day job suddenly takes a few extra days.

 

[dudio]

I'm running an i7 6700K and it's the first major upgrade I've done in years.. I don't want to lose any performance !
When is the update supposedly coming thru, and can it (and should it) be blocked?

 

[cryoburner]

Haswell Chips And Older
No matter which version of Windows, machines that use 2015-era Haswell CPUs or older will experience “significant slowdowns,” according to Microsoft. The company also expects that most users will be able to notice that performance degradation. Skylake CPUs and newer have more refined branch prediction, so the Spectre patch doesn’t influence them to the same degree.

I suspect what you meant to say is "Broadwell Chips and Older". I believe Broadwell was primarily just a process shrink of Haswell as far as the CPU logic is concerned, so it is also likely affected to a larger degree. When I started reading that, I thought "Oh good, at least the laptop won't see quite as much of a performance impact" before noticing that you completely ignored the existence of Broadwell, and just skipped to saying how Skylake made some changes to branch prediction that results in less performance degradation.

Also, I would argue that these CPUs can still be considered "modern hardware". They were the latest generation of processors up until a little over two years ago, and the year-on-year improvements to Intel's CPUs have been quite minimal for much longer than that.

I ran a test on my haswell system it said it wasn't vulnerable to the defects? win7 4790k cpu

I believe there was another article on here that incorrectly linked to a test for a different recent hardware exploit that affects certain Intel CPUs. Unlike that exploit, these particular hardware exploits should affect all Intel CPUs stretching back the last couple decades, so your CPU is most definitely vulnerable.

 

[Benjamin_20]

Once they drop I'd love to see some numbers for 2600k on windows 10, I know a lot of gamers still rocking sandy-bridge.

 

[btmedic04]

Looks like sandy bridge users finally have a reason to upgrade

 

[drtweak]

::Looks at my FX-8320@4.3Ghz:: Now we are on even playing fields! XD haha

I know Spectra affects AMD and from the reports isn't affected by Meltdown. I still have a lot of clients with older Core 2 Duo/Quads and everything in between now and then.

I guess Intel is going to have to take its tick tock tock and do a tick to fix this issues probably lol. I will have to do a LOT of testing. I have a few Core 2 E7200, i5-680, and a i5-6700 along with a AMD B55 and a FX 8320E at my office. Also got my Core 2 P8600 in my laptop and another laptop with a second gen i3 (Which oddly even with a SSD seems slower than my 8 year old Dell P8600 haha)

 

[tpi2007]

Haswell Chips And Older
No matter which version of Windows, machines that use 2015-era Haswell CPUs or older will experience “significant slowdowns,” according to Microsoft. The company also expects that most users will be able to notice that performance degradation. Skylake CPUs and newer have more refined branch prediction, so the Spectre patch doesn’t influence them to the same degree.

I suspect what you meant to say is "Broadwell Chips and Older". I believe Broadwell was primarily just a process shrink of Haswell as far as the CPU logic is concerned, so it is also likely affected to a larger degree. When I started reading that, I thought "Oh good, at least the laptop won't see quite as much of a performance impact" before noticing that you completely ignored the existence of Broadwell, and just skipped to saying how Skylake made some changes to branch prediction that results in less performance degradation.

Also, I would argue that these CPUs can still be considered "modern hardware". They were the latest generation of processors up until a little over two years ago, and the year-on-year improvements to Intel's CPUs have been quite minimal for much longer than that.

I ran a test on my haswell system it said it wasn't vulnerable to the defects? win7 4790k cpu

I believe there was another article on here that incorrectly linked to a test for a different recent hardware exploit that affects certain Intel CPUs. Unlike that exploit, these particular hardware exploits should affect all Intel CPUs stretching back the last couple decades, so your CPU is most definitely vulnerable.

If you read Microsoft's blog post you'll notice that they forgot to mention Broadwell. Having said that, Broadwell's branch predictor has more commonalities with Skylake that mean that for Google's Retpoline software approach to work it also needs a microcode update. Haswell and below work just fine with Retpoline without a microcode update.

Read more here:

https://newsroom.intel.com/wp-content/uploads/sites/11/2018/01/Intel-Analysis-of-Speculative-Execution-Side-Channels.pdf

"For Intel® Core™ processors of the Broadwell generation and later, this retpoline mitigation strategy also requires a microcode update to be applied for the mitigation to be fully effective. "


Also here:

https://security.googleblog.com/2018/01/more-details-about-mitigations-for-cpu_4.html

"Variant 2 (CVE-2017-5715), “branch target injection”. This variant may either be fixed by a CPU microcode update from the CPU vendor, or by applying a software mitigation technique called “Retpoline” to binaries where concern about information leakage is present. This mitigation may be applied to the operating system kernel, system programs and libraries, and individual software programs, as needed."

 

[abbadon_34]

How convenient ...

 

[thor220]

So how long did Intel exactly know about this issue? It entirely possible that Intel designed the chip with this flaw on purpose to get the extra performance. Imagine if the 2000 series had a 30% drop in performance. It puts them pretty close to AMD at the time in fact. These finding just make everything seem even more sketchy. It's entirely possible that after Intel's bribing of OEMs, they pulled off an even scummier move in intentionally speeding up their processors regardless of security.

 

[InvalidError]

It entirely possible that Intel designed the chip with this flaw on purpose to get the extra performance.

Meltdown and Spectre are relatively unexpected problems and the fact that they spawn several generations of Intel, AMD and even ARM hardware should tell you that they are pretty much certainly unintentional - how many chip designer would foresee that someone would find a way to do a side-channel attack on a CPU based on the work speculative execution does in the background? For all intents and purpose, that work is supposed to be invisible to the software but people found a way to measure it and leverage it to infer data in other processes.

If you'd told that to chip designers 10 years ago, I bet most of them wouldn't have believed you and even told you it sounds impossible until you showed them proof that it is actually possible. I can't blame them for not seeing this sort of convoluted exploit coming.

 

[thor220]

It entirely possible that Intel designed the chip with this flaw on purpose to get the extra performance.

Meltdown and Spectre are relatively unexpected problems and the fact that they spawn several generations of Intel, AMD and even ARM hardware should tell you that they are pretty much certainly unintentional - how many chip designer would foresee that someone would find a way to do a side-channel attack on a CPU based on the work speculative execution does in the background? For all intents and purpose, that work is supposed to be invisible to the software but people found a way to measure it and leverage it to infer data in other processes.

If you'd told that to chip designers 10 years ago, I bet most of them wouldn't have believed you and even told you it sounds impossible until you showed them proof that it is actually possible. I can't blame them for not seeing this sort of convoluted exploit coming.

They don't span multiple generations of AMD hardware at all

https://www.amd.com/en/corporate/speculative-execution

Direct report from AMD says they are only vulnerable to meltdown, which was already fixed with the windows and linux patch.

 

[boju]

Will the update be labelled in a way to identify its the spectre update or will it just be the usual naming of 'security update'? Id like to block the update on win7.

Nm i got it, KB 4056894, its there in the update list, think i'll wait. Not liking what im reading about this, blue screens/needing to flash bios. MS pulled the update at one stage, im not going to bother with it.

http://news.softpedia.com/news/microsoft-s-windows-7-meltdown-and-spectre-patch-kb4056894-failing-with-bsod-519264.shtml

 

[Kenton82]

So me running a i7 950 on windows 10 will get royally shafted then? This is confusing the hell outta me! Only just fixed a browsing speed issue due to Norton 360...

 

[jabliese]

So how long did Intel exactly know about this issue? It entirely possible that Intel designed the chip with this flaw on purpose to get the extra performance. Imagine if the 2000 series had a 30% drop in performance. It puts them pretty close to AMD at the time in fact. These finding just make everything seem even more sketchy. It's entirely possible that after Intel's bribing of OEMs, they pulled off an even scummier move in intentionally speeding up their processors regardless of security.

It is highly unlikely that a hardware fix for this problem will have the same performance penalties as the software patch does. Also, Intel has had no competition from AMD for the last 10(?) years, so had no reason not to go ahead and fix this, had they discovered it.

 

[bigdragon]

As an Ivy Bridge-E (i7-4930k) user, I'm ticked. The performance impact from these processor flaws looks huge. I'm not seeing a serious upgrade option from Intel, nor do I want to spend the money right now.