Personal computing discussed
Moderators: renee, Flying Fox, morphine
chuckula wrote:I think some of the FUD being spread around doesn't survive the light of actual testing, and that includes synthetic benchmarks where the KPTI performance hit is supposed to be "catastrophic".
Glorious wrote:the wrote:Intel will likely push Cascade Lake into 2019 due to this bug or simply cancel it in favor Cannon Lake-SP in 2019 which is still on the roadmap last I heard. The rumors are pointing toward Cascade Lake as being a Sky Lake-SP update still on 14 nm due to delays in 10 nm production. One big new feature in Cascade Lake was to fix to support Optane DIMMS, a feature Intel removed at the last minute for Sky Lake-SP due to a bug found in validation. I see it as incredibly unwise to release Cascade Lake and Optane DIMMs if it is still susceptible to Meltdown or Spectre.
Why would Intel cancel anything? They're going to be shipping "flawed" silicon in the interim anyway.
the wrote:Intel will likely push Cascade Lake into 2019 due to this bug or simply cancel it in favor Cannon Lake-SP in 2019 which is still on the roadmap last I heard. The rumors are pointing toward Cascade Lake as being a Sky Lake-SP update still on 14 nm due to delays in 10 nm production. One big new feature in Cascade Lake was to fix to support Optane DIMMS, a feature Intel removed at the last minute for Sky Lake-SP due to a bug found in validation. I see it as incredibly unwise to release Cascade Lake and Optane DIMMs if it is still susceptible to Meltdown or Spectre.
Microsoft wrote:Microsoft has reports of customers with some AMD devices getting into an unbootable state after installing recent Windows operating system security updates. After investigating, Microsoft has determined that some AMD chipsets do not conform to the documentation previously provided to Microsoft to develop the Windows operating system mitigations to protect against the chipset vulnerabilities known as Spectre and Meltdown. To prevent AMD customers from getting into an unbootable state, Microsoft will temporarily pause sending the following Windows operating system updates to devices with impacted AMD processors at this time: (a long list of KBs followed that I won't add)
DancinJack wrote:https://support.microsoft.com/en-us/help/4056892/windows-10-update-kb4056892?ranMID=24542&ranEAID=nOD%2FrLJHOac&ranSiteID=nOD_rLJHOac-T4HFDD4IA8m_R5rpVHsPKw&tduid=(4f8fd051ba6f771f06d1cb7f4161fc1a)(256380)(2459594)(nOD_rLJHOac-T4HFDD4IA8m_R5rpVHsPKw)()
https://support.microsoft.com/en-us/hel ... based-devi
Some people with AMD CPUs were apparently applying the updates for these bugs and the result was a computer that wouldn't boot anymore. Microsoft...responded.Microsoft wrote:Microsoft has reports of customers with some AMD devices getting into an unbootable state after installing recent Windows operating system security updates. After investigating, Microsoft has determined that some AMD chipsets do not conform to the documentation previously provided to Microsoft to develop the Windows operating system mitigations to protect against the chipset vulnerabilities known as Spectre and Meltdown. To prevent AMD customers from getting into an unbootable state, Microsoft will temporarily pause sending the following Windows operating system updates to devices with impacted AMD processors at this time: (a long list of KBs followed that I won't add)
Not a great look for Microsoft, but probably a worse look for AMD.
DancinJack wrote:Not a great look for Microsoft, but probably a worse look for AMD.
setaG_lliB wrote:From the above thread, it looks like the update is only bricking really old K8-based systems (Athlon 64 and A64-X2). K10/Phenom and newer appear to be just fine.
just brew it! wrote:Most people will blame Microsoft, since the patch came from them.
ludi wrote:A human-interest piece on the various researchers that found these flaws, from Wired:
https://www.wired.com/story/meltdown-sp ... -discovery
SuperSpy wrote:just brew it! wrote:Most people will blame Microsoft, since the patch came from them.
It's Microsoft's perpetual curse. Vendor does something undocumented or stupid, MS updates Windows, previously harmless but stupid thing breaks computer, user blames MS. It's a large part of why MS is so scared of changing anything.
mudcore wrote:SuperSpy wrote:just brew it! wrote:Most people will blame Microsoft, since the patch came from them.
It's Microsoft's perpetual curse. Vendor does something undocumented or stupid, MS updates Windows, previously harmless but stupid thing breaks computer, user blames MS. It's a large part of why MS is so scared of changing anything.
And it wasn't found in testing? Sounds 100% like a shared screw up. Users pay Microsoft in large part to vet these things for them.
just brew it! wrote:mudcore wrote:SuperSpy wrote:It's Microsoft's perpetual curse. Vendor does something undocumented or stupid, MS updates Windows, previously harmless but stupid thing breaks computer, user blames MS. It's a large part of why MS is so scared of changing anything.
And it wasn't found in testing? Sounds 100% like a shared screw up. Users pay Microsoft in large part to vet these things for them.
SHIP IT!!!
chuckula wrote:PC Perspective ran a few storage tests (including Optane): https://www.pcper.com/news/Storage/Melt ... ally-Issue
Short version: There's an impact but it's not catastrophically bad and ironically the 900P SSD actually saw a small reduction in latency that I'm not saying is caused by the Meltdown fix but is more indicative of the Meltdown fix not being catastropically bad.
Waco wrote:chuckula wrote:PC Perspective ran a few storage tests (including Optane): https://www.pcper.com/news/Storage/Melt ... ally-Issue
Short version: There's an impact but it's not catastrophically bad and ironically the 900P SSD actually saw a small reduction in latency that I'm not saying is caused by the Meltdown fix but is more indicative of the Meltdown fix not being catastropically bad.
I don't trust those results for one second. Not to dig on PC Perspective, but there's no way what they're claiming is true - something else it at work there.
We're seeing 5-25% drops in performance for a lot of the workloads we run. Anything that does intensive I/O (network or disk) is seeing a pretty significant impact.
Waco wrote:chuckula wrote:PC Perspective ran a few storage tests (including Optane): https://www.pcper.com/news/Storage/Melt ... ally-Issue
Short version: There's an impact but it's not catastrophically bad and ironically the 900P SSD actually saw a small reduction in latency that I'm not saying is caused by the Meltdown fix but is more indicative of the Meltdown fix not being catastropically bad.
I don't trust those results for one second. Not to dig on PC Perspective, but there's no way what they're claiming is true - something else it at work there.
We're seeing 5-25% drops in performance for a lot of the workloads we run. Anything that does intensive I/O (network or disk) is seeing a pretty significant impact.
chuckula wrote:That article uses a Kaby Lake CPU that presumably has full PCID support (including INVPCID). Are your servers also of skylake or newer generation or are they Haswell/Broadwell era (or older) machines? Additionally, I'm assuming your storage system is not just hanging off the chipset like on a consumer device. Is there an interaction with the storage controller that could be unhappy with the patches?
just brew it! wrote:BTW, here's the BSOD you got if you're one of the unfortunate people who still have a K8-based system and installed the MS Meltdown/Spectre patch before MS blacklisted them:
Waco wrote:It's almost as useful as the Windows 10 "Something Happened" bsod.
Topinio wrote:Intel published client impact assessment benchmarks, impact is +6 to -21% AFAICT.
Obviously, these are close to besst-case as they are all with CPU well above mid-range let alone low-end, and all Skylake or newer.