HURRR GAEMS ARE FINE

So... I manage a cluster running multiple VMs for SAP applications. The storage system is Sun ZFS 7000. The database system is Oracle Exadata. Everything is overloaded and overbooked. CPU, network, I/O and all. And everything is Intel.

Fuck.

Other urls found in this thread:

techspot.com/article/1554-meltdown-flaw-cpu-performance-windows/
lkml.org/lkml/2018/1/3/770
twitter.com/SFWRedditVideos

Thanks for the update, keep us posted

rip in pepperonies OP

shouldn't have bought intel

suck it

Not my fault. I've suggested going for the SPARC-based Supercluster instead of Exadata but fucking Oracle was hellbent on convincing the higher ups to get the intel-based Exadata.

Why would you lie on the internet?
The Meltdown patch doesn't affect CPU performance by that much. We're talking less than single digits here. Fuck off, AMDshill

Fuck off Intel shill

>it's only a 10% performance drop goy, you don't even need 100% of your cpu!

KYS AMDrone
Where's the proof? None of the real world benchmarks and tests show any significant performance downgrade after the patch. Windows 10 sees a positive bump in some instances.
Proof or fuck off.

>oracle trash
Got what you deserved, Kumar. Go bitch about it on your blog.

It's over 60% in some usage cases.

Only in benchmarks, not in the real world.

>most debilitating hardware bug in recent history
>"Windows 10 sees a positive bump in some instances."

>10% drop is okay, you don't need all those speeds anyways
>None of the real world benchmarks
>>implying any tests shown to you, by us, won't have you claiming the tests are fake, not from official sources, not from Intel, etc etc
>only in benchmarks, not the real world
Make up your fucking mind. You're so desperate to be right you can't even keep your arguments straight.

Patently false. People have been posting real world pre and post patch performance cases all day.

>10%
Not even 0.6% in the most severe examples (aka margin of error).
Oh, you mean the ones that show almost no difference at all?

Sorry user, those are benchmarks not the real world

No. The ones that show massive 200% increases in servers under the same workload, and much more.

>Oh, you mean the ones that show almost no difference at all?
Try FDE on a C2D and tell me about this "no difference at all."

You mean the nonexistent ones that a redditor posted?

You wanna compare it to actual benchmarks?
techspot.com/article/1554-meltdown-flaw-cpu-performance-windows/

>look at these benchmarks!!!
>only in benchmarks, not the real world
You can't claim benchmarks are not reflective of the actual results in one post and then post nothing but benchmarks to prove your point in others

>show me real world cases not benchmarks
>I reject those LOOK AT THESE BENCHMARKS THOUGH!

Are you fucking retarded or what?

At least I've got proof that AMDrones are overhyping and underdelivering yet again

user, your benchmarks aren't a good representation of what's happening in the real world. They are meaningless

Hearsay and rumors are more meaningless.

You're going to love retpoline.
lkml.org/lkml/2018/1/3/770

In all fairness, this horrific bug makes windows 10 look good in comparison.

Look at your benchmarks. Then read the op post. Then look at your benchmarks.

Then see how performance drops rapidly in some cases.

It's not hearsay or rumours if you understood how the patch reworks how operations are performed you would instantly realize how serious the negative performance impact would be. You're just a stupid shill or delusional fanboy.

>I've suggested going for the SPARC-based Supercluster
sparc is dead

This is some next-level nasty shit.
What teh fuck.

By your own posts, benchmarks do not accurately reflect the (supposed) performance loss happening "in the real world". Following your own logic, benchmarks then should also not count for reflecting no performance loss (or even performance gain) "in the real world". If you claim that benchmarks do indeed reflect accurate loss or gain due to the update, as evidenced by the fact that you are posting them, then you can't get upset and claim they are not valid if others post verified benchmarks from others proving you wrong. Now, I don't have any because I don't give a flying fuck due to never using Intel, but I do have a problem with your intellectual dishonesty.
>inb4 AMD shill
I honestly just want people to be honest and consistent with their arguments

He's using it for a professional workload you fucking moron. Not gaymz.

Yeah the big news is still incoming.

This is going to rip Intel based servers a whole shiny new asshole

RIP in peace Intel CPUs

>SPARC
Didn't SUN die a decade ago?

SPARC processors are almost dead. They don't offer a sufficient advantage over x86 chips to warrant the price.

>No
>Working
>Exploit
>Exists

That teach you to fucking not plan your design so you have no headroom. Fucking retard.

Oracle kept SPARC to honor contracts with Sun customers, and to please Fujitsu and a few other large customers
But they basically abandoned R&D on it when they killed Rock, leaving the field free for Intel and to a lesser degree AMD

The information is public and it's one hell of a target.