Jump to content

Intel CPU? Prepare to lose some frames


Recommended Posts

  • Aerosoft

If you run a modern Intel CPU you have to be aware of the rather large security issue found late last year. The only way to fix it seems to be to patch the OS (Windows/Linux) but that will have the drawback that your machine will be at least 5% slower. Not patching it simply is not an option, this leak is to big to ignore.

 

The internet is awash with articles about it, but for a quick heads-up I advise this one: https://www.howtogeek.com/338269/a-huge-intel-security-hole-could-slow-down-your-pc-soon/

Link to comment
Share on other sites

This is not good. 

 

https://www.theregister.co.uk/2018/01/02/intel_cpu_design_flaw/

 

"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. "

Link to comment
Share on other sites

  • Aerosoft

The news on this issue is still very confusing, with many conflicting reports. But for sure all IO, both to disk and network will be heavily affected. And as the issue seems to be in software (or even hardware parts) that can't be patched, the OS will have to avoid these calls and use different, less effective methods.

 

Spoken to a database manager just now. He got 150 servers, all with i7's and all running almost constant network and disk I/O. He simply can't afford to lose 20%  (that seems o be a very moderate guestimate now for that work) performance. He is full blown panic mode. 

Link to comment
Share on other sites

2 hours ago, Mathijs Kok said:

 

Spoken to a database manager just now. He got 150 servers, all with i7's and all running almost constant network and disk I/O. He simply can't afford to lose 20%  (that seems o be a very moderate guestimate now for that work) performance. He is full blown panic mode. 

He's using i7s to run a database farm with 150 servers rather than XEON CPUS??? 

 

But in any case, this is something on my radar as well.  I'm the Chief Technology Officer for a global company - we utilize both on-prem and cloud compute platforms.  What will also be interesting is to see how cloud providers manage to any performance hits with their customers...

 

PS - look at AMD stock after this went public!

Link to comment
Share on other sites

  • Aerosoft

Oops indeed... not i7 but 18 core Xeon, same problem though. Apart from the fact it is a massive issue, it is very interesting to see how it is developing in the news. There is still a massive amount of nonsense going round and none of the big sites seems to be able to patch it all together. The latest reports however are not good. 

 

I got one older Intel test machine here that runs the Fast ring of win 10/64 and I remember I had a video showing the startup of XP-11. Now I am not 100% sure if  it ran the same build as I got now on that SSD, but a late December Fast ring compared to the June video showed it now takes 20% longer to load.  From 19 seconds to 23 seconds (core i7-965 Extreme with a Samsung 512Gb SSD). Seems it will become very important to be frugal with the thousands of small files that are so often part of our sims.

 

 

Link to comment
Share on other sites

I have asked my architects to begin testing in our labs and will report back any performance differential results we see. 

 

We have 24,000 employees however who run i5 & i7 laptops...LOL, our Information Security department has slowed them down so much as it is with existing encryption, anti virus and other security runtime measures I think this patch may turn them into genuine bricks.

 

Interesting point you bring up about managing the thousands of small files in the flightsim development community going forward Mathijs.... again we will have to see to what extent any final patch will have on CPU and overall performance.

Link to comment
Share on other sites

Hey all,

So one of my engineers has been doing some serious research on this issue and found this article:  https://support.microsoft.com/en-us/help/4072698/windows-server-guidance-to-protect-against-the-speculative-execution

 

Note that this article is in reference to servers only for now.  It goes on to say that the provided mitigations impact to performance is largely dependent on the type of workload running and the particular processor....  It may be negligible, or it could be significant.  I am providing this as informational only.

 

In our labs we saw the most impact to measurable performance on virtualized and storage I/O related workloads, but due to confidentiality and security requirements I can't provide much detail on our architecture.  But we aren't panicking just yet.  The desktop environment will be different and for normal office work we didn't see much if it all degraded performance (I wish I knew more about how flight simming files and code execution is handled by the CPU in order to say if our little community would be impacted)  Again, the patch alone will not be the factor that degrades performance....it is the type of workload that the patch will negatively impact.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. Privacy Policy & Terms of Use