Page 1 of 1

Intel chip solution: performance data available

PostPosted: Thu Jan 11, 2018 7:14 am
by stix

Re: Intel chip solution: performance data available

PostPosted: Thu Jan 11, 2018 3:25 pm
by Ian
Thanks stix. That's been a big topic at work as of late. Thankfully, most of our VM nodes aren't that old and only about 1/3 of our desktops will see a decrease in performance. Most of our users won't notice anyway. :)

Re: Intel chip solution: performance data available

PostPosted: Thu Jan 18, 2018 2:33 pm
by stix
I still have a couple of Intel Core i5 (32nm Clarkdale) based machines that serve me well, and a Core 2 Duo based laptop I still like. Wonder if there will even be a fix for these machines. Any thoughts?
Also wonder when it will be practical to buy a new CPU and board, given the vulnerabilities.

Re: Intel chip solution: performance data available

PostPosted: Sat Jan 20, 2018 12:32 am
by Ian
There really haven't been any vulnerabilities in the wild so I'm not sold on the idea of updating desktops yet. It's really going to depend on the motherboard manufacturer and whether or not they care.

Intel: new "microcode update guidance"

PostPosted: Mon Apr 23, 2018 12:21 am
by stix

Re: Intel chip solution: performance data available

PostPosted: Mon Apr 23, 2018 9:26 pm
by Ian
I've been dealing with Meltdown on my test system recently. I intentionally haven't updated the BIOS but Microsoft shoved through a fix with one of their updates. I really hadn't noticed a dip in performance with SATA and slower NVMe SSDs, but it really has an impact on drives like the Samsung 960 PRO. Results on the left are from my original review. Results on the right were last week. Same drive. Same firmware. Same test system. Only difference was that I had swapped out the boot drive for another 960 PRO and that Windows had been updated.

Samsung 960 PRO MeltDown.png
Samsung 960 PRO MeltDown.png (55.45 KiB) Viewed 18942 times


I can disable the Meltdown fix in Windows using Inspectre and results improve somewhat but I think the microcode in the CPU has already been updated since I can't reproduce the results from a year ago. ](*,)