Puppy Linux Discussion Forum

[rufwoof]

Linux (and Windows) Intel set to get 20% slower

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

[Sailor Enceladus]

[fabrice_035]

bad new!

I try with intel tool detection vulnerability

[fabrice_035]

The are new performance after patch :

[Sage]

Yes, the WIntel cartel scores again! If you weren't at least a late teenager in the late end of the 1980's, you may be unaware that Intel has consistently screwed up for three decades? Always with the cooperation of their $$$-sapping companions in Redmond, themselves guilty felons by the USDoJ.
Nothing, but nothing, has stopped the punters from snapping up their cr*p, particularly in their home nation which has had other more successful fabricators, some of which were sent to the wall by this misguided bunch of greedy capitalists.

[8Geee]

I find it disturbing that Intel denies improper programming of its CPU. Basically, Intel says their chips have the acceleration flaw, and its due to trying to predict the next call to HW/SW/Graphics. After all it IS unknown, and the CPU is looking for a performance edge over its rival.

IMHO not only do desktop/laptop CPU's have this "flaw" all smartphones definately have it, and HERE is where other vendors are at risk... ARM especially, since it farms some work to Intel.

Well, thats Intel Management Engine, and now predictive (ahem, accelerated) processing.

BTW, now is a good time to check your browser's predictive behavior... things like "auto-complete", for example.

Regards
8Geee

[belham2]

....Better re-think those thoughts that AMD is not affected by processor gobblins peaking their heads up. It sort of stinks that both AMD and Intel, it appears, have known about their specific cpu vulnerabilities for some time and have chosen to do nothing. There is no other explanation as there is NO excuse in any universe for not knowing about possible bypasses of "memory isolation mechanisms". Damn, this behavior almost makes Microsoft look saintly all these decades releasing OSes with full knowledge they all were (and still are) vulnerable.

http://www.securityweek.com/intel-amd-chip-vulnerabilities-put-billions-devices-risk

[Sky Aisling]

Here's another article about the 'holey' chip.

https://discuss.howtogeek.com/t/a-huge-intel-security-hole-could-slow-down-your-pc-soon/66686

[bark_bark_bark]

[Sailor Enceladus]

[bigpup]

Now you know how they are going to talk you into buying the next generation newest processor and Windows 11.

If you build it so it never breaks. They will only buy it one time!!

[Keisha]

Hello all, long time no see.

According to
https://www.postgresql.org/message-id/20180102222354.qikjmf7dvnjgbkxe@alap3.anarazel.de

you can do what the patchset will do by setting pti=on (pti = page table isolation) in the kernel line of the grub bootloader.

At work I use a dual xeon e5-2696 v3, running Fedora 26, nice fast number-cruncher.

Hardinfo 0.6-alpha benchmark results on this machine with and without pti=on, average of three runs:

Without pti=on:
CPU Blowfish 0.28 seconds
CPU CryptoHash 3008.91 MiB/second
CPU Fibonacci 1.13 seconds
CPU N-Queens 26.97 seconds
CPU ZLib 5.75 seconds
FPU FFT 0.78 seconds
FPU Raytracing 29.84 seconds
GPU Drawing 21,815.72 HIMarks

With pti=on:
CPU Blowfish 0.29 seconds
CPU CryptoHash 2652 MiB/second
CPU Fibonacci 1.13 seconds
CPU N-Queens 27.04 seconds
CPU ZLib 5.86 seconds
FPU FFT 0.78 seconds
FPU Raytracing 42.42 seconds
GPU Drawing 22,494 HIMarks

CPU CryptoHash shows a 12% performance loss.
FPU Raytracing shows a 30% performance loss.

All other benchmarks, essentially the same. Subjectively, the machine doesn't feel any slower.

Could this translate to a significant advantage of AMD over Intel in product sales going forward?

[8Geee]

based upon the previous post... is this something that can be added to extlinux_conf?

Regards
8Geee

[s243a]

[Keisha]