Zombieland exploit...will you DISABLE hyper-threading???

Discussion in 'Politics, Religion, Social Issues' started by Crunch, May 16, 2019.

  1. Crunch macrumors 6502a

    Crunch

    Joined:
    Jun 26, 2008
    Location:
    Crazy L.A.
    #1
    So you bought yourself a nice expensive MacBook Pro or iMac and have an Intel CPU that supports hyper-threading. How many of you will disable HT thanks to the latest backdoored-in-Israel chip and lose up to 40% in performance??
     
  2. SpeedyTheSnail, May 16, 2019
    Last edited by a moderator: May 17, 2019

    SpeedyTheSnail macrumors member

    SpeedyTheSnail

    Joined:
    Sep 22, 2018
    Location:
    Caprica
    #3
    Also it is Zombieload not Zombieland.

    But the answer is no. It's not publicly exploited and there's really no point to worry at this point time.

    I also would expand your question to:
    You own a modern computer that has an Intel CPU. Would you disable hyperthreading?

    This is in the same lines as Spectre and Meltdown.

    These academics should be hired or paid by Intel to unscrew their designs, but these speculative execution and side-channel attacks are very new and have not been seen in the wild.

    Edit: and SGX.
     
  3. treekram macrumors 68000

    Joined:
    Nov 9, 2015
    Location:
    Honolulu HI
    #4
    Back when news of the Spectre and Meltdown HT vulnerabilities broke, there was a podcast where the announcer said that hyperthreading added 10% performance. (Keep in mind that you only get extra performance when your simultaneous thread count exceeds the number of physical CPU cores in your system.) At the time, I tested using HandBrake on my 2012 Mini and I got 15% extra performance. Keep in mind to get this performance increase, I was using 90-100% CPU utilization on all (physical and virtual) cores. I'd be curious to know if anybody can actually get 40% extra performance.

    People got unduly (IMO) freaked out the last time and they will again this time. If you're concerned, turn off hyperthreading. Be more concerned about getting any kind of malicious software on your system rather than the specific nature of that malicious software.
     
  4. throAU macrumors 603

    throAU

    Joined:
    Feb 13, 2012
    Location:
    Perth, Western Australia
    #5
    Hyperthreading can add a LOT More than 10% performance depending on workload, and the more recent cores (than you 2012 era mini) gain more from it in general due to having more "idle" resources spare in the later model processors at any given time.
     
  5. jerwin macrumors 68020

    Joined:
    Jun 13, 2015
    #6
  6. Crunch, May 16, 2019
    Last edited by a moderator: May 17, 2019

    Crunch thread starter macrumors 6502a

    Crunch

    Joined:
    Jun 26, 2008
    Location:
    Crazy L.A.
    #7
    Unless and until y'all realize that Intel, Microsoft, and now Apple ALL are moving OUR (U.S.) high-tech industry to ISRAEL, you know exactly NOTHING.

    https://www.timesofisrael.com/bill-gates-israeli-tech-changing-the-world/

    https://www.israel21c.org/microsofts-bill-gates-israel-is-a-vital-resource-for-us/

    Or watch this: How Israel backdoored everything: youtube.com/watch?v=gLQNWWyI5iQ

    Or how about Trump an asset of Israeli milintelligence: youtube.com/watch?v=lKe32JerYws&t=1s

    Here's more: How Israel rules: Barbarians inside the gate: youtube.com/watch?v=gLQNWWyI5iQ
     
  7. Crunch, May 17, 2019
    Last edited by a moderator: May 17, 2019

    Crunch thread starter macrumors 6502a

    Crunch

    Joined:
    Jun 26, 2008
    Location:
    Crazy L.A.
    #8
    Alright, and sorry to be dramatic about this but this is so important that I just put the WiFi back on just to come back to this thread and really attempt to instill in you and everybody else who cares...hopefully everyone.

    We are under attack. It nevertheless has to be said: Israel is nothing like we're constantly being told what it is and we're in BIG trouble. Everything's back-doored but in a serious way. This little Zombieload, or whatever the latest exploit pushed by the media onto the masses for the purpose of keeping you in a constant state of fear, is, while very real, meaningless in the grand scheme of things.

    Anyone who's interested, please send me a PM (or whatever it's called now) and I will put a few things together for you. I have noticed that every last one of those videos is a dead link. I suspect this topic will be shut off shortly. Every other one always has in the past even when posted in the appropriate sub-forum, which this topic, admittedly, is not.
     
  8. SpeedyTheSnail, May 17, 2019
    Last edited by a moderator: May 17, 2019

    SpeedyTheSnail macrumors member

    SpeedyTheSnail

    Joined:
    Sep 22, 2018
    Location:
    Caprica
    #9

    I honestly read your post and didn't catch it at first, they did choose a poor name for this vulnerability. I very much doubt that these processors are intentionally backdoored, but of course I could be wrong. Companies that intentionally backdoor their products are bound for failure once discovered.
     
  9. Crunch, May 17, 2019
    Last edited by a moderator: May 17, 2019

    Crunch thread starter macrumors 6502a

    Crunch

    Joined:
    Jun 26, 2008
    Location:
    Crazy L.A.
    #10
    I respect you for responding instead of doing what most everyone else would have done.

    Do you work in the tech industry?? I can tell you're at least a little bit receptive and curious. You just want evidence and that's only fair and to be expected. I'd love your opinion on a few key videos that do a great job of fretting this out and they go into depth as to how Intel and QualComm are two key players in this.

    None of this is my view, by the way. It's just information that most people don't know and that the people in power don't want you to know. I'll see if I can upload them and send you the link before it gets taken down.

    I'll try it do it today. Meantime, have a great day! :)
     
  10. flowrider macrumors 603

    flowrider

    Joined:
    Nov 23, 2012
  11. blackfox macrumors 65816

    blackfox

    Joined:
    Feb 18, 2003
    Location:
    PDX
    #12
    wtf my dude. Seriously, chill.
     
  12. treekram macrumors 68000

    Joined:
    Nov 9, 2015
    Location:
    Honolulu HI
    #13
    After doing some testing on my various macos computers, I'm left wondering how Apple came up with the 40% loss from disabling hyperthreading figure. Using the H.264 and H.265 encoders in HandBrake (x264 and x265), the most I get is 20% (just tested one video for each encoder on the various computers). On my 2018 base-model 15" MBP (6-core), I get 16.6% for x264 and 8% for x265. I triple-checked the x265 figure and did a second video. Now, if I compare x265 with 10.14.5 vs 10.14.3 I get 17.2% loss (so either there was mitigation in the 10.14.3 supplemental and 10.14.4 or the later versions are less efficient - I also used HandBrake 1.2.0 which was the latest version at the time vs. 1.2.2 for my current tests).

    Using the Geekbench multi-core tests, which has multiple benchmark tests rolled into one score, I get 12.0% peformance loss on my 2012 quad-core Mini, 7.5% on my 2018 6-core MBP and that 7.5% is awfully close to the x265 performance loss.

    So did Apple use figures from the iMac Pro (which uses Xeon and Intel says performance loss is greater in data-center computers) and/or some hyperthreading benchmark that is not realistic in real-world computer use? Or perhaps the 40% was using completely non-mitigated systems (there have been hyperthreading mitigations for the 2017 bug and the 2018 Meltdown/Spectre vulnerabilities). Surely somebody in PR-savvy Apple would have known that using that 40% was sure to grab headlines. At this point, Apple should really disclose more details on how they got the 40%.

    If anybody has done testing where they've come close to the 40% hyperthreading loss, I'd like to get details.
     
  13. thekev macrumors 604

    thekev

    Joined:
    Aug 5, 2010
    #14
    Apple always conditions these things as "up to" without providing useful guidelines or recommendations.

    See this post for an example from VMWare.

    No MDS-style attack patching for older MacPros - now what?

    As I pointed out, their big losses are under workloads that have a high level of compulsory latency due to multiple fetches from memory or disk when run at a high level of utilization. These are all running server side workloads on top of hypervisors to maximize utilization. At 90% of cpu cycles in active use, they're hitting a > 30% slowdown.

    As I point out in the same thread, if your main thread can saturate a single core over the majority of its execution time, there's very little that hyperthreading can actually do for you. I would guess that your examples fall closer to my reasoning.
     
  14. throAU macrumors 603

    throAU

    Joined:
    Feb 13, 2012
    Location:
    Perth, Western Australia
    #15
    It REALLY depends on what you are doing and how the workload scales with hyperthreading.

    Geekbench attempts to be an overall measurement. The badly hurt hyperthreading situations will be PART of it's benchmark process. Some workloads don't gain much with hyperthreading, some do.
     
  15. cube macrumors P6

    Joined:
    May 10, 2004
    #16
    I would say that if you're going to buy an Intel CPU, you should not pay more than a few bucks to get HyperThreading until they actually fix the hardware.

    If you don't plan to resell it later, I would not even bother.
     

Share This Page

15 May 16, 2019