[DISCUSSION] Meltdown and Spectre

Discussion in 'PC Hardware' started by scaramonga, Jan 3, 2018.

  1. Joe C

    Joe C MDL Guru

    Jan 12, 2012
    3,507
    2,082
    120
    Obviously they do have a bios fix but it's up the the OEM's to take the time to put that into a new bios update and that is not profitable for them on older boards.
    It'll probably come down to those that are tech savy to provide bios patches for older boards when the bios microcode becomes available for all
     
    Stop hovering to collapse... Click to collapse... Hover to expand... Click to expand...
  2. Yen

    Yen Admin
    Staff Member

    May 6, 2007
    13,101
    14,047
    340
    #262 Yen, Jan 18, 2018
    Last edited: Jan 18, 2018
    Stop hovering to collapse... Click to collapse... Hover to expand... Click to expand...
  3. Joe C

    Joe C MDL Guru

    Jan 12, 2012
    3,507
    2,082
    120
    Maybe someone will/can do that here but there's one hell of a lot of older boards out there today. All the current boards that have a modified SLIC bios will need to have an update, and a new SLIC (or the old SLIC re-added to the updated bios)
     
    Stop hovering to collapse... Click to collapse... Hover to expand... Click to expand...
  4. Daz

    Daz MDL Developer / Admin

    Jul 31, 2009
    9,530
    67,271
    300
    Stop hovering to collapse... Click to collapse... Hover to expand... Click to expand...
  5. Yen

    Yen Admin
    Staff Member

    May 6, 2007
    13,101
    14,047
    340
    Stop hovering to collapse... Click to collapse... Hover to expand... Click to expand...
  6. Daz

    Daz MDL Developer / Admin

    Jul 31, 2009
    9,530
    67,271
    300
    When they say "as prioritized by our customers" it makes me think they'll only patch a few of them if there's a demand for it.

    Personally I think Intel might release a few updates for older Xeon processors, but beyond that they're probably hoping that Microsoft adds Retpoline support. They've already said that the Retpoline technique may perform better than their own solution.

    If you do notice a difference in performance it'll be with storage I/O benchmarks, but it should still give a similar or better result than an updated BIOS.

    Code:
    spectre_v2=     [X86] Control mitigation of Spectre variant 2
                    (indirect branch speculation) vulnerability.
    
                    on   - unconditionally enable
                    off  - unconditionally disable
                    auto - kernel detects whether your CPU model is
                           vulnerable
    
                    Selecting 'on' will, and 'auto' may, choose a
                    mitigation method at run time according to the
                    CPU, the available microcode, the setting of the
                    CONFIG_RETPOLINE configuration option, and the
                    compiler with which the kernel was built.
    
                    Specific mitigations can also be selected manually:
    
                    retpoline         - replace indirect branches
                    retpoline,generic - google's original retpoline
                    retpoline,amd     - AMD-specific minimal thunk
                    ibrs              - Intel/AMD microcode feature
    
                    Not specifying this option is equivalent to
                    spectre_v2=auto.
     
    Stop hovering to collapse... Click to collapse... Hover to expand... Click to expand...
  7. Tiger-1

    Tiger-1 MDL Guru

    Oct 18, 2014
    7,894
    10,734
    240
    yep I remember once you make comment about :D
     
    Stop hovering to collapse... Click to collapse... Hover to expand... Click to expand...
  8. Joe C

    Joe C MDL Guru

    Jan 12, 2012
    3,507
    2,082
    120
    #269 Joe C, Jan 19, 2018
    Last edited: Jan 19, 2018
    We are just SOL if we have older boards, Looks like they could care less but I'm thinking that the industry as a whole might pressure board oem's to patch older boards to prevent any spreading of the flaw

    Edit:
    Then again maybe not?
     
    Stop hovering to collapse... Click to collapse... Hover to expand... Click to expand...
  9. Yen

    Yen Admin
    Staff Member

    May 6, 2007
    13,101
    14,047
    340
    ...not necessarily...not more than others....it is a general and major vulnerability based on a design flaw.

    We should not forget that the cause resides in the CPU (hardware) itself.
    AND branch prediction which is affected is responsible to have today's performance.
    I assume that spectre never gets fully patched/eliminated except by redesign of CPU.

    Even if the 'tools' report some day no vulnerability to spectre 1/2 and meltdown you cannot say that you are safe now. It concerns a basic CPU feature 'prediction' and speculation and also caching.

    Intel could disable the entire prediction, but then you would just have a CPU performance which is comparable to the millennium 2000.

    Anything will lead to a compromise where performance loss is still acceptable but spectre attacks still possible....
    The fight against spectre will finally be made by recognition of malicious code. It will be a cat and mouse game until the real cause vanishes...(the affected CPUs)...

    Btw: 32-bit windows versions are still vulnerable to both. M$ did nothing so far.....
     
    Stop hovering to collapse... Click to collapse... Hover to expand... Click to expand...
  10. Mikorist

    Mikorist MDL Member

    Dec 26, 2012
    205
    145
    10
    Stop hovering to collapse... Click to collapse... Hover to expand... Click to expand...
  11. Michaela Joy

    Michaela Joy MDL Crazy Lady

    Jul 26, 2012
    4,068
    4,649
    150
    It's going to vary from program to program. The amount of performance loss will be inversely proportional to what steps the programmer(s) took to optimize their software.
    Programs like CAD and DAWs might take a bad hit. CAD because of the graphics calculations (Although these can and will be minimized by use of the on board Floating Point coprocessor.)

    DAWs might not be so lucky, in that they typically use their own Floating point format (32 bit) which doesn't (IIRC) work with the FPU.

    Not only is every OS affected, every compiler is affected too. That means that every program that is time intensive will need to be recompiled, and re-optimized for performance.

    There's going to be a -LOT- of software consulting work coming up over the next year or so. Just like Y2K.
     
    Stop hovering to collapse... Click to collapse... Hover to expand... Click to expand...
  12. John Sutherland

    John Sutherland MDL Addicted

    Oct 15, 2014
    866
    1,390
    30
    #273 John Sutherland, Jan 22, 2018
    Last edited: Jan 22, 2018
    Like I said in another post, it's not a question of what Intel is able to do in order to resolve the problem. The real question is what are they willing to do? And at what cost? I'm sure their accountants and their lawyers will have the final say in this.
     
    Stop hovering to collapse... Click to collapse... Hover to expand... Click to expand...
  13. Michaela Joy

    Michaela Joy MDL Crazy Lady

    Jul 26, 2012
    4,068
    4,649
    150
    @john: The standard corporate line is "Duck and Weave."

    The CPU makers are going to blame it on the BIOS and the OS. The BIOS makers are only going to be able to do so much. So the rest of the mess falls on the OS makers and the compiler writers.
     
    Stop hovering to collapse... Click to collapse... Hover to expand... Click to expand...
  14. Mikorist

    Mikorist MDL Member

    Dec 26, 2012
    205
    145
    10
    #275 Mikorist, Jan 22, 2018
    Last edited: Jan 22, 2018
    Stop hovering to collapse... Click to collapse... Hover to expand... Click to expand...
  15. John Sutherland

    John Sutherland MDL Addicted

    Oct 15, 2014
    866
    1,390
    30
    Stop hovering to collapse... Click to collapse... Hover to expand... Click to expand...
  16. Yen

    Yen Admin
    Staff Member

    May 6, 2007
    13,101
    14,047
    340
    Stop hovering to collapse... Click to collapse... Hover to expand... Click to expand...
  17. Mikorist

    Mikorist MDL Member

    Dec 26, 2012
    205
    145
    10
    #278 Mikorist, Jan 23, 2018
    Last edited: Jan 23, 2018
    Stop hovering to collapse... Click to collapse... Hover to expand... Click to expand...
  18. Joe C

    Joe C MDL Guru

    Jan 12, 2012
    3,507
    2,082
    120
    Stop hovering to collapse... Click to collapse... Hover to expand... Click to expand...
  19. Mikorist

    Mikorist MDL Member

    Dec 26, 2012
    205
    145
    10
    :rofl: I never saw him so angry.
     
    Stop hovering to collapse... Click to collapse... Hover to expand... Click to expand...