Announcement

Collapse
No announcement yet.

Spectre Mitigation Performance Impact Benchmarks On AMD Ryzen 3700X / 3900X Against Intel

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • elatllat
    replied
    Cleaning up Michael's format fail;
    • Intel
      • l1tf: Mitigation of PTE Inversion
      • mds: Mitigation of Clear buffers; SMT vulnerable
      • meltdown: Mitigation of PTI
      • spec_store_bypass: Mitigation of SSB disabled via prctl and seccomp
      • spectre_v1: Mitigation of __user pointer sanitization
      • spectre_v2: Mitigation of Full generic retpoline IBPB: conditional IBRS_FW STIBP: conditional RSB filling.
    • AMD Zen/Zen+
      • l1tf: Not affected
      • mds: Not affected
      • meltdown: Not affected
      • spec_store_bypass: Mitigation of SSB disabled via prctl and seccomp
      • spectre_v1: Mitigation of __user pointer sanitization
      • spectre_v2: Mitigation of Full AMD retpoline IBPB: conditional STIBP: disabled RSB filling.
    • AMD Zen 2
      • l1tf: Not affected
      • mds: Not affected
      • meltdown: Not affected
      • spec_store_bypass: Mitigation of SSB disabled via prctl and seccomp
      • spectre_v1: Mitigation of __user pointer sanitization
      • spectre_v2: Mitigation of Full AMD retpoline IBPB: conditional STIBP: always-on RSB filling.
    It looks like there is a decline in Zen 2 security ... maybe they are just late with the patches.
    Last edited by elatllat; 08 August 2019, 06:08 AM.

    Leave a comment:


  • reavertm
    replied
    Originally posted by scottishduck View Post

    The article clearly states intel has a 50% IPC lead over Zen 2 with their Celeron G4900?
    IPC stands for instructions per clock.
    And clock to clock single thread comparisons show over 10% Zen 2 advantage over flagship 9900k. Benchmarks are all over Internet, in ex Hardware Unboxed.
    Intel regains single thread performance in brute force - higher clocks on 14+++ process than with 7nm recently and in single process multi-thread in cross-core delay (lower on Intel due to its monolithic architecture).

    Those are the facts but hey, you can argue all you want.
    Edit: you are on crack (seeing things like the mention of Celeron) therefore my post is irrelevant.
    Last edited by reavertm; 16 July 2019, 07:19 AM.

    Leave a comment:


  • nranger
    replied
    Originally posted by Mark625 View Post
    Michael, in the intro page you list the two Threadrippers as "Ryzen 9 2950X" and "Ryzen 9 2990WX". Shouldn't those be just "Ryzen 2950X", or maybe "Ryzen Threadripper 2950X"? It took me a few seconds to figure out what you were talking about.

    It would be nice if the charts listed them as "TR 2950X" etc, too.

    Thanks.
    Indeed, my pedantic brain kicked in and started thinking there was some AMD press release renaming their workstation processors that I missed. Before I even read past the 2nd paragraph of the article I went to https://www.amd.com/en/products/spec...ons/processors to check I wasn't confused and the Threadripper name had been retired or that the "Ryzen 9" branding had been ret-conned.

    Michael if you value accuracy I'd recommend going back and removing the "Ryzen 9" when actually referring to Threadripper parts.

    Leave a comment:


  • torsionbar28
    replied
    Originally posted by scottishduck View Post
    The article clearly states intel has a 50% IPC lead over Zen 2 with their Celeron G4900?
    Lolwut? You drunk bruh?
    Last edited by torsionbar28; 15 July 2019, 09:53 PM.

    Leave a comment:


  • audir8
    replied
    The 2990WX (32C64T) is barely keeping up with the 3900X and the 2950X is far behind in the geometric mean/llmv compilation. You know, I would like to buy an R9 sometime, and these numbers are not helping if I can't find it in stock.

    Leave a comment:


  • scottishduck
    replied
    Originally posted by devius View Post

    Which are also included in that article.
    The article clearly states intel has a 50% IPC lead over Zen 2 with their Celeron G4900?

    Leave a comment:


  • devius
    replied
    Originally posted by scottishduck View Post
    Single core tests are the only valid way to judge it.
    Which are also included in that article.

    Leave a comment:


  • scottishduck
    replied
    Originally posted by devius View Post
    Not sure in what world you think 8 core tests are a bar on IPC. Single core tests are the only valid way to judge it.

    Leave a comment:


  • devius
    replied
    Originally posted by scottishduck View Post
    ...AMD IPC is still slower when using a lithography process half the size of Intels
    Except it isn't: https://www.techspot.com/article/187...en-vs-core-i9/

    Leave a comment:


  • scottishduck
    replied
    Originally posted by Jumbotron View Post
    Correct on both counts as I did not make that claim for either of us. That said George Carlin was and still is unparalleled in skewering the bullshit that makes up most American life including "businessmen" and "marketing / advertising" each of which received PARTICULAR scorn from George. I am QUITE certain that if George had knowledge of and/or interest in the corporate and technical shenanigans Intel has perpetrated to unfairly and illegally (research lawsuits lost by Intel over this very subject) curtail competition ESPECIALLY against AMD, that he would call Intel's antics..."Grade A Prime American Bullshit". For OVER 15 years Intel has pushed a processor design that inflated benchmarks only to p0wned by hackers using that very same design. Most of 15 years of advancement by Intel particularly amongst their HUGE Corporate customers in their CPU products have been crapped away in an instant only to show their clients they WAY OVERSPENT on Intel vs AMD with hardly ANYTHING to show for it AFTER mitigations are put into place.

    THAT...is truly..."Grade A Prime American Bullshit"
    Not sure I would say the advancement has been "crapped away" when AMD IPC is still slower when using a lithography process half the size of Intels

    Leave a comment:

Working...
X