Meltdown & Spectre potpourri

Discussion in 'General Chat' started by DrinkLyeAndDie, Jan 4, 2018.

  1. DrinkLyeAndDie

    DrinkLyeAndDie Retired Moderator

    I happened to both be smart and make some mistakes when I decided to build my system. Timing-wise luck had it that I bought my hardware right before Kaby Lake, Ryzen & then Coffee Lake dropped. It was still far enough off that I didn't get my CPU at a steal, however. That kinda sucks but such is life. On the plus side, I was paying attention to talk about supply, demand, and pricing. Word was RAM prices would be going up so I purchased my needed hardware sooner and didn't actually build the system for about 5 months. I believe I bought everything in December of 2016. I paid $229 for 32 GB of G.Skill TridentZ DDR4 3200 [F4-3200C16Q-32GTZB]. Currently it's selling for $418 on Amazon and $381 on Newegg.

    I'm impressed with Ryzen but, personally, I'd probably wait until the 2nd generation which I believe is supposed to be released in Q2 of 2018 if I were to go that route.
     
  2. RedFox 1

    RedFox 1 Super Moderator

    Wait for the 10nm Ice Lake from Intel.
     
    DrinkLyeAndDie likes this.
  3. DrinkLyeAndDie

    DrinkLyeAndDie Retired Moderator

    If I were to go the Intel route I would definitely wait for Ice Lake. It won't be the first 10 nm release so Intel can work out any kinks in Cannon Lake after its release. Cannon Lake & Ice Lake will supposedly contain in-silicon fixes for Meltdown & Spectre which sounds great but the details are lacking. I'd like to know how and do those changes result in the loss of "potential performance" were they not to exist. Obviously, if performance goes up over existing processors that is great but will the increase be reduced versus what it could have been?
     
  4. gereral1

    gereral1 Well-Known Member

    Anyone seen iops benchmarks for ryzen online with these patches added?. Intel takes a big hit on these updates.

    Second question: do you Intel users feel the performance hit with editing video or games or are these things so fast it's only benchmark data which really means squat?
     
  5. DrinkLyeAndDie

    DrinkLyeAndDie Retired Moderator

    AMD released updates and I have seen people claim "zero performance loss" after said BIOS updates but I haven't seen any actual empirical data. I'm sure it's out there somewhere but I have to see it and trying to do a search results in a lot of unhelpful results.
     
  6. gereral1

    gereral1 Well-Known Member

    My gut says amd does not do prediction caculations like Intel. If I'm right then I'll be going as ryzen 7 route as I'm not a heavy gamer. I play cs source that's it lol.
     
  7. gereral1

    gereral1 Well-Known Member

    I've found squat for hdd 4k transfers after patches for amd. Like to see iops as this new pc will be for 4k capture and editing and hdr htpc and YouTube videos. HHD fast writes is important for 4k and that Intel drop could mean extra drive in raid 0 to make up for Intel stupidity
     
  8. DrinkLyeAndDie

    DrinkLyeAndDie Retired Moderator

    The loss of IOPS will definitely impact game load times. I'm thinking games with large textures, etc. Not sure how that would directly connect to fps. No idea about capture and editing. AMD definitely designed their processors with differences in speculative prediction and compartmentalization of the process hence why they are much less susceptible to Spectre. Don't quote me on the technical wording of what I said, I'm saying that generally as a layman.

    That said, Intel has been far more vocal and open about what they are doing or contemplating with respect to current CPUs on the market. I really haven't seen AMD explain anything other than going from "there is zero chance of Spectre v2 impacting our processors" to "Spectre v2 can impact our processors but it's near zero and would be near impossible to accomplish." That's really general paraphrasing, mind you. Nonetheless, they've not exactly been forthcoming which I do understand but while they are the winner in all of this they aren't helping themselves playing word games and making missteps.
     
    Last edited: Mar 11, 2018
  9. DrinkLyeAndDie

    DrinkLyeAndDie Retired Moderator

    Last edited: Mar 13, 2018
  10. gereral1

    gereral1 Well-Known Member

    My goodness......this is crazy the hard drive iops. Question is this 30 percent loss in writes and read can you guys feel it in every day loading of apps and copying? Can't believe Intel Microsoft and usus is doing this. I'm not building pc tell they fix this crap
     
  11. RedFox 1

    RedFox 1 Super Moderator

    I do not notice it at all in everyday use. I have a 1TB 950 Pro NVMe M.2 and it works fine.
     
    DrinkLyeAndDie and gereral1 like this.
  12. DrinkLyeAndDie

    DrinkLyeAndDie Retired Moderator

    The issue really isn't ASUS. It's the microcode update. And, as I stated to note Mr.Wolf's observations, he is saying that Samsung Magician only uses a single core vs other benchmark software that uses multiple cores. Where multiple cores are used the huge hit isn't seen. This brings in to question the value of the results from Samsung Magician with respect to real-world impact.

    Now, the person who began the thread noted two things:

    1. IOPS dropped when BIOS 3504 was flashed.
    2. IOPS dropped further when BIOS 3703 was flashed.

    What was present in both of these BIOSes? Microcode updates for Intel processors.


    Btw, I updated to 3504 when it was released on my motherboard after having used a prior BIOS. I noticed zero difference in performance in real-world use. I didn't do any benchmarks because I had no reason to. I re-enabled KB4090007 a couple days ago. I haven't noticed any observable difference in performance thus far.
     
  13. DrinkLyeAndDie

    DrinkLyeAndDie Retired Moderator

    Last edited: Mar 13, 2018
  14. RedFox 1

    RedFox 1 Super Moderator

    I dont even use the Samsung Magician. The whole thing is a mess. I am happy the way it works without the Magician software
     
  15. Ch3vr0n

    Ch3vr0n Translator NL & Mod

    Doubtful mate. An article on a local tech site and even by AMD themselves says AMD never heard of the company before they published their 'research'. Contradicting to standard bug policies they supposedly were only given 24H! to address the vulnerabilities/reports. That's even on the CNET site.

    Official AMD response: http://ir.amd.com/news-releases/news-release-details/view-our-corner-street-0

    Other sources believe it's an attempt to manipulate AMD stock.

    Sent from my Nexus 6P with Tapatalk
     
    eviltester likes this.
  16. DrinkLyeAndDie

    DrinkLyeAndDie Retired Moderator

    Indeed, that is possible. Or something akin to the idiotic Skyfall hoax that was an "experiment".
     
  17. Ch3vr0n

    Ch3vr0n Translator NL & Mod

    I'm no AMD fan/user and never have been, but even I'm not buying that crap. I'll be surprised if actual evidence of those claims turns up and even more if it's true.

    Sent from my Nexus 7 with Tapatalk
     
  18. DrinkLyeAndDie

    DrinkLyeAndDie Retired Moderator

    At this point I agree.
     
  19. DrinkLyeAndDie

    DrinkLyeAndDie Retired Moderator

  20. RedFox 1

    RedFox 1 Super Moderator