Announcement

Collapse
No announcement yet.

Linux 5.5-rc7 Kernel Released

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

  • #11
    Originally posted by ktecho View Post

    Yeah, it's a pitty when you read all that criticism about Intel in this forums. It seems like Intel does everything wrong and if you buy AMD your computer starts to fly. Then, in the real life people needs to wait for 1 or 2 years after buying an AMD system until it's stable enough to use all that amount of cores for everyday tasks.

    15 years back when I was at the university I had an AMD card and they promised to release a new driver each two months. "Now we're getting serious about Linux". "Oh wow!" I thought. "Now I'm going to be able to play without problems and unstable patches by the community!" But then you waited for two months and the changelog was reading "Update to make it work with kernel 3.1.4".

    Don't get me wrong: I want AMD to succeed, as they're contributing a lot to open source and Linux. But they're not betting enough money as to support new products. How comes you have to wait one year+ to have a driver in the kernel to read CPU temperatures when it's just a bunch of lines of code? At least with Intel you can be confident that when you buy it it will work and drivers for CPU and GPU will be stable from day 1. Yes, for sure they have mitigations that make it no so powerful than advertised, but hey, I can plug it and it works. And that's what business value, not "OMFG!1!! it has 700 cores!!!" but then you have to wait until you can use it.
    AMD wasn't making GPU's 15 years ago

    Comment


    • #12
      Originally posted by anarki2 View Post

      I meant 3900X, whatever.

      It works perfectly fine on 5.4, it fails consistently on 5.3. How on Earth did you come to the conclusion that it's a hardware issue then? Elementary logic.
      and what is the problem ? 5.4 is current stable since month. my xeon 2286 also runs better on 5.4. with 5.3 it is more likely to get thermal throttled. btw b450 pro auros might also be working on the edge of its power delivery abilities.

      Comment


      • #13
        Originally posted by ktecho View Post

        Yeah, we don't know all the Intel vulnerabilities, but we know all the AMD ones, right?
        Would you buy a car where the brakes were known to fail...or one that had been through the same tests and the brakes were proven not to fail?

        Comment


        • #14
          Originally posted by anarki2 View Post
          (I can't edit my prev comment) I also read A LOT of random complaint threads about these. The workarounds are even more staggering, like disable all CPU scaling in UEFI. Or disable PSU whatever states, I can't even remember all this random [email protected] Like we live in the year 2020 but CPU scaling still is a challenge?

          Then again, that's exactly the reason why I already gave up on AMD once many many years ago.
          Because you can't go in BIOS and find an option to disable it?

          The "PSU whatever states" actually Power Supply Idle Control is the actual firmware-level workaround for hardware bugs in first and second gen Ryzens, and it's only partially disabling C6 state, which is completely inconsequential for a desktop system.

          Comment


          • #15
            Originally posted by ktecho View Post
            In the end you need stuff to work. I read each benchmark that Michael does when a new Ryzen or Threadripper goes to market because I like how powerful they are. But last year I needed to build a PC and I went for a 8700k as I needed to start working on it. Didn't have the time to skim through forums for RAM sticks that work with a specific mobo.
            Sorry what? As long as you aren't overclocking RAM (which is mostly pointless for a CPU nowadays) you can install whatever branded RAM you want.

            All the complaints are from overclockers, and they are free to whine if something can't OC as high as they wanted to, but it's never guaranteed so whatever.

            You sure you know how to troll properly? Because it seems you don't.

            Comment


            • #16
              Originally posted by starshipeleven View Post
              Sorry what? As long as you aren't overclocking RAM (which is mostly pointless for a CPU nowadays)
              Isn't RAM overclocking the only useful overclocking anymore on Ryzen?

              Comment


              • #17
                Originally posted by geearf View Post
                Isn't RAM overclocking the only useful overclocking anymore on Ryzen?
                Useful overclocking died long ago. Overclocking now is a sport.

                Comment


                • #18
                  Originally posted by geearf View Post
                  Isn't RAM overclocking the only useful overclocking anymore on Ryzen?
                  Yes, also timing tuning. I could run some tests with a Ryzen 3600X and Shadow of the Tomb Raider saw +30% more fps in total CPU bound scenario with 3600 tuned RAM vs. 3200 XMP CL14. Overclocking the CPU just runs into voltage and clock walls.

                  Comment


                  • #19
                    Originally posted by aufkrawall View Post
                    Yes, also timing tuning. I could run some tests with a Ryzen 3600X and Shadow of the Tomb Raider saw +30% more fps in total CPU bound scenario with 3600 tuned RAM vs. 3200 XMP CL14. Overclocking the CPU just runs into voltage and clock walls.
                    That's a seriously unrealistic scenario. Seriously man, I would only be expecting around 12% better throughput (DDR4-3600 vs 3200) in a synthetic memory bandwidth benchmark. The impact on CPU bound tasks might be 1-2%, while games, otoh, are often not CPU bound at all.

                    Comment


                    • #20
                      Originally posted by caligula View Post
                      That's a seriously unrealistic scenario. Seriously man, I would only be expecting around 12% better throughput (DDR4-3600 vs 3200) in a synthetic memory bandwidth benchmark. The impact on CPU bound tasks might be 1-2%, while games, otoh, are often not CPU bound at all.
                      Apparently you don't understand how Ryzen cores are connected in non-monolithic designs and what the implications of this are for games with heavy multithreading + heavy thread dependencies.
                      That the SotTR game reacts by a lot to this is also proven by others (use search machine), so perhaps keep your uneducated opinion to urself or ask kindly so that someone might explain it to you.

                      Comment

                      Working...
                      X