Announcement

Collapse
No announcement yet.

AMD FX-8350 "Vishera" Linux Benchmarks

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

  • It's not like Intel has 8 FPUs and 4 integer cores...

    Comment


    • Ivy Bridge was shown not to measurably scale with RAM speeds on Windows. So not surprising you didn't see it on Linux either.

      Comment


      • Originally posted by Kano View Post
        ddr3-2133 is not really cheap, the money you spend for that you could instead invest in a nv gfx card without onboard vga faster ram over ddr3-1333/1600 is hardly noticeable. OEM boards often run with ddr3-1333 setting all the time (you can not select the ram speed with those boards). Lately i got 4gb ddr3-2133, it was 1-2 fps faster with TF2/mesa 9.1 and intel hd 4000. I would not say that when it jumps from 51 fps to 53 fps it was a needed upgrade (usually i use 8gb ddr3-1600 with that board) - well I switched back to the slower but more ram after some benchmarks.
        In my country the difference between 8GB 1333 and 8GB 2133 is pretty ridiculous: $5.

        Intel chips have bad performance gains, but AMD FX chips do it better. If you look to the figure in my previous post you can see that going from 1333 to 2133 can generate up to 10 extra frames (10% more performance) on a bulldozer chip. And some people claim greater improvements in games such as civ 5.

        2133 RAM seems a must for new builds.

        Most windows benchmarks report the brand and timings of the memory used. Why does phoronix not provide that basic info?
        Last edited by juanrga; 04-19-2013, 11:27 PM.

        Comment


        • Originally posted by juanrga View Post
          In my country the difference between 8GB 1333 and 8GB 2133 is pretty ridiculous: $5.
          It's about the same where I live. It's possible to find really cheap no-name 1333 memory, but quality 1333 is almost the same price as 2133.

          Comment


          • Originally posted by juanrga View Post
            Moreover, how many users will be overclocking the FX chip @ 4.8 but underclocking the ram? Memory @ 2133 seems a more adequate timing and then differences would be a bit larger.
            For overclockers these benchmarks are usually useless anyway. And those who really want performance go with a i7-3930K at 4.5-5 GHz or a i5-3570K.

            If the reviewer forgot to check memory timings, this should be an advantage for AMD, since they will be more often be running DDR34-1600 at full speed than Intel without configuration.

            Originally posted by juanrga View Post
            As you say, the APUs are much more sensitive to memory bandwidth. I read reports where the gain in performance is of up to 20% by using faster memory under windows. I know AMD chips usually run faster under linux. That is why I asked about what memory is being used in phoronix tests. This is important info (at least for me) which is lacking.
            Compiler optimizations for AMD makes AMD CPUs faster on Linux than Windows, that's true. But how do you think this affects the performance of Catalyst? This is not related to the performance gain for higher memory bandwidth for the graphics parts of the APU.

            As regarding the graphics performance of the A10 APUs, even though you will get more performance with higher memory speeds and AMD outperforms Intel's integrated GPUs with a factor of 2, they still are way too weak to run newer commercial games at FullHD at approx 60 FPS (windows games), so people would be better off with a cool Intel dual core and putting some money in a cheap dedicated graphics card. For Windows anything from 7750-7790 and up would be great.

            Unfortunately benchmarks of SandyBridge-E on Windows does not utilize all four memory channels. I don't think Windows is aware, so it's practically impossible to benchmark this since physical addresses are hidden for programs.

            But I do agree on one thing, test should provide all the technical details for those interested.

            Originally posted by juanrga View Post
            So far as I know AMD mobos support XMP via emulation. Enthusiasts users say me that the best results are obtained with AMD optimized RAM. I do not know more about this issue, and this is why asked to test some AMP profile.
            Enthusiasts may say a lot of suff, enthusiasts also claim vinyl and tubes yield higher audio quality even when proven wrong.
            I suggest you read the XMP specs from Intel, and you'll see it's just some extra information to help choose memory clocks and timings. Once you set these settings, the performance will be equal regardless of XMP or AMP.

            Originally posted by juanrga View Post
            In my country the difference between 8GB 1333 and 8GB 2133 is pretty ridiculous: $5.
            2133 RAM seems a must for new builds.
            Even my new workstation with i7-3930 and 64GB DDR3-1600 does not use faster memory, because tests show non-significant performance improvements. You might see higher scores in memory benchmarks, but in real-world performance the gain is negligible.

            Comment


            • Originally posted by efikkan View Post
              Compiler optimizations for AMD makes AMD CPUs faster on Linux than Windows, that's true. But how do you think this affects the performance of Catalyst? This is not related to the performance gain for higher memory bandwidth for the graphics parts of the APU.
              I don't know what do you mean really neither which is the relation with what I said.

              Originally posted by efikkan View Post
              Enthusiasts may say a lot of suff, enthusiasts also claim vinyl and tubes yield higher audio quality even when proven wrong.
              I suggest you read the XMP specs from Intel, and you'll see it's just some extra information to help choose memory clocks and timings. Once you set these settings, the performance will be equal regardless of XMP or AMP.
              I know lots of AMD users having problems with XMP memory kits. This may be related to XMP being optimized for intel chipset motherboards. I also read this

              http://sites.amd.com/us/game/downloa...Pages/amp.aspx

              Originally posted by efikkan View Post
              Even my new workstation with i7-3930 and 64GB DDR3-1600 does not use faster memory, because tests show non-significant performance improvements. You might see higher scores in memory benchmarks, but in real-world performance the gain is negligible.
              As stated above Intel chips performance is rather insensible to memory speed. This is not the case with AMD chips which can see 8--20% benefits in real-world performance; I did not even mention memory synthetic benchmarks, therefore I fail to see your point again.
              Last edited by juanrga; 04-23-2013, 08:22 AM.

              Comment


              • Originally posted by juanrga View Post
                In my country the difference between 8GB 1333 and 8GB 2133 is pretty ridiculous: $5.

                Intel chips have bad performance gains, but AMD FX chips do it better. If you look to the figure in my previous post you can see that going from 1333 to 2133 can generate up to 10 extra frames (10% more performance) on a bulldozer chip. And some people claim greater improvements in games such as civ 5.

                2133 RAM seems a must for new builds.

                Most windows benchmarks report the brand and timings of the memory used.
                If you take AMD chip, then its a money waste to get any non-ecc memory. Even for gaming, whats more important - a 10% more fps or no crashes? Mind you RAM memory is the ONLY random access memory that is NOT ecc checked and is proven to bit-rot 8% chance in a month.


                Originally posted by juanrga View Post
                Why does phoronix not provide that basic info?
                Because civ5 is not available as native Linux/Unix game.

                Comment


                • Originally posted by juanrga View Post
                  I know lots of AMD users having problems with XMP memory kits. This may be related to XMP being optimized for intel chipset motherboards. I also read this
                  Your friends might have actual problems, but none of these are related to XMP. XMP is not an optimization (even if advertised as so), it's just a set of "recommended" settings and timings. If the motherboard does not support it, it will not interpret this extra information stored in the flash on the memory chip, XMP does not affect memory operation.

                  An analogy would be EDID on Screens, without it you would need to manually adjust the configuration, but it does not affect picture quality.

                  Comment


                  • Originally posted by efikkan View Post
                    Your friends might have actual problems, but none of these are related to XMP. XMP is not an optimization (even if advertised as so), it's just a set of "recommended" settings and timings. If the motherboard does not support it, it will not interpret this extra information stored in the flash on the memory chip, XMP does not affect memory operation.

                    An analogy would be EDID on Screens, without it you would need to manually adjust the configuration, but it does not affect picture quality.
                    I physically owned a dual-bank set of 1600 ddr3 ram that refused to start on AMD system, that I had to softreset on every cold start, otherwise it will fall through memtest86. But after several softresets, the chips behaved perfectly.
                    On Intel system, they worked perfectly out of the box. So it could be that some RAM brands are incompatible, better always to keep invoice handy for RMA of memory if that happens.

                    The same AMD system is working right now perfectly with GSkill 8G dual bank set. No softresets, nothing, perfect. So neither AMD hardware nor particular memory were faulty, there maybe problems in configuration or timings in particular MB/memory cases.
                    Last edited by brosis; 04-23-2013, 06:28 PM.

                    Comment


                    • Originally posted by brosis View Post
                      I physically owned a dual-bank set of 1600 ddr3 ram that refused to start on AMD system, that I had to softreset on every cold start, otherwise it will fall through memtest86. But after several softresets, the chips behaved perfectly.
                      On Intel system, they worked perfectly out of the box. So it could be that some RAM brands are incompatible, better always to keep invoice handy for RMA of memory if that happens.
                      It's not uncommon with obscure bugs in BIOSes, some memory modules have problems on some motherboards or chipsets. But lacking of XMP has nothing to do with this, as it has nothing to do with memory operation. XMP is just an aid to set recommended settings, nothing more.

                      Comment


                      • Originally posted by efikkan View Post
                        It's not uncommon with obscure bugs in BIOSes, some memory modules have problems on some motherboards or chipsets. But lacking of XMP has nothing to do with this, as it has nothing to do with memory operation. XMP is just an aid to set recommended settings, nothing more.
                        Yes, thats true.

                        Comment


                        • Originally posted by brosis View Post
                          Because civ5 is not available as native Linux/Unix game.
                          What have to do this with phoronix not giving timings and brand of the memory used in their tests? I want to know if they used 1300, 1600, 1866...

                          Comment

                          Working...
                          X