Announcement

Collapse
No announcement yet.

AMD Ryzen 8500G / 8600G / 8700G Performance @ 35 Watt & 45 Watt cTDP

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

  • #31
    Originally posted by WiR3D View Post
    Am I the only one that looked at the graphics results and was like wtf how can the same chip running at lower power (45w) perform better (that the default 60w)? Smells like firmware/driver issues.
    It happens when I undervolt the CPU. The temperature and power limits allow higher frequency with undervolt if the CPU can stay stable

    Comment


    • #32
      I like the ASRock STX platform with AMD SOC chips. They only support SOC chips since those boards do not even have a south bridge. I have a Deskmini X300 system with 5700G. Maybe I can try the newly announced X600 with 8700G. They can even overclock, it is so much fun trying to overclock with limited cooling in such a small chassis. Almost impossible to hit all cores at max boost unless I am lucky to have a CPU with great undervolt potential.

      Comment


      • #33
        Originally posted by quaz0r View Post
        Am I wrong to feel existential dread whenever I see people abuse contractions and write sentences like "I've two motherboards" ? 💀
        You're 'one of those Pommie bastards' whom who gets on well with, well, everyone, eh 😁

        Yeah I need to work on my Ingrish. Health has me degenerating in to a right-ol ape.
        Hi

        Comment


        • #34
          Originally posted by sophisticles View Post

          Textures are always pre-rendered and layered on the rendered polygons, do if you have the option of using high quality textures, then yes more ram should be beneficial.
          That question was so poorly framed, I'm not surprised it wasn't answered, and you helpfully answering and making me realise how poorly phrased it was.

          I was taking a cue from an earlier post about limited bandwidth from the APU's onboard graphics processor being severely limited (dGPU comparison) by the RAM speed, with the follow on effect limiting the real world AA and texture resolution. Go higher, and performance drops irrespective of RAM ceiling.
          Hi

          Comment


          • #35
            Originally posted by sophisticles View Post

            The problem with many benchmarks from nearly all sites is that they only look at a small section of time, i.e. the benchmark runs for a few minutes and a result is obtained.

            This is not how people normally use their computers they edit video, record video and audio, watch videos and play games for hours at a time and while 35w may be stable for a 2 minute benchmark it may not be stable for a 5 hour gaming session.
            In fact short (in terms of minutes) benchmarks tends to favour Intel chips due to their often insane Power Limit levels. This is even more evident on laptops, where power constraints are more stringent. Those chips can't sustain high loads for longer times because they would draw too much power, that's why on Phoronix benchmarks usually there are the cumulative energy usage accounts.

            Comment


            • #36
              Originally posted by sophisticles View Post

              Because the power consumption may have more to do with stability than performance.

              The problem with many benchmarks from nearly all sites is that they only look at a small section of time, i.e. the benchmark runs for a few minutes and a result is obtained.

              This is not how people normally use their computers they edit video, record video and audio, watch videos and play games for hours at a time and while 35w may be stable for a 2 minute benchmark it may not be stable for a 5 hour gaming session.
              No, it's just a power limit. These parts are all running 100% on the factory specified voltage-frequency curve. No undervolting, no loss of stability. The only things putting the reliability of these machines at risk are the overclocked memory (assuming Michael enabled it), and whatever motherboard vendor idiocy is present (usually some).

              Originally posted by WiR3D View Post
              Am I the only one that looked at the graphics results and was like wtf how can the same chip running at lower power (45w) perform better (that the default 60w)? Smells like firmware/driver issues.
              I see no tests like that that are outside margin of error. Only the 8500G has any questionable sorts, and in the graphics tests it doesn't seem to be running against the power limits at all, because it's GPU is so small.

              Originally posted by mrg666 View Post
              It happens when I undervolt the CPU. The temperature and power limits allow higher frequency with undervolt if the CPU can stay stable

              Like I said, no undervolting here.

              Comment

              Working...
              X