Announcement

Collapse
No announcement yet.

CompuLab Turns An 8-Core/16-Thread Xeon, 64GB RAM, NVIDIA Quadro RTX 4000 Into Fan-Less Computer

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

  • #31
    This looks like a good case for the ryzen 3000 series. They clock down to 0.2 volts(stall) at idle. And 7nm doesn't hurt.

    Comment


    • #32
      Originally posted by schmidtbag View Post
      Because we already know roughly how good that CPU and GPU performs. It was very clear that they weren't thermal throttling.
      Ah, how naive!

      The BIOS can throttle the CPU to stay within a predefined power envelope. Unless you know what the CPU clocks are doing, you cannot say that it's running as fast as this CPU was benchmarked to run, in other systems.

      I don't know as much about the GPU, but Nvidia certainly allows for similar configurability, as they use the same base GPUs in laptops.

      Don't you think it's at all suspicious that Michael didn't post any benchmark results?

      Comment


      • #33
        Originally posted by schmidtbag View Post
        Fair enough, though, it didn't peak that high for very long either. I don't think most people who get a PC like this are going to push the CPU under 100% load for extended periods of time.
        Why would you get an 8-core workstation CPU if you weren't planning on loading up the cores?

        Most buyers probably have in mind some rendering, simulation, or compilation workloads.

        Comment


        • #34
          This would be extremely cool with Zen 2 and Navi mid-tier GPU. Even a 95W Zen 2 would be cooler than a 95W Xeon, since AMD use a TDP value closer to the actual usage.

          Comment


          • #35
            Originally posted by pegasus View Post
            Numbers come from server vendors I'm talking to. Those fans can easily go over 30W per fan at full power and your typical pizzabox has 8-12 of them.
            Because they need a lot of airflow! Heatsinks add drag, as well. You're nuts, if you think convection can deliver that kind of airflow. All that would change by orienting the server vertically is at most a couple % improvement in fan efficiency.

            Originally posted by pegasus View Post
            Some major rethinking would be needed for a bottom to top airflow design and careful arrangement of components inside such rack would be needed (most heat sensitive at the bottom, least heat sensitive at the top), but I'm sure that on a rack level energy savings would be substantial.
            It doesn't work. There aren't enough heat-insensitive components you can afford to bake at the top of the rack. You design is not stackable, unless you add extra ducting, which directly impacts density.

            Comment


            • #36
              Originally posted by coder View Post
              The BIOS can throttle the CPU to stay within a predefined power envelope. Unless you know what the CPU clocks are doing, you cannot say that it's running as fast as this CPU was benchmarked to run, in other systems.
              *rolls eyes* You can say that for just about anything niche. Seems in this case, PTS identified the CPU as being 5GHz (obviously boost clocks) so worst-case scenario, you loose a few hundred MHz here and there.
              Wow, what a big loss.
              If you agree with that statement and didn't interpret it as sarcastic, then don't buy something passively cooled (let alone one this small). I'm sure most reasonable people would realize some sacrifices have to be made, which in the case for this PC, means getting its full potential while putting under 100% load for hours at a time.
              Don't you think it's at all suspicious that Michael didn't post any benchmark results?
              No, because I'm not a pessimist and Michael is usually pretty good about telling us when things aren't settling right. Although it would be nice and appreciated to see benchmarks (particularly something like framerate over a given time, to see when/if turbo speeds start to drop), you don't buy a PC like this for the sake of raw performance under full load.
              Originally posted by coder View Post
              Why would you get an 8-core workstation CPU if you weren't planning on loading up the cores?

              Most buyers probably have in mind some rendering, simulation, or compilation workloads.
              You could ask Apple the same thing about their high-end Mac Pros, and yet people are still buying those. A device like this could handling such workloads just fine, so long as they're not going for hours at a time. Again - if getting the maximum performance potential is important to you, perhaps you shouldn't be getting a small passively cooled PC.

              Comment


              • #37
                I also wish Compulab would equip an Airtop3 with something like Ryzen 7 3700x and an AMD GPU - I would definitely purchase one. I think Ryzen would be an excellent option for Airtop3.

                Comment


                • #38
                  Originally posted by schmidtbag View Post
                  *rolls eyes* You can say that for just about anything niche. Seems in this case, PTS identified the CPU as being 5GHz (obviously boost clocks) so worst-case scenario, you loose a few hundred MHz here and there.
                  Wow, what a big loss.
                  If you agree with that statement and didn't interpret it as sarcastic, then don't buy something passively cooled (let alone one this small). I'm sure most reasonable people would realize some sacrifices have to be made, which in the case for this PC, means getting its full potential while putting under 100% load for hours at a time.

                  No, because I'm not a pessimist and Michael is usually pretty good about telling us when things aren't settling right. Although it would be nice and appreciated to see benchmarks (particularly something like framerate over a given time, to see when/if turbo speeds start to drop), you don't buy a PC like this for the sake of raw performance under full load.
                  I don't understand your position. You're clearly not willing to sacrifice everything for the sake of passive cooling, yet you seem unconcerned with just how much you are actually losing. What if you lost 1 GHz of all-core clocks? Would that get your attention?

                  Without the benchmarks, we can't make an informed tradeoff of how much performance is being sacrificed (and extra $ charged) for sake of silence.

                  Originally posted by schmidtbag View Post
                  You could ask Apple the same thing about their high-end Mac Pros, and yet people are still buying those.
                  I don't see what Mac Pros have to do with this. Do they throttle or run at below-spec clocks? AFAIK, the only bad thing about them is their ridiculous price.

                  Originally posted by schmidtbag View Post
                  A device like this could handling such workloads just fine, so long as they're not going for hours at a time.
                  No, we don't have the data to support that.

                  Originally posted by schmidtbag View Post
                  if getting the maximum performance potential is important to you, perhaps you shouldn't be getting a small passively cooled PC.
                  Most people don't have one exclusive priority. They want to know how much performance they would be giving up. If you didn't care at all about performance, then why bother with an 8-core Xeon, in the first place?

                  Michael is definitely burying something. I think he favors certain vendors - especially smaller companies that give him stuff.

                  Comment


                  • #39
                    ​Without the benchmarks, we can't make an informed tradeoff of how much performance is being sacrificed (and extra $ charged) for sake of silence.​​​​​​
                    ​​​​​As far as I can tell, for most users the key benefit of fanless is not silence, but rather consistent, maintenance free reliability.

                    ​​​​​​
                    Michael is definitely burying something.​​​​​​
                    Speaking as a vendor, for getting a skewed review one should look elsewhere. Phoronix benchmarks and publishes the findings. It's a site where quantitive results are more important than the cover story.

                    Best regards,
                    Irad Stavi
                    Compulab

                    Comment


                    • #40
                      Originally posted by Compulab View Post
                      ​​​​​As far as I can tell, for most users the key benefit of fanless is not silence, but rather consistent, maintenance free reliability.

                      ​​​​​​

                      Speaking as a vendor, for getting a skewed review one should look elsewhere. Phoronix benchmarks and publishes the findings. It's a site where quantitive results are more important than the cover story.

                      Best regards,
                      Irad Stavi
                      Compulab
                      Thanks for adding your perspective.

                      I'm not saying your products are bad, in any way - I just want to see the complete data. I'm not averse to sacrificing some performance for silence - I'd just want to know how much.

                      Comment

                      Working...
                      X