5 Replies Latest reply on Sep 26, 2016 6:27 PM by brucer

    FX 8350 great results: Arctic liquid freezer 120 + Liquid Pro

    volumetricsteve

      I thought I'd throw it out there that I recently had great results combining the Arctic liquid freezer 120 and CoolLabratory's Liquid Pro.

      Under air cooling, I was (supposedly) getting over 67C, though I understand there's a lot of confusion reading temperatures from FX series chips.

       

      In my case, I'm using 64-bit linux and the 'lmsensors' package.

       

      after re-lapping my cpu, applying a healthy amount of liquid pro to both the cpu and the liquid freezer 120, my highest temperature claims 38C under continuous load for ten minutes.

       

      In the wider scope of things, I'm not sure how good these numbers really are compared to the results others have based on the confusion with reading the real value in the first place...but, 38C seems pretty good to me.  That would mean I'm maxing out around 100F, and some people seem to think the temp reading is artificially increased by about 10C, so I may actually be maxing out around 82F, which would be pretty amazing.

       

      I should also emphasize the vital importance of ending up with a flat cpu surface that isn't just flat and planar with itself, but the flatness needs to be fully perpendicular to the socket so the pressure being applied by the heatsink is straight up and down, making evenly pressured contact.  When I was running air-cooled, I didn't realize that I'd lapped my cpu down to a mirror shine, but it was really badly slanted.  Once I corrected the slant, cooling performance took way off! 

       

      I also replaced Arctic's default fans with a pair of Noctua's industrial 120mm fans, which I believe push a good deal more air.  They're louder, sure, but they also do a great job.

      If anyone else is using lmsensors to get readings from FX series chips, i'd be curious to see what results you're getting.

        • Re: FX 8350 great results: Arctic liquid freezer 120 + Liquid Pro
          black_zion

          Temperatures below 40°C aren't reported by the sensors correctly. The CPU socket temperature will read higher than the core temperature, is an estimate, and shouldn't be used (in the 10 minute stress below, my CPU temp hit 5°C higher than the core temp). Also have to wonder why you went so overkill on the TIM yet used a lower mainstream 120mm radiator unit. Using a Corsair H105, two COUGAR CF-V12HP Vortex Hydro-Dynamic-Bearing (Fluid) 300,000 Hours 12CM Silent Cooling Fan with Pulse Width Modulation-N… spinning at 1100-1200 range (both tied to Chasis #2 header), and some IC Diamond TIM, with a 27°C/80°F ambient temperature while drawing even warmer case air through the radiator, under extended stress I top out at 52°C while remaining whisper quiet.

           

          stabilitytest.png

            • Re: FX 8350 great results: Arctic liquid freezer 120 + Liquid Pro
              volumetricsteve

              "Temperatures below 40°C aren't reported by the sensors correctly." - I've read this as well.

              "The CPU socket temperature will read higher than the core temperature, is an estimate, and shouldn't be used (in the 10 minute stress below, my CPU temp hit 5°C higher than the core temp)."  - This is where it gets harder to tell what's going on, especially with lmsensors.  I haven't heard this before, but you're saying there's a separate socket (like the physical socket, right?) thermal register, and then there's a different thermal register in the cpu...built into the die?  I had always thought there was just the one in the chip after about the pentium4 era.  If the temp I'm reading is the socket temperature, it could be my actual core temp is even lower?  That sounds like good news.

               

              "Also have to wonder why you went so overkill on the TIM yet used a lower mainstream 120mm radiator unit. Using a Corsair H105, two COUGAR CF-V12HP Vortex Hydro-Dynamic-Bearing (Fluid) 300,000 Hours 12CM Silent Cooling Fan with Pulse Width Modulation-N… spinning at 1100-1200 range (both tied to Chasis #2 header), and some IC Diamond TIM, with a 27°C/80°F ambient temperature while drawing even warmer case air through the radiator, under extended stress I top out at 52°C while remaining whisper quiet."  - The computer I built couldn't possibly have fit a radiator that big, though that would be nice.  My system sits in a fairly well air conditioned room, I pretty much treat it like it's part of a data-center, it's just the case it's in is physically small-ish.  I horrifically mangled a Cooler Master Elite 360 to fit all my stuff in so it became a balancing act of finding what parts would get it to run well, while also being relatively small and extremely dense.  I should probably post pictures around here somewhere.

              In the end, it sounds like if your core temp (I gather...this is the one that matter slightly more) is about 27 and I'm maxing at 38, I'm not breaking any records but it seems like I've done passably well for the size of case I'm working with...Though again, I'm not sure if the reading from lmsensors is misleading me or not, all I can tell for sure is once I finished my upgrade the readings it's giving me are just under half of what they were.

            • Re: FX 8350 great results: Arctic liquid freezer 120 + Liquid Pro
              brucer

              I5-6600k @4.5ghz running Cryorig H5universal all air cooled.

               

              stabilitytest.png

                • Re: FX 8350 great results: Arctic liquid freezer 120 + Liquid Pro
                  volumetricsteve

                  That's a good sized heatsink you have there.  Looks like my FX-8350 consumes a fair deal more power than a i5-6600K, so between that difference that the fact that lmsensors is almost certainly giving me a higher temp measurement than I'm actually at, it seems like I'm running fairly cool considering our few degrees difference.  Looking over your results again, it looks like you go as high as 51.8C on one of your cores.  I need to do more testing on my rig, but my first stress test showed that as high as anything in my cpu got was 38C.

                   

                  What I really need is one of those FLIR heat visualizer things so I can get a solid 3rd perspective on how much heat is actually building up.