I could use a sanity check on a cooling situation. It doesn’t seem right or possible. Sorry for the long post.

CPU: AMD 9700X MB: Asus TUF Gaming B650E Wifi RAM: 4x16GB T-Force CL30 Case: Antec P182 (Giant quiet gaming case from 2007)

The system has been together for about 5 months. When I got the components, I picked up a $30 CAD generic cooler:

“iCan 6-Heatpipe CPU Cooler | 2*120mm PWM Fans with Rainbow Color | TDP - 265W | Fan Speed upto 1800RPM | Fan Noise - 31DBA”

It all ran OK, was quiet but not silent. Only has fan noise when working hard. In other words: Great.

I’m doing a bunch of ffmpeg/x265 transcode benchmarking. With the iCan cooler, it would thermal throttle (75C throttle, 76C measured) at 5350MHz. It would bounce between about 5200 and 5450.

Anyway, a buddy told me the Noctua D15 is the best air cooler going. I did some reading. He seems to be right. It is a highly regarded cooler.

The D15 fits into the system fine but only one of the 140MM fans will fit. The 140mm fan won’t fit in the back, due to the VRMs. The 140mm fan won’t fit in the front, due to RAM. So, the D15 has a 140mm stock fan and a 120mm screamer (140CFM). That’s a lot of air.

It doesn’t cool nearly as well. Not even close. I’ve bumped up the thermal throttle to 80C (where it sits 100% of the time while transcoding). Fans are set to 100%, all the time. It barely makes 5GHz with the D15. Average is about 4900MHz.

I put the iCan cooler back on and it went back to almost boost clock frequency.

The thermal paste is very well done. I was super careful between cooler changes and I clean the old stuff with alcohol before applying fresh and mounting another cooler.

Why does the smaller cooler with a far worse machined CPU contact area cool better than the gigantic D15?

I will say, when using QFan and the iCan cooler, the fans ramp in about 2 seconds after launching ffmpeg. QFan with the D15, it takes about 4~5 seconds for the fans to fully ramp. The D15 has a substantially heavier CPU block and this does seem to add considerable inertia (which is nice).

Any thoughts?

  • TomB19@lemmy.worldOP
    link
    fedilink
    arrow-up
    3
    ·
    13 hours ago

    Major update:

    I forgot that I reset the BIOS, not long before the cooler change. Previously, the system hat a 0.35v undervolt. That had been removed in the BIOS reset.

    With the 0.35v undervolt restored, the D15 performance is essentially identical to the iCan dual tower cooler.

  • TomB19@lemmy.worldOP
    link
    fedilink
    arrow-up
    2
    ·
    14 hours ago

    Minor update:

    I pulled the D15 and discovered it was not interfacing with the CPU in the center in the block. The mounting brackets have two sets of holes so I shifted it to put the heat closer to the center of the cooling block.

    It didn’t make much difference. I won’t know until I’ve run my entire test suite. I hate to jump to any conclusions after 15m of testing. So far, it’s hitting 5.14 GHz, where it was not quite hitting 5GHz before.

    I’ll let it run some benchmarks overnight and then analyze the speeds and temps.

    Thank you for the ideas and advice. I really appreciate it.

  • CameronDev@programming.dev
    link
    fedilink
    arrow-up
    7
    ·
    edit-2
    19 hours ago

    The d15 is normally a fantastic cooler, so something isnt right. Is it brand new? Long shot, but are there any cracks on the heatpipes? If the liquid has leaked out it won’t work well?

    I have one with only a single fan in the middle, and it worked fine, so unless you put the two fans pointing air at themselves, its not the fans.

    I’d return it for a replacement, it sounds like you’ve tried all the reasonable steps.

    • TomB19@lemmy.worldOP
      link
      fedilink
      arrow-up
      1
      ·
      14 hours ago

      This is such a great idea. When I installed the D15, I looked at the pad on the interface block for a while. It is so shiny, I thought it was plastic. After a minute of trying to peel it off with my nail, I figured it was aluminum that is really, really beautifully finished. It must be machine ground with a diamond hone and maybe even lapped or polished.

      When I read your comment, I figured it must be plastic. I went at it with a razor knife but it simply isn’t plastic. If I had left the plastic protector on, it wouldn’t be the dumbest thing I’ve ever done. Not by a long shot.

      Great idea. Just not the problem in this situation.

  • Steve
    link
    fedilink
    English
    arrow-up
    4
    arrow-down
    1
    ·
    edit-2
    14 hours ago

    You can offset the 140mm in the front a little higher, so it clears the memory. The 150mm in the middle.
    Shouldn’t make a big difference though.

    Outside that? Mounting pressure of the cooler?
    The D15 is a heavy beast, pulls on the board. Double check it’s properly snugged down?

    That’s about all I can think of.

    • TomB19@lemmy.worldOP
      link
      fedilink
      arrow-up
      3
      ·
      14 hours ago

      I appreciate the thought. Unfortunately, my case won’t allow for raising of the front fan.

      If I decide to stay with the D15, I’ll pick up a new case that has room for this massive cooling tower and the windmills that operate it. lol!

  • zurohki@aussie.zone
    link
    fedilink
    English
    arrow-up
    1
    ·
    17 hours ago

    Is the D15 putting out hot air?

    If the air coming out is really hot, it’s a fan/airflow issue.

    If it’s just warmish, it’s a problem transferring heat from the CPU into the cooler’s fins. Could be the contact or bad heat pipes.