Overclock.net banner
16,921 - 16,940 of 20,233 Posts
Which heatsinks would be best for air?

Bykski

Bitspower

Iceman
Just for those interested in pad thickness setup for Bitspower DDR5 kit.

I had to tear down my heatsink twice because it was not sitting centre with the heatsink.

I figured out the optimal thickness so that under clamping pressure, the sticks do not slant.

The most important pad for this setup is the empty side of the stick. There is a PMIC trace layout and you need to put a 1.5mm thick pad around 2cmx1.3cm (Be careful it does not overlap with the lower part of the heatsink which sits higher or you will have problems clamping). If you do not have back support on the PMIC with enough thickness, you will end up slanting the modules under pressure so this PMIC pad levels the entire stick once you start to tighten it down.

Just by adding the PMIC thermal pad on the rear with 1.5mm thickness allowed me to drop 3 degrees more on SPD Hub temps. I believe there is better pressure on the PMIC on the front due to this back support. The entire setup now is solid as a rock and does not wiggle or move despite repeated installing and removing of the rams.

Summary of what I used :

RAM Kit - G.Skill 7800C36 Kit

Front RAM chips - 1.0mm Gelid Extreme Thermal Pad (4.5cm x 1.0cm)
Front PMIC chip - 1.5mm Gelid Extreme Thermal Pad (2cm x 1.5cm)

Rear Ram Chips (Empty) - 0.5mm Gelid Extreme Thermal Pad (4.5cm x 1.0cm)
Rear PMIC traceout - 1.5mm Gelid Extreme Thermal Pad (2cm x 1.3cm)

Also, I pasted some thermal paste just above the green line as outlined below for better thermal transfer. Hope this helps you guys out if you are using the same kit.
View attachment 2589545
View attachment 2589548
View attachment 2589549
View attachment 2589547
I'm using the Bitspower. I've put out this guide which has helped many get the temps optimally. And they are very straight forward to work with if you use the correct thermal pad thickness.

Hope this helps!
 
  • Rep+
Reactions: Papusan and morph.
That 63.5°C reading is a sensor bug in HWiNFO64 that many have reported, but even 50-55°C is more likely to result in errors. If your system is stable in normal use (benching, gaming and anything else) the memory stress tests can error out if the RAM starts getting warmer than 45°C.
So how likely will the memory be in the 45-55C range with a fan with RefI at 131K with regular net surfing or a short gaming session? Seems unlikely to get anywhere close to a temp seen with a stress test.
 
I know it’s messy. Test bench:

View attachment 2656747
Look nice. Test bench is definitely the ideal way to go. I am thinking about pulling my Praxis WetBench out of closet for the Encore and selling the Dark Base Pro 901. I love the case, but no case is as good as an open bench unless aesthetics are more important than function. I also feel the horizontal motherboard orientation is superior to vertical. It is much easier to lay the O11 XL on its back for maintenance than it is the Dark Base Pro 901. The Dark Base is gorgeous, but it is too big and bulky. It would be ideal if I used in-chassis water cooling but all of my pumps and radiators are external, so it is just a lot of bulk that I don't need.
Electricity Art Electric blue Computer hardware Space

Computer hardware Gas Technology Machine Display device
 
Look nice. Test bench is definitely the ideal way to go. I am thinking about pulling my Praxis WetBench out of closet for the Encore and selling the Dark Base Pro 901. I love the case, but no case is as good as an open bench unless aesthetics are more important than function. I also feel the horizontal motherboard orientation is superior to vertical. It is much easier to lay the O11 XL on its back for maintenance than it is the Dark Base Pro 901. The Dark Base is gorgeous, but it is too big and bulky.
View attachment 2656752
View attachment 2656753
I wish my wife agreed with you. She still wants to kill me….
 
So how likely will the memory be in the 45-55C range with a fan with RefI at 131K with regular net surfing or a short gaming session? Seems unlikely to get anywhere close to a temp seen with a stress test.
I doubt you will see temperatures above 40-45°C gaming, benching or web surfing if you install some Byski or Bitspower RAM jackets on them and keep your fans blowing over them.
 
  • Rep+
Reactions: Papusan
I doubt you will see temperatures above 40-45°C gaming, benching or web surfing if you install some Byski or Bitspower RAM jackets on them and keep your fans blowing over them.
Even without the RAM jackets I have a hard time thinking the memory temps will be even close to 45C since most of the rendering is done on the video card itself. But I am going to keep HWInfo open and see what temps look like on my regular machine just for comparison.
 
  • Helpful
Reactions: MrFox
  • Rep+
Reactions: Papusan and MrFox
  • Rep+
Reactions: Papusan and MrFox
That fan isn't doing much placed where it is, set it right on top of the memory.
I agree. The 120MM fan on the flex arm would be more beneficial directly above the memory and DIMM.2 area.

Alternative, one of these blowing air directly down on both of the modules would work better than the fan on the end. I use one of these in my SFF setup with the Byski jackets and ribbed aluminum heat sink bolted on top of the RAM jackets.

Image
 
For air cooled DDR5 with factory heatsinks above 8200 and semi tight timings I would stick to 65k TREFI.
Even gaming in a game like BF2042 DIMM temps can get quite high.
 
I agree. The 120MM fan on the flex arm would be more beneficial directly above the memory and DIMM.2 area.

Alternative, one of these blowing air directly down on both of the modules would work better than the fan on the end. I use one of these in my SFF setup with the Byski jackets and ribbed aluminum heat sink bolted on top of the RAM jackets.

Image
I have one of those if I can remember where I put it…..Thanks.
 
  • Rep+
Reactions: Papusan and MrFox
Even without the RAM jackets I have a hard time thinking the memory temps will be even close to 45C since most of the rendering is done on the video card itself. But I am going to keep HWInfo open and see what temps look like on my regular machine just for comparison.
Your memory modules would run cooler naked with a fan blowing on them than they do with stock heatsinks with a fan blowing on them. The stock heatsinks are more for aesthetics than anything else. They do a piss poor job of cooling and the thermal pads and adhesives they use to attach them are utter garbage. The exposed LEDs are ugly, but you can cover them with some Testers black modeling/hobby paint or a strip of electrical tape to kill the bright lights.

I really hate RGB memory, but it's hard to find good overclocking modules without RGB rainbow puke because that's all the kiddos want. TeamGroup Xtreem are one of a few good options.
 
  • Rep+
Reactions: Papusan
For air cooled DDR5 with factory heatsinks above 8200 and semi tight timings I would stick to 65k TREFI.
Even gaming in a game like BF2042 DIMM temps can get quite high.
Will likely reduce the RefI. Thanks!
 
Your memory modules would run cooler naked with a fan blowing on them than they do with stock heatsinks with a fan blowing on them. The stock heatsinks are more for aesthetics than anything else. They do a piss poor job of cooling and the thermal pads and adhesives they use to attach them are utter garbage.
Thanks!
 
I agree. The 120MM fan on the flex arm would be more beneficial directly above the memory and DIMM.2 area.

Alternative, one of these blowing air directly down on both of the modules would work better than the fan on the end. I use one of these in my SFF setup with the Byski jackets and ribbed aluminum heat sink bolted on top of the RAM jackets.

Image
I'm using this too and works plenty well!

Green Motor vehicle Gas Computer hardware Engineering
 
Just found mine. I am going to install it tomorrow and retest.
It won't do much if you're on stock heatsinks. I've tried back to back with stock gskill heatsinks at 1.35v on 8000c40 2x24gb kit vs my 7800c36 2x16gb kit with bitspower heatsinks at 1.47v and the difference is more than 10 degrees. The problem is the heatsinks more than getting air over it. Even if you're at 45 degrees during TM5, you still can be unstable and throw errors. The heat buildup and conductivity in the stock heatsinks is absolutely horrid.
 
  • Rep+
  • Helpful
Reactions: Papusan and MrFox
It won't do much if you're on stock heatsinks. I've tried back to back with stock gskill heatsinks at 1.35v on 8000c40 2x24gb kit vs my 7800c36 2x16gb kit with bitspower heatsinks at 1.47v and the difference is more than 10 degrees. The problem is the heatsinks more than getting air over it. Even if you're at 45 degrees during TM5, you still can be unstable and throw errors. The heat buildup and conductivity in the stock heatsinks is absolutely horrid.
Understood. Thanks.
 
Understood. Thanks.
Getting the stock heatsinks off, even if you don't have the aftermarket heatsinks in-hand and ready to install, will improve your temperatures with that Corsair fan blowing over the naked modules. You need to get them off to install the aftermarket heatsinks anyway, so I would start there. I suspect you will see a measurable improvement by simply getting the crappy stock heatsinks heating blankets off of the modules. They are just trapping heat.
 
I think I’m going to try your laquer thinner method when I have time at the end of the week.
 
  • Rep+
Reactions: Papusan
16,921 - 16,940 of 20,233 Posts