Prototype DAN HSLP-48: A powerful sub 50mm heatsink

[dondan]

@HeroLazer: Backplate + Screwhead = 4mm.

After some test I made the decision to go with the two test scenarios. Default clock and 3,6Ghz @1,275V. I made this decision to make sure that also the L9i can handle the test. What I can say, the 5 heatpipe version is 15°C better as the L9i

 

[HeroXLazer]

@HeroLazer: Backplate + Screwhead = 4mm.

After some test I made the decision to go with the two test scenarios. Default clock and 3,6Ghz @1,275V. I made this decision to make sure that also the L9i can handle the test. What I can say, the 5 heatpipe version is 15°C better as the L9i

That's nice, some companies make the backplate too thick.

 

[Kmpkt]

Out of curiosity is the HSLP-48 15 degrees ahead of the L9i in open air or inside a case? Also if I'm not mistaken, you're planning on Aluminum fins bound to copper heatpipes and a copper baseplate, correct?

 

[dondan]

In the first step I test open air to get an overview. On the materoal you are right.

 

[dondan]

So here are the first results.

Setup:
Environment: open air (benchtable)
Room Temp: 22°C
Hardware: Ryzen 7 1700, Asus ROG Strix B350I-Gaming, Micron VLP DDR4 2400
Test Tool: Prime95 v.29.4, 8K Test FFTs in-place
Duration: each 15min

Ryzen 1700 default

• HSLP-48 4HP (CJ) + A12x15 = 48°C
• HSLP-48 5HP (LL) + A12x15 = 49°C
• HSLP-48 4HP (LL) + A12x15 = 51°C
• HSLP-48 6HP (LL) + A12x15 = 51°C
• Noctua L9i + A9x14 = 62°C

Ryzen 1700 @3,6Ghz 1,26V

• HSLP-48 4HP (CJ) + A12x15 = 62°C
• HSLP-48 5HP (LL) + A12x15 = 67°C
• HSLP-48 6HP (LL) + A12x15 = 69°C
• HSLP-48 4HP (LL) + A12x15 = 77°C
• Noctua L9i + A9x14 = 82°C

* LL = LianLi
* CJ = CoolJag
* HP = Heatpipe

Yes you see right the 4 heatpipe CoolJag sample from summer is still the best. Maybe CoolJag use better heatpipes, or a better soldering material to connect heatpipe and CPU plate or maybe the brushed bottom surface is better as the polished LL surface.

The test show us, that heatpipe isn't the most important thing. There is a point where surface rules against heatpipe count. On the Ryzen CPU I used for the test you can see that on default clock nearly every heatsink perform the same. So the heatpipe performance or count has not a big impact on the results. The overclocked CPU generates more heat so the heatsink from LianLi with more pipes are better, but 5 heatpipes are enough. The 6 heatpipe sample has less surface as the 5 heatpipe sample because of the extra pipe this result in the 2°C difference. Maybe some of you will think a CoolJag sample with 5 heatpipes will be also much better as the 4 pipe sample. I don't think this is true because the CoolJag sample reach already the sweetpot and more pipes will reduce the surface.

I will do more tests with different fan sizes and also inside the case in the next days.

 

[HeroXLazer]

So here are the first results.

Setup:
Environment: open air (benchtable)
Room Temp: 22°C
Hardware: Ryzen 7 1700, Asus ROG Strix B350I-Gaming, Micron VLP DDR4 2400
Test Tool: Prime95 v.29.4, 8K Test FFTs in-place
Duration: each 15min

Ryzen 1700 default

• HSLP-48 4HP (CJ) + A12x15 = 48°C
• HSLP-48 5HP (LL) + A12x15 = 49°C
• HSLP-48 4HP (LL) + A12x15 = 51°C
• HSLP-48 6HP (LL) + A12x15 = 51°C
• Noctua L9i + A9x14 = 62°C

Ryzen 1700 @3,6Ghz 1,26V

• HSLP-48 4HP (CJ) + A12x15 = 62°C
• HSLP-48 5HP (LL) + A12x15 = 67°C
• HSLP-48 6HP (LL) + A12x15 = 69°C
• HSLP-48 4HP (LL) + A12x15 = 77°C
• Noctua L9i + A9x14 = 82°C

* LL = LianLi
* CJ = CoolJag
* HP = Heatpipe

Yes you see right the 4 heatpipe CoolJag sample from summer is still the best. Maybe CoolJag use better heatpipes, or a better soldering material to connect heatpipe and CPU plate or maybe the brushed bottom surface is better as the polished LL surface.

The test show us, that heatpipe isn't the most important thing. There is a point where surface rules against heatpipe count. On the Ryzen CPU I used for the test you can see that on default clock nearly every heatsink perform the same. So the heatpipe performance or count has not a big impact on the results. The overclocked CPU generates more heat so the heatsink from LianLi with more pipes are better, but 5 heatpipes are enough. The 6 heatpipe sample has less surface as the 5 heatpipe sample because of the extra pipe this result in the 2°C difference. Maybe some of you will think a CoolJag sample with 5 heatpipes will be also much better as the 4 pipe sample. I don't think this is true because the CoolJag sample reach already the sweetpot and more pipes will reduce the surface.

I will do more tests with different fan sizes and also inside the case in the next days.

 

[Sandez]

They are some very interesting and promising results @dondan

You are getting very close to optimal levels by the looks of it. I saw another CPU cooler test where they used the Noctua NH-D15 on the 1700 with a room temp of 20°C and they were getting 42°C (in AIDA64 not Prime95, don't know if it's much of a difference), effectively making it 44°C vs your 48°C. A 4°C difference is very impressive for such a compact cooler vs that monster. Given the temp difference from 4 to 5 HP's on the LL, going from 4 to 5 HP's on the CJ could be that crucial 4°C difference!

Either way, I'm already sold on this. Looking forward to seeing more tests and waiting for it to be released

 

[T_Tank]

Can we get some pictures of this heatsink mounted to the am4 board I"m just having a hard time visualizing how it looks mounted got me curious as I"m jerry rigging a big shuriken 2 revB right now on my am4 board... really wish someone would make new brackets for it for native am4 mounting @_@

 

[LeMoNH3aD]

Thrown for a loop in seeing the 4HP version take the cake...

Really excited to see the progress!

 

[EdZ]

I will do more tests with different fan sizes and also inside the case in the next days.

If you can rent/borrow/acquire a thermal camera (e.g. one of the relatively low-cost Lepton-based FLIR cameras that conenct to a phone) you can look at the sides of the heatpipes exposed between the baseplate and fin stack, and get an idea of how much of the thermal load each heatpipe is carrying.

With the current very small die sizes for consumer chips, any heatpipes not over the die itself will only be moving what heat is absorbed and then reemitted by the baseplate & heatpipes that are over the die. If that retransmission is less than the los in fin area from poking extra heapipes through them, that would be a net performance loss.

 

[Garrzilla]

Will this be ready by the time the A4-SFX V2 is shipped out? I'm looking to get the LP53, but would prefer this one

 

[Ionrent]

 

 

[theGryphon]

Can we get some pictures of this heatsink mounted to the am4 board I"m just having a hard time visualizing how it looks mounted got me curious as I"m jerry rigging a big shuriken 2 revB right now on my am4 board... really wish someone would make new brackets for it for native am4 mounting @_@

I'm with you bro. Please PM me with how you get it installed on AM4, or make a new thread?


By the way, 4HP version beating the rest out must be about build quality, so maybe you should work with Cooljag, Dan?

 

[LeMoNH3aD]

Anyone's thoughts on this "hold-me-over" until this heat sink drops?

Heat sink: http://www.coolermaster.com/cooling/cpu-air-cooler/geminii-m5-led/

+

Fan: http://noctua.at/en/products/fan/nf-a12x15-pwm

 

[LeMoNH3aD]

Good translatable review:
https://www.hardwareluxx.de/communi...leiser-itx-kuehler-mit-potenzial-1162594.html

 

[T_Tank]

Anyone's thoughts on this "hold-me-over" until this heat sink drops?

Heat sink: http://www.coolermaster.com/cooling/cpu-air-cooler/geminii-m5-led/

+

Fan: http://noctua.at/en/products/fan/nf-a12x15-pwm

nope cause I don't see AM4 in the list of compatible sockets ;-;

 

[absolut_zero]

The heatsink itself at 45.5mm will fit but it will be passively cooled. With a 120mm slim fan on top of it at 62.7mm, it's too tall for Dan A4-SFX unless you will leave the side panel off.

 

[Qrash]

The heatsink itself at 45.5mm will fit but it will be passively cooled. With a 120mm slim fan on top of it at 62.7mm, it's too tall for Dan A4-SFX unless you will leave the side panel off.

... but it will fit in the new Louqe Ghost S1 case which has a 66 mm height limit for the CPU cooler.

 

[slimjim]

Anyone's thoughts on this "hold-me-over" until this heat sink drops?

Heat sink: http://www.coolermaster.com/cooling/cpu-air-cooler/geminii-m5-led/

+

Fan: http://noctua.at/en/products/fan/nf-a12x15-pwm

why not get a cooljag falcon or a titan?

 

[Ca11idus]

... but it will fit in the new Louqe Ghost S1 case which has a 66 mm height limit for the CPU cooler.

This thread is for the HSLP-48 that is being made for the Dan Case A4. Let's stay on topic.