Drop Temperatures on RYZEN 3000 CPUs: der8auer

[Kenrou]

New mounting bracket specific for Ryzen 3000 chiplet placement, 6c-8c lower temp on Cinebench20, possibly less for Prime95

 

[mackerel]

If it helps at all with cooling, it should help more the bigger the load.

While an interesting idea, unless you're pushing for that last bit of overclock I'm not sure it is worth getting.

 

[bassnut]

Looks like 35 euro + shipping, it's in german. I'll have to check at home my browser here on my phone won't translate or I don't know how.

Interesting though.

EDIT >>>>

Only available in Europe there is no option for US or Canada shipping.

 

[ShrimpBrime]

The problem with this nonsense is that when you remove the solder and use paste, the temps go up, not down. Even LM isn't as good as the soldered IHS plate.
However, you can de-lid these Ryzen chips, use a decent thermal paste, re-install the stock cooler and use the chip like nothing happened. Again, yes higher temps, but slightly.

The only benefit to de-lidding is by running naked waterblock to die. That's where the temp drops come from. Not some fancy mounting garbage that won't bring temps 8c lower even if you pray for it.

 

[EarthDog]

The problem with this nonsense is that when you remove the solder and use paste, the temps go up, not down. Even LM isn't as good as the soldered IHS plate.
However, you can de-lid these Ryzen chips, use a decent thermal paste, re-install the stock cooler and use the chip like nothing happened. Again, yes higher temps, but slightly.

The only benefit to de-lidding is by running naked waterblock to die. That's where the temp drops come from. Not some fancy mounting garbage that won't bring temps 8c lower even if you pray for it.

True...

However, this product isn't made for delidded CPUs or designed to delid in the first place. It was made to replace the stock amd mounting with a new one aligned to the CPUs hotspot center. This has nothing to do with delidding.

 

[ShrimpBrime]

True...

However, this product isn't made for delidded CPUs or designed to delid in the first place. It was made to replace the stock amd mounting with a new one aligned to the CPUs hotspot center. This has nothing to do with delidding.

Click the video and watch the first 120 seconds EarthDog. It's explicitly designed for the de-lidded cpu which they could not run naked direct die cooler because the die sits lower than the cpu socket.

It's all about the de-lidding actually.

 

[Mr.Scott]

The project started out for delidded CPU's, but progressed into an all around solution for even the lidded processors.

 

[trents]

The project started out for delidded CPU's, but progressed into an all around solution for even the lidded processors.

This^

 

[ShrimpBrime]

The project started out for delidded CPU's, but progressed into an all around solution for even the lidded processors.

Interesting. I wonder what the real world figures look like.

8c drop in temps is a pretty big claim and what I consider outside the margin of error in measurements.

Any real world numbers out there yet?? Or we just going to take the word that shifting the pressure (slightly) would make any difference on a cpu with solder under the lid?

Speculation?

 

[Mr.Scott]

No idea. I fancy the idea though. Centering the cooler over the hot spot seem logical for a temp decrease.

 

[bassnut]

I would buy the kit but it isn't available for oversea shipping.

 

[Railgun]

Shall I send you one?

 

[EarthDog]

The project started out for delidded CPU's, but progressed into an all around solution for even the lidded processors.

Yep!


That first 120s shows it started out as a design for delidded prcoessors but since it failed due to die height difference and socket height over it, the mounting/product is specifically designed for lidded processors. They are just saying to use it for better temps, not to delid the soldered processors.

 

[bassnut]

Shall I send you one?

if you have or can get one of the …… yes if you have or can get the custom mount version or at least let me know what the final cost would be to Ontario K0H 2L0. I just bought a new MSI 2080 Super my wife might flip if I spend too much more.

 

[ShrimpBrime]

No idea. I fancy the idea though. Centering the cooler over the hot spot seem logical for a temp decrease.

Aye, this could assist the upward motion of heat travel to some extent. Even enough to drop 8c in temps. Being the heat spreader at that point does little to "spread" heat in an even fashion, the hot spot slightly to one side of the cpu.
With board upright in a case, the heat source would be located near the bottom of the cooler or block, then besides the heat traveling straight up, in this position would heat out and up, if this makes any sense. This comment solely based on the idea that heat rises.

This reminds me of the convo to turn your cooler 90 degrees and it will magically drop temps by 10c...... Lets now mount the cooler 3 mm (guessing this number) lower on the "heat spreader".....

 

[Zerileous]

This looks cool to me. Most water blocks force water into the center of the cold plate first, that extra turbulence as it contacts the cold plate perpendicular would transfer more heat. At least that's what I'm imagining here. I don't think it would do anything for air coolers. Noctua tested the heatpipe orientation and found no difference, I'm guessing again that it is because the heatpipes are so efficient that the easily transfer the heat from one to another.

 

[ShrimpBrime]

This looks cool to me. Most water blocks force water into the center of the cold plate first, that extra turbulence as it contacts the cold plate perpendicular would transfer more heat. At least that's what I'm imagining here. I don't think it would do anything for air coolers. Noctua tested the heatpipe orientation and found no difference, I'm guessing again that it is because the heatpipes are so efficient that the easily transfer the heat from one to another.

Ah good point, while many use a waterblock that is not full copper, I suppose the orientation or location of that plate could possibly make a huge 8c difference!!! Maybe even more than that like 12c!!! wouldn't that be something!!

Ryzen chips run fine on the stock coolers. I can't imagine even 8c drop in temps to affect these chips while you want to stay within FIT parameters, the heat almost means nothing even at 85c.

 

[Woomack]

Every single Ryzen 3000 that I was testing was working fine with a stock cooler ... however, every single Ryzen 3000 that I was testing was dropping CPU clock because of too high temps as in every case it was reaching 95°C. The difference between a stock cooler under load and above average 3rd party cooler was around 100-200MHz for constant higher load work. Hard to call it throttling as it fits CPU's clock range specified by AMD but for sure causes the CPU to run slower than it could at a lower temp.

I still see no point in a special mounting as 6-8°C (as ShrimpBrime said) is not affecting OC in any way unless the CPU is already reaching 95°C. Even then you can set a higher Tj like +20°C (at least on higher series motherboards). Going from a small cooler like Noctua D9L/single fan to U12S helped to OC my CPUs all... 0-50MHz depends on the chip. The difference in temps was around 10°C.

It looks like another stuff full of marketing. The price is ~€30 but have to add ~€15 for shipping to near countries like Poland. Still better than €90 EUR(+shipping) delidding tools.

 

[trents]

Every single Ryzen 3000 that I was testing was working fine with a stock cooler ... however, every single Ryzen 3000 that I was testing was dropping CPU clock because of too high temps as in every case it was reaching 95°C. The difference between a stock cooler under load and above average 3rd party cooler was around 100-200MHz for constant higher load work. Hard to call it throttling as it fits CPU's clock range specified by AMD but for sure causes the CPU to run slower than it could at a lower temp.

I still see no point in a special mounting as 6-8°C (as ShrimpBrime said) is not affecting OC in any way unless the CPU is already reaching 95°C. Even then you can set a higher Tj like +20°C (at least on higher series motherboards). Going from a small cooler like Noctua D9L/single fan to U12S helped to OC my CPUs all... 0-50MHz depends on the chip. The difference in temps was around 10°C.

It looks like another stuff full of marketing. The price is ~€30 but have to add ~€15 for shipping to near countries like Poland. Still better than €90 EUR(+shipping) delidding tools.

I think Woo laid the case out well. This offset system would only have benefit for those pushing an overclock to the edge. Otherwise, existing cooling solutions using the stock mount that keep temps at 85c or below are perfectly adequate.

 

[Kenrou]

Keep in mind that the tool is for ALL the 3000 Ryzen, those 6c-8c might be very interesting for someone with a 3900+ looking for a decent overclock an air cooler for example. Granted, small population but still might be useful.