Pro Overclocker der8auer Delids the AMD Ryzen 5 2600 Processor

[Chino]

In his latest Youtube video, famous overclocker der8auer has delidded his AMD Ryzen 5 2600 processor for the sole purpose of evaluating whether the benefits justify the risk. Since the IHS in the new Pinnacle Ridge processors is soldered directly to the die with Indium, delidding the processors is a tricky but not impossible task. Everything melts when it gets warm enough, and indium starts melting around 156.60 °C. Therefore, der8auer had to use a modified version of his popular Delid Die Mate 2 tool by replacing the acrylic pieces with aluminum while also removing the rubber washer. After baking his Ryzen 5 2600 chip in the oven between 170 °C to 180 °C, Der8auer removed the IHS easily with his delidding tool. For his testing, he replaced the indium solder with Thermal Grizzly liquid metal thermal compound. As expected, the results weren't very impressive. With the Ryzen 5 2600 overclocked to 4.1 GHz with 1.35V, the difference was a mere 4 °C under load. So, there you have it. Don't delid your Pinnacle Ridge processor. It's not worth the effort.


View at TechPowerUp Main Site

 

[R-T-B]

I am actually somewhat surprised liquid metal beats straight Indium solder...

 

[xorbe]

I am actually somewhat surprised liquid metal beats straight Indium solder...

Maybe reduced gap.

 

[Supercrit]

I am actually somewhat surprised liquid metal beats straight Indium solder...

Is the liquid metal compound used with or without the lid?

 

[_JP_]

I am actually somewhat surprised liquid metal beats straight Indium solder...

I am not, because the delid means there's one layer less until a heat transfer surface is reached. If it wasn't delidded and instead only the solder replaced with liquid metal, I wager results wouldn't be better.
Next-up, dies soldered to heatsinks.
NH-D15, step forward please.

 

[laszlo]

Is the liquid metal compound used with or without the lid?

at these cpu's is hard to make direct die contact (without lid) due the socket build-up as i know so he must used the lid

 

[QinX]

at these cpu's is hard to make direct die contact (without lid) due the socket build-up as i know so he must used the lid

Are you referring to the Intel socket in this case? Because that retention bracket is indeed in the way for direct die cooling, although not impossible.
AMD however have full clearance around the socket for the heatsink. The only issue would be proper mounting pressure on the die itself and the retention system used by the heatsink manufacturer.

 

[ghazi]

at these cpu's is hard to make direct die contact (without lid) due the socket build-up as i know so he must used the lid

Part of the problem is also mounting a heavy heatsink onto such a thin, flimsy PCB...

 

[eidairaman1]

Part of the problem is also mounting a heavy heatsink onto such a thin, flimsy PCB...

Erm cpu pcbs are not flimsy, the dies in all cpus are sensitive to crushing though.

 

[Johan45]

Erm cpu pcbs are not flimsy, the dies in all cpus are sensitive to crushing though.

Been there done that 6700K chalked up to a bad experience.

 

[eidairaman1]

Been there done that 6700K chalked up to a bad experience.

Perfectly good Athlon XP3200+ that was unlocked (before Week 39 i believe in 2003/2004)

 

[ghazi]

Erm cpu pcbs are not flimsy, the dies in all cpus are sensitive to crushing though.

Maybe not in the past, maybe not in general, but Intel's mainstream chips since Skylake have had very flimsy PCBs. I've heard of people breaking them before but this is the best source I could find on the fly: http://www.legitreviews.com/some-cpu-coolers-are-bending-intel-skylake-cpus_175933

Can't speak for Ryzen specifically though.

 

[eidairaman1]

Maybe not in the past, maybe not in general, but Intel's mainstream chips since Skylake have had very flimsy PCBs. I've heard of people breaking them before but this is the best source I could find on the fly: http://www.legitreviews.com/some-cpu-coolers-are-bending-intel-skylake-cpus_175933

Ok thats intel then, apples/oranges are not the same.

I can make my mobo board flex if i tighten the hsf screws too much, did it upon initial setup, backed them off till board no longer flexed before putting board in case

 

[FreedomEclipse]

well... Only 4'c difference. Some might say that its almost negligible and that CPU is not worth the effort delidding.


If it was just a test or experiment to see if lower temps could be had then fair beans. I guess people have their answer now.

 

[eidairaman1]

well... Only 4'c difference. Some might say that its almost negligible and that CPU is not worth the effort delidding.


If it was just a test or experiment to see if lower temps could be had then fair beans. I guess people have their answer now.

Thats the whole thing he saved us the hardship of it by proving its not worth it.

 

[Hood]

So after thousands of people went crazy about how messed up of Intel to save pennies by not soldering, and how bad "toothpaste" TIM was for temps, that goop beats solder by 4c? That was my touchstone for ignore list candidates - loud bitching about TIM vs solder.

 

[MrGenius]

I am not, because the delid means there's one layer less until a heat transfer surface is reached.

Wrong. He relidded it. Well...effectively at least. Though he doesn't mention whether or not he glued the IHS back on. He does mention still using the IHS on the CPU.

Maybe reduced gap.

Correct.

The only issue would be proper mounting pressure on the die itself...

Not an issue with PGA CPU PCBs. They're not supported just from their edges like LGA CPU PCBs. PGA CPU PCBs are supported almost entirely, except for a small area in the center where there are no pins. You would have to apply enough pressure to crack the die itself before it would be a problem. Not a realistic scenario. See replies to quotes below.

Part of the problem is also mounting a heavy heatsink onto such a thin, flimsy PCB...

Not applicable. See replies to quotes above and below.

Erm cpu pcbs are not flimsy...

Although it's a relative statement to say they're "flimsy"(compared to what?). IMO, yes, they are(since you can bend them easily). Also see replies to quote above and below.

...the dies in all cpus are sensitive to crushing though.

Not really. It would take a RIDICULOUS amount of pressure to damage the die itself. Silicon is not a weak material. Brittle yes. But extremely strong under compression.

...for Ryzen specifically though.

Is why the statement regarding the "flimsiness" of the PCB is irrelevant.

Ok thats intel then, apples/oranges are not the same.

1. Apples/oranges precisely. No comparison to be made whatsoever in this instance.

I can make my mobo board flex if i tighten the hsf screws too much, did it upon initial setup, backed them off till board no longer flexed before putting board in case

Also not really a comparable scenario between LGA and PGA CPUs. You're not ANYWHERE NEAR as likely to flex the board enough for the pins to lose contact, as your are for the lands to lose contact. I won't say it's impossible to do with a PGA CPU. But it's pretty damn close to it.

So after thousands of people went crazy about how messed up of Intel to save pennies by not soldering, and how bad "toothpaste" TIM was for temps, that goop beats solder by 4c? That was my touchstone for ignore list candidates - loud bitching about TIM vs solder.

This is not about TIM vs. solder. This is about solder vs. liquid metal with a reduced gap between the IHS and die.

 

[bug]

So after thousands of people went crazy about how messed up of Intel to save pennies by not soldering, and how bad "toothpaste" TIM was for temps, that goop beats solder by 4c? That was my touchstone for ignore list candidates - loud bitching about TIM vs solder.

Depends on who you listen to.
Intel CPUs have been delidded and replacing TIM yielded no better results, so that TIM does its job as well enough. But haters will keep on doing their thing...

 

[Raptor_i]

See Intel that's how you do thermal conductor ... 4 degrees.

 

[RealNeil]

Intel CPUs have been delidded and replacing TIM yielded no better results, so that TIM does its job as well enough

I'm using an i7-6700K that keeps to low temps all of the time. (if it works, don't fix it)
I have a pair of i7-7700K CPUs that didn't until I did a DeLid on them. Now I'm seeing a 15-20C lower temp on both of them. I'm using liquid metal between the lid and the silicone.

I then did my i9-7900X CPU and got vast temp reductions with it. It was throttling from high heat, but now, not so often.
DeLids, done right are a good thing with Intel CPUs these days.

 

[Bones]

So after thousands of people went crazy about how messed up of Intel to save pennies by not soldering, and how bad "toothpaste" TIM was for temps, that goop beats solder by 4c? That was my touchstone for ignore list candidates - loud bitching about TIM vs solder.

Just more proof in how effective AMD's way of making them is vs what Intel has been doing. No real need to delid or even worry about it for the AMD's, the Intels can and sometimes do run hot at or close to stock settings if left "As Is" from the factory - Many will run hot period if a moderate OC is applied.

Had to delid my 7700K to have decent temps for all around use, didn't have this issue with any of my AMD's.

 

[nickbaldwin86]

haha yeah on my intel it was a LOT more than 4*c LOL

This is a HUGE risk for VERY little, but good for him to show us it can be done,... but DONT do it, well do it if you want 4*c.... HOnestly would love to see the temp change with a higher OC, 4.1Ghz just seems low for that amount of effort?

Depends on who you listen to.
Intel CPUs have been delidded and replacing TIM yielded no better results, so that TIM does its job as well enough. But haters will keep on doing their thing...

Delidded with LM my Intel CPU, it is a MUST, I run 5Ghz at or below 60c, on a custom loop. without a delid that would be remotely possible.

 

[eidairaman1]

haha yeah on my intel it was a LOT more than 4*c LOL

This is a HUGE risk for VERY little, but good for him to show us it can be done,... but DONT do it, well do it if you want 4*c.... HOnestly would love to see the temp change with a higher OC, 4.1Ghz just seems low for that amount of effort?



Delidded with LM my Intel CPU, it is a MUST, I run 5Ghz at or below 60c, on a custom loop. without a delid that would be remotely possible.

Already been down that road for 3 years now on Air. 43 idle 55 Gaming.
Signature below.

 

[bug]

I'm using an i7-6700K that keeps to low temps all of the time. (if it works, don't fix it)
I have a pair of i7-7700K CPUs that didn't until I did a DeLid on them. Now I'm seeing a 15-20C lower temp on both of them. I'm using liquid metal between the lid and the silicone.

I then did my i9-7900X CPU and got vast temp reductions with it. It was throttling from high heat, but now, not so often.
DeLids, done right are a good thing with Intel CPUs these days.

haha yeah on my intel it was a LOT more than 4*c LOL

This is a HUGE risk for VERY little, but good for him to show us it can be done,... but DONT do it, well do it if you want 4*c.... HOnestly would love to see the temp change with a higher OC, 4.1Ghz just seems low for that amount of effort?



Delidded with LM my Intel CPU, it is a MUST, I run 5Ghz at or below 60c, on a custom loop. without a delid that would be remotely possible.

Well, yeah, for Intel you can get lower temps, but rarely higher overclocks. If you delid a lot of second generation Ryzens, some will overclock better. Doesn't mean the default compound isn't doing its job.

 

[iO]

Must be the reduced gap width since a 73 W/(m·K) metal TIM shouldnt be able to beat a ~80 W/(m·K) Indium alloy...