• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Ryzen 2700x de-lid. (I did it a while ago...)

  • Thread starter Deleted member 185158
  • Start date

Which Cpu is the 2700X? (The other is Ryzen 1200)


  • Total voters
    11
D

Deleted member 185158

Guest
Just wanted to show the under hood of a Ryzen 2700X that's actually used daily with it's hat removed.

Method: Soldering torch
Time: Including cutting the glue around the edges, oh 15 minutes or so. Actual fire to plate, under 60 seconds.
Use Time: While de-lidded... a few months now. (still in use daily)
Use Method: Now Gaming Started with testing chip behavior.
Cooling Used: Air, Liquid and TEC cold plate to die. (not chill loop)
Recent Chips delidded: 200ge (no solder) 220ge (no solder) Ryzen 1200 and Ryzen 2700X (many more previous generations soldered AMD processors)

Using a small model vice, I heat sink the pins to a large based heatsink and clamp the PCB in the vice. To give size, Jaw opening 1.5 inch and 3 inch. It's small fits in your hand. Vice grip IHS plate. Add heat with solder torch on med to high flame. On average the solder melts in 48-52 seconds. Quiet room, can hear when the solder melts, it happens all at once. Pull lid off, quickly cool the chip with a small fan and done. Pretty simple process, been pretty successful.

So it's a long story, but I've found no real reason to delid Ryzen processors. There is no gains to be seen even running lower than ambient core temps. The chilling does help to reduce leakage and allows for a lower v-core to be obtained and remain stable. However removing the solder hasn't helped the cause for cooler temps..... Unless it's applied for fun and testing purposes. I always have the most challenging fun trying to cool processors with a cold plate and TEC. Not as easy as it sounds. But this also depends on what your looking to do! I try to maintain a frozen processor, but also test going above freezing to see if it caused any reaction. Those tests have been done some time ago, but never the less didn't have any bad experience with it at all. I can touch on cooling and such more if requested.
In the past, de-lidding helped quite a bit as cooling increased headroom (typically Phenom and FX chips) had an average of 10c drops in temps and 100mhz increased frequency. Well with these new chips, the average is a tad different.
On air cooling, re-using the IHS plate substitute the solder with TIM, made matters a tad worse, but not as bad as expected!! I run a daily clock of 4.1Ghz at 1.412v just the same as if the solder was there. Temps are a little higher for sure when loading the IBT or Prime95 and I also enjoy OCCT too. Reaching throttle point happens much faster when running 4.2ghz and 1.43750v and that is where the ill return came in with the stock cooling. Kiss high clocks and higher voltage by by. Before the de-lid on stock cooling, reached a max clock of 4500mhz and a pretty stable x4 4.4ghz. Max Cinebench stable run was at 4300mhz All threads, now only to 4100mhz. So on ambient air coolers and AIO, this is actually not going to fair well. But don't let me leave out that 4.1Ghz on this 2700X is actually it's most stable overclock. PBO won't even pass this frequency. So manual clocking it is.

Well I ramble a little. Pics or didn't happen right?
A little TEC frost there. That copper plate is directly on the core.

0.jpg



TEC frozen.jpg


20190409_214953.jpg
 

eidairaman1

The Exiled Airman
Joined
Jul 2, 2007
Messages
40,435 (6.61/day)
Location
Republic of Texas (True Patriot)
System Name PCGOD
Processor AMD FX 8350@ 5.0GHz
Motherboard Asus TUF 990FX Sabertooth R2 2901 Bios
Cooling Scythe Ashura, 2×BitFenix 230mm Spectre Pro LED (Blue,Green), 2x BitFenix 140mm Spectre Pro LED
Memory 16 GB Gskill Ripjaws X 2133 (2400 OC, 10-10-12-20-20, 1T, 1.65V)
Video Card(s) AMD Radeon 290 Sapphire Vapor-X
Storage Samsung 840 Pro 256GB, WD Velociraptor 1TB
Display(s) NEC Multisync LCD 1700V (Display Port Adapter)
Case AeroCool Xpredator Evil Blue Edition
Audio Device(s) Creative Labs Sound Blaster ZxR
Power Supply Seasonic 1250 XM2 Series (XP3)
Mouse Roccat Kone XTD
Keyboard Roccat Ryos MK Pro
Software Windows 7 Pro 64
On the right lol
 
D

Deleted member 185158

Guest
Lol, I didn't pay attention. 1st picture gives it away doesn't it.

Here check this TEC experiment out.
4Ghz @ 1.188v temps went above frozen to 13.8c.
That cold sure does slow down the leakage doesn't it?
 

Attachments

  • 2700X TEC CBR15 temps 4Ghz.png
    2700X TEC CBR15 temps 4Ghz.png
    616 KB · Views: 523
Joined
Jul 26, 2019
Messages
418 (0.24/day)
Processor R5 5600X
Motherboard Asus TUF Gaming X570-Plus
Memory 32 GB 3600 MT/s CL16
Video Card(s) Sapphire Vega 64
Storage 2x 500 GB SSD, 2x 3 TB HDD
Case Phanteks P300A
Software Manjaro Linux, W10 if I have to
Nice. Any chance you will try it on 3000 series?
 
Joined
May 12, 2017
Messages
2,178 (0.87/day)
I will do a Ryzen 3000 series, but do not expect any major positive outcome from doing so.
Wanting to wait till the prices come down or used maybe even a trade.

Since you like doing experiments try this. Lap the IHS down to the copper. Then lap a air cooler also down to the copper, then solder the two together. Now use LM between die & IHS & see what temperature you get. This is what I do here but on much older processors (AMD FX-60). I do have a Ryzen CPU but no AM4 motherboard.

EDIT: If you try this, you may have to do the soldering in the kitchen over under the grill unless you have the correct equipment. (At your own risk).
 
Last edited:
D

Deleted member 185158

Guest
If I solder a cooler the the IHS plate, there will be no way to open the socket clasp. The lever would hit the heatsink.

One of the Ryzen issues is the plate used. Its quite small in mass and retains very little heat for dissipation.
 
Joined
May 12, 2017
Messages
2,178 (0.87/day)
If I solder a cooler the the IHS plate, there will be no way to open the socket clasp. The lever would hit the heatsink.

One of the Ryzen issues is the plate used. Its quite small in mass and retains very little heat for dissipation.

I'm using Noctua C14 I have two of them with solder IHS, but I have other heatsink with soldered IHS. Look at the Noctua heatsink to see if it will work on AM4 platform as I have not checked if it will work on the AM4 platform, but it looks like it will work due to the way they are mounted. I'm testing on the 939 platform. You have to be creative when mounting any heatsink, this is what I do. Most heatsinks are custom bolt-down, when IHS is soldered to heatsink.

You will get improved temperature with a solder IHS & LM on the die.
 
Last edited:
D

Deleted member 185158

Guest
I'm using Noctua C14 I have two of them with solder IHS, but I have other heatsink with soldered IHS. Look at the Noctua heatsink to see if it will work on AM4 platform as I have not checked if it will work on the AM4 platform, but it looks like it will work due to the way they are mounted. I'm testing on the 939 platform. You have to be creative when mounting any heatsink, this is what I do. Most heatsinks are custom bolt-down, when IHS is soldered to heatsink.

You will get improved temperature with a solder IHS & LM on the die.

I have improved temps with removing the IHS plate and running direct die cooling.

This was known before you did it... but kudos for doing it anyway!

Indeed Deh8auer beat me to it. Shame he killed a couple before having success. Kudos to him for doing it first..... On a Ryzen chip only.
But thanks for the Kudos. It means something to someone out there Im sure.

My mother passed about 9 years ago, the wake on my daughters 4th birthday. Needless to say I dont do the mamma jokes any more.

On the topic, hows those 90c temps working for ya? Did you buy a real water loop yet?
Go to Warp9 systems and see how Zila1 rig is doing now. Sent him some hardware, stess test maxed at 82c. Believe he has the same chip as you. His ambient temp was 80f btw.
 
Joined
May 12, 2017
Messages
2,178 (0.87/day)
I ran direct die with LM but this solution destroyed my R9 Nano. While cleaning with compress air LM was blown from the die & landed on the back of the GFX card. Did not see it. Turn on after cleaning & bang, card destroyed. With the IHS in place this protects anything below if it happens to drip/leak.

Test here has shown there is no difference whatsoever between direct die & soldered IHS/heatsink. I also carried out a test to see if reducing the IHS height improved temperature. The answer to this is no, in-fact it's around 2c worse off. This could be down to the heatsink heatpipe soldering or something else as both heatsink are Noctua C14. The only difference is one IHS is lapped super thin where the other is not. I just don't understand how the super thin IHS performed around 2-3c worse.
 
Last edited:
D

Deleted member 185158

Guest
When you lapped the plate thin, you removed the mass of the copper. Having more copper allows you to store more BTU and giving a little extra time to dissipate that heat as its stored in the plate.
My largest cold plate as seen in the picture at OP is about doube the size and mass of the IHS plate. It performs much better for sure.
 
Joined
Dec 31, 2009
Messages
19,366 (3.72/day)
Benchmark Scores Faster than yours... I'd bet on it. :)
On the topic, hows those 90c temps working for ya? Did you buy a real water loop yet?
lol, it's for a review test system... a 3x120mm AIO is beefy for most users. You forget why things are the way they are for my Ryzen rig. Short term memory of a goldfish, lol! That said, the cpu is just fine at 90c when stress testing.

When you lapped the plate thin, you removed the mass of the copper. Having more copper allows you to store more BTU and giving a little extra time to dissipate that heat as its stored in the plate.
But once its saturated, which is pretty quick/similar amount of time considering the small difference in mass between lapped and unlapped processor, it will perform the same. In fact, the thinner it is, the quicker it gets from A to B not having to move through a thicker medium.
 
Last edited:
D

Deleted member 185158

Guest
Its not the rad size on AIO that makes them marginally better than an air cooler, its the small little cold plate.
Design is better, but lacks the copper.
I perfer all copper blocks. Can cool the top side of an all copper block helping dissipation, something you cannot do with waterblocks that are made mostly from insulator producs like plastic and acrylic.

I did lap any IHS plates. That was a response to the other fello there
 
Joined
Dec 14, 2013
Messages
2,603 (0.69/day)
Location
Alabama
Processor Ryzen 2700X
Motherboard X470 Tachi Ultimate
Cooling Scythe Big Shuriken 3
Memory C.R.S.
Video Card(s) Radeon VII
Software Win 7
Benchmark Scores Never high enough
Good job!
Just waiting to see how you do with a 3000 series chip. :cool:
 
Joined
May 12, 2017
Messages
2,178 (0.87/day)
When you lapped the plate thin, you removed the mass of the copper. Having more copper allows you to store more BTU and giving a little extra time to dissipate that heat as its stored in the plate.
My largest cold plate as seen in the picture at OP is about doube the size and mass of the IHS plate. It performs much better for sure.

You can remove some mass. When I said one was lapped super thin & the other is not, in-fact the other one is lapped about a quarter & the other about three quarters. The idea here is the heat getting to the cold plate faster. The thinner IHS does seem to show signs of faster reaction, but it's like it gets soaked in heat. Temperature is still impressive for both heatsink.

I also forgot to say I also played around with real silver sheet between CPU & heatsink. I need to complete this test, but again I got worse results, but it's almost up there with soldered IHS. I think the problem is I never solder the silver sheet. The good news is LM loves to stick to it, so this is a good sign. So if you have a alloy heatsink, not a problem, solder silver sheet to it & you can now use LM.
 
Last edited:
D

Deleted member 185158

Guest
I have a silver dollar lapped both sides for a cold plate.

I should try LM. Ive heard a lot of mixed feelings about it though.
 

eidairaman1

The Exiled Airman
Joined
Jul 2, 2007
Messages
40,435 (6.61/day)
Location
Republic of Texas (True Patriot)
System Name PCGOD
Processor AMD FX 8350@ 5.0GHz
Motherboard Asus TUF 990FX Sabertooth R2 2901 Bios
Cooling Scythe Ashura, 2×BitFenix 230mm Spectre Pro LED (Blue,Green), 2x BitFenix 140mm Spectre Pro LED
Memory 16 GB Gskill Ripjaws X 2133 (2400 OC, 10-10-12-20-20, 1T, 1.65V)
Video Card(s) AMD Radeon 290 Sapphire Vapor-X
Storage Samsung 840 Pro 256GB, WD Velociraptor 1TB
Display(s) NEC Multisync LCD 1700V (Display Port Adapter)
Case AeroCool Xpredator Evil Blue Edition
Audio Device(s) Creative Labs Sound Blaster ZxR
Power Supply Seasonic 1250 XM2 Series (XP3)
Mouse Roccat Kone XTD
Keyboard Roccat Ryos MK Pro
Software Windows 7 Pro 64
I have a silver dollar lapped both sides for a cold plate.

I should try LM. Ive heard a lot of mixed feelings about it though.

Etching the die is 1, just like diamond too
 
Joined
May 12, 2017
Messages
2,178 (0.87/day)
20190409_214953.jpg


Those unused pads solder capacitors to them if you are skilled enough. It may give you enhance performance. Get the capacitors from old dead CPUs.

ESR metre recommended for this mod.
 
D

Deleted member 185158

Guest
Etching the die is 1, just like diamond too

I actually use nano-diamond pastes, prefer to actually. Nothing conducts heat better than diamonds! Occasionally I relap the cores, not too often though. Can't see the scoring from the paste like you do on the IHS plates.

Those unused pads solder capacitors to them if you are skilled enough. It may give you enhance performance. Get the capacitors from old dead CPUs.

ESR metre recommended for this mod.
I have not the talent for that! But have seen it done on socket A chips. Pencil mods too.
 

eidairaman1

The Exiled Airman
Joined
Jul 2, 2007
Messages
40,435 (6.61/day)
Location
Republic of Texas (True Patriot)
System Name PCGOD
Processor AMD FX 8350@ 5.0GHz
Motherboard Asus TUF 990FX Sabertooth R2 2901 Bios
Cooling Scythe Ashura, 2×BitFenix 230mm Spectre Pro LED (Blue,Green), 2x BitFenix 140mm Spectre Pro LED
Memory 16 GB Gskill Ripjaws X 2133 (2400 OC, 10-10-12-20-20, 1T, 1.65V)
Video Card(s) AMD Radeon 290 Sapphire Vapor-X
Storage Samsung 840 Pro 256GB, WD Velociraptor 1TB
Display(s) NEC Multisync LCD 1700V (Display Port Adapter)
Case AeroCool Xpredator Evil Blue Edition
Audio Device(s) Creative Labs Sound Blaster ZxR
Power Supply Seasonic 1250 XM2 Series (XP3)
Mouse Roccat Kone XTD
Keyboard Roccat Ryos MK Pro
Software Windows 7 Pro 64
I actually use nano-diamond pastes, prefer to actually. Nothing conducts heat better than diamonds! Occasionally I relap the cores, not too often though. Can't see the scoring from the paste like you do on the IHS plates.


I have not the talent for that! But have seen it done on socket A chips. Pencil mods too.

Yes make an AXP-M into an AXP-M DP unlocked multi lol
 
Joined
May 12, 2017
Messages
2,178 (0.87/day)
I don't think that will work. I think you will end up with worse temperature.

You need to cover the whole circumference with silver to cover the losses bonding between the silver/copper.
 
Top