AMD Ryzen 9 7950X AM5 processor open-cover cooling performance test / air-cooled, integrated water-cooling, liquid metal and more

AMD has always had a unique PBO (Precision Boost Overdrive) mechanism that makes it easier for the CPU to routinely exceed temperatures above 90°C under high loads. For those interested, you can google the principle of the PBO (Precision Boost Overdrive) mechanism. Without going into details, gamers who are more familiar with the related information should know that apart from the PBO mechanism of AMD AM5 series processors, there are also theories that it is due to the IHS (Integrated Heat Spreader) of the top cover and the TIM (Thermal Interface Material) of the inner layer, or the CCD (Cold Cold Chip) of the heat source of AMD CPUs. AMD CPU heat source CCD (Core Chiplet Die) position is obviously offset and not centered and so on, which further leads to poor heat conduction and heat dissipation under extreme conditions. Therefore, the author will use AM5 offset fasteners, CPU open cover + AM5 offset fasteners, CPU open cover + replacement of high-performance top cover and so on, to look at the different ways of heat dissipation. What is the difference in cooling performance of AMD Ryzen 9 7950X processor under PBO Auto setting?

 

Currently, the mainstream opinion on the Internet is that AMD processors have more serious heat buildup:

1. PBO (Precision Boost Overdrive) Mechanisms
2. CPU Top Cover IHS (Integrated Heat Spreader) Design and Thickness
3. CCD (Core Chiplet Die) Heat source position shifted southward

 

In fact, this test theme is provided by a friend of mine, originally this friend aka 錯字王 is want to run these test items, but he became a poor social animal this year to work overtime to not have their own life, so I was transferred to run this project, this post was developed by the settings and test items are specified by him, because he wanted to see these data.

Once again, a reminder! The action of uncovering the CPU processor "100% is a human damage action and will absolutely lose the warranty", it is not advocated, recommended, or recommended for anyone to uncover their CPU to avoid losing the product warranty.

 

Testbed Introduction

The test was conducted using an AMD Ryzen 9 7950X processor with two CCDs and one I/O Die, and an ASRock X670E Taichi Carrara motherboard. The processor was not additionally overclocked or overclocked, and the test was conducted in the motherboard's Auto setting, which was specified by the lender (the sponsor of the project) to be in the Precision Boost Overdrive (PBO)_Auto setting. The lender (project sponsor) specified the PBO (Precision Boost Overdrive)_Auto setting for the test.

ASRock X670E Taichi Carrara motherboard was updated to the latest BIOS version 2.02 at the time of author's test. The motherboard only enabled the XMP Profile in the BIOS for Kingston FURY Renegade DDR5 RGB 6000 MT/s 16GB x2, and manually set the fan and water pump to full speed in the cooler. The motherboard will only enable the XMP Profile in the BIOS, and manually set the fan and water pump of the cooler to full speed, and keep the rest of the motherboard in Auto by default.

The test software uses the Cinebench R23_Multi Core multi-core rendering test to perform a short stress test and a 30-minute long stress test, and HWiNFO64 collects the maximum CPU (Tctl/Tdie) temperature, the maximum power consumption of the CPU package, and the frequency during the test.

Cooler Master CryoFuze CF14 was used throughout the entire test process, with a thermal conductivity of 14 (W/m.K), one of the few coolers with a higher thermal conductivity currently available in Taiwan.

 

Testing Platform

Processor: AMD Ryzen 9 7950X (PBO Auto)
Cooler: Noctua NH-D15 / LIAN LI Galahad II LCD 360 (full speed)
Water-cooled fan: LIAN LI UNI FAN P28 x3 (full speed)
Thermal Balm: Cooler Master CryoFuze CF14
Motherboard: ASRock X670E Taichi Carrara(BIOS version: 2.02)
Memory: Kingston FURY Renegade DDR5 RGB 6000 MT/s 16GB x2(XMP On)
Graphics: MSI GeForce GTX 970 GAMING 4G Golden Edition
Operating System: Windows 11 Professional 22H2
System Drive: WD BLUE 3D NAND SATA M.2 2280 SSD 500GB
Case: STREACOM BC1 Benchtable V2
Power Supply: MONTECH Century 850W
Graphics driver: GeForce Game Ready 546.17

∆ CPU-Z platform view.

∆ AMD Ryzen 9 7950X, Kingston FURY Renegade DDR5 RGB 6000 MT/s 16GB x2 Open XMP profile.

∆ PBO(Precision Boost Overdrive)_Auto.

 

Noctua NH-D15 Dual Tower Air Cooling Cooling Test

First of all, we are going to use the classic air-cooled model with dual towers, dual fans and six heatpipes: the Owl Noctua NH-D15. Why did we use the Noctua NH-D15 to join the test? Because Owl has launched the Noctua AMD offset mounting bars explained this year, that is, the AM5 offset mounting bars can be purchased and used.

As mentioned at the beginning, the AMD Ryzen 7000 series processor heat points are significantly shifted to the south compared to the center of Intel processors, and AMD has separated the package into two major blocks, the I/O Die and the CCD (Core Chiplet Die), with some of the AMD Ryzen 7000 series processors (including models starting from 7700X and below) having only one CCD. Some AMD Ryzen 7000 series processors (including models starting with the 7700X and below) have only one CCD, while higher-end processors have two CCDs, such as the Ryzen 9 7950X, which has a combination of two CCDs and an I/O Die.

The Noctua AMD offset mounting bars explained that by moving the bottom mounting base -7 mm south/down (towards the M.2 SSD mounting position), the Noctua cooling tower further achieves better heat dissipation, but according to the Noctua AMD offset mounting bars, the Noctua AMD offset mounting bars will not be as good as the Noctua AMD offset mounting bars. Noctua Official ArticlesThe difference in heat dissipation is only 1~3°C, says Malaysian Youtuber: "The difference in heat dissipation is only 1~3°C," says Malaysian Youtuber.Kenny Cool KENNYThe difference is really not that big." I've also tested it in real life.

∆ AMD Ryzen 9 7950X Heating Point Schematic, image from: Owl Noctua website.

∆ vs. Intel hotspot from: It should be a movie from China 51972.

 

Using an unpacked AMD Ryzen 9 7950X processor with Noctua AMD offset mounting bars explained (NM-AMB12) to simulate the NH-D15's original thermal installation, the unoffset NH-D15 held down the maximum temperature to 92.8 °C and the maximum power consumption to 230 W in a round of Cinebench R23_Multi Core tests at 5 Gigahertz. In a round of Cinebench R23_Multi Core (multi-core) testing, the unoffset NH-D15 suppressed the maximum temperature to 92.8 °C, with a maximum power consumption of 230 W and an AMD Ryzen 9 7950X clocked at 5 GHz.

Under the 30-minute long term Cinebench R23_Multi Core stress test, the maximum temperature of the unbiased NH-D15 is 95.4 °C, the maximum power consumption is 220 W, and the AMD Ryzen 9 7950X is clocked at 4.9 GHz. Under short term rendering, the "unbiased NH-D15" is able to keep the Ryzen 9 7950X below 95 °C, but the long term stress test is not able to do so. NH-D15" was able to keep the Ryzen 9 7950X below 95 °C in short-term rendering, but not in long-term stress tests.

∆ Noctua AMD offset mounting bars explained(NM-AMB12) Unoffset mounting setup that simulates the original heat dissipation mounting of NH-D15.

∆ NH-D15 Unoffset test, R23_Multi Core (Multi Core) test result: 37724 pts.

 

Noctua NH-D15 Dual Tower Air-Cooled / Plus -7 mm Offset Buckle Cooling Test

Next, the Noctua NH-D15 is tested with Noctua AMD offset mounting bars explained (NM-AMB12). The test platform will mount the Noctua NH-D15 directly on the Noctua AMD offset mounting bars explained (NM-AMB12) and use the “-7 mm” holes on the NM-AMB12 for temperature test comparison. The test platform will mount the Noctua NH-D15 directly onto the Noctua AMD offset mounting bars explained(NM-AMB12) and use the "-7 mm" holes on the NM-AMB12 for temperature test comparison.

∆ Noctua has introduced various AMD offset mounting bars explained for its various air-cooled models, and the NH-D15 is to be paired with the NM-AMB12.

∆ Displacement direction indication for offset fasteners.

∆ The entire centerline moves down -7 mm.

 

The Noctua NH-D15 was offset tested with Noctua AMD offset mounting bars explained (NM-AMB12), and in a round of Cinebench R23_Multi Core tests, the offset NH-D15 reached a maximum temperature of 93.1 °C and a maximum power consumption of 229 W with AMD Ryzen 9 7950X at 5 GHz.

"The temperature alone has increased by 0.3°C instead of decreasing, but I personally think this is within the margin of error and as mentioned earlier, even Noctua's official use of the NH-D15 with the 190 W Ryzen 9 7950X only decreased the temperature by 3.1°C. Different test scenarios and platforms can't be compared, and my results only represent the difference between my platform and the environment, not a direct comparison. The results are not directly comparable.

But there is one "small detail"! The R23_Multi Core result for the offset test has improved slightly from 37724 to 37991 pts.

Under the thirty minute long Cinebench R23_Multi Core stress test, the offset NH-D15 has a maximum temperature of 95.5 °C and a maximum power consumption of 222 W. The AMD Ryzen 9 7950X is clocked at 5 GHz, which is a slight increase in both power consumption and temperature.

∆ -7 mm offset mounting locking holes.

The ∆ temperature did not drop, but the R23 score increased a bit.

 

LIAN LI Galahad II LCD 360 thermal test

The next step is to use the stock AMD Ryzen 9 7950X processor with Asetek's 8th generation pump solution. Galahad II LCD 360(GA II LCD 360) All-in-one water-cooling, and replace the cooling fan with three LIAN LI UNI FAN P28, also set the PUMP and fan to full speed for testing.

Currently, the Asetek 8th generation water pump solution is still the most popular water-cooling solution. In the short R23 rendering test, the 7950X was only 88.6°C, and during the test, the frequency was 5.1 GHz with a maximum power consumption of 233 W. The R23 result was 38,062 pts, so it's true that water-cooled cooling is still much more simple and violent.

The GA II LCD 360's maximum temperature came in at 91.4 °C under the 30-minute Cinebench R23_Multi Core stress test, which is slightly higher but still much better than air-cooling, and the maximum power consumption was 232 W and the frequency was 5 GHz, both of which were slightly lower.

∆ with 360 AIO has increased both cooling performance and results.

 

AMD Ryzen 9 7950X processor open-cap surgery performed!

After testing the Ryzen 9 7950X in its original form, it was time to craniotomize... I mean craniotomize, for extreme overclockers and temperature sensitive users, it's always a good idea to craniotomize the processor, either by letting the CPU's Die touch the heatsink or replacing the top cover of the IHS to get better cooling efficiency.

Here with the audience to supplement the entire processor usually installed in the section schematic diagram as follows, the top Heat Sink is your usual use of the processor, such as air-cooled tower cooling, integrated water-cooled, down-blowing radiator and other equipment, the following TIM 1 is our usual use of thermal paste, thermal paste is the majority of people will use as the outer layer of the thermal conductivity of the medium TIM (Thermal Interface Material). Thermal Interface Material (TIM).

The Heat Spreader is the IHS (Integrated Heat Spreader) mentioned in the introduction. The IHS is used to protect the CPU Die (Chip) underneath to avoid the processor from being blown up by the pressure of the heatsink grommets, and the gray cover that you see printed on the processor with the words Ryzen 9 7950X or i9-14900K is the IHS. The gray cover you usually see printed on the processor with the words Ryzen 9 7950X or i9-14900K is the IHS, the copper top cover of the IHS is usually nickel plated copper to avoid oxidation of the copper or chemical reaction with other materials and that's why the gray top cover you see on the processor.

The inner thermal conductor (TIM 2) under the IHS is responsible for transferring the high heat from the CPU die to the IHS, and then through the outer thermal conductor (thermal paste / schematic TIM 1) to the heat sink. The inner thermal conductor (TIM 2) used in the AMD Ryzen 9 7950X uses a multi-layer, multi-material alloy thermal conductor, which is composed of gold, platinum, vanadium, titanium, and other materials. The alloy thermal conductive layer consists of various materials such as gold, alloy, vanadium, titanium, etc. The method of applying this multi-layer, multi-material alloy thermal conductive layer to the inner thermal conductive dielectric is commonly known as the soldering process, and the PC community often refers to the soldering process when they say that a processor is a soldered CPU.

∆ Processor cutout schematic.

 

Processor Craniotomy (Capping) Surgery Performed by: Me, Patient: AMD Ryzen 9 7950X, Surgical Gear: Thermal Grizzly Ryzen 7000 Delid-Die-Mate (cost $79.99).

 

Things to prepare for this opening

• Thermal Grizzly Ryzen 7000 Delid-Die-Mate
• AMD Ryzen 9 7950X
• IKEA electric screwdriver
• Plastic Scraper
• Polishing block (sanding sponge)
• abrasive
• Metal Polishing Paste
• WD-40 SPECIALIST Precision Electrical Cleaner
• Liquid Metal

∆ AMD Ryzen 9 7950X has 16 cores and 32 threads, new price 18150.

The ∆ Thermal Grizzly Ryzen 7000 Delid-Die-Mate, which doesn't look quite the same as the Intel processor.

∆ Go to the execution platform.

 

Again, I do not recommend, suggest, or promote opening the cover of the processor, and opening the cover 100% the CPU will be ruled out of warranty by the original manufacturer due to acts of human damage.

The Thermal Grizzly Ryzen 7000 Delid-Die-Mate is a cover opener for AMD Ryzen 7000 series processors. Unlike the cover opener used by Intel, the Thermal Grizzly Ryzen 7000 Delid-Die-Mate repeatedly pushes and squeezes the IHS to open the cover. The Thermal Grizzly Ryzen 7000 Delid-Die-Mate, unlike Intel's, keeps pushing the IHS side to side to complete the opening.

I didn't have to heat up the processor to open the Ryzen 9 7950X, because the Ryzen 7000's fixing vinyl only coats the corners, top and bottom, which is pretty good for opening the lid. I've even seen Koreans open the lid with dental floss directly .....

Thermal Grizzly Ryzen 7000 Delid-Die-Mate comes with a free L-shaped hexagonal screwdriver that can be used by hand, but the author suggests to prepare an electric screwdriver for use, so I prepared an IKEA small electric screwdriver! Start ~ the result is to get to half of the electric screwdriver no power, causing me to hand-rotate to get an hour and a half strongly recommend that you use the electric screwdriver before charging.

∆ Thermal Grizzly Ryzen 7000 Delid-Die-Mate traces after use.

∆ IHS push marks.

 

After an hour and a half of hand spinning hell the electric screwdriver was finally charged and ready to go in 10 minutes.

∆ After opening the cover.

The CPU Die layout of ∆ 7950X has two CCDs and one IO Die.

∆IHS uses a soldering process to remove heat from the CPU Die.

∆ IHS nickel-plated copper top.

 

Then, let's install and test the 7950X to see if there are any problems with the processor, and power up the machine! Power up! Take it easy .... Boss? Why isn't it powering up?

∆No response at all after installing without powering up ....

∆ Finished, branding contest.

 

Then I used WD-40 SPECIALIST Precision Electrical Cleaner to spray it over and then it lit up and booted up smoothly, there should be some small dust on the processor joints that caused it not to boot up, and then I disassembled it again to remove the alloy heat-conducting layer on the CPU Die, I see that a lot of people will use liquid metal to paint on the alloy heat-conducting layer first, so that the liquid metal can help to soften the alloy heat-conducting layer before wiping it off. Then wipe it off.

I also use this procedure, but I personally don't recommend it because it's very easy to wipe the liquid metal beyond the CPU Die, which means you have to clean it again with Precision Electrical Cleaner to make sure there's no liquid metal left before repeating the removal process.

I later used a polishing block (sanding sponge) and abrasive paste, metal polishing paste, direct sanding and polishing program is much more efficient, but this also suggests that the abrasive paste and metal polishing paste do not buy too water-based, choose a higher viscosity is better.

∆ 1000 mesh sanding.

∆ 4000 mesh sanding.

∆ Cleaned with WD-40 SPECIALIST Precision Electrical Cleaner after polishing! One of the CCDs seems to be oversharpened.

The ∆ AM5 IHS should also be scraped of any residue on it.

∆ IHS High Display.

 

Noctua NH-D15 Dual Tower Air-Cooled Direct Contact CPU Die / Plus -7 mm Offset Fixture Cooling Test

After the open-cap grinding was completed, the heat dissipation performance test was continued by switching back to the Noctua NH-D15 and Noctua AMD offset mounting bars explained (NM-AMB12) offset fixture combination for the test, as well as Noctua's direct-contact fixture set for the open-cap processor for the test.

In the middle of this year, Noctua Owl cooperated with the famous extreme gamer der8aue to launch the “NM-DD1”. In fact, the NM-DD1 was exhibited at COMPUTEX 2023, and the NM-DD1 is applicable to the AMD AM5 processor with the "open cover" on Ryzen 7000 series. The NM-DD1 is suitable for "after cover" AMD AM5 Ryzen 7000 series processors. The NM-DD1 fixing fastener set is used to install Noctua's CPU cooler on the AM5 processor after the cover is opened. Noctua explained to the author that after the cover is opened, the temperature can be lowered by up to 10~15°C by touching the CPU die directly with the NM-DD1 fixing fastener set.

In addition to the Noctua NH-D15, which requires the addition of the NM-DD1 Retainer Kit, the Thermal Grizzly AMD Ryzen 7000 Direct Die Frame is also required for open-cap Ryzen 7000 processors, replacing the CPU Retainer Kit that was originally installed on the motherboard to hold the open-cap CPU in place, and preventing the risk of the CPU Die being blown out by the pressure of the heat sink. It also prevents the risk of the CPU Die being blown out by the pressure of the heat sink.

The black one in the center of the ∆ is a replacement for NM-DD1, which is used with the CPU after the cover is opened.

∆ The black frame that holds the processor in place after the cover is opened is the Thermal Grizzly AMD Ryzen 7000 Direct Die Frame, which is used to secure the processor and reduce the pressure on the CPU Die to prevent the CPU Die from being directly crushed.

 

In the middle of the year, Noctua provided 10~15 °C lowering data based on der8aue's "How this 120$ Noctua Cooler beats a 360 AIO" video, which uses open-cover direct contact with liquid metal to perform the test results, and in the past, we all know that open-cover direct contact with the CPU can significantly lower the temperature, but also because of the extreme gamers who use open-cover direct contact with "liquid metal" as a thermal conductor to test the results. We all know that opening the CPU cover and touching it directly can greatly reduce the temperature, which is also tested by the extreme gamers who use "liquid metal" as a heat-conducting medium after opening the cover.

But since der8aue has already done this test, I don't want to repeat it again (it's a bit boring). For those who are interested, you can just watch the video directly, and I've changed the CPU to use Cooler Master CryoFuze CF14 thermal paste as a thermal conductor to see how it works with less IHS.

∆ covers 7950X, Noctua AMD offset mounting bars explained (NM-AMB12), Thermal Grizzly AMD Ryzen 7000 Direct Die Frame, Noctua NH-D15, Noctua NM-DD1, Cooler Master CryoFuze CF14.

∆ Since it's kind of boring to look at everyone with liquid gold after opening the cap, I switched to a cooling cream to see how it works.

The ∆ Cooling Paste squeezes out very well with no problems at all.

 

However, the cooling seems to be even worse than the 7950X original, with the R23 reaching a maximum temperature of 95 °C in one round of testing and a multi-core score of 37,508 pts, which is even hotter and a little lower.

But this is entirely to be expected, and it is most intuitively obvious that the thermal conductivity of thermal paste is in no way comparable to that of the soldering process and the copper top, and even less so to that of a liquid metal.

Another possibility is that the total area of heat conduction after the CPU Die touches the heatsink directly is smaller than the IHS, resulting in a lower cooling effect. Combining the two possibilities mentioned above, the cooling performance of the open-cover direct contact with thermal paste is not ideal.

∆ Combination of open-cover direct contact plus thermal paste up to 95 °C Multi-core performance 37508 pts.

 

Noctua NH-D15 Dual Tower Air-Cooled Replacement High-Performance CNC Top Cover w/ -7mm Offset Fixture Thermal Testing

Next, the 7950X is coated with Thermal Grizzly Conductonaut Liquid Metal as the inner TIM and replaced with the Thermal Grizzly AM5 High Performance Heatspreader top cover, CNC's nickel-plated copper Thermal Grizzly AM5 High Performance Heatspreader top cover that increases the surface area by 240 %. Performance Heatspreader top cover can increase the surface area by 240 %, which can increase the heat dissipation performance, but it will also use more heat dissipation paste.

∆ Apply Thermal Grizzly Conductonaut Liquid Metal, but I couldn't find any solder mask or green oil to protect the sides, so I used paper tape for the time being (it's just for a little while anyway, but I don't recommend it).

∆ Noctua AMD offset mounting bars explained(NM-AMB12) -7 mm Installation, Thermal Grizzly AM5 High Performance Heatspreader.

 

The surface of the Thermal Grizzly AM5 High Performance Heatspreader top cover is quite delicate, and the surface area of the 240 % top cover has been increased considerably, but this increase in surface area also means that more heat spreading paste has been used.

∆ The amount of cooling cream to be applied is really too much.

∆ D15 Pressure paste condition.

∆ Paste condition of this side of the top cover.

 

The TIM (Thermal Interface Material) between the CPU Die and the top cover is made of Thermal Grizzly Conductonaut Liquid Metal, and the top cover is replaced with a Thermal Grizzly AM5 High Performance Heatspreader, which is tested by Cooler Master CryoFuze CF14 as the TIM (Thermal Interface Material) between D15 and D15. Cooler Master CryoFuze CF14 was used as the TIM (Thermal Interface Material) between D15 and D15 for thermal performance testing.

With a maximum temperature of 91 °C in the R23 round, the AMD Ryzen 9 7950X clocked at 5.1 GHz, the second best combination of temperature and clock performance so far, with an R23 score of just under 38,000 pts.

∆With the Liquid Metal and Thermal Grizzly AM5 High Performance Heatspreader top cover, the D15's cooling and performance is close to that of the 360 AIO.

 

LIAN LI Galahad II LCD 360 with Thermal Grizzly AM5 High Performance Top Cover Thermal Testing

After seeing the performance of the Noctua NH-D15 with the Thermal Grizzly AM5 High Performance Heatspreader top cover and the Thermal Grizzly Conductonaut Liquid Metal inside the top cover, I am looking forward to seeing how the LIAN LI Galahad II LCD 360 all-in-one watercooling with this combination!

After an actual round of R23, the AMD Ryzen 9 7950X topped the 95.1 °C temperature wall and was significantly downclocked to 4.5 GHz and 173 W. The test software R23 scores also dropped to 34,391 pts.

The ∆ temperature performance was significantly lower, and the R23 results also dropped significantly.

 

I don't think so. It's not possible that the cooling performance of Noctua NH-D15 has been improved by using this combination, and then dropped dramatically when I switched to LIAN LI Galahad II LCD 360, right? After troubleshooting and reapplying the thermal paste several times, the problem is still the same.

Finally, I found that the LIAN LI Galahad II LCD 360 moved a little bit when I pushed the water cooler head by hand with the fastener nut locked down, which is very strange, normally it should not move, so I pushed the water cooler head down with my hand and my weight and ran it again, and the temperature was 95.3 °C, but the power consumption and the frequency increased a little bit, and the R23 result increased a little bit to The R23 results increased slightly to 36854 pts.

∆ After using a water-cooled head with weight suppression, the results have recovered a little but still not normal.

 

If the cooling performance is a little bit better by compressing the water cooling head with weight, then it is obvious that the water cooling head doesn't fit perfectly on the Thermal Grizzly AM5 High Performance Heatspreader top cover, but I have disassembled and reassembled the LIAN LI Galahad II LCD 360 so many times that I am sure that all four screws are locked to the top cover. I'm pretty sure all four screws are locked in place.

Later on, I found out that the height of the Thermal Grizzly AM5 High Performance Heatspreader top cover is lower than the original IHS height of the Ryzen 9 7950X, and the Asetek eight-generation pump solution grommet set used in the LIAN LI Galahad II LCD 360 is adjusted only for the original IHS height, so it can't reach a lower depth in terms of screws, studs, and fasteners. The Asetek 8-generation water pump solution fastener set used in the LIAN LI Galahad II LCD 360 is only adjusted for the stock IHS height, so it is not possible to achieve lower depths at the heights where screws, studs, and fasteners can be fixed.

∆Thermal Grizzly AM5 High Performance Heatspreader Top Height Problem Point Schematic Reference Diagram, the screws, studs, and fasteners, even when tightened to the maximum, still do not allow the cooler head to fully contact the high performance top cover, thus causing the LIAN LI Galahad II LCD 360 to underperform in this combination.

Schematic diagram of ∆ Asetek's eight-generation water pump solution AM5 fastener set.

 

I found out that der8auer, the owner of Thermal Grizzly, used another kind of Thermal Grizzly AM5 Adapter & Offset Mounting Kit in the video introducing the Thermal Grizzly AM5 High Performance Heatspreader top cover. This screw set can be used for the Thermal Grizzly AM5 High Performance Heatspreader top cover which is lower in height and locks the water cooler a little lower than the original fasteners, but we missed this screw set this time, so the water cooler will be mounted on the top cover of the Thermal Grizzly AM5 High Performance Heatspreader. Performance Heatspreader top cover, we have to give up on this combination.

 

When using the Thermal Grizzly AM5 High Performance Heatspreader top cover with an integrated water cooler, ∆ der8auer has to use it with the Thermal Grizzly AM5 Adapter & Offset Mounting Kit in order to comply with the mounting height.

 

LIAN LI Galahad II LCD 360 with original top cover / PTM7950 phase shifter replacement thermal test

Next is a small test combination for my personal curiosity, if the TIM (Thermal Interface Material) in the AMD Ryzen 9 7950X CPU Die with the stock IHS is replaced with Honeywell PTM7950 Honeywell Phase Change Heat Sink and the same full speed LIAN LI Galahad is used. II LCD 360 all-in-one water-cooled cooler, I wonder how effective it is?

However, the thermal conductivity of the Honeywell PTM7950 phase shifter must not be as good as the soldering workmanship, and according to the project sponsor, he cannot guarantee that this phase shifter is a genuine Honeywell PTM7950 because he bought it from Shrimp.

∆ Using Honeywell PTM7950 Honeywell Phase Change Thermal Conductors under AM5 original IHS (not sure of authenticity).

 

Just by replacing the TIM (Thermal Interface Material) of the CPU Die and the inner layer of the IHS with Honeywell PTM7950, the highest temperature in one round of R23 test was 92.4 °C and the multi-core result was 37690 pts, which is about 4 °C higher than the original Ryzen 9 7950X and LIAN LI Galahad II LCD 360, and the multi-core result was about 372 pts lower. Ryzen 9 7950X and LIAN LI Galahad II LCD 360 are about 4°C higher in thermal performance, and the multi-core result is also about 372 pts lower, which means if you just replace the TIM (Thermal Interface Material) of the inner layer of the IHS with Honeywell PTM7950, you will be able to get a better thermal performance. That is to say, if you simply replace the TIM (Thermal Interface Material) thermal media on the inner layer of the IHS with Honeywell PTM7950, the heat dissipation performance will not be as good as the original soldering process.

After replacing the ∆ internal TIM with Honeywell PTM7950 phase shifter, the heat dissipation performance is not as good as the original soldering process.

 

Cinebench R23_Multi Core Rendering One Round Temperature Test Chart

Next, we have compiled the maximum temperature, maximum power consumption, and frequency of AMD Ryzen 9 7950X processor read by HWiNFO64 after one round of Cinebench R23_Multi Core multi-core rendering for all the test items, and organized these three information into a chart for your reference.

A round of Cinebench R23_Multi Core rendering with the AMD Ryzen 9 7950X usually finishes rendering an image in about a minute, so it's more in line with most short rendering scenarios where the stress time is shorter and therefore you can see optimal cooling from the various coolers.

The best cooling performance is still the unopened LIAN LI Galahad II LCD 360 (full speed) combination, which is the only combination in the room with a maximum temperature below 90 °C. Perhaps the cooling performance can be further improved by replacing the internal TIM with Liquid Metal after the cover is opened, but this time, unfortunately, the combination of the GA II 360, Liquid Metal, and Thermal is not implemented. Grizzly AM5 High Performance Heatspreader top cover.

In second place is the combination of Thermal Grizzly AM5 High Performance Heatspreader top cover, Liquid Metal, Noctua NH-D15, and offset grommets with the cover on, which are also the only two combinations capable of running the Ryzen 9 7950X at 5.1 GHz in Cinebench R23_Multi Core multi-core rendering.

∆ Cinebench R23_Multi Core Multi Core Rendering One Round Temperature Test Chart.

 

Cinebench R23_Multi Core Rendering Thirty Minute Temperature Chart

The second test scenario is a Cinebench R23_Multi Core multi-core rendering run for 30 minutes, this scenario is a longer rendering stress test so the performance of the heatspreader will be even more demanding, especially with the Thermal Grizzly AM5 High Performance Heatspreader top cover. For the GA II LCD 360 project, because of the low height of the top cover, the heatspreader could not fully fit on the top cover, so the author did not conduct the long time stress test here.

The best cooling performance in the long term stress test was still the same as the short term, the unopened LIAN LI Galahad II LCD 360 (full speed) with the D15+ high performance top cover combination, but the maximum temperatures for all combinations were slightly higher, with all combinations exceeding 91°C. The maximum temperature for all combinations exceeded 91°C. The best cooling performance for the long term stress test was still the same as the short term.

∆ Cinebench R23_Multi Core Multi Core Rendering Thirty Minute Temperature Test Chart.

 

Cinebench R23_Multi Core Rendering Test Results Chart

A round of Cinebench R23_Multi Core multicore rendering was performed and the multicore results obtained by the AMD Ryzen 9 7950X were also recorded, and the ranking is still similar to the thermal performance ranking.

∆ Cinebench R23_Multi Core Multi Core Rendering Test Results Chart.

 

Conclusion

Once again, a reminder! The action of uncovering the CPU processor "100% is a human damage action and will absolutely lose the warranty", it is not advocated, recommended, or recommended for anyone to uncover their CPU to avoid losing the product warranty.

This time, there are seven different cooling combinations after opening the cover, the whole process of opening the cover is quite cumbersome and troublesome and may even damage the processor, and different air-cooled and water-cooled coolers have to buy different related fastener sets and peripheral accessories, and the cost of buying a bunch of things is also not low.

Taking the Noctua NH-D15 air-cooled tower as an example, I feel that the Noctua AMD offset mounting bars explained (NM-AMB12) is not a must-have, and looking at the various unboxing tests, the actual temperature performance is not that much different; and even in the author's test scenario, the temperature did not drop but increased (only for the platform I have in my hands). I can't guarantee the same for other platforms, but the R23 did show a slight increase in performance.

If you have already opened the cover of the processor, the author would recommend using liquid metal for both the inner and outer layers of the TIM (Thermal Interface Material), which would be more in line with the cost of the application, because the thermal conductivity of the thermal paste or the phase change sheet has a big difference with the original soldering process. But this part is left to the discretion of the extreme players, after all, the liquid metal after treatment is also very troublesome and will lead to electricity.

Replacing the AMD Ryzen 9 7950X's original IHS (Integrated Heat Spreader) nickel-plated copper top cover with the Thermal Grizzly AM5 High Performance Heatspreader top cover does provide a little bit of a thermal performance boost (240 % of additional thermal area after all). ), so if you have enough money in your pocket, the author suggests that you can buy it across the sea, but it's really a waste of thermal paste.