AMD Ryzen Threadripper Waterblock Comparison #1

We have been waiting for AMD Threadripper CPU custom cooling parts to make their way to us. We have our first two purpose-built Threadripper waterblocks from XSPC and Bykski. We put both these coolers to the test with our 4GHz overclocked Threadripper in hour long stress tests to see how our temperatures fare.

continued...

XSPC TR4 RayStorm and Bykski A-RYZEN-Th-X Results

So how did our new Threadripper waterblocks perform? Actually, both of these blocks worked extraordinarily well. Again, keep in mind that we are looking for long term stability, not a quickie benchmark score that some will point to when suggesting they got their system "stable." Our results below are obtained by running the Threadripper for an hour using Prime95 with Small FFTs. We monitor the ambient temperatures on the test bench with Sperry thermocouples.

Article Image

We monitor radiator intake temperatures as well as radiator exhaust temperatures. Our intake temperatures fluctuate about 1.5 degrees F during testing (76F to 77.5F). I wish I was able to regulate this better, but that is simply not obtainable without a proper lab environment. This is another reason we run long workloads. We generally saw a 10F to 11F delta between intake and exhaust temperatures.

The temperatures noted below are peak temperatures during testing. Possibly in the future we may move to an average over time after the systems show to be heat-soaked, then start taking temperature data from there. However, these temperatures will surely make for good baselines in order to start our Threadripper waterblock testing. Note that I did lose my 1 hour screen shot on the Bykski block, but it had already peaked in temperature, hence the timestamp difference. That said, we generally saw peak temperatures reached at the 33 minute mark during the 1 hour Prime95 workload.

All the testing was run on the ASUS ROG Zenith Extreme. We were able to lock in low levels of CPU LLC in order to get a static vCore (1.439v) on our Threadripper CPU. Better cooling has helped dial that in significantly. (Review of that motherboard is on the way.)

Article Image

Article Image Article Image

Also worth discussing is how well these blocks unload heat. Below is a screenshot of the Threadripper temperature after 60 seconds of Prime95 being unloaded. The picture below is of the XSPC TR4 RayStorm block, but the Bykski gave similar results.

Article Image

Threadripper TIM...Again

The temperatures represented above were obtained using Prolimatech Pk-1 Nano Aluminum Thermal Compound. We did also run full tests on our Threadripper using my MG Chemicals Super Thermal Grease II "testbench TIM" as well. I use the MG on the bench because it is easy to apply, easy to clean up, and inexpensive...and I use a lot of it. However for temperature testing I would consider it not appropriate for the job. The MG STG II has a thermal conductivity of 1.78W/(m-K). The Prolimatech Pk-1 has a thermal conductivity of 10.2W/(m-K). That is not the highest in the industry, but it is up there with the best. The Pk-1 is non-corrosive, non-electrically conductive, and requires no burn-in time. The qualities along with its very high heat conductivity rating make it excellent for our test bench.

I did check for temperature deltas between our MG and Pk-1 TIM. Given Threadripper's huge surface area, it seemed as though it might be significant. Our delta was a little over 3.2C between TIMs with the Pk-1 being the better. I actually thought it would be bigger, but it the delta was pretty much the same with both XSPC and Bykski blocks.

The temperatures above were all obtained using our "Spread with 5 Dots" method and the Pk-1 TIM.

Overall Experience

I want to take a few paragraphs to discuss my subjective experiences. I will continue to use the XSPC RayStorm on our test bench due to the fact that it has delivered better temperatures.

I do however like the installation procedure on the Bykski waterblock better. It is much simpler in its function as you can see in our videos; no mounting bracket, no springs, not mounting posts, half the washers. The Bykski simply uses allen-keyed bolts that mount directly into the TR4 socket. Now the downside to all the simplicity is the fact that if you get out of hand with torquing that system down you could damage the socket/motherboard and you could screw up your mate as well. Worth remembering is that I am using these on an open test bench as well; once you get to working inside a case simplicity may not be "better." The XSPC system is more complicated, but certainly geared more towards safety and making sure you get a good mount as well as pressure on the waterblock is mitigated by the mounting system. The modular mounting system, included TIM, and LEDs, is one of the reasons the XSPC waterblock is going to be a bit more expensive than the bare-block Bykski. ($64.99 vs "sub-$100 per XSPC" for the RayStorm)

Also the Bykski block got coolant under the colored surface of the block while filling it. Not a great thing, and I have not yet tried to remedy this as I did not want to damage the block before testing was done. This was not a big deal for me, but most likely your waterblock would not be sitting flat on a test bench but rather vertically, which might be an issue with rogue moisture. The modularity of the XSPC TR4 RayStorm allows you to easily clean every part of the block easily and fully.

The Bottom Line

The XSPC TR4 RayStorm proved to be the superior waterblock in our testing here today. That said, the differences in temperature were very small. The RayStorm is the superior product overall. The Bykski TR4 block is a more stripped down no-frills version that goes for the "less is more" approach. Both of these waterblocks certainly have great value depending on what exactly your specific needs are.

At these links you can see some of the records we have accomplished with this system in Gooseberry and LuxMark rendering.

Article Image

XSPC TR4 RayStorm Neo / Bykski A-RYZEN-Th-X

Dicussion

If you like our content, please become a Patron for as little as $1 per month.


AM4 Testing with the XSPC TR4 Block

As mentioned earlier, our prototype RayStorm mounting system gives us the ability to use it on an AM4 X370 motherboard. The retail TR4 RayStorm will not have this ability.

We are in middle of revamping our air cooler and AIO review platform and are moving this to an AM4 platform. After getting our platform stable with the Thermalright True Spirit 140 Direct, I just wanted to see what delta the RayStorm would supply us. Just for fun.

Article Image Article Image

We turned out getting a 15C delta in temperature under full Prime95 and Furmark loads that showed our CPU package wattage up to 200 watts. We do have the XSPC AM RayStorm block in the office and will be running some comparison tests as well in the near future...just for fun.