Corsair AIO Liquid Cooler Round-Up: Ryzen Overclocking Explored
Corsair Hydro Series Test Setup and Intel Burn Test
Our test system revolves around the AMD Ryzen R5 1600X. To reflect a more typical mainstream gaming system we've outfitted it with an EVGA GeForce GTX 970 SSC, 16GB (2x8GB) Dual-Channel DDR4 3200MHz Geil EvoX RGB, a Toshiba OCZ RD400 512GB M.2 NVMe SSD, and ASRock Fatal1ty AB350 Gaming K4 motherboard. See the full system specs below:
All testing is performed on Windows 10 Pro Build 1709 with the latest BIOS, firmware, and drivers available. At the time of testing these were as follows:
On the testing front, our primary concerns are thermals and noise levels. We’ve selected Cinebench R15 to gauge pure CPU performance, PC Mark 10 Extended and 3D Mark Time Spy for mixed workloads, and a trio of stress tests with 3D Mark Time Spy’s Stress Test option, Intel Burn Test, and AIDA 64’s Stability Test.
Whenever we test thermals like this we have to see how far we can push an overclock and keep things cool. We've previously overclocked a Ryzen R5 1600X chip to 4.025GHz at 1.35V, however this time we wanted to push things a bit harder on a per cooler basis. Using the Ryzen Master Utility, we locked in our memory's profile at 3200MHz (16-16-16-36), then for each cooler dialed in the best speeds we could while verifying stability with the Intel Burn Test using a 40 loop run on High - a duration of about 30 minutes. We also worked up the voltage as we went until we settled at 1.425V. Our stable frequency results for each cooler are listed below:
The default all-core turbo speed with XFR and this processor is 3.7 GHz so in all cases we've managed to pick up some nice frequency gains. The H110i and H150i Pro were both able to deliver the highest clocks at 4.075 GHz. The H115i Pro could hold 4.075 GHz for a few minutes under Intel Burn Test, but would invariably fail until we dropped the clock speed to 4.05 GHz.
The other coolers did not fair so well. The H60 required speeds to be cut by another 25 MHz and the Wraith MAX cooler trailed the leaders by a whole 100 MHz.
While the Burn Test ran we also took note of thermals as reported in CPUID's HWMonitor application. Results below are measured in Celsius:
Under stock conditions, the H110i and H150i Pro deliver our best results with a peak sustained temperature of 51C. The H115i Pro lags slightly with a peak temperature of 53C but still strongly outperforms the H60 at 60C and Wraith MAX at a toasty 67C.
Once overclocked, the sequence mostly holds the same. The H150i Pro loses some ground to the H110i and holds the same temperature as the H115i Pro this time around, but the H115i Pro was unable to sustain the same clock speeds so that needs to be taken into account as well. The H60 held temperatures to about the highest level we'd be comfortable with in regular usage while the Wraith MAX registered a scorching 83C under load. That's a 22C increase over the H110i while also running at a 100 MHz lower frequency.
Intel Burn Test ramps up and down for each loop of the run, so we also tested using AIDA64 and Time Spy's Stress Test mode for a more complete picture...
| HotHardware Test Bench | |
| Processor | Ryzen R5 1600X (3.6GHz @ 95W TDP) |
| OS | Windows 10 Pro 64-bit (V1709 Build 16299.192) |
| Graphics | EVGA GeForce GTX 970 SSC (4GB) |
| Memory | 16GB (2 x 8GB) Geil EvoX DDR4-3200MHz |
| Storage | Toshiba OCZ RD400 512GB M.2 SATA SSD |
| Motherboard | ASRock Fatal1ty AB350 Gaming K4 |
| Power Supply | EVGA Supernova 850 G2L 850W PSU |
| Chassis | Streacom BC-1 OpenBenchTable |
All testing is performed on Windows 10 Pro Build 1709 with the latest BIOS, firmware, and drivers available. At the time of testing these were as follows:
| Version | |
| Windows 10 Pro x64 | 1709 (OS Build 16299.192) |
| AMD Ryzen Master | 1.0.1.0239 |
| Ryzen R5 1600X | 17.30 |
| GeForce GTX 970 Driver | 390.65 |
On the testing front, our primary concerns are thermals and noise levels. We’ve selected Cinebench R15 to gauge pure CPU performance, PC Mark 10 Extended and 3D Mark Time Spy for mixed workloads, and a trio of stress tests with 3D Mark Time Spy’s Stress Test option, Intel Burn Test, and AIDA 64’s Stability Test.
Whenever we test thermals like this we have to see how far we can push an overclock and keep things cool. We've previously overclocked a Ryzen R5 1600X chip to 4.025GHz at 1.35V, however this time we wanted to push things a bit harder on a per cooler basis. Using the Ryzen Master Utility, we locked in our memory's profile at 3200MHz (16-16-16-36), then for each cooler dialed in the best speeds we could while verifying stability with the Intel Burn Test using a 40 loop run on High - a duration of about 30 minutes. We also worked up the voltage as we went until we settled at 1.425V. Our stable frequency results for each cooler are listed below:
| Cooler | CPU Frequency |
| AMD Wraith MAX | 3.975 GHz |
| Corsair H60 | 4.025 GHz |
| Corsair H110i Extreme | 4.075 GHz |
| Corsair H115i Pro | 4.050 GHz |
| Corsair H150i Pro | 4.075 GHz |
The default all-core turbo speed with XFR and this processor is 3.7 GHz so in all cases we've managed to pick up some nice frequency gains. The H110i and H150i Pro were both able to deliver the highest clocks at 4.075 GHz. The H115i Pro could hold 4.075 GHz for a few minutes under Intel Burn Test, but would invariably fail until we dropped the clock speed to 4.05 GHz.
The other coolers did not fair so well. The H60 required speeds to be cut by another 25 MHz and the Wraith MAX cooler trailed the leaders by a whole 100 MHz.
While the Burn Test ran we also took note of thermals as reported in CPUID's HWMonitor application. Results below are measured in Celsius:
Under stock conditions, the H110i and H150i Pro deliver our best results with a peak sustained temperature of 51C. The H115i Pro lags slightly with a peak temperature of 53C but still strongly outperforms the H60 at 60C and Wraith MAX at a toasty 67C.
Once overclocked, the sequence mostly holds the same. The H150i Pro loses some ground to the H110i and holds the same temperature as the H115i Pro this time around, but the H115i Pro was unable to sustain the same clock speeds so that needs to be taken into account as well. The H60 held temperatures to about the highest level we'd be comfortable with in regular usage while the Wraith MAX registered a scorching 83C under load. That's a 22C increase over the H110i while also running at a 100 MHz lower frequency.
Intel Burn Test ramps up and down for each loop of the run, so we also tested using AIDA64 and Time Spy's Stress Test mode for a more complete picture...
