Test methodology
What we've got is essentially a bigger, better Hydro Series cooler (aside from the pump, of course).
Therefore, in order to get an idea of performance, we're comparing the H70 with last year's H50 and two highly-regarded air coolers; the Noctua NH-D14 and Prolimatech Megahalems Rev B.
Using the coolers on a high-end system equipped with an Intel Core i7 980X processor, we'll also throw in temperature readings taken from Intel's reference cooler, too.
Why stop at that? Users already running last year's H50 will have one obvious question; should I upgrade to the H70 or add another fan to my existing cooler? To find out, we'll also provide H50 numbers with two 120mm fans.
That gives us a six-way comparison that breaks down as follows:
| Cooler | Corsair Hydro Series H70 | Corsair Hydro Series H50 | Corsair Hydro Series H50 | Noctua NH-D14 | Prolimatech Megahalems Rev B | Intel stock cooler |
|---|---|---|---|---|---|---|
| Fan(s) | 2x Corsair 120mm | Corsair 120mm | Corsair 120mm & Akasa Apache Black 120mm | Noctua NF-P14 & Noctua NF-P12 | 2x Akasa Apache Black 120mm | Intel 100mm |
| Approx. price (inc. fans) | £77 | £57 | £67 | £61 | £63 | £10 |
| Processor | Intel Core i7 980X (3.33GHz, 12MB L3 cache, hexa-core) | |||||
| Motherboard | ASUS P6X58D Premium Deluxe | |||||
| Motherboard BIOS | 0808 | |||||
| Memory | Corsair Dominator 6GB (3 x 2GB) DDR3 | |||||
| Memory timings | 8-8-8-24-1T @ 1,600MHz | |||||
| Graphics | AMD ATI Radeon HD 5850 | |||||
| Graphics driver | Catalyst 10.7 | |||||
| Disk drive | 120GB OCZ Vertex SSD | |||||
| Optical drive | Generic 24x DVD-RW | |||||
| Chassis | Corsair Obsidian Series 800D | |||||
| Power supply | Corsair HX1000W | |||||
| Operating system | Windows 7 Ultimate 64-bit | |||||
Benchmarks
Putting numbers to the names, we will be providing temperature readings in various scenarios.
We start by running the Core i7 980X at its default frequency, and record the temperature each minute for 30 minutes whilst the system idles. We then repeat the process with the system under load (defined as Prime95 tasking all 12 threads).
We then do it all over again, only this time with the Core i7 processor overclocked to a modest 4.125GHz and CPU voltage increased from 1.16V to 1.35V.
Temperature for each of the CPU's six cores is logged every minute, and in order to provide stabilised readings, we then calculate an average across all cores from the last 10 minutes of testing.
Actual CPU temperature is recorded, but to take into account the fluctuating British weather, we also graph the delta temperature (that's CPU temperature minus ambient temperature). Readers should note that ambient temperature fluctuated between 21.4°C and 23.8°C during testing.
Finally, we overclock the Core i7 chip to find the limits of the H70 cooler and provide a human-ear assessment on noise levels.
Notes
During testing, our Corsair Obsidian Series 800D chassis is equipped with its default fan configuration; a 140mm drive-bay fan, a 140mm lower compartment fan and a 140mm rear exhaust. The rear exhaust is removed and replaced by the Corsair radiator/fan when testing the H50 and H70 coolers.
All dual-fan coolers are installed in a push/pull configuration, and five of the six coolers are configured to exhaust hot air to the rear of the chassis. The only exception is the Noctua NH-D14, which, when used in a large 800D chassis with an ASUS P6X58D and Corsair Dominator memory, is simply too big to implement horizontally.
As a result, Noctua's gargantuan solution is configured vertically and has one fan in close proximity to the GPU. With no top exhaust attached to the chassis, we suspect the NH-D14 will deliver sub-optimal results in this configuration.
