Intel Core i9 10900K
Zak Storey straps himself in, ready to ride Intel’s fastest chip yet!
Zak Storey straps himself in, ready to ride Intel’s fastest chip yet! The fresh Canon Lake architecture promises flaw-free, super-fast processing, but high power use.
We’re long past the days of monolithic chip designs. Now it’s all about providing a modular solution by carefully weaving core complexes together, and having different transistor sizes across different components. That way you create a processor that not only provides impressive performance, but is also affordable and simple to mass-produce.
This is where Intel has come unstuck in the great core war because its monolithic designs are far harder to produce, especially to add more cores to. To get around the loss in overall number of cores and lack of a die shrink, the company has been aggressively targeting clock speed to bolster its multi-core performance and hold on to the single-core IPC and gaming crown it has held for so long.
Intel’s 10th generation of processors is aptly named Cannon Lake, because its core architecture is not too dissimilar from that of its great-grandfather, Sky Lake (2015). The company’s made some significant changes not only to how it manufactures its chips, but also on a hard-encoded software level. This time around Intel’s introduced its new Turbo Boost Max Technology 3.0. The chip itself will identify which cores perform best, and then increase clock speed on those individual cores – up to 5.2GHZ on the Core i9-10900k – to improve performance in single- and dual-core applications. What’s impressive about this is that the chip doesn’t increase the voltage to do it. On top of that it also has its Thermal Velocity Boost tech. If temperatures and voltages are within certain parameters, the processor will push that clock speed further, increasing a single core to 5.3GHZ, or all cores up to 4.9GHZ at stock.
Turning up the heat
When Intel first reacted to AMD’S Ryzen processors, the simple answer was to add more cores. That in turn required more power and produced more heat. To continue that rising clock speed trend, Intel has introduced a “Thin Die STIM.” The CPU die itself is far thinner compared to previous generations, enabling Intel to use a thicker soldered heat-spreader across a wider surface to remove heat more efficiently.
On top of that it’s also packed in hyper-threading across the entire range (from the i3 up to the i9), improved native DDR4 support up to 2,933MT/S, included Wi-fi 6 AX and 2.5G Ethernet as standard, and of course introduced a new chipset, which supports up to 40 PCIE 3.0 lanes.
In the case of the flagship Core i9-10900k, all this comes together to produce a heck of a processor. In our testing we saw single-core Cinebench R15 figures up at 228, multi-core at 2,608, and an incredible x264 encode score of 54.64fps. Now don’t get us wrong, this thing is a hot chip, and it’s incredibly power-hungry, too. Intel rates the Core i9-10900k with a TDP of 125W. Under load our sample was pulling 301W from the wall, and reaching 85 degrees C with a 280mm Corsair H115i AIO.
From an engineering perspective, these chips are impressive. However, it’s still missing the mark in a few key areas. Although Z490 motherboards have support for PCIE 4.0, that’s not supported on 10th gen (which could be a deal-breaker for those future-proofing for next-gen GPUS and storage). The temperatures and power-draw for the flagship are incredibly high, and right now you can pick up the Ryzen 9 3900X, 12-core processor from AMD with a cooler for almost £100 less than the 10900K. It’s a tough one.
On paper the Core i9-10900k is stunning, but in the real world it’s tricky to justify. If Intel’s history is anything to go by, once it breaks that 10nm barrier, with an architecture as refined as this, we should be looking at one hell of a processor. We’re just not quite there yet.