View from the Labs
Intel’s new architecture is doing something significantly different, but I suspect it won’t have AMD quivering nervously at the knees
After years of churning out iterative updates to its architecture, the threat from both AMD and Apple has proven too much for Intel, and it has radically redesigned its new chips. Opting for a combination of big and little cores (Pand E-cores), Intel’s 12th generation processors are fast and efficient.
The disappointing news is that
Intel doesn’t use E-cores on all models: drop down to the Core i5-12600 and below, and there aren’t any. Sure, there’s a performance increase over the last generation, but the difference is far less pronounced.
Where a 12th gen chip has E-cores and P-cores, the performance is hard to ignore. That can be seen in the main benchmark results, where the Core i9-12900K largely outdoes everything AMD has with fewer cores and threads; even the Core i7-12700K outperforms a lot of AMD’s chips for less money.
So, is it curtains for AMD? Not quite yet. For starters, it’s working on its next-generation Zen 4 architecture, coming via the AM5 socket later this year. Its current range of processors is also still pretty good, particularly at the high end and entry-level.
AMD’s advantage has been that it’s simply thrown more cores at the problem. Take the Ryzen 9 5900X, a well-priced CPU with 12 cores and 24 threads. To get the same on an Intel processor, you must spend £100 more.
At the entry level, the Ryzen 5 5500 has six cores and 12 threads; Intel’s Core i3 chips have just four cores. While each Intel core may be more efficient than AMD’s, the simple fact is that more cores are generally better for modern applications.
What’s more, the current range of AMD chips can be installed in older AM4 motherboards, making for a simple and cost-effective upgrade; with Intel, you have to buy a new board at a minimum.
AMD’s architecture generally has more L3 cache than its Intel counterparts. While the Core i9-12900K has 30MB of smart cache, even the budget Ryzen 5 5600 has 32MB of L3 cache. In many cases, AMD chips have even more: the 5800X3D squeezes 92MB of cache onto one chip.
While more cache doesn’t make a huge difference in most applications, it does help AMD consistently perform well in gaming benchmarks. Games benefit from fast access to objects, which is what lots of L3 cache does. Look at the gaming benchmarks below and the results are far closer than with general applications, with one exception: the 5800X3D is miles ahead in Shadow of the Tomb Raider.
Intel is currently leading the way overall, but the fact that AMD’s significantly older chips hold their own mean that it can still walk with a swagger. Once its new generation comes along, I suspect that swagger will turn into a cocky strut.
“Is it curtains for AMD? Not quite yet. For starters, it’s working on its nextgeneration Zen 4 architecture”