The specs and benchmarks are out and once again AMD continues to hold the crown. AMD has really put faith into its chiplets over 3D stacking and it shows in the performance over their Intel rivals.
The war rages on
Intel has, for some time, absolutely dominated the CPU market. Their flagships - the i7 and now the i9, have consistently held larger parts of the processor market than their AMD counterparts. The problem is that AMD is fighting two wars. One for CPU performance and the other for GPUs. While they may be losing their war for GPU benchmarking – when it comes to CPUs both server grade and desktop, they’re starting to pull out ahead.
There’s also the speed of which AMD engineers seem to be able to create the technology. Released in November/December the chip that has now beaten the Intel benchmarks has already been available for four months. In CPU land that’s a lifetime, with new chips being announced roughly every 12 months, so AMD still has a significant advantage with its releases.
Cinebench is typically the most stringent test a CPU has thrown at it, as it renders at varying resolutions and uses all the cores available instead of single core processes. Generally speaking it means the CPU gets a good workout.
(source: Tom's Hardware)
With more cores, there’s always going to be a strong bias towards AMD here, but what’s stark is the fact that when using single core processes the AMD is almost as strong as Intel, with an almost unnoticeable difference. It’s when all the cores work together that the AMD shines. Proof that the chiplet design favoured over Intel’s 3D stacking is still paying dividends.
Typically it’s well known that Intel charge more for their chips – the same is true of its flagships tested above. AMD has always managed to come just under the Intel price range, mainly due to the fact they wanted to attract a bigger market share of those unwilling to pay the extra for the performance. Now however, the tables have turned in terms of performance and we can really see AMD taking larger market shares based on performance AND price.
This doesn’t necessarily mean the end for Intel, they’ll come back bigger and better. They’re still working with the 14nm and 10nm technologies, which they’re adamant they will squeeze every drop of performance out of. AMD made the jump some time to 7nm which also extended to their EPYC chips for servers, where Intel are still using 14nm for Xeon processors.
It means for cloud hosting though, that’s notoriously hard on processors due to the constant level of activity and variance in server loads throughout any period of time – that you’ll want a chip that can handle multiple processes on a consistent basis. AMD EPYC rule the roost here as Intel still go for a lower core count in favour of the 14nm. But this needs to change as 14nm uses a far greater amount of power and is consistently being outscored by the AMD counterpart in tests.
It’ll be interesting to see the next generation of Intel chips, but currently AMD reigns.
To speak to us about tech or anything else, please do get in touch.