Today we can finally show you how the new 9th generation Intel Octa-Core processors work. There is still a lot to talk about, so we won't be spending a lot of time checking the technical data, since the CPUs have been available for pre-order for 10 days and nothing in the technical data sheet is a mystery.
For testing, we have the Core i9-9900K on hand – in fact we have some of them, with more tests and comparisons in the works for later – and we also have the i7-9700K, which is basically the same CPU, but crucial, with disabled Hyper threading.
The Core i9-9900K is an 8-core processor with activated Hyper-Threading for 16 logical threads. It works with a base frequency of 3.6 GHz, but will increase on all cores with a maximum single core frequency of 5 GHz up to 4.7 GHz. The L3 cache has been increased from 12MB of the 8700K to 16MB. Despite packing 2 more cores and 4 MB more cache, the TDP rating remains at 95 watts, which was already a suspiciously low rating for the 8700K. We'll investigate the effects shortly.
The Core i7-9700K contains the same eight cores, but can only process eight threads at a time. It is clocked at the same base frequency of 3.6 GHz, while the all-core and single-core clock rates decreased by a marginal 100 MHz. The L3 cache capacity is also reduced to 12 MB.
For testing we use the MSI Z390 Godlike, but also the Asrock Z390 Taichi Ultimate to confirm the results. Both cards were tested with DDR4-3200 CL14 memory, and this memory was used on all platforms without manually tuned timings. The graphics card of choice is Gigabyte's RTX 2080 Ti Gaming OC. We have a lot of results ahead of us, so let's get started!
First, we have the memory bandwidth results and it is not surprising that the new Coffee Lake Refresh CPUs are comparable to previous models like the Core i7-8700K. So everything here is as expected. Let's look at some Cinebench results.
As expected, the 9900K and 9700K offer the highest standard single-thread values we've seen so far, and thanks to a clock speed of 5 and 4.9 GHz, they easily break the 200-point limit when using a single core.
When all cores are active, the 9900K breaks the 2000 point barrier, 14% faster than the Ryzen 7 2700X. In the meantime, the 9700K achieved a score of just over 1500 points, which placed it directly behind the old 1800X and just before the 8700K. That also meant that it was 26% slower than the 9900K.
Given what we've seen in Cinebench, it's no surprise that the 9900K outperforms the 2700X in Blender and cuts completion time by 23%. However, the 8-core Ryzen CPU was a fraction faster than the 9700K.
If we move to Corona and find a similar story here, the 9900K has reduced the render time by 20% compared to the 2700X and takes only 96 seconds. However, if you mainly render, the Threadripper 2950X makes more sense and I'll talk about it a little later.
The last rendering application we tested with is V-Ray. Here the 9900K reduced the rendering time by 18% and only took 62 seconds compared to 76 seconds with the 2700X. The 9700K was much less impressive. It took a few seconds longer than the 8700K, making it slower than the 1800X and 2700X.
The PCMark 10 benchmark for synthetic games depends heavily on the clock speed and the number of cores. It is therefore interesting to see that the 9900K only matches the 2700X here, while the 9700K was a step ahead of the older 1800X.