A very interesting SiSoft Sandra entry (via TUM APISAK) confirms something that had long been suspected over the past few weeks – Intel is going to be ditching Hyper-Threading for its i7 parts in the upcoming Coffee Lake processors. This is now going to be a feature limited to the i9 branding, led by the flagship Core i9-9900K whose existence I exclusively revealed almost a month ago.
Intel’s Core i7-9700K spotted in SiSoft Sandra database – will not support Hyper-Threading
There is something very interesting happening with Intel’s branding strategy and I will get into that in a bit, but first let’s take a look at the leak. SiSoft Sandra is one of the more reliable benchmark databases out there since engineers usually use it for testing and almost always forget to uncheck that pesky upload-to-them-interwebs option. This allows us to get a reliable and early sneak peak on engineering samples and such beforehand.
The Core i7-9700K will have 8 cores and 8 threads (this also automatically means that the i5 variant will have 6 cores and threads) which means that Hyper-Threading is not supported on this part. The chipset in question will be the Z390 chipset, which is something we have covered a while back as well. The Core i7-9700K scores a cool 157.24 GOPs in the benchmark but please keep in mind that this is most likely a non-final sample and the performance could improve significantly.
The most important takeaway from this leak is not the score but rather the shift in branding strategy – so let’s talk about that for a bit. Intel is currently executing a strategy that would shift their customer base towards the higher end. Depending on how they price this, they have essentially introduced an additional tier in the mainstream segment to further tap the market.
The Core i5 and i7 will represent the stepping stones towards the full-fledged Core i9-9900K. Also, keep in mind that there are design based reasons for this decision as well. Most applications can usually handle a maximum of 8 threads efficiently, any more than that is diminishing returns anyways. So for most use cases, the Core i7-9700K should perform similarly to its Hyper-Threaded counterpart clock for clock.
For rendering and multi-threading optimized applications, however, the Core i9-9900K will take a clear lead over the i7-9700K at the same clocks. Once again, it is too early to decide whether this is a good decision or not – it will depend entirely on how Intel prices it. If the company plays its cards right, this would be a very powerful tool to expand the market depth and tap more customer bases without sacrificing any significant amount of performance (remember, most applications right now aren’t designed to take advantage of more than 8 threads).
On a side note: We still have not heard anything about an i3 variant, which should be a quad-core part. Yield theory makes it one of the most lucrative parts to sell and I find it hard to believe that Intel would give that up. The most likely scenario is that we will see a staggered launch sometime soon after (unless the company’s yield for 14nm++ has increased to such an extent that they can comfortably sell 6-core parts like they would quad cores).