Here's what the Intel Core i7-9700K scores in Geekbench

[HEXUS]

Eight-thread chip appears to have the edge over 12-thread Core i7-8700K.

Read more.

[LSG501]

Wow call me shocked.... 8 real core are faster than 6 real cores plus 6 'fill in the gap' hyperthreading cores....

[Troopa]

I personally hope that Intel doesn't diferentiate the i7 and neglect the i5 platform too much. I was hoping to get one of the new i5 9600k'series and I have put off getting an i5 8600k because of the upcoming refresh.

[devBunny]

Wow call me shocked.... 8 real core are faster than 6 real cores plus 6 'fill in the gap' hyperthreading cores....

If you'd expect 2 cores to beat 6 hyperthreads then those hyperthreads would be 30% as productive or worse. I'd hope, and have found, that a hyperthread provides a lot more than that, although it would depend on the application.

[scaryjim]

If you'd expect 2 cores to beat 6 hyperthreads then those hyperthreads would be 30% as productive or worse. I'd hope, and have found, that a hyperthread provides a lot more than that, although it would depend on the application.

Hmm, I genuinely wouldn't be surprised to find that Intel's HT threads can't manage more than 33% boost in throughput. COmparisons are tricky to find since Intel stopped doing the top rank Core i5 and Core i7 at the same clock speeds, but a quick look back to Haswell/Devil's Canyon gave up a nice direct comparison between the i5 4690k and the i7 4770k (both have 3.5GHz base and 3.9GHz turbo): https://hexus.net/tech/reviews/cpu/7...aswell/?page=3

What you find in that review is that the i7 has < 35% higher throughput than the i5 in Cinebench, so yes, those Hyperthreads are genuinely worth that little in the real world.

As an interesting aside, one of the strengths of Ryzen is it's relatively good SMT throughput. You can do the same comparison between the Ryzen 3 1300X and Ryzen 5 1500X (https://hexus.net/tech/reviews/cpu/1...3-1200/?page=3) and the SMT adds around 45% more throughput. So an 8C/8T Ryzen probably wouldn't beat a similarly clocked 6C/12T Ryzen....

[Iota]

So instead of listening to the criticism of the product lineup fragmentation of features, Intel have decided to double down on their stupidity. In comparison to AMD who try to make features more broadly available across their product lineup.

We're eager to see how the 9700K compares to AMD's eight-core, 16-thread Ryzen 7 2700X using our own benchmark suite

I'm more interested to see the price difference between the two, taking into account the above point.

[Ironbuket]

Do these cpus do anything at all for Meltdown and Spectre in their hardware? Im not knowingly buying a CPU architecture that has unfixed security flaws in it

[DanceswithUnix]

Do these cpus do anything at all for Meltdown and Spectre in their hardware? Im not knowingly buying a CPU architecture that has unfixed security flaws in it

AIUI these still have a Meltdown performance hit. Not that most users would notice.

[LSG501]

If you'd expect 2 cores to beat 6 hyperthreads then those hyperthreads would be 30% as productive or worse. I'd hope, and have found, that a hyperthread provides a lot more than that, although it would depend on the application.

Actually what you've likely found is that with hyperthreading active is that the full cores are actually working at their full potential, because the hyperthreading is filling in the gaps, rather than them being a massive performance gain in their own right. With a benchmark like geekbench where it will be maxing out every thread it could easily be 30% or less, pretty sure Intel even said as much on release. You also gain less from hyperthreading if the core is being fully utilised and has no 'gaps to fill', something geekbench would likely be doing.

[preter_s]

"...this time around ... Intel appears to be using hyper-threading as a means of setting the Core i9 apart from the rest of the pack; ... K-Series Core i7 and Core i5 desktop CPUs will have the feature disabled, marking a surprise shift from previous generations."

Chuckle LOL! So no new tricks except disabling/crippling feature sets like hyper threading on existing parts? What a loser!

[scaryjim]

... no new tricks except disabling/crippling feature sets like hyper threading on existing parts? What a loser!

Erm, no. The 9000 series parts are not existing parts. There's no 8-core mainstream i7 currently.

They're shuffling the feature line up without changing performance, so they can change the naming for the top-tier parts without having the rather problematic issue of the new i7 being slower than the old one. The Core i5 and i3 9000 series are likely to have the same configuration as their 8000 series counterparts - that is 6C/6T and 4C/4T. I guess the decision would've been whether they made the i7 6C/12T (with an 8C/16T i9) or 8C/8T. From this benchmark it seems likely that 8C/8T would perform slightly better than 6C/12T, so going 8C/8T makes sense.

[ByteMyAscii]

If geekbench is supposed to be a benchmark of "real world" usage, then it shouldn't be a surprise that faster cores is going to excel there.
This is a product aimed squarely at a specific audience, those not needing more threads.

What will be telling is workloads which do require that.

[haryyray]

I have core i7 6700...i am thinking to upgrade. Does it make sense to upgrade?

[badass]

Hmm, I genuinely wouldn't be surprised to find that Intel's HT threads can't manage more than 33% boost in throughput. COmparisons are tricky to find since Intel stopped doing the top rank Core i5 and Core i7 at the same clock speeds, but a quick look back to Haswell/Devil's Canyon gave up a nice direct comparison between the i5 4690k and the i7 4770k (both have 3.5GHz base and 3.9GHz turbo): https://hexus.net/tech/reviews/cpu/7...aswell/?page=3

What you find in that review is that the i7 has < 35% higher throughput than the i5 in Cinebench, so yes, those Hyperthreads are genuinely worth that little in the real world.

IIRC the original suggestion from Intel when they first introduced it was a 25% boost in performance in the right kind of applications. However that was a marketing claim so probably based on cherry picked applications.

Also, it's important to remember that the i7 has 33% more L3 cache than the i5 so even at the same clock speed with hyperthreading disabled I would expect the i7 to be (a bit) faster.

[Jack-0]

I have core i7 6700...i am thinking to upgrade. Does it make sense to upgrade?

OK then! according to geekbench my i7-6700K which is clocked at 4.8GHz has a single core score of 6058 and a multi core of 21264.

[DanceswithUnix]

IIRC the original suggestion from Intel when they first introduced it was a 25% boost in performance in the right kind of applications.

If you are referring to the P4 hyperthreading, that is a really different beast to what modern CPUs do (or IBM processors of the day where it was quite well established).