Intel's 10th-gen Core family cracks 5GHz barrier with H-series laptop processors Intel has announced its tenth-generation Core i5, i7, and i9 H-series microprocessors for laptops, which max out at 5.3GHz. Chipzilla pitched the chip family at “gamers and creators,” and advances the usual reasons: games need a decent amount of CPU power as well as GPU, and artists need to render at 4K resolution without …
These are all basically last years chips (same architecture and node) just with the thermal boost tech added, and one of last years i9s now being branded as an i7 in order to add an 8 core chip to the mid range, in order to compete with AMDs 8 core chips.
Also how relevant is a single core boost speed these days? Especially when it's thermally managed and in a laptop form factor. I can't imagine those 5GHz+ speeds lasting very long without some sort of very clever and probably noisy cooling solution.
All core boost would be more useful info, but as far as I can see Intel haven't published those figures yet, and as this will be thermally limited, I doubt it will be much different, if at all, from last years chips.
According to Anandtech's press-release review these CPU's are last years Coffee-Lake rebranded and binned for voltage profile to allow the high clock, with any silicon changes being minimal and most likely hardware mitigations for Intel's own security bugs.
Those high clocks will only work on 2 cores on the CPU too, so it's basically a baseless headline of "look we OC'd a laptop to 5Ghz+", it is highly dependant on the way laptop manufacturers opt to cool their laptops, it has to be operating within it's secondary power envelope and running Intel's Thermal Velocity Boost before it will ever boost that high.
Throw in Intels recommended PL2 turbo power limit for the i9-10980HK of 135w and your going to be hunting down power outlets most of the time to recharge the thing.
This is Intel trying to protect it's mobile computing marketshare with a rebranded flawed product, this announcement is not dissimilar to the 2018 Computex demo they did of the 28 core Xeon (rebranded for PR as a desktop part) running a highly overclocked cpu to all-core 5Ghz and just happened to not mention the thing was using an industrial grade water chiller to prevent it going on fire.
"According to Anandtech's press-release review these CPU's are last years Coffee-Lake rebranded and binned for voltage profile to allow the high clock,"
Another review site pointed out that the "gains" posted over older systems with a vastly inferior video card and most of the supposed gains are a result of the video card, not the CPU
"games need a decent amount of CPU power as well as GPU, and artists need to render at 4K resolution without needing to pop out for a cup of tea while things process"
Gamers want to play games for more than 1 minute at the PL2 clock (at 107-135W TDP - what happened to Intel's 1.25x TDP boost for turbo?) which allows for the >5GHz operation. Switching between turbo and non-turbo in a game would be more jarring than just having a lower max clock. And once the CPU is hot, there's no max 5.3GHz turbo any longer (if the laptop design supported that in the first place). It'll look good in benchmarks, but reality could be different.
And I suspect renderers want to render at max clocks more than one minute every 10 minutes.
Fact is the 16C 65W ecoMode Ryzen 3950X laptops (e.g,. Clevo) are using less power than this chip, and have higher base clocks.
Trading standards needs to take a long hard look at Intel's TDP and clock speed claims.
Intel defines TDP as minimum power needed whereas AMD defines TDP as typical power needed. Intel is not being dishonest because they are clear how they define it. However, the problem is the documentation which they state that is not going to be found by the average person and Intel is not upfront about it. Misleading, but not dishonest. The documentation also states that their CPU's boost based on well the cooling is. If the cooling fan is rated for 45W, then it will boost very little. But if the cooling fan is rated for about 130W, then this CPU will reach max boost.
We just need to remember that to reach boost, Intel processors may use 250% more power but AMD processor use just 50% more.
> ... is this the fancy "one core can supercharge if the others turn off and act as heat sinks" stuff?
Yes, note that the extra speed is bought by spending a silly amount of extra power. If you like your system silent (or battery last) then disable speed boost. Btw. it also makes benchmarking essentially impossible.
"core speeds had stopped rising once they moved the focus to multiple cores"
In the SPARC world, core count rose while raising clock speed:
2011 - T4 - 3.0 GHz - 8 cores - 1+ instruction per clock cycle
2013 - T5 - 3.6 GHz - 16 cores - 1+ instruction per clock cycle
2015 - M7 - 4.1 GHz - 32 cores - 2 instructions per clock cycle
2017 - M8 - 5.0 GHz - 32 cores - 4 instructions per clock cycle
These systems run at top speed, not just the occasional "boost".
I would say their integrated GPUs are pretty impressive certainly compared to those from the past. Of course sheer grunt isn't available on something so small but the featureset in terms of shader model means you can actually DO GPU stuff on a business laptop now. In the past, that was basically impossible.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
