New Tiger Lake U leak: 2.2 Ghz and 2.7 GHz configs; massive CPU performance delta hints at future OEM cooling shenanigans

[Redaktion]

Leaked Timespy scores hint at the existence of lower-tier Tiger Lake U SKUs. A massive difference between the 2.2 GHz and 2.7 GHz config scores implies that OEM cooling solutions could make or break performance.

https://www.notebookcheck.net/New-Tiger-Lake-U-leak-2-2-Ghz-and-2-7-GHz-configs-massive-CPU-performance-delta-hints-at-future-OEM-cooling-shenanigans.453581.0.html

[Valantar]

15W vs. 25W cTDP-up versions? That's been possible with all U-series Intel CPUs for the past few years at least, with significant base clock speed boosts for the 25W configurations despite having the same model name.

[francescovv]

vague article about performance which could imply anything.. but also I don't  understand why you are saying that "These figures demolished the Ryzen 7 3780U." doing so you are comparing next gen intel with prev gen ryzen..

ryzen 4000 will be on market before this new tiger lake U and have more accredited benchmark than this one .. why not compare the ryzen 4000u series?
also how can be cpu compared this way without tdp or prices ? a cpu that run 10% faster and costs 60% more consuming 30% more  is better because is 10% faster? (and maybe you are also comparing the top of one company with the bottom of the other )  that's not a good way to make comparison, IMHO if you need to give stats in so obscure way without any other metrics, at least.. leave comparison out..

this seems to follow the same author related article.. published some hours earlier.. (
Leaked-12-core-Ice-Lake-server-ES-nearly-100-percent-faster-than-Cascade-Lake-in-Geekbench-takes-the-fight-to-EPYC-Milan.453582.0.html)   same obscure benchmark comparison and very positive (cough biased cough) conclusion for intel


we all know that intel marketing has more budget than whole amd revenue....

[william blake]

so far all tiger's leaks/data/information tells us about the same. ice lake with slightly higher frequencies. so, irrelevant. tiger, no tiger, the laptop market picture is:
amd-technology leadership
intel-volume leadership

[william blake]

15W vs. 25W cTDP-up versions? That's been possible with all U-series Intel CPUs for the past few years at least, with significant base clock speed boosts for the 25W configurations despite having the same model name.

you can count up to 10 different versions with the same model name, at least. mobile cpus a total mess.

[ascariss]

10 nm is still a mess at Intel, regardless of what they claim.

[Valantar]

15W vs. 25W cTDP-up versions? That's been possible with all U-series Intel CPUs for the past few years at least, with significant base clock speed boosts for the 25W configurations despite having the same model name.

you can count up to 10 different versions with the same model name, at least. mobile cpus a total mess.

Yep, there's a lot of leeway for OEMs to configure parts as they please. After all, power limits are enforced by BIOS and can even be adjusted by end users. My current laptop for some reason limits its i7-8650U to 12W (though it still maintains turbo above 15W base clocks at that power). Still, there should  roughly be two performance classes per SKU for 15W and 25W (possibly with a third for 9W cTDP-down), but there seems to be too much variance in cooler design/quality and system configurations for things to align that well. It's a good thing sites like NBC do thorough testing so at least some people can make informed decisions though.

[Ariliquin]

Whats the point of comparing this next generation Intel to last generation Ryzen laptop CPU, seems purposefully disingenuous.

Thermal bottleneck is going to be all too common for Intel upcoming chips and cannot be solved easily for Intel until they move to 7nm or better. Too much grunt in a package that is not designed for it leading to lower performance than on paper.

[_MT_]

Whats the point of comparing this next generation Intel to last generation Ryzen laptop CPU, seems purposefully disingenuous.

Thermal bottleneck is going to be all too common for Intel upcoming chips and cannot be solved easily for Intel until they move to 7nm or better. Too much grunt in a package that is not designed for it leading to lower performance than on paper.

Are there benchmark results for 4000 series Ryzen? Honestly, I don't know. I have no problem waiting for a finished product. It would make more sense to me to compare them to their predecessors. And frankly, it's not Intel's fault that 3000 series mobile Ryzen is not build on Zen 2 cores.

Actually, Intel's 10 nm node is pretty close to TSMC's 7 nm node. Intel's original 10 nm was denser than TSMC's original 7 nm, I believe. But that's the process that gave Intel a lot of headaches (poor yields, I believe) and competition doesn't sleep. Not sure how they stack up right now. Frankly, it's just marketing designation these days. It has lost its original meaning. When Intel reaches 7 nm, it will be more like 5 nm TSMC node.

This article is really about nothing. Not only is the source questionable, nice scaling with frequency doesn't imply thermal limitation. It implies that the workload is processing power bound, rather than cache, RAM or I/O bound. At 4 GHz, I should process twice the number of instructions compared to 2 GHz. As long as I can feed instructions and data fast enough. And while they seem to have the same designation, unless I'm missing something, there is no proper designation. Only TigerLake U and 0000, which means nothing. Reported turbo clocks also look suspicious. Ice Lake tops out at 2.3 with all core turbo up to 3.6, I think. A part with 2.7 base and 2.7 turbo seems weird.

[dsajdisajoidas]

If these results are indeed run at 2.7Ghz, then it is very impressive. 3780u runs at 4Ghz and it still gets lower CPU score. That should give you an idea of how better the IPC might be. I expect 5-10% better than Ice Lake.