Intel "Coffee Lake" (2017/2018, 14 nm)

Intel Core i7-8700K overclocked to 7.4GHz for SuperPi 32M record
Pro overclocker Dancop, AKA Daniel Schier, has scalped a new SuperPi 32M record. The achievement is a little unusual as Dancop overclocked the popular Intel Core i7-8700K processor for his record but leveraged a 200-Series Intel chipset motherboard- something that isn't supposed to work.

The unlocked Intel Core i7 8700K Coffee Lake-S processor used by Dancop operates out of the box at 3.7GHz base / 4.7GHz boost frequencies, but is of course designed to be easy to overclock as signified by the 'K' suffix. The overclock achieved was 7,344.53MHz (recorded by HWBot as +98.50 per cent). However, the i7-8700K was run with just 3C/3T active rather than its default 6C/12T config.
http://hexus.net/tech/news/cpu/119765-intel-core-i7-8700k-overclocked-74ghz-superpi-32m-record/
 
Coffee Lake R is out, Core i5-9600K, i7-9700K and i9-9900K. They include same Meltdown & L1TF fixes as Whiskey Lakes released couple months ago.

Also, for whatever reason Intel decided to employ "3rd party" called Principled Technologies to run benchmarks early and referenced to those results in their launch press release.
Only thing is, the tests are completely bogus:
- Ryzen 7 2700X run in "Gaming Mode" meant for Threadrippers. Works the same way too, it makes 8-core 2700X a 4-core CPU.
- XMP profiles used on Intel but not on AMD (for which they make a bigger influence, too)
- Core i7-8700K results are higher than they should be at claimed settings, they match more or less DDR4-3200 XMP results instead of the claimed DDR4-2666 XMP

Obviously media can't confirm that Core i9-9900K wouldn't be as fast as claimed since NDA isn't lifted for a little over a week or so, but when the other results are bogus it makes you doubt those too

Hardware Unboxed who raised the issues up: https://www.patreon.com/posts/21950120
GamersNexus on same thing:

 
Talk about totally fucking up the tests producing entirely bogus results while stacking the playfield to be as uneven as possible … the AMD systems used the included retail AMD coolers while the Intel systems used Noctua high-end coolers shoved into a case with not great cooling dynamics.
 
Coffee Lake R is out, Core i5-9600K, i7-9700K and i9-9900K. They include same Meltdown & L1TF fixes as Whiskey Lakes released couple months ago.

Also, for whatever reason Intel decided to employ "3rd party" called Principled Technologies to run benchmarks early and referenced to those results in their launch press release.
Only thing is, the tests are completely bogus:
- Ryzen 7 2700X run in "Gaming Mode" meant for Threadrippers. Works the same way too, it makes 8-core 2700X a 4-core CPU.
- XMP profiles used on Intel but not on AMD (for which they make a bigger influence, too)
- Core i7-8700K results are higher than they should be at claimed settings, they match more or less DDR4-3200 XMP results instead of the claimed DDR4-2666 XMP

Obviously media can't confirm that Core i9-9900K wouldn't be as fast as claimed since NDA isn't lifted for a little over a week or so, but when the other results are bogus it makes you doubt those too

Hardware Unboxed who raised the issues up: https://www.patreon.com/posts/21950120
GamersNexus on same thing:


Looks like the third party should be called Unprincipled Technologies.
 
Back
Top