Why dishonest? Did they say they will sell a 28 core processor with 5Ghz frequency in the future? Or did they clearly say that it is a demo?
Please, lets use our brains, it is easy to understand, given current performance in CPUs that it is not possible to sell that kind of a processor.
I hope we are not as dumb as the americans that put a cat in the microwave oven and then complained that it is manufacturers fault that it didn't say it is not safe for cats... Come on...
There's such a thing as Grey areas, and Intel were clearly aware of what these were here.
They went onstage, placed undue stress on 5GHz as a number during their demo. They didn't mention overclocking at any point. They *hid* their exotic cooling solution and attempted to make it look like an ordinary watercooling loop - or at least, close enough to one that it wouldn't be exposed the moment they took the stage.
If they had been interested in showing a raw performance, overclocked number, they would have used LN2 - Hell, Der8auer and EVGA were at the show, both with LN2 in tow and on show - Der8auer was showing off some 8086K overclocks at 7+GHz, and EVGA were showing off a closed LN2 system that Kingp1n has been using for the last while.
Most people on the periphery of the enthusiast scene, will see an article about something making a big splash on every site - they will *NOT* see the followup articles explaining how bullshit that is or how dishonest it was - they will repeat the lie unaware of this, and that's exactly what Intel wants them to do - to tell their friends "Don't buy Threadripper 2, Intel showed something way cooler at computex" until one of those friends actually stops them and explains their error. Add in the chinese whispers effect you get amongst the uninformed and Intel have a lot to gain out of KNOWINGLY MISLEADING and then peddling some weak excuse like "we forgot to mention" - the misleading information will reach far more people than the clarifications ever will.
Intel probably knew they were going to get caught, but it didn't matter to them as long as the tech press within those first 24 hours, ran a bunch of articles with "28 Core 5GHz" in the headline so that it would take the wind out of the AMD Threadripper 2 announcement.
Remember, this sort of thing causes lasting ripples far beyond the savvy tech press like Gamers Nexus, who immediately ran a video calling this demo out as bullshit, and didn't report the headline Intel clearly wanted. Even sites like Tom's and TPU have a vested interest in breaking the story the way Intel wants them to, also - there's a far larger potential for clicks and traffic, if you're reporting something outlandish and incredible, rather than something that exposes the lie of that outlandish or incredible story.
It's not overclocked to the max, that would need LN2. I would be OK if the 5GHz result has been achieved with custom water cooling loop or AIO kit(thus that if they have not forgot to mention it's OC). From coffee lake S the 14nm process can do over 5GHz, so the chance that monstrous chip can achieve that too with sufficient cooling. But in the end the cooling needed is not for 24/7 systems people would actually build, which makes this whole exercise very much pointless.
Gamer's nexus made the very good point here, that an All-core overclock on a high corecount CPU is disproportionately more difficult to achieve than the same overclock on a lower corecount CPU.
The reason is that even if temps and process are identical, the larger the piece of silicon is, the larger the chance that one of those cores contains an imperfection. On a 4Core CPU, that would just make it a poor overclocker, but there'd be a larger number of 4core dies out there without imperfections, and so it's easier to test a whole whackton of CPUs and find a golden chip.
Ramp that up to 28 cores, and suddenly it's not good enough if 4 of those cores are perfect, flawless silicon - the moment any of the other 24 are flawed in any way, your all-core-overclock is limited. Less 28 core CPUs are made per wafer, they cost much more to obtain before you can even test them, and finding 28 flawless cores is way less likely than finding 4 flawless cores.
End result? Even on the same process under the same temperature constraints, with the same architecture, you're unlikely to get similar clocks on HCC or XCC chips versus consumer chips.
3300W is more than you can pull from the wall here in the UK with its 240V mains, let alone the USA with its 120V mains.
You just know what's coming, and still... lmao
