They should do their homework if they pretend to be genuine tech sites..
What homework? You take the chip, you install it, you install the HSF, you install the OS + software, you boot, you tune the UEFI, and then you bench it and report results.
You don't stick your head in the sand because the voltage scaling doesn't match expectations from Broadwell-U and Broadwell-D results from months earlier.
As said it is impossible that the voltage trend is reversed to this point as frequency increase, this is where the mistake was done and this put under question their honnesty.
Of course it's possible! They did a respin of the process, making of it a high-performance node that would be unsuitable for low power ICs. Why do you think Broadwell-C took so long to reach market? It's a test-case for Skylake.
It is the only feasible way for Intel to continue covering such a broad spectrum of the computing market with the same basic core design.
We have brains to check what is realistic and what is not.
Your perceptions are skewed. What you WANT is for Intel to be stuck at 4 ghz and below for the next 2-3 years so that AMD can play catch-up. Sorry, but it didn't happen that way.
In this case they should release the whole datas rather than displaying what seems more to be organised marketing, and i have no other priority than discussing tech, generaly in the AMD threads but currently it s Intel that is under scrutinity by the tech crowd, hence my presence here, to discuss specificaly about tech.
Then go and get your own ES and run your own benchmarks, if you don't like the presentation. Or go run your own investigation of Lam and HKEPC and prove to us that they are fudging numbers. Right now you have no proof.
Lol, i m just analysing the datas, what are you speaking of..?.
That's bs and you know it. We should all be glad you weren't around here in '05 when the Conroe leaks started to roll in. You would have dredged up a bunch of 65nm Presler and Cedarmill voltage scaling charts to show us why Conroe couldn't possibly do so well on Intel's 65nm process, and in so doing, you would have accused a bunch of actual PC enthusiasts of lying about their ES samples.
Many were expected/wanting Intel to continue to backslide on peak clockspeeds as their process shrinks conitinued. They FINALLY bucked the trend. Get over it.
On topic they cant reverse this behaviour
Yes they can, and apparently they did. If someone like pm came in here and said, "Sorry guys, there's 99.99% chance that these leaks are fake, my bosses just told me so", I might listen to that guy, given what he does for a living. Anyone less than that? No, unless they also had an ES and had gotten different results.
a process as such means nothing as it s not a start but a end, that is, the mean to manufacture transistors with the required caracteristics
So, the most logical conclusion here is that it is not the exact same process.
Yes, you are. What you are spreading is the very definition of FUD :
Fear, Uncertainty, and Doubt
You want to plant fear in the mind of others; fear that Intel will forever lower clockspeeds as the continue to pursue the mobile market
You create uncertainty by presenting data gathered from different processors coming from different wafers, without entertaining the possibility that Intel did a respin of the process.
You doubt that Intel actually has within their vast, well-paid R&D department that consumes billions of dollars a year the minds capable of maintaining two different 14nm processes aimed at different market segments.
Any RATIONAL human being would have deducted that Intel engineers detected problems with clockspeed/voltage scaling with their low-power 14nm that we first saw in Broadwell-Y and Broadwell-U well before any of us forum junkies. So they would have used their considering resources and talent to fix the problem. Now they likely have a 14nm that is suitable for desktop, HEDT, and high-clockspeed server chips (read: not Broadwell-D).
Such deductive reasoning is only useful in the presence of data obtained through direct physical experimentation with a suitable sample (in this case, an i7-5775C). Anything less than that simply isn't good enough. It would be like trying to examine chickens to learn more about the nature of living turkeys, pigs, and cows.
You are using deductive reasoning to refute inductive reasoning. If you do not like the presentation of data, politely contact the source and request further data. Be curious and interested, not curmudgeonly and combative. Be scientific! Test up close, don't just analyze from a distance. If you can not obtain the sample for yourself, you are stuck with whatever results you get from the person that does have access to the sample.
For you to impugn the reputation of the tester as being ignorant and biased (or fundamentally compromised) is an incredible insult. You probably don't even know Lam, much less know anything about the test setup or chosen methodology. What empowers you to criticize something you haven't even the curiosity to examine objectively?
I don't know Lam either, and I only know what Kitguru and other sources have chosen to share with us. Until I get more data points, this is the best I've got, and I'm certainly not going to pretend that a boatload of old, irrelevant data empower me to accuse him of fraud and/or stupidity.
quite the contrary, i m warning the tech crowd that there s eventual deceptive marketing campaign, would it be the first time that such thing would occur..??.
Okay, show me the last time Intel used an ES leak to deliberately mislead the public about an upcoming product. Conroe? Oh wait, no, that really was a good product. Nehalem? No, all those Gulftown leaks pretty-much turned out to be true too. Got anything else?
Intel has played dirty games before, but not with ES leaks.
At first sight i would say that teh marketing idea is to induce people thinking that the chip overclock well, it s exactly what was made when they launched Dcanyon and displayed over optimistic screen prior to launch...
You mean
leaks like this one? All the 4790k leaks pointed to 4 ghz base clock and 4.4 ghz turbo, which is what happened. You had some people wishing for 5 ghz Haswell and 5 ghz Pentium G3258, but neither Intel nor those in possession of ES chips actually showed that happening.
I didnt negate the bench, the screenshot at 3.7 and 1.227V look genuine, it s the 4.8 and 5.0 screens that are totaly made up, so for one you didnt read accurately waht i posted and are hence doing wrong claims...
I read what you posted. Nothing you say means anything compared to concrete results, which is what Lam is saying he's got. You have no empirical evidence showing the contrary, only deductive nonsense.
If you think he's full of crap, buy an i7-5775C and show the world why it won't do suicide runs @ 4.8Ghz or 5Ghz or even 4.6-4.7 ghz ala Haswell. Show us that Intel is still using their low-power node for Broadwell-C. What we don't need is endless speculation and prattling. We need good old-fashioned benchmarking with real hardware. We need hard science. What we don't need are more people chained to the wall of Plato's cave, grasping at shadows.