In the Investor meeting Intel was up to their old tricks. This is one of the slides that they posted:
View attachment 19109
Can you spot the deliberate mistake?
View attachment 19109
Can you spot the deliberate mistake?
Intel's fake news
- Thread starter lefty
- Start date
There are quite a few glaring issues with this slide. Omission of foundry 7nm due next year, and then extrapolating the foundry line without it. Extrapolating their own line at a rate that isn't consistent with their roadmap. Claiming a 3 year process advantage based on the past, when in the present their advantage is perhaps 1 year, soon to be 0 years. List goes on.
I can't believe Murthy presented that slide. He should know better. I have attached his presentation in case someone is interested. I guess it takes about one year to brain wash a semiconductor professional. Good to know.
"Intel will have enjoyed a lead of three years when competitors launch 10nm process"
Seriously? competitors have ALREADY launched 10nm processes and they are better than Intel 14nm. Based on what we know today TSMC 7nm will lanuch about the same time as Intel 10nm and again it will be better.
What is Intel really trying to accomplish here?!?!?!? Besides angering the fabless semiconductor ecosystem yet again.
Last edited:
True, but foundry 7nm and Intel 10nm are very similar (same CPP and MMP) so there's little timescale difference -- but Intel being way in front is clearly bulls*it.
Daniel, why do you say that TSMC 7nm will be "better" than (same geometry) Intel 10nm? Better PPA? Better IP? Cheaper?
Not trying to favour Intel or TSMC, just get a level playing field in discussions...
Last edited:
Here is a transcript. I found the whole thing less than transparent even though they said it would be transparent. I guess it is subjective but I was seriously disappointed in the entire day in regards to the information provided. Hopefully AMD will provide Intel with a long overdue wake-up call.
P
Portland
Guest
I think AMD will succeed in supporting niche thing like webhosting, cdn, digital advertising agency, ... but won't make mainstream. We'll feel the impact but intel's the best product for the typical consumer.
No discussion about AMD in the PPT. Perhaps this is doctrine--never provide free advertising/legitimatize a competitor by discussing.
The foundry world remains relatively incomparable to the Intel world. It is frustrating. It lets Intel claim whatever they want, and they are taking advantage of this opportunity. This is capitalism though, not a conspiracy; Intel tech is ubiquitous but not a commodity.
Eventually, Chromebooks will provide the comparisons needed. Samsung is introducing low-end Chromebooks with ARM processors and high-end ones with Intel Y processors with everything else being pretty identical. Interesting to note: The 28nm ARM A72 chip performs nearly as well as the 14nm Y chip, in early "how many tabs in Chrome can I keep open" kind of rough comparisons.
The foundry world remains relatively incomparable to the Intel world. It is frustrating. It lets Intel claim whatever they want, and they are taking advantage of this opportunity. This is capitalism though, not a conspiracy; Intel tech is ubiquitous but not a commodity.
Eventually, Chromebooks will provide the comparisons needed. Samsung is introducing low-end Chromebooks with ARM processors and high-end ones with Intel Y processors with everything else being pretty identical. Interesting to note: The 28nm ARM A72 chip performs nearly as well as the 14nm Y chip, in early "how many tabs in Chrome can I keep open" kind of rough comparisons.
Intel have announced that their 8th generation PC line up will be mostly on 14nm:Intel's 8th-gen Core CPUs will again be 14nm - TechSpot
Intel's 2017 10nm is basically a "paper launch" to save face and look like they are not behind TSMC and Co. It won't be for another year before 10nm yields are at a level where they can use it economically on their high volume products.
Intel's 2017 10nm is basically a "paper launch" to save face and look like they are not behind TSMC and Co. It won't be for another year before 10nm yields are at a level where they can use it economically on their high volume products.
hist78
Well-known member
Or, maybe there is another platform that can provide a real world comparison: Windows 10 for ARM which first announced in the December 2016.
Microsoft-Qualcomm deal finally puts Windows 10 and Win32 apps on ARM devices | PCWorld
My guess is that Qualcomm is the first one but won't be the only one. Although the traditional PC market is declining but there're still a lot of PCs need to be replaced in the coming years and Windows 10 for ARM can provide a large scale of market for ARM's manufacturers.
PC sales in 2016: CHART - Business Insider
Yep, that will turn tick-tock into tick-tock-tock-tock. And I just saw a benchmark of a 32nm i3 Sandy Bridge from 2011 vs. a 14nm Kaby Lake i3 from 2017 (both sold at USD 120 when launched) and in most benchmarks the Sandy Bridge managed to stay only some 35% below the Kaby Lake and at its worst it showed half the speed.
Now, if Moore's law is doing well, why in 6 years from sandy bridge to kabylake we see "only" a 1.5-2.0x speed improvement? Is Intel raising their gross margins in the sense that they are using the lower cost/transistor to improve their financials rather than to improve their product's performance? Maybe they are doing it to counter the increasing need for more and more R&D and CAPEX?
That (and the lack of real competition) could explain why we are not seeing Intel deliver as much as we grew used to expect.
On the other hand, it makes 2017 even more exciting, maybe Intel has left a lot of room for Ryzen, OpenPower and ARM to flourish. Let's wait and see.
What everybody seems to miss is that Intel's "tick-tock" power-hungry 7nm is not the same as TSMC's (or Samsung's) low-power CMOS 7nm. Intel's announced $7bn investment in Fab 42 here in Chandler AZ will be (in my opinion) for low-power CMOS, so they don't have to go to TSMC for their 4G modems (as they do now) and they will undoubtedly grow it into a major foundry to compete directly with TSMC and Samsung. That's also likely why they did the major licensing of ARM's recipe. So, Intel's Apples vs. Oranges comparison is total BS.
W
wow&wow
Guest
The anal_ysts allowed the Intel management talking about the revenues and profits like the coming new AMD chips either do not exist or won't affect anything at all, the proof of how stupid, short-sighted those bean-counting monkeys are : )
W
wow&wow
Guest
The presentation was for the Street monkeys, not for people like you or me : )
W
wow&wow
Guest
"I was seriously disappointed in the entire day in regards to the information provided."
It's for the anal_ysts, so as long as they were happy ...
In addition, it's great for AMD investors, the more denials the better and the greater ROI!
....consider this........Intel (Holt...now gone) at their Nov 2013 dog and pony show showed 4 so called 14nm processes.....all with differing gate densities.....saying the least dense process is used for high performance (ex server) applications and the most dense for something like mobile SOC. First....how can 4 processes with differing gate densities all be called the same thing....in this case 14nm?.........AMD Ryzen on Samsung 14nm LPP at Global Foundries is 195mm^2 for 4.8B transistors. Intel Broadwell on 14nm (must be least dense one) is 246mm^2 for 3.2B Transistors. ANY Intel claim (past/present and future) regarding process density is pure BS. I suspect Intel has NEVER shipped any product using their so called densest 14nm process but uses metrics from it to DUPE the analysts. Intel should be called out in no uncertain terms and EXPOSED for the liars they are.