Intel is splitting its high-end Tiger Lake mobile chip lineup to meet two TDP targets: 35 Watts and 45 Watts. Tiger Lake-H35 chips have been launched, with 4 cores, 8 threads, and 96 graphics "Xe" (Gen12) execution units. Later in Q1, Intel will launch 45 Watt TDP Tiger Lake with up to 8 cores, 16 threads.
Intel has also launched new "Jasper Lake" Celeron/Pentium chips on a "10nm" process node. Jasper Lake uses the Tremont Atom core previously used in Lakefield. TDPs range from 6-10 Watts. 16 GB of memory is explicity supported, up from 8 GB of Gemini Lake Refresh (although boards like the ODROID-H2+ supported 32 GB, go figure). Graphics performance of the Pentium Silver N6005 should be substantially higher than its predecessor due to the use of Gen11 graphics and an increase to 32 execution units.
See also: Intel says the Iris Xe Max isn't really for gaming. They're not wrong
Intel Confirms 10nm Ice Lake Xeon Production Has Started
Intel Appoints Pat Gelsinger as New CEO, From Feb 15th
Intel Launches 11th Gen vPro For Tiger Lake Mobile CPUs, Adds CET Security Tech
An Interview with Intel CEO Bob Swan: Roundtable Q&A on Fabs and Future
Also at Ars Technica, The Verge, Notebookcheck, and Wccftech.
Related: Intel Details Tremont Microarchitecture; x86 Atom Could be Used in Tablets Again
(Score: 2) by Snotnose on Friday January 15 2021, @03:16AM (2 children)
Based on what I've heard of their shrinking failures, and their "oops, we assumed a smaller tech and now we have heat/performance issues", one has to wonder.
These may sound well and good, but can Intel actually make the things? Or will they need to subcontract TSMC and friends (well, no friends. TSMC is the leader).
Enquiring minds want to know.
Relationship status: Available for curbside pickup.
(Score: 3, Informative) by takyon on Friday January 15 2021, @04:13AM
Indications are that "10nm" is working better than ever before. But notice:
1. They have been coming out with mobile "10nm" first. These are easier to make and don't have to reach the same clocks.
2. They are backporting "10nm" Willow Cove cores as "14nm" Cypress Cove cores on Rocket Lake in March.
3. When they do release a true "10nm" desktop CPU, Alder Lake (supposedly in late 2021), that will have 8 big "Golden Cove" cores and 8 small "Gracemont" Atom cores. They are changing their strategy dramatically and part of the reason could be high power consumption and heat.
Intel has been planning to make some GPU products at TSMC, but TSMC could begin making Intel CPUs as well:
TSMC To Produce Intel Core i3 CPUs on 5nm Process Node in 2H 2021, 3nm Mainstream & High-End CPUs Enter Mass Production in 2H 2022 [wccftech.com]
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by fakefuck39 on Friday January 15 2021, @04:34AM
actually intel's mobile lineup has been extremely successful since they got into that space. it's what's running all the little laptops, convertible touchscreen laptops, and netbooks. things like lenovo yoga. they have batteries that last a full day, they're thin and light as paper, and they run a full blown windows, and it's quite fast.
arm is not the competition for these processors. there only competition is amd mobile, and intel beats them in every category. and there are thousands of these little laptops sold daily. i take one with me when I travel. don't care if i break or lose it. you can get work done on them just fine too - visio, powerpoint, xwindows apps, browsing - just not video editing.
so why would they pay someone else to help them lead more, when they are currently leading in the target market?
now, as far as their fab scaling down issues - i'm not seeing any. first, intel 10nm is not measured on the same scale as samsung 5nm. different metrics for what that "nm" measures. what you're seeing is a lot of detailed info on how they get a process to work, and assume those are failures. that's like standing behind someone while they write code, seeing constant compilation errors, and assuming they're having issues. they're not.
in fact, one thing intel does that others don't, is be conservative. this comes after they got burned by the original pentium, and had to exchange all those CPUs for free. AMD and ARM release stuff that's brand spanking new and on the latest fab. Intel does not. Heck, all their datacenter CPUs is on 14nm - because it's more proven, and companies place more cost on risk than saving a few thousand bucks on CPUs for a million dollar server farm. And still beats the crap out of AMD, despite AMD using a less proven shiny new fab.
(Score: -1, Offtopic) by Anonymous Coward on Friday January 15 2021, @04:48AM
Gotta ask, why did they do this? Self-loathing? Or was Runaway1956 involved?
(Score: 3, Insightful) by RedGreen on Friday January 15 2021, @01:15PM (4 children)
The situation is pretty bad when even the trolls are not out commenting on their problems in masses. And apparently the fan boys have abandoned them too not a peep in sight from them, these type of postings use to be full of both. You know you have lost them, could not happen to a better bunch of scumbags who have held the computer industry hostage for years. Enjoy the bad times assholes.
"I modded down, down, down, and the flames went higher." -- Sven Olsen
(Score: 2) by Freeman on Friday January 15 2021, @04:16PM (1 child)
What's happened is that Intel hasn't hit a homerun recently. Whereas AMD has been knocking it out of the park for the last couple of releases. AMD seems to be cranking them out like hot cakes while Intel's offerings are just kind of sitting there. Nvidia is pushing AMD as well, but you can still generally put your hands on newer Nvidia GPUs for reasonable prices. Which seems to indicate that they also don't have as pressing of a supply/demand issue.
Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
(Score: 2) by takyon on Friday January 15 2021, @04:34PM
AMD is the one pushing Nvidia to put 12 GB of VRAM in the RTX 3060.
BOM costs of some of the RTX 30-series (RTX 3080 in particular) are high due to components needed for power management. Which is why Nvidia sold off many cards to cryptominers. MSRP pricing is a fantasy.
Prices are kind of ludicrous. There are supply/demand "issues", they just aren't a bad thing for AMD or Nvidia. All cards will be sold, all costs will be passed on to the consumer. One hidden cost: air shipping is more expensive during a pandemic because less people are flying.
AMD and NVIDIA confirm that GPU stock levels will remain non-existent for the foreseeable future [notebookcheck.net]
NVIDIA Allegedly Sold $175 Million Worth of Ampere GeForce RTX 30 GPUs To Crypto Miners, Could Be A Contributing Factor Behind Immense Shortages [wccftech.com]
Nvidia’s Ultimate Play [mooreslawisdead.com]
Nvidia Mulls Restarting Cryptomining GPU Production [tomshardware.com] - This is new. I wonder if the bubble will burst again.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 0) by Anonymous Coward on Saturday January 16 2021, @12:50PM (1 child)
Consumer sentiment means jack shit to Intel. I don't think you're grasping how huge they are, and how many pies their fingers are in.
(Score: 2) by RedGreen on Saturday January 16 2021, @07:49PM
Yeah, they probably could care less that Apple and AMD are kicking their ass, those many millions of processors are nothing for their revenue anyways, probably just a rounding error.
"I modded down, down, down, and the flames went higher." -- Sven Olsen