"Intel Processor" Replaces Pentium & Celeron Brands

[Vistaus]

Looks like we don't solely have to make fun of GNOME anymore for choosing generic names like "Web" and "Files".

[ClosedSource]

Looks like we don't solely have to make fun of GNOME anymore for choosing generic names like "Web" and "Files".

It's actually the opposite if you look closer. GNOME does it for cohesion and prioritizing app usability over their own branding (I'm not referring to gnome-shell). Nautilus doesn't particularly imply file browsing but it does say "GNOME" branding. It's the same with Microsoft Edge. At no point does the name tell users that this is where you access the web. For new computer users, Microsoft can improve usability by making Edge shortcuts say "Web Browser". Branded names are better left out to self-advertise Google Chrome and Mozilla Firefox.

On the contrary, Intel looks to be doing so to emphasize their non-Celeron/Pentium higher end processors. The lower end processors will have generic brandless names.

[Waethorn]

Here's a question: are the E cores in the hybrid models the same design as what was in the "Atom-architecture" budget Pentium and Celeron CPU/SoC family, like the Cherry Trail to Jasper Lake era stuff, or are they an entirely new core design just for the mainstream hybrid CPU's?

[Waethorn]

I know. Most normal people can't even remember the i5 or whatever that is written on the sticker on the front of their PC.

Did you just out yourself as an American who lives in the South? Been my experience that calling everything "Coke" is a Southern thing. Guess who lives in the South to know that's true?



I'm full of useless knowledge like that.

If you go into a bar and ask for a vodka soda, you'll get soda water or club soda in it, not any other pop, and certainly not Coke. This is why you don't call pop/soda-pop just "soda".

[Waethorn]

Here's another question: why do they only use odd numbers for the Core iX branding?

[jaxa]

Here's a question: are the E cores in the hybrid models the same design as what was in the "Atom-architecture" budget Pentium and Celeron CPU/SoC family, like the Cherry Trail to Jasper Lake era stuff, or are they an entirely new core design just for the mainstream hybrid CPU's?

Yes, they are Atom cores. The recent generations are:

Goldmont Plus (Gemini Lake/Refresh)
Tremont (Jasper Lake, also used in the experimental Lakefield CPUs)
Gracemont (Alder Lake)
Crestmont (future Meteor Lake)

We haven't seen the chips with only Gracemont E-cores yet. Leaks are calling it Alder Lake-N with 8 cores per die instead of 4 cores of previous Atom generations. Maybe there will be some other subtle differences relating to, e.g. the cache. Check the Gracemont article to see what changed from Tremont. Support for instruction sets like AVX and AVX2 is pretty notable. They basically threw in everything that wasn't AVX-512 to make it work seamlessly with the big cores, and then AVX-512 was (eventually) physically disabled on the big cores.

https://www.tomshardware.com/news/alder-lake-n-takes-shape-gracemont-xe-lp

One difference I just noticed: Raptor Lake will double L2 cache per quad-core Gracemont cluster to 4 MB, while it will stay 2 MB for all of the Alder Lake CPUs including N.

[kylew77]

Let's hope they'll use this opportunity to deprecate/stop selling CPUs having fewer than four physical cores. Nowadays systems with dual-core CPUs are a pain to use and Intel dares sell Celeron G6900 (from the same generation as Alder Lake) which has just two cores and no HT. That's just ugly.

I don't mind the lack of SMT especially with all the security vulnerabilities it brings, but yeah dual and even quad core processors from the Pentium and Celeron lineup can be slow! I got a Chromebook with quad core Jasper Lake and the COG system manager shows the cores are almost ALL over 50% utilization. I have since rectified the problem by getting an 8 core AMD ThinkPad.

[Vistaus]

It's actually the opposite if you look closer. GNOME does it for cohesion and prioritizing app usability over their own branding (I'm not referring to gnome-shell). Nautilus doesn't particularly imply file browsing but it does say "GNOME" branding. It's the same with Microsoft Edge. At no point does the name tell users that this is where you access the web. For new computer users, Microsoft can improve usability by making Edge shortcuts say "Web Browser".

I was just making a joke, as implied by my use of emoticons.

But now that we're seriously discussing this: I understand GNOME's approach, but I wish they would've done it the way Deepin does it. Deepin uses generic names too, but precedes it with "Deepin", i.e. Deepin File Manager, Deepin Movie, Deepin Screenshot. Still generic enough for new users to understand, but not so generic that it looks weird or out of place (since it's clear which DE the apps belong to) to more experienced users.

[direc85]

So, instead of associating "Celeron" as a no-good processor, it's now going to be "Intel Processor" instead? Talk about shooting yourself in the foot...

[Waethorn]

Yes, they are Atom cores. The recent generations are:

Goldmont Plus (Gemini Lake/Refresh)
Tremont (Jasper Lake, also used in the experimental Lakefield CPUs)
Gracemont (Alder Lake)
Crestmont (future Meteor Lake)

We haven't seen the chips with only Gracemont E-cores yet. Leaks are calling it Alder Lake-N with 8 cores per die instead of 4 cores of previous Atom generations. Maybe there will be some other subtle differences relating to, e.g. the cache. Check the Gracemont article to see what changed from Tremont. Support for instruction sets like AVX and AVX2 is pretty notable. They basically threw in everything that wasn't AVX-512 to make it work seamlessly with the big cores, and then AVX-512 was (eventually) physically disabled on the big cores.

https://www.tomshardware.com/news/alder-lake-n-takes-shape-gracemont-xe-lp

One difference I just noticed: Raptor Lake will double L2 cache per quad-core Gracemont cluster to 4 MB, while it will stay 2 MB for all of the Alder Lake CPUs including N.

So these'll be like the Avoton server chips of old then (aside from not running ECC RAM). Here's a related question to that then: is LPDDR, with its faster speeds and non-modular design (i.e. soldered) any more resilient to errors than socketed RAM? In simplistic terms, is it any closer to ECC than regular socketed slower DDR non-ECC RAM? I used to run an Avoton server with 16GB of RAM and a couple of VM's for deployment testing back when those were still current. If the cores are any more efficient, I may look at doing that again. It may be good for power efficiency in low volumes, although that's always been hotly contestable against mainstream and higher-power server chips that scale better, even when dividing single cores into multiple low-requirement workloads.

I'm going back to my earlier statement then: for the next-next gen they should just name then as such:

Generation-model (example for desktops given current specs for 12th gen):

14-1xx -> "Celeron/Atom class" 0 P cores, all E cores
14-2xx -> "Pentium class" 1-2 P cores
14-3xx -> "Core i3 class" 4 P cores
14-5xx -> "Core i5 class" 6 P cores
14-7xx -> "Core i7 class" 8 P cores
14-9xx -> "Core i9 class" 8+ P cores, enthusiast class

Now you have 1 product lineup. No division between low-end "Atom-only" architecture and mainstream, since they're already using Atom cores in their mainstream chips. It's just all Intel CPU's. Missing model numbers like 14-400, 600, 800, etc., can be used for alternate series or maybe mobile chips or something else.