The Intel Atom Thread

[NTMBK]

Sorry, I forgot to mention my use case... So this laptop is for office, netflix and some old games (like Prey, Fear, or 1st far cry).

Don't worry about the PCIe x16 vs x6. PCIe is used for connecting devices such as GPUs and modems to the CPU- since your new laptop does not have a separate GPU, it does not need as many PCIe connections, and having fewer connections helps make the laptop less power hungry.

When you say "Prey", I presume you mean the original Prey, not the game from 2017? You should be able to run that alright. And it should handle Office and Netflix just fine. Let us know how the laptop works out for you!

 

[Eykxas]

Yes I mean the Prey of 2006.

I'll receive the laptop (a Dell 14 3482) at the end of the month. I hope it is more powerful than my old 2010's laptop (Alienware M11x R2, i5-520um, GT335m, 4gb of ram and no SSD). apparently yes, but I'm not sure.

 

[jpiniero]

https://browser.geekbench.com/v4/cpu/compare/3161188?baseline=9099549

N5000 is a tad faster than the 520um in single thread but decently faster in multi due to having 4 cores.

 

[FanlessTech]

Elkhart Lake looks very promising (Gen11 Graphics). The first products are expected in Q1 2020...

 

[jpiniero]

It'll have to be sliced and diced of course. Has 32 EUs as the top model but also 16 and 8 in the drivers. (By comparison, Gemini Lake Pentium/Celeron is 18 or 12)

 

[Kaloi48]

The product that Lakefield is going to is going to be a low volume high margin device. Like Microsoft's version of the Galaxy Fold and/or the Courier. Intel is probably getting Coreish prices for the chip too.

Plus, Lakefield's 10 nm portion might just be the cores and the GPU, and nothing else. Elkhart Lake would most likely have to be a monolithic chip.

The next-generation Atom will no longer be monolithic.
Elkhart Lake is a multi-chip package processor that includes a Tremont die and a PCH chipset.

 

[DrMrLordX]

The next-generation Atom will no longer be monolithic.
Elkhart Lake is a multi-chip package processor that includes a Tremont die and a PCH chipset.

Is the iGPU on the Tremont die or the chipset?

 

[thigobr]

The GPU should be together with the memory controller/CPU complex.

 

[IntelUser2000]

Elkhart Lake looks very promising (Gen11 Graphics). The first products are expected in Q1 2020...

You sure about the timeframe? That slide just tells us about when the eMB version will launch. Other roadmaps had it mid to Q3 this year.

That ~6 months difference seems to jive well with the Coffeelake-S version that came out in January of this year when the chip itself arrived Q2 last year.

 

[jpiniero]

The next-generation Atom will no longer be monolithic.
Elkhart Lake is a multi-chip package processor that includes a Tremont die and a PCH chipset.

Well, if the 10 nm portion is just 4C+32EU+memory controller, that wouldn't be so bad. It'd be pretty tiny, and being able to cut to 2C and 16/8 EU would mean they would get a nice amount of usable chips per wafer.

 

[FanlessTech]

Other roadmaps had it mid to Q3 this year.

Any link to these roadmaps? Things are moving fast at Intel nowadays.

 

[Shivansps]

32EU in a Atom means things will get very interesting for IGPs next year.

 

[FanlessTech]

Skyhawk Lake will have to wait because Gemini Lake Refresh is not canceled. Six new processors will launch in November :
The 10W (desktop) Pentium J5040, Celeron J4125, Celeron J4025 and the 6W (mobile) Pentium N5030, Celeron N4120, Celeron N4020.

 

[DrMrLordX]

I am still shaking my head over Gemini Lake Refresh. Is Intel having so much trouble with 10nm that they can't do Atom cores with it? Or is this a wafer allocation issue (Icelake-U/Y and Icelake-SP taking up all the wafers)?

 

[FanlessTech]

It's coming soon https://ark.intel.com/content/www/us/en/ark/search.html?_charset_=UTF-8&q=Gemini+Lake

Honestly, Gemini Lake is excellent, with impressive single thread performance. A refresh is fine. The main problem is availability.

 

[DrMrLordX]

It's coming soon https://ark.intel.com/content/www/us/en/ark/search.html?_charset_=UTF-8&q=Gemini+Lake

Honestly, Gemini Lake is excellent, with impressive single thread performance. A refresh is fine. The main problem is availability.

Impressive compared to what?

 

[moinmoin]

Just a note that the B-1 stepping of N3350, J3355, J3455 and N4200 have degradation issues and need firmware/BIOS update to mitigate :

Intel Moonwalks on Apollo Lake, Says Consumer Chips Aren't Dying After All (Updated)

Intel says that, contrary to its own documentation posted yesterday, the Apollo Lake chips are not dying faster than expected.

www.tomshardware.com

 

[Bouowmx]

As long as Windows-ARM continues to be awful (expensive hardware and low software support), 14% frequency bump (2.8 to 3.2 GHz on top Pentium J) should do.

Though, the micro-server line (Atom C3000) hasn't been updated since 2017.

 

[Dayman1225]

As long as Windows-ARM continues to be awful (expensive hardware and low software support), 14% frequency bump (2.8 to 3.2 GHz on top Pentium J) should do.

Though, the micro-server line (Atom C3000) hasn't been updated since 2017.

Pretty sure that’s being updated with Tremont/10nm atom. Iirc the platform is called Tanner Ridge

 

[NTMBK]

How does Gemini Lake stack up to historic "big core" CPUs? Are they at Core 2 Quad performance yet?

 

[Nothingness]

How does Gemini Lake stack up to historic "big core" CPUs? Are they at Core 2 Quad performance yet?

It seems so: https://browser.geekbench.com/processor-benchmarks
J4105 at 417/1310
Q9550 at 424/1369

J5005 is at ~480/1500.

 

[Brunnis]

It seems so: https://browser.geekbench.com/processor-benchmarks
J4105 at 417/1310
Q9550 at 424/1369

J5005 is at ~480/1500.

Yep, IPC seems to be better than Core 2 and quite close to Nehalem.

 

[Abwx]

It seems so: https://browser.geekbench.com/processor-benchmarks
J4105 at 417/1310
Q9550 at 424/1369

J5005 is at ~480/1500.

10x the AES perf provide 42% better overall perf all else being equal...

 

[Nothingness]

10x the AES perf provide 42% better overall perf all else being equal...

Huh no. You should look at how the score is computed

Integer score (that is without crypto) for a Q9550 is 480, vs 473 for J5005.
Q9550: https://browser.geekbench.com/v5/cpu/155946
J5005: https://browser.geekbench.com/v5/cpu/89276

 

[Abwx]

Huh no. You should look at how the score is computed

Integer score (that is without crypto) for a Q9550 is 480, vs 473 for J5005.
Q9550: https://browser.geekbench.com/v5/cpu/155946
J5005: https://browser.geekbench.com/v5/cpu/89276

But your J5005 has the AES instructions and they are used since it run at 1.39 GB/s while the J4105 is a little short of 1GB/s, Q9550 being at 0.13 GB/s.

So actually you brought even more AES in the overall result, but if that s how it should be computed then i have nothing to add...