The Intel Atom Thread

Page 203 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Oct 14, 2003
5,979
163
126
The 7y30 does 105/250 in Cinebench compared to 73/236 for N5000 although the reviews I saw the Core M throttles hard with MT load.
This is a big deal. They are saying at 7W, the scores are at 210. Looking at the chip alone, M3s suck. Core M chips are sucky chips Intel pretended was totally optimized for fanless, but nothing more than binning and marketing so they could sell a lower performance chip for $300.

Kabylake based m3-7Y30 does much better in Geekbench or Cinebench SC.
After doing my own Geekbench runs, I wonder whether its worth running that thing at all. It hardly loads the CPU. Cinebench, is a good benchmark though. The only N5000 result seems to be somewhat hobbled by the single channel memory too.

Only 4GB RAM and 64GB storage o_O
But if they can really put a 8114Y chip in a $400 system, neither the less MT performance nor the RAM/storage configuration is a problem. Nowadays you spend that much for a crappy laptop anyway.
 

mikk

Platinum Member
May 15, 2012
2,396
25
126
Mar 11, 2000
22,595
1
106
Dunno about the m3-7Y30, but this is how our m3-7Y32, i5-7Y54, and i7-7Y75 fanless MacBooks behave with repeated runs of Cinebench. Note thought that the base clocks in the MacBooks are 100 MHz higher than usual because presumably Apple uses TDP up.



The graph has the Cinebench R15 scores on the Y axis and the run number on the X axis. I also ran my m3-7Y32 on a granite counter and a wood tabletop. The wood scores a bit worse because it acts as an insulator.

Yes, throttling occurs, but IMO it's not that bad overall. On a wood table, the m3-7Y32 drops from 265 on the first run to 246 on the last run, which is a drop of 7%. That's around half an hour of continuous Cinebench multi-core workload, with maybe 1-2 seconds of downtime in between runs.
 
Jul 5, 2013
83
1
66
So the ASRock J5005 GLK beats an i7 based MacBook @ Cinebench R15 with a score of 302?

Dunno about the m3-7Y30, but this is how our m3-7Y32, i5-7Y54, and i7-7Y75 fanless MacBooks behave with repeated runs of Cinebench. Note thought that the base clocks in the MacBooks are 100 MHz higher than usual because presumably Apple uses TDP up.



The graph has the Cinebench R15 scores on the Y axis and the run number on the X axis. I also ran my m3-7Y32 on a granite counter and a wood tabletop. The wood scores a bit worse because it acts as an insulator.

Yes, throttling occurs, but IMO it's not that bad overall. On a wood table, the m3-7Y32 drops from 265 on the first run to 246 on the last run, which is a drop of 7%. That's around half an hour of continuous Cinebench multi-core workload, with maybe 1-2 seconds of downtime in between runs.
 
Mar 11, 2000
22,595
1
106
So the ASRock J5005 GLK beats an i7 based MacBook @ Cinebench R15 with a score of 302?
Yes. It shouldn't come as an enormous surprise though as it's got 4 CPU cores, but an inferior GPU, and over twice the TDP. The i7-7Y-75 TDP is 4.5 W (with max TDP up at 7 W), while the J5005 TDP is 10 W.

It does much worse in PassMark and Geekbench 4 multi-core though, and scores in the same ballpark as m3-7Y32 for Geekbench 4 multi-core.
 
Last edited:
Jul 5, 2013
83
1
66
I thought the 6W Gemini Lakes performed roughly the same as the 10W ones. It's not clear that comparing TDPs is the right thing to do.

Yes. It shouldn't come as an enormous surprise though as it's got 4 CPU cores, but an inferior GPU, and over twice the TDP. The i7-7Y-75 TDP is 4.5 W (with max TDP up at 7 W), while the J5005 TDP is 10 W.

It does much worse in PassMark and Geekbench 4 multi-core though, and scores in the same ballpark as m3-7Y32 for Geekbench 4 multi-core.
 
Mar 11, 2000
22,595
1
106
I thought the 6W Gemini Lakes performed roughly the same as the 10W ones. It's not clear that comparing TDPs is the right thing to do.
6 W N5000 does significantly worse in Cinebench R15 and Geekbench 4 than 10 W J5005 (and 4.5 W i7-7Y75) in benchmarks, so no, the 6 W Gemini Lake parts do not perform as well as the 10 W Gemini Lake parts.

BTW, FWIW, N5000 is a Q4 2017 part (as is J5005), whereas, i7-7Y75 is a Q3 2016 part.

The latest Y series part may or may not show up soon as the Cannon Lake m3-8114Y, which supposedly has a base clock of 1.5 GHz, which should give it a boost over the 2017 Kaby Lake Y models, but unfortunately, it would just be a low volume part.
 
Oct 14, 2003
5,979
163
126
Do i miss something? Results look very underwhelming when compared to J3350 (which is Apollo Lake 2 core, 6W TDP, 2.4GHz Turbo). With J4005 we are talking Gemini Lake 2 core, 10W TDP, 2.7GHz Turbo.
Remember what I said back in post #4948? https://forums.anandtech.com/threads/the-intel-atom-thread.2341969/page-198#post-39347630

Now here's the J5005 review: https://www.cnx-software.com/2018/05/16/intel-nuc-kit-nuc7pjyh-review/

The performance is 30-50% better than the best Apollo Lake result. It even beats the Kabylake Celeron miniPC on single thread.
 

NostaSeronx

Platinum Member
Sep 18, 2011
2,356
145
126
So my awkward sources; Mercury Lake is Goldmont Plus Plus.
Quad-core MCL = Gemini Lake Plus (GML-R) <-- has an integrated XMM™ 7560 modem.
Octo-core MCL = Mercury Lake <-- does not have a modem.

Apollo -> Gemini -> Mercury = 14nm
Apollo = 14nm
Gemini = 14nm+
Mercury = 14nm++ and Intel's RFA FinFET. (No RF FinFET = more cores and more GPU)

Intel's Jasper Tremont is on-par with AMD's Zen. (10nm+ Atom is on par with AMD's Zen and Samsung's Mongoose3) ((tntver1 trm <-- not this one, tntver2 trmp <-- this one))

One of the features of the enhanced Analog FinFET is the device-level FIVR is back in town. (WHL->ICL same FIVR and MCL->JPL same FIVR) // Allows sub-microsecond Speedshift.
 
Last edited:
Mar 27, 2009
12,911
27
106
Some info I found on Chromebook "Octopus":

https://chromeunboxed.com/upcoming-chromebooks-new-devices-2018

Octopus

Much like its Apollo Lake predecessor, the Gemini Lake-powered Octopus is not only a Chromebook but a baseboard for upcoming Intel Atom devices. Just as we saw with the Reef baseboard, Octopus will end up being the platform on which many an educational device will be built.

The really cool part is that current chipsets like Gemini Lake and Kaby Lake are now being used for new Chromebooks unlike in the past where Chrome OS was generally 6-12 months behind its Windows counterparts.
Very interesting that an Intel produced motherboard (reference design) is being used in production laptops. Windows "Cloud book"/ "Stream book" notebooks as well? Linux laptops?

https://chromeunboxed.com/chromebook-development-still-going-strong-meet-our-next-contestant-pyro/

We haven’t much information on this device. But, the overlay for ‘Pyro’ clearly shows a Chromebook built on the ‘Reef’ baseboard.

As Robby reported last week, ‘Reef’ is Intel’s baseboard carrying the new Apollo Lake Atom processor.
 
Last edited:
Oct 14, 2003
5,979
163
126
Goldmont Plus in Gemini Lake platform is 14, not 14+.

Also Nosta, while your latest post is the most convincing yet, linux patch notes say Whiskey Lake is a Coffeelake-U at 15W with CNL(300 series chipset). The rumors that Cannonlake had McIVR makes WHL having one plausible, but there needs to be evidence.
 
Last edited:

NostaSeronx

Platinum Member
Sep 18, 2011
2,356
145
126
Goldmont Plus in Gemini Lake platform is 14, not 14+.

Also Nosta, while your latest post is the most convincing yet, linux patch notes say Whiskey Lake is a Coffeelake-U at 15W with CNL(300 series chipset). The rumors that Cannonlake had McIVR makes WHL having one plausible, but there needs to be evidence.
Intel is super confusing in regards to what is actually happening. Since, most of their big guns left in 2016-2017 when they didn't get Cannonlake or Icelake out on time.

Gemini Lake is on 14nm+ transistors, but it is re-using a lot of the 14nm IP. (No additional CPP or Vts or logic cells)
Gemini Lake Plus and Mercury Lake on 14nm++, will not be re-using any of the IP from previous generation. Bottom[FEOL]-up[to BEOL] remake with new CPP, Vts, and track height(9t4f, 7.5t3f, 6t2f, etc) and revised metal stack.

Coffee Lake is on 14nm++ transistors, but it follows mostly the same with Gemini Lake.
Whiskey Lake is also on 14nm++ transistors, but it utilizes the new CPP, Vts, track height, and metal stack.

(Also, there is two Analog VTs, the old one and the new one. The new one is a lot better, but the older one has been improved with lower leakage.)

2019 Icelake-Y is not the 2016 Icelake-Y that was meant to replace the 2015 Cannonlake-Y. So, 2018 Cannonlake-Y is based on 10nm which has similarities with Gemini and Coffee. While, 2019 Icelake-Y is on 10nm+, but is getting forwarded 10nm++ CPP, Vts, track height, hyper-scaling, revised metal stack, etc. Making it more similar to GML+/MCL and WHL.

If the codenames wrong are don't blame me, I had enough of Intel's quirky roadmap.

Parts on 14nm++ with the full suite includes all known and unknown Meltdown and Spectre fixes.
 
Last edited:

mikk

Platinum Member
May 15, 2012
2,396
25
126
Oct 14, 2003
5,979
163
126
Mar 27, 2009
12,911
27
106
Lenovo Flex 6 (11'")


Celeron N4000 (dual core), 2GB RAM (soldered on, not upgradeable), 64GB eMMC, Good keyboard and trackpad, 8.5 hour battery life (doing light tasks), fanless.

(This is more powerful than my old Core 2 Thinkpad which I use quite often....although my Thinkpad has a much larger screen which would make it easier to use than this one with its 11.6"*.)

Also available as Pentium N5000, 4GB, 128GB eMMC---> https://www3.lenovo.com/us/en/laptops/ideapad/flex-series/Lenovo-FLEX-6-11IGM/p/88IP8FX0889

*Lenovo does have an IdeaPad 330 with 15.6" screen, Celeron N4000 (dual core), 4GB RAM, 500GB HDD for $299.99---> https://www3.lenovo.com/us/en/laptops/ideapad/ideapad-300-series/Ideapad-330-15-Intel/p/81D1002GUS (Will be on the lookout for a review)
 
Last edited:
Mar 27, 2009
12,911
27
106
Intel Atom finally beats Q6600 in passmark:

https://www.cpubenchmark.net/compar...Q6600-vs-Intel-Pentium-J4205/3144vs1038vs2877

(Keep in mind this is only from a single sample though)

And the N5000 (6W) and J4150 (10W) beat the A8-7410 (25W):

https://www.cpubenchmark.net/compar...ron-J4105-vs-AMD-A8-7410-APU/3204vs3159vs2537

P.S. I would have posted up Celeron N4100 (6W) results but they are not available yet.
Celeron N4100 passmark results finally show-up:

https://www.cpubenchmark.net/compare/Intel-Celeron-N4100-vs-Intel-Celeron-N4000/3270vs3239

(2554 for N4100. This is about 1000 points higher than the 1542 of the N4000 (the processor found in the laptops from the previous post)....however, the N4000 does have a faster single thread of 1135 vs. 1016 for the N4100)

(For Firefox browsing I would much rather have the N4100, but the N4000 might actually work better for some older programs)
 
Last edited:

ninaholic37

Golden Member
Apr 13, 2012
1,880
2
91
@cbn looks interesting. Cool how they have a Pentium N5000 4GB RAM edition in the 11.6" Flex too, and that it is fanless. I wonder how often it could hit it's max turbo in my 30C room with no AC in the summer without it having a fan. I would prefer 10.1" but I guess 11.6" is the lowest size these things go these days.
 

IntelUser2000

Elite Member
Oct 14, 2003
5,979
163
126
Notebookcheck has the result for N5000 in the Clevo N750GU. Search indicates the 8GB configuration is dual channel. It's 10% faster in both single and multi-threaded compared to the Swift 1's N5000.

Cinebench R15
ST: 73 vs 80
MT: 236 vs 260

That compares very well to the J5005, which has 4W more TDP(6W vs 10W). It gets same in ST and J5005 is 12% better in MT.
 

mikk

Platinum Member
May 15, 2012
2,396
25
126
Acer Swift 1 with Pentium Silver N5000 "Gemini Lake" review: https://www.notebookcheck.net/Acer-Swift-1-SF114-32-N5000-SSD-FHD-Laptop-Review.303606.0.html

Like I said before, Gemini Lake has higher performance but also reaching the higher price bracket. The performance is very good despite the low thermals and having only single channel memory. I wonder how it'll do with dual channel? The laptop doesn't get very warm. It uses nice high end components otherwise, Wireless AC-9560 which is CNVi, and SATA3 SSD from Micron. Over 12 hours of battery life.

A new Surface is also coming based on Gemini Lake.


There is another test: https://notebooks-und-mobiles.de/acer-swift-1-n5000-fullhd-laptop-im-test


This device will be also available with 8GB RAM, probably the first Gemini Lake notebook with option for 8 GB. Difference between 4GB only 50€. Unfortunately this is a singlechannel config.

https://geizhals.de/acer-swift-1-sf114-32-p0l3-schwarz-nx-h1yeg-002-a1844660.html?hloc=at&hloc=de

The price is pretty high for some Atom based device but with some of the features features it is hard to find Core based devices below 1000€.

- fanless and low temperature
- only 1.4 Kg for a 14" device
- very thin only 14.95 mm
- very long battery life
- PWM free
- non-glare display
 

IntelUser2000

Elite Member
Oct 14, 2003
5,979
163
126
The price is actually not that bad. I'm seeing on Amazon.ca 4GB, 64GB eMMC version going for $500 cdn, which is $385 US. Though its a Windows 10 S mode, you can now switch that to Home for free. There are many in the $500 range including at best buy.

The performance is good enough that it can be an alternative to the Core m3 based devices.

So it seems the high cost is due to recent inflated costs on the SSDs. The 256GB configuration will probably make it close to the $700-800 range hinted on reviews.

On another news:

Rumors are swirling the cheap Surface will be Pentium Silver(Gemini Lake) or Kabylake Y, specificially the 4410/4415Y chip. Though rumor sites are pegging the Kabylake Y option as the higher end part, its actually a downgrade. N5000 will beat the 4415Y by 20-30% in both ST and MT, offer better battery life, and throttle less.
 


ASK THE COMMUNITY

TRENDING THREADS