The Intel Atom Thread

[dark zero]

Laughs why would you want to use power vr with atom?

Power vr drivers for Windows has been a joke and always been a joke. For example in 2012 atom released it's first semi serious tablet hardware with clovertrail. It used power vr tech and trust me it sucked due to drivers. You were lacking basic functionality like if you had a 1280x800 screen and you did a full screen app it could not strech the screen for an older full screen program thus if the app was coded for 800x600 or a smaller resolution you lost 33% of your vertical height for 1280x800 screen. Only 1:1 pixel mapping was allowed in the driver's. And there were far more limitation with power vr drivers including constant crashing it was just bad. Crappy netbook hardware and the GPU was half the mess. So much better with the switch to baytrail for the Intel you drivers actually work.

Now power vr graphics is good for Intel but atom is dead more or less in Android. Furthermore who knows the future for imagination now that it is acquired by that Chinese company (got a new CEO a few weeks ago). Remember imagination lost half of it's revenue when apple left. Before being acquired imagination did 350 layoffs and when the acquisition was announced they said there was going to be more layoffs with the restructuring. Furthermore before apple pulled the plug on imagination it was hiring away talent from imagination including the coo and the heads of several divisions involving both hardware and software in powerVR. Beside probably the best talent being poached how do you attract and retain talent with things like stock options when a company takes a beating and the acquirer does not see the division as a growth opportunity and is thus more stingy for it is downsiding so both less people and less talented people and the ones who remain have worse moral.

So remind me why power vr for Intel atom is a good idea again?

Because their own option is a bigger mess and can't solve most basic things now due the programmers screwups.
And even with Raja, Intel won't catch AMD or nVIDIA up on the foreseeable future.

And that's why they should bought Power VR tech in order to make them decent. They were the most advanced of the bunch.

 

[mikk]

Power VR was a big mess, drivers and the hardware side, almost unusable. Intel wasn't ready at this time with an own GPU solution for Atom, things have changed since then, you are living in a cave it seems. It is plain stupid saying they should use PowerVR.

 

[IntelUser2000]

Preliminary Pentium Silver N5000(6W TDP) result can be seen on Notebookcheck. Its being used in Acer Swift 1.

The increase over the predecessor N4200 seems to be about 30% in average: https://www.notebookcheck.net/N4200-vs-N5000_8180_9420.247596.0.html

It does noticeably lower than J5005 in multi-threading but quite close to it in single thread.

 

[dark zero]

Power VR was a big mess, drivers and the hardware side, almost unusable. Intel wasn't ready at this time with an own GPU solution for Atom, things have changed since then, you are living in a cave it seems. It is plain stupid saying they should use PowerVR.

Intel should had bought Power VR. It was easier to fix something with potential than making something from scratch seeing how fierce are the competitors.

And they are not Apple in order to pull a leapfrog to catch them up.

 

[cbn]

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.

 

[IntelUser2000]

cbn: Don't forget about the result from Brunnis. His system scores even better.

 

[cbn]

N4000 (Gemini Lake) vs. C2D E6600 vs. Apollo Lake N3350:

https://www.cpubenchmark.net/compar...-E6600-vs-Intel-Celeron-N3350/3239vs912vs2895

N4000 (6W) within a hair of C2D E6600 (65W) in MT (and 25% stronger than C2D E6600 in ST).
N4000 (6W) is also a very noticeable improvement (37% gain in MT) over N3350 (6W).

This really bodes well for low end laptops.

P.S. With this higher level of performance perhaps we might see N4000 with 2GB RAM/16GB M10 Optane* (with swappiness of OS set at 100) replacing the classic 4GB RAM/16GB eMMC memory storage combo for Chromebooks (as well as encourage the entry of linux laptops**)

*running at a lower power state.

**Firefox >> Chrome. (Mainly for privacy reasons)

EDIT: With these atom processors becoming more powerful Intel should also consider enabling Optane Memory for use with Windows. (As I mentioned here, I am particularly interested to see if 32GB Optane (as cache) can replace DRAM for the non-compute intensive activity of Browsing)

 

[cbn]

6W Pentium Atom (N5000) vs. 15W Pentium Core (4415U) vs. (thrown in for good measure) 15W Core i3 7100U:

https://www.cpubenchmark.net/compar...tium-4415U-vs-Intel-i3-7100U/3204vs2977vs2879

N5000 does well (even in single thread) for a processor only having 6W.

I wonder what the battery tests will be like if comparing ~like for ~like?

EDIT: Very impressed also when I compare N5000 to Core i5 4200U (a common CPU found in refurb business laptops):

https://www.cpubenchmark.net/compare/Intel-i5-4200U-vs-Intel-Pentium-Silver-N5000/1947vs3204

 

[IntelUser2000]

I wonder what the battery tests will be like if comparing ~like for ~like?

There is a significant advantage, but its generally not shown in products, because the Core-based platforms are sold at a higher price, and use higher quality components that somewhat offset the differences.

The Qualcomm Windows 10 laptops likely use high quality components(with a large battery) to achieve excellent battery life, but it jacks up the price to a point that's not reasonable.

Because the Apollo Lake laptops are sold at the bottom barrel price point, it not only uses cheaper, lower quality components, but use smaller batteries and that in return results in lower battery life than most Core devices. It also seems to be featured in lot of 15.6-inch screen systems that negate any advantages in power use Apollo Lake might have. Intel also seems to deliberately keep attention away from their low power/cost chips to favor Core.

Gemini Lake likely won't change things much. Well, maybe slightly with the Pentium Silver range. Some indications suggest Pentium Silver is aimed at higher end than upper class of Apollo Lake chips.

 

[cbn]

Comparing J5005 (10W) to N5000 (6W) in passmark the scores are very similar:

https://www.cpubenchmark.net/compar...5005-vs-Intel-Pentium-Silver-N5000/3144vs3204

However, I do wonder how much different the CPU clocks would be if the iGPU were simultaneously run under load?

Also how different would the iGPU clocks be on the J5005 vs. N5000 (during simultaneous load with CPU)?

 

[cbn]

Interestingly this N4100 laptop was available as soon as March 6.

 

[cbn]

HP T430 thin client:

http://www.fanlesstech.com/2018/05/hp-t430-thin-client.html

https://liliputing.com/2018/05/hp-t...lly-a-small-fanless-low-power-desktop-pc.html

Specs are strong enough to make it basically a MIni-PC:

It’s powered by an Intel Celeron N4000 dual-core processor and ships with 4GB of RAM and either 16GB or 32GB of flash storage.

OS is either HP Thinpro , Windows 10 IoT Enterprise , HP Smart Zero Core.

Windows 10 IoT Enterprise
Windows 10 IoT Enterprise is a full version of Windows 10 that delivers enterprise manageability and security to IoT solutions. It is designed for powerful industry devices used in retail, manufacturing, healthcare, and other industries. Note: Windows 10 IoT Enterprise is a binary equivalent to Windows 10 Enterprise.

The following slides came from this video.

 

[dark zero]

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.

Actually Apollo Lake ARE on levels of Intel Core 2 Quad Q6600... heck, even Qualcomm's Kryo despite the lack of power effiency, was a great competitor of that old chip.

Interestingly this N4100 laptop was available as soon as March 6.

DAMN! I just got a Jumper EZPad 6 Pro with an Apollo Lake inside... it was a decent deal (and is a decent deal!)

Still... runs as great as a phone with Qualcomm Snapdragon 820... with Adreno from the SD 805 ...

 

[IntelUser2000]

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.

 

[Brunnis]

A new Surface is also coming based on Gemini Lake.

Oh, I had no idea. Do you have any links or more info? I haven’t seen anything about this yet.

 

[IntelUser2000]

I perhaps should have put a rumor tag. It is from Bloomberg so I reckon quite reputable.

https://www.bloomberg.com/news/arti...id-to-plan-low-cost-tablet-line-to-rival-ipad

"$400"
"Intel-based"
"2nd half"

Only option is Gemini Lake.

 

[jpiniero]

Only option is Gemini Lake.

The 8114Y would have been an option if yields weren't so bad.

 

[Dayman1225]

The 8114Y would have been an option if yields weren't so bad.

I guess it still could be an option depending on how many of these MS expects to sale.

 

[geoxile]

The 8114Y would have been an option if yields weren't so bad.

Wouldn't the cost of the 8114Y be prohibitive for a $400 tablet? The surface pro with the most basic options has ultra low-performance cores but starts at $800 and it's not fantastic imo

 

[Dayman1225]

Wouldn't the cost of the 8114Y be prohibitive for a $400 tablet? The surface pro with the most basic options has ultra low-performance cores but starts at $800 and it's not fantastic imo

Considering its a 10nm part that is essentially a yield learning part Intel probably sells it for pennies

 

[IntelUser2000]

You know what? ZDNet is reporting it may be based on Core M. That means Cannonlake 8114Y is a possibility.

Graphics may be better on the 8114Y, but CPU is better on Gemini Lake.

 

[mikk]

You know what? ZDNet is reporting it may be based on Core M. That means Cannonlake 8114Y is a possibility.

One of my contacts said Microsoft has been working on a low-end Surface device that would be priced at $399 without the separately-priced keyboard/cover. According to that contact, the device will be an Intel Core m device with 4GB of RAM and 64GB of storage.

https://www.zdnet.com/article/microsoft-to-launch-smaller-low-cost-surface-devices/


Only 4GB RAM and 64GB storage I hope we get other configs with 8 GB and 256 GB.

Graphics may be better on the 8114Y, but CPU is better on Gemini Lake.

Singlethread performance could be better on CNL-Y. Kabylake based m3-7Y30 does much better in Geekbench or Cinebench SC.

 

[jpiniero]

You know what? ZDNet is reporting it may be based on Core M. That means Cannonlake 8114Y is a possibility.

Graphics may be better on the 8114Y, but CPU is better on Gemini Lake.

Wikichip speculates the the boost would be 2.6 Ghz. That was sort of the reason I brought it up, because that's the same turbo clock as the 7y30, which is what the cheapest current Surface Pro comes with. 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.

After playing with a die yield calculator, it does seem plausible even with horrific yield if MS's needs are not too much. How many wafers/month do you think Intel is producing?

 

[IntelUser2000]

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

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]

https://www.elefacts.de/test-67-asrock_j5005_itx_mit_passiv_gekuehltem_4_kern_prozessor_im_test


Under full load with Cinebench R15 MT and Furmark, J5005 consumes 3W less than J4205 despite having a vastly better CB R15 performance. I haven't seen a proof for this:

Intel did add quite a few gates for these gains making the whole thing much less power efficient.