Apollo Lake (Platform) / Goldmont (CPU core Codename) Info

[Shivansps]

Im not so sure about Mullin IGP, or at least the 6400T igp to be better than a dual channel Cherry Trail.

Z8700/N3700/A4-6400T should about the same on all fronts.

I do hope for Stoney Ridge to be dual channel, AMD has been cripping their cat apus in the most stupid way possible.

 

[DrMrLordX]

It is the Atom thread and will continue to be.

So are we getting Pentium, Celeron, i3, i5, and i7 threads in the future? 'cuz we ain't got em now. Instead we have Broadwell, Skylake, etc. Granted those threads are also huge.

a dual channel Cherry Trail.

How many Cherry Trail devices out there use dual channel RAM configurations? I'm not asking to be contentious: I genuinely don't know the answer to that, at least not across a broad range of SKUs.

 

[Abwx]

I was talking of the 6700T, of course, the 6400T doesnt make sense in comparison safe for dedicated usages, as are most of the Braswells if we except the N3700.

As for Stoney ridge it wont replace Mullins adequatly, at least that s my opinion, 4 cores at lower frequency are better than 2 higher clocked ones, i hope they ll get back to Jaguar cores at 14nm for thoses devices, otherwise they wont be competitive against Intel s next Atoms iterations.

 

[Shivansps]

How many Cherry Trail devices out there use dual channel RAM configurations? I'm not asking to be contentious: I genuinely don't know the answer to that, at least not across a broad range of SKUs.

Its hard to tell, but most Z8500 and Z8700 devices should be dual channel, Surface 3 it is too, Notebooks with Braswell it will depend on the number of slots.

But who knows, some oems are so retarded that they even use UEFI x86 on 4GB Z8300 tablets.

 

[DrMrLordX]

As for Stoney ridge it wont replace Mullins adequatly, at least that s my opinion, 4 cores at lower frequency are better than 2 higher clocked ones, i hope they ll get back to Jaguar cores at 14nm for thoses devices, otherwise they wont be competitive against Intel s next Atoms iterations.

Don't sweat it. Stoney Ridge is just a stopgap until they go to low-power Zen APUs (Basilisk?)

Its hard to tell, but most Z8500 and Z8700 devices should be dual channel, Surface 3 it is too, Notebooks with Braswell it will depend on the number of slots.

But who knows, some oems are so retarded that they even use UEFI x86 on 4GB Z8300 tablets.

I'm not going to say that it's a bad thing for Apollo Lake devices to get dual channel RAM configs . . . I certainly hope that they do. The tech is there, use it. But it would say a lot about OEMs in general if they gave Apollo Lake dual channel after what they did to some other platform-to-remain-unnamed.

 

[Shivansps]

Don't sweat it. Stoney Ridge is just a stopgap until they go to low-power Zen APUs (Basilisk?)



I'm not going to say that it's a bad thing for Apollo Lake devices to get dual channel RAM configs . . . I certainly hope that they do. The tech is there, use it. But it would say a lot about OEMs in general if they gave Apollo Lake dual channel after what they did to some other platform-to-remain-unnamed.

OEMs has been doing DC on both Cherry Trail and Bay Trail since day one if the cpu supports it, so i dont see why that whould not move to Apollo Lake, there are probably several exceptions, just i dont know one, all those are running SC atoms.

 

[DrMrLordX]

That's . . . interesting. And informative. I guess we should expect Apollo Lake to be the same way.

 

[nvgpu]

https://build.opensuse.org/package/...el-driver/vaapi-intel-driver.changes?expand=1

Update to version 1.6.2:
+ Add support for Broxton:
- Decoding: H.264/MPEG-2/VC-1/JPEG/VP8/HEVC/HEVC 10-bit/VP9.
- Encoding: H.264/JPEG/VP8/HEVC.

Good news for HTPC users, Apollo Lake should have fixed function HEVC 10-bit & VP9 hardware decoding if VA-API changelog is correct.

 

[Arachnotronic]

https://build.opensuse.org/package/...el-driver/vaapi-intel-driver.changes?expand=1



Good news for HTPC users, Apollo Lake should have fixed function HEVC 10-bit & VP9 hardware decoding if VA-API changelog is correct.

Silver lining to Apollo Lake/Broxton being late...they probably had time to swap in the newer media block.

 

[sm625]

Yes, but the Atom in question, is what, a 4-5W TDP? Whereas the G3258 is what, a 58W TDP + overclocking factor? (Ok, realistically, when I had my G3258 OCed to 4.0Ghz @ 1.2 or 1.3V, CoreTemp reported Package Power at ~54W at full load.)

Still, that's 10X the TDP / Wattage, and the Atom has better than 1/10 the ST performance.

The G3258 doesnt run anywhere near its TDP when only one core is loaded. I have tested it on my machine using a Kill-A-Watt; the actual at-the-wall power increase is only about 26W above idle. Which means one haswell core at 4.4GHz is only pulling around 22W given my PSU's 85% efficiency. So in reality atom isnt that much more power efficient than haswell.

 

[mikk]

22W is a lot for one core, Atom consumes 1W for one core roughly.

And he didn't say G3258 consumes 58W for one core, your answer is confusing.

 

[reb0rn]

If atom can catch phenom II and core2due IPC it would be gr8 upgrade, + better GPU and you have decent office PC....

In the shit hole I live, 90% are with bay trail or E-xxx AMD crap, and its paint to do basic work for my clients...

 

[IntelUser2000]

If atom can catch phenom II and core2due IPC it would be gr8 upgrade, + better GPU and you have decent office PC....

That's only 60% difference in IPC from Skylake. It's too close, unless they decide to make such a core and make it 2 cores.

Intel wants to make lower Core chips perform 2x Atom. That means Core 2 Duo IPC is still years away.

 

[reb0rn]

ppl are pissed of by low end performance, if AMD worked as it should we would have a lot more, even phone segment has more power then 250$ laptop or 200$ desktop

 

[mikk]

https://browser.primatelabs.com/geekbench3/search?q=+ID5AE8

Pentium N4200 is Broxton based I think. MT score looks poor, although Geekbench reports a low clock (1.2 Ghz) and based from the low digit numbering it may be an entry SKU option. Or maybe it's a fanless SKU for tablets as a Broxton replacement.

 

[ksec]

https://browser.primatelabs.com/geekbench3/search?q=+ID5AE8

Pentium N4200 is Broxton based I think. MT score looks poor, although Geekbench reports a low clock (1.2 Ghz) and based from the low digit numbering it may be an entry SKU option. Or maybe it's a fanless SKU for tablets as a Broxton replacement.

How did a 1Ghz Broxton had the same result as the 2.3Ghz Broxton at the bottom of the page?

Do Geekbench report Turbo Pref numbers?

 

[VirtualLarry]

So, clue me in, has this chip been cancelled? Are all Atom CPUs dead? Or just the phone / tablet-oriented ones?

I just bought some Gigabyte Brix J1900 mini-PCs, and I think that they're great. If Intel could release a quad-core Atom, with low TDP, and close to Core2Quad IPC, and an iGPU that could decode HEVC and VP9, and possibly with 4K decode / output and HDMI2.0, they would have a REALLY potent media-consumption / web-browser / HTPC box using that CPU.

 

[Roland00Address]

So, clue me in, has this chip been cancelled? Are all Atom CPUs dead? Or just the phone / tablet-oriented ones?

I just bought some Gigabyte Brix J1900 mini-PCs, and I think that they're great. If Intel could release a quad-core Atom, with low TDP, and close to Core2Quad IPC, and an iGPU that could decode HEVC and VP9, and possibly with 4K decode / output and HDMI2.0, they would have a REALLY potent media-consumption / web-browser / HTPC box using that CPU.

Apollo Lake is not canceled, it can be used in small netbook like pcs and tablets, though the tablets will be mostly convertibles and detachables.

Broxton the one for phones and some tablets (the smaller tablets like 7") is canceled.

 

[frozentundra123456]

Too bad. Those small, cheap tablets are where atom really belongs.

 

[Gunbuster]

Typing this on an Atom Z8500 (Asus T100HA). It almost works but it's got that laggy feeling always lurking in the background. When do the new atoms come out?

 

[SAAA]

How did a 1Ghz Broxton had the same result as the 2.3Ghz Broxton at the bottom of the page?

Do Geekbench report Turbo Pref numbers?

Geekbench reads speed from the OS and Windows doesn't show turbo values.
Just made a comparison, look at individual single thread scores and aside from a few variations both are likely running at the same turbo speed, probably 2.3GHz.

Pentium has much lower SHA scores (maybe disabled on this part) and it looks like they were only using 2-3 cores out of 4.

Btw I seriously hope atoms can reach 3+GHz this time, at least on top desktop parts, cause they aren't that bad and at 10-15W TDP it should be doable.
Right now they score less than half a Skylake score at the same clock, for a budget laptop/desktop that's really limiting considered ~2.4GHz has been top speed for a while...

 

[Paul-Jackson]

Intel unraveled some detail on APL. I do some translation and post here：

http://b166.photo.store.qq.com/psb?...fSXLYGgc!/c/dK0AAAAAAAAA&bo=IANYAgAAAAAFAFo![

Harvey
Senior AT Mod/Admin

 

[Paul-Jackson]

Some detailed information：

On the left are current SKUs of braswell, comparing to the future SKU of APL on the right.

 

[dark zero]

Seems that Apollo Lake went up on wattage required. At least jumped from 4 to 6 and is a good sign.
Also the base clock might be higher.

Is on par of Stoney Ridge.

 

[Paul-Jackson]

Other details from conference ，I do some translation：

CPU performance seems to be clock to clock comparison， since APL has almost identical frequency to braswell.

Seems not to be used on TABLETs.But convertible and 2 in 1 are covered.