Intel Pentium And Celeron (Braswell) 14 nm SoCs Now Shipping

[scolaner]

At IDF Shenzen, Intel made a flurry of announcements, including the news that its new Braswell SoCs are now shipping. The four Braswell chips include a new Pentium chip (quad core) and three Celerons (one quad core, two dual core).


Intel Pentium And Celeron (Braswell) 14 nm SoCs Now Shipping : Read more

 

[TNT27]

4 core pentiums coming? how about something like a 4 core g3258 on desktops? at the same prces?!

 

[artk2219]

"4 core pentiums coming? how about something like a 4 core g3258 on desktops? at the same prces?!"

Or hell, why not just have an unlocked I3? No it isnt going to be like an I5, but dammit I want to tweak stuff! Honestly it would be nice if the i7's moved to 6 core 12 thread chips, the i5's were 4 core 8 thread chips, the i3's were quad cores, the pentium were 2 core 4 thread, and the celerons were just as they are now, dual cores. But i'm asking too much and i have no idea what their yields are like or what the sales would be like. They could also sell a 6 core with no hyperthreading thats unlocked, but i have no idea where that would slot in. It would also be nice if mobile i5's were real i5's and not rebadged and higher clocked I3's, but it is what it is, dont even get me started on the 2 core 4 thread "i7's".

 

[Ninjawithagun]

Enough with the hooplah crap processors. What we really want to know is when will Skylake CPUs be released!!

 

[CaedenV]

4 core pentiums coming? how about something like a 4 core g3258 on desktops? at the same prces?!

These are dual and quad core Atom chips branded as Pentium and Celery... these are not the same as the g-series Celery and Pentium chips for desktop products.

Confused? Well so is everyone else.

 

[CaedenV]

Enough with the hooplah crap processors. What we really want to know is when will Skylake CPUs be released!!

100% agree! Skylake is supposed to be the first real focus on desktop in a while, so it will be interesting to see what they put out. Personally, I am still quite happy with my 4 year old i7 humming away at 4.2GHz, and I fear that a faster or more efficient chip is not going to be enough to get me to upgrade.
What will make me upgrade? More cores at the same (or better) performance and same price point (~$350-400). Onboard 10Gb/s Ethernet. Board level support and headers for mobile tech like BT, WiDi (RX and TX), and NFC/Tap-to-send with wifi handoff for faster than BT file transfers between mobile devices and the desktop. Multiple M.2 or PCIe SATA connections. DDR4.
Sadly, I don't think Skylake is going to offer most of these.

 

[OC_DeMoN]

4 core pentiums coming? how about something like a 4 core g3258 on desktops? at the same prces?!

These are dual and quad core Atom chips branded as Pentium and Celery... these are not the same as the g-series Celery and Pentium chips for desktop products.

Confused? Well so is everyone else.

Celery?

lol

 

[Wisecracker]

Ahhh, Springtime.

Sun is shinning. It's getting warm. Birds are singing as thoughts turn to love.

And Intel hoists some more billion-dollar losing Atoms on an unsuspecting future back-to-school market.

 

[Pherule]

It's Rockwell Intel! Use the original name, I don't know why it was changed in the first place. Rockwell sounds way better.

 

[knowom]

"4 core pentiums coming? how about something like a 4 core g3258 on desktops? at the same prces?!"

Or hell, why not just have an unlocked I3? No it isnt going to be like an I5, but dammit I want to tweak stuff! Honestly it would be nice if the i7's moved to 6 core 12 thread chips, the i5's were 4 core 8 thread chips, the i3's were quad cores, the pentium were 2 core 4 thread, and the celerons were just as they are now, dual cores. But i'm asking too much and i have no idea what their yields are like or what the sales would be like. They could also sell a 6 core with no hyperthreading thats unlocked, but i have no idea where that would slot in. It would also be nice if mobile i5's were real i5's and not rebadged and higher clocked I3's, but it is what it is, dont even get me started on the 2 core 4 thread "i7's".

4 core g3258 would actually be faster than a unlocked i3 and have 1MB cache more cache on top of not having 4 physical cores for threading rather than hyper-threading.

 

[PaulBags]

If these are ready why put bay trails in their compute sticks?

 

[Wisecracker]

It's Rockwell Intel! Use the original name, I don't know why it was changed in the first place. Rockwell sounds way better.

Rockwell Automation likely thought so, too. Or, whatever is left of Rockwell defense industries -- previous designer of the Space Shuttle and Hellfire missiles, among other things.

This ain't rocket science, this is the Silvermont Bay Trail-M Atom replacement. The big thing these days is HEVC decode, and they didn't mention it. Not a good sign, especially given that these babies will hit 14- to 15w under 'duress' ... regardless of what Intel says. It starts to bump up against LP Haswell-U, and sure can't hang in graphics.

Double the graphics power of Bay Trail-M Atom lands roughly between the AMD A6-1450 Temash and the a4-5000 Kabini SoCs. AMD will answer with Carizzo-L and whatever comes after the 6700T.

If these are ready why put bay trails in their compute sticks?

You are not supposed to notice those things.

 

[photonboy]

Intel has done amazing things with POWER EFFICIENCY. Apparently using 25% the power in some cases with recent CPU's as what they offered just FOUR years ago.

Yes they also get more powerful on the desktop but the focus really is mobile. That said, I think most of the mobile advancements end up benefiting desktop power users.

I'm looking for an SoC in the near future that has H.265 decode, and works with "KODI" to play my video collection. Very few media players support the latest H.264 profiles. WD TV did but they dropped Netflix support (my WD box died).

Intel has lost MILLIONS of dollars trying to break into x86 mobile and it's unsure if they'll get back their money. Profit margins for x86 mobile devices will continue to shrink so they need to make a lot of profit now but competition is really high.

I am interested in a 12" Windows 10 transformer laptop/tablet device though which would have an Intel SoC like the ones mentioned above.

 

[mx_mp210]

Intel has done amazing things with POWER EFFICIENCY. Apparently using 25% the power in some cases with recent CPU's as what they offered just FOUR years ago.

They did with advantage of getting ahead in manufacturing process. 22nm v 32nm & 28nm , now 14nm vs existing 28nm & future 16nm chips from competitors. The architecture itself may not be equal efficient as on same die process when you compare.
The chips stated here are perfect candidates for tablet sized computers but I really have doubt about laptop sized machine.

The desktop consumer industry is declining every year with increasing usage of mobiles and their hand held availability vs laptops and desktops for daily & normal tasks. Plus Intel will not focus that much on desktop since they are pretty much happy with no competition, all they need is to keep upgrade path with little or no performance gains with die shrinks.

I agree with H.265 support but it'll take several years before it goes mainstream and supported by all hardware vendors in their chips.

 

[dextermat]

Junk for the fools.... too bad

 

[InvalidError]

Skylake is supposed to be the first real focus on desktop in a while, so it will be interesting to see what they put out. [...]
Sadly, I don't think Skylake is going to offer most of these.

I am not expecting all that much out of Skylake either, aside from the continued push for power-efficiency, slightly better IPC, some updated IOs and that's about it. While Skylake was supposed to be Intel's next push on the desktop after "dedicating" Broadwell to mobile, I doubt Intel intends to stray away from mobile focus in any meaningful way since that would mean more work for the next mobile-centric die shrink assuming Intel decides to favor mobile on die shrinks as they intended to with Broadwell.

The single biggest update/upgrade between Broadwell and Skylake might be Iris 6100 dropping from 23W on Broadwell to 15W on Skylake, hinting at major efficiency gains in the IGP department. Also of interest is that some people have benchmarked Iris 6100 and found that it is only slightly behind Nvidia's GT750.

 

[cats_Paw]

Im still waiting for software to catch up to multicores...maybe dx 12 will do it.
I mean, we got 8 core CPUS... and they run 10% faster than 4 cores and 20% faster than dual cores in some cases...
That is not a good sign .

 

[MumblinBerk]

Well,... on the bright side, this could leave a window of opportunity wide open for AMD to leap-frog Intel in the PC arena. Nothing like a little competition to motivate. Hell, I'm still rocking a 3570-K. Starting to feel a bit "dusty" to me. What shall we build next....

 

[Wisecracker]

... major efficiency gains in the IGP department. Also of interest is that some people have benchmarked Iris 6100 and found that it is only slightly behind Nvidia's GT750

If they would only put it in a reasonably-priced chip The eDRAM overlay is a pretty slick trick -- 85mm2 or so added to 177mm2 die (?) at 22nm. Hopefully, it hits the mainstream though I highly suspect Crystalwell and what follows will remain high-end extra process stuff not for the masses.

They did with advantage of getting ahead in manufacturing process. 22nm v 32nm & 28nm , now 14nm vs existing 28nm & future 16nm chips from competitors ...

The thing is ... we are reaching amazing parity in transistor density yield and mix. All these Fab guys are good. There is not really such a thing as "node-shrink" anymore as much as "process enhancement." Intel transistor density at 14nm is likely not substantially different than GlobalFoudries 28nmSHP. TSMC 20nm with the Apple A8 easily tops both of those.

The Samsung 20- 14nm FinFET, even better.

...I'm still rocking a 3570-K. Starting to feel a bit "dusty" to me.

You've been orphaned. Like AM3+, only better

Hard to say where AMD will end up with Zen next year but it will likely come from the Cat family (Puma+ cores with Excavator dense libraries?) with serious 256-bit 'Fusion Control Links' (and eDRAM last-level cache?).

That's the other thing. CPU core logic is becoming a smaller and smaller portion of any given die. That effects transistor density more so than anything. GPU transistors can be much more dense, and being that they're 60%+ of any given die these days, it only makes sense to find the density sweet spot.

 

[mx_mp210]

The thing is ... we are reaching amazing parity in transistor density yield and mix. All these Fab guys are good. There is not really such a thing as "node-shrink" anymore as much as "process enhancement." Intel transistor density at 14nm is likely not substantially different than GlobalFoudries 28nmSHP. TSMC 20nm with the Apple A8 easily tops both of those.

The Samsung 20- 14nm FinFET, even better.

"Other" foundries are just starting to move on to FinFET on their existing processes and with node shrinks. While Intel is already using FinFETs since past several years. Node shrink allows lower voltage depletion layers and as an effect takes "lower power" to operate.In addition it allows more transistors to be embed in the chip (Same die size with improved logic ). Regardless what they call their process (14,16,20,22 nm) There's a difference between Intel's and Other foundries Nodes.
Note that there's significant difference between Finned Transistor and Planar transistor. The process is done on types of FETs selected to build these chips. "Process" refers to applying impurities, Diffusion techniques, interconnection and manufacturing processes etc.

 

[InvalidError]

Regardless what they call their process (14,16,20,22 nm) There's a difference between Intel's and Other foundries Nodes.

And regardless of how they call their processes, many of the physical dimensions are not as different as the numbers imply. There was a time where the "node size" was actually related to physical feature size but for much of the last decade, it has become some arbitrary blended index of physical properties and performance that does not really mean anything anymore.