Intel To Expand Core M Broadwell Line With Faster Dual-Core Processors

MojoKid writes: Intel didn't waste much time following-up on its initial Core M lineup launch. The company has added 4 more Core M models to its roster. Like the launch chips, these four are dual-core designs that support HyperThreading to enable an effective four logical threads for processing. Also like those earlier chips, these are spec'd with a TDP of 4.5W. These new chips, however, are generally faster than the launch models, with a new top-end processor called the M-5Y71. This chip has a base clock speed of 1.2GHz, but is burstable through Turbo up to 2.9GHz. What really sets these chips apart from the initial Core M models is that their TDP is scalable, based on what the builder is looking to do with it. If the chip is set to be used in a notebook with very little free space, the OEM could opt to drop the chip down to 3.5W and lose 600MHz in the process. By contrast, a bulkier notebook could handle a hotter chip better, so a higher TDP could be decided upon. If that route's taken, any one of these new chips could peak at 6W and add 200MHz to the base and top-end clocks.

Comments

Intel's new Tock-Tick release cycle ...

[systemBuilder]

For so many years in a row, Intel has been making faster and faster processors. This year, for a change, they have decided to focus on making only slower processors, and the Broadwell series is the result! This year they are slowing down the CPUs, next year they will slow down the system bus ...

Re:Intel's new Tock-Tick release cycle ...

[Anonymous Coward]

all to better compete head on with amd

Re:Intel's new Tock-Tick release cycle ...

[WarJolt]

AMD? Arm is the real competition.

Re:Intel's new Tock-Tick release cycle ...

[ArcadeMan]

If they keep lowering their CPU clock speeds like that, their competition is going to be Atmel.

Re:Intel's new Tock-Tick release cycle ...

[hairyfeet]

You make a funny but frankly there realy is no reason for AMD to follow Intel down the rabbit hole of ever lower nm sizes (which if rumors are true is ending up with worse yields and lower clocks do to leakage with each rev) because what Intel doesn't want to admit that AMD seems to have accepted is that CPUs have gone waaay past "good enough" and into insanely overpowered for all but the handful that are doing jobs that stress a CPU to its limit and those folks would be better off with a dual socket workstation anyway.

What jobs do Joe and Jane Average have that won't be well served by a C2Q or Phenom X4 from 7 years ago? None, not a damned thing, in fact many can get by just fine on a C2D or Athlon X2 and never notice any difference because they just aren't stressing the chips. Hell its even true of the gamers who traditionally were the first adopters, with the first gen i5s and Phenom II X4s and X6s able to play pretty much any game out there when paired with a $150 GPU. When the MHz war was in full swing I was getting rid of my PC for a new one every other year with a major upgrade at the halfway point, but now why bother? I got 8GB of RAM, a 6 core CPU, a board that will take up to 4 GPUs in crossfire and 3TB of storage for my games so why waste money when it'll already play everything and do everything I want?

This is why I'm not worried about what Intel does even though I'm an AMD exclusive shop, because a dual or quad APU laptop or an APU or CPU based desktop does everything my customers want it to do and will last them for many years. this is why I've branched into home networking and HTPC setups as the days of the 3 year upgrade cycle are well and truly over. Intel can fill their bins with bad chips chasing 0nm all they want, the simple fact of the matter is computers have become like washers and dryers, no need to replace 'em until the previous one dies. I predict ARM will be in the same boat in 2 years or less as like X86 they are ramming into the thermal/power wall and soon won't have anywhere else to go, hence why we are seeing 8 core phones and tablets and the local Walmart is selling dual core tablets for $48. ARM devices are already becoming overpowered compared to the jobs ordinary folks have so all they can do is throw more cores or lower the prices. But that will only take you so far before you are in the same boat as X86, more cores than most need and prices so cheap everybody has more than they can use.

Re:Intel's new Tock-Tick release cycle ...

[mlts]

What I think we will see once Intel and ARM start hitting a wall with nm sizes is a race to throw as many cores on a die as possible.

Next step after that will be bigger caches and better caching algorithms.

After that, the next step will be to have special purpose cores. For example, some phones have two low-power/low-speed cores, and two faster/more energy using cores. I wouldn't be surprised to see that, as well as cores that are dedicated to specific tasks, but general tasks can be put on those cores if need be. For example, a core for AES, cores that are optimized for floating point operations, but can do integer math, GPUs, a specialty core that is intended to be just for security sensitive processes (perhaps even running a Harvard architecture so that heap smashing is not possible), maybe even FPGAs.

Having a lot of special purpose cores in addition to general purpose tasks, perhaps with a scheduler that can tell the OS to use proper weighting when selecting core affinities will be the next performance gain. However, this will come later, as it requires "knowledge" of tasks to not just be at the CPU branch prediction level, but at the OS scheduler level.

Re:Intel's new Tock-Tick release cycle ...

[FlyHelicopters]

What jobs do Joe and Jane Average have that won't be well served by a C2Q or Phenom X4 from 7 years ago? None, not a damned thing, in fact many can get by just fine on a C2D or Athlon X2 and never notice any difference because they just aren't stressing the chips.

That is true to a point...

Tossing a SSD and more RAM into such a machine, Windows 8 runs just fine on it, for most people...

Except, it also would suck down more power than it needs to... The same performance that took 65w or 95w of power in 2006 today can be had at 15w of power...

Those AMD chips you love so much suck down power like it was going out of style...

Re:Intel's new Tock-Tick release cycle ...

[tlhIngan]

After that, the next step will be to have special purpose cores. For example, some phones have two low-power/low-speed cores, and two faster/more energy using cores. I wouldn't be surprised to see that, as well as cores that are dedicated to specific tasks, but general tasks can be put on those cores if need be. For example, a core for AES, cores that are optimized for floating point operations, but can do integer math, GPUs, a specialty core that is intended to be just for security sensitive processes (perhaps even running a Harvard architecture so that heap smashing is not possible), maybe even FPGAs.

Most ARM SoCs already have that, actually.

AES cores are called "security accelerators" and practically all modern ARM SoCs have them for security - either with full disk encryption or just offloading the crypto tasks. Floating point - ARM has the Vector Floating Point (VFP) unit for some time, though most people use NEON, the SIMD version of same.

Security sensitive processes? ARM has that too - TrustZone, and many SoCs even have a low-end ARM7 or ARM11 that stays in secure mode and handles everything involved in bringing the system up. It's the first processor that starts up, which then initializes peripherals, RAM, then boots the big powerful cores that you associate with the SoC. The big powerful cores handle the user level stuff, while system level stuff like power control is often passed to that system controller core.

Re:Intel's new Tock-Tick release cycle ...

[chmod a x mojo]

What jobs do Joe and Jane Average have that won't be well served by a C2Q or Phenom X4 from 7 years ago? None, not a damned thing, in fact many can get by just fine on a C2D or Athlon X2 and never notice any difference because they just aren't stressing the chips.

Bullshit. Maybe, MAYBE they are not stressing the C2D / C2Q chips if they only browse the net / read and reply to emails. Otherwise stuff like watching 1080P, and to a lesser extent 720P H.264 10bit videos ( or even 8bit, without GPU acceleration ), which more and more videos are in now, pushes a C2D into the 80-90% utilization range. Then they can't do any background tasks on the CPU when watching movies.

Even if it does do everything you need, the thermal profile is horrid... I moved away from my Prescott that did everything I needed to a C2D simply because I didn't want the thermal profile of the throat of an active volcano on my motherboard anymore, or the fans that sounded like jet engines that cooled it. The benefits of less heat output and less power input by far outweighed the performance gains ( those had been only a bonus ).
No one can know if we are going to hit a thermal wall anytime soon, the next breakthrough could be happening right as we discuss. Add that to the fact that Intel is reaching ARM power profiles with similar, or higher, clocks, and that x86 / x86_64 vastly outperforms ARM on a per clock-cycle basis and Intel has nothing to fear.

And congratulations on being an AMD only shop, heaven forbid your customers have any kind of choice in the matter. AMD hasn't made anything more than an "adequate" chip since the K6. Can you name one x86 based laptop / net-top / tablet that uses AMD chips and boasts long battery life made in the last few years? I've seen quite a few Intel based ones that boast 6-7+ hours VS. the AMDs that are usually rated at ~5+.
You may think I'm "trolling" asking that, but I am actually quite serious, if there is some AMD chip out there that is really power efficient I would like to know, it's just that the only really efficient stuff I have seen recently was all Intel.

Ouch

[harryjohnston]

Very cool technically speaking, and good for system designers ... it will, however, make it that much harder to comparison shop, if the same CPU has a different speed depending on how it's wired up.

Re:Ouch

[Anonymous Coward]

Realistically speaking... designs with constrained cooling options have been thermally limited for quite a time now. This is most obvious on thin (and of course fanless) tablets (where ARM rules), but becomes increasingly visible also on thin laptops. Thus, you can't really judge all performance measures of a sub-laptop device by CPU model number alone; it is already a puzzle which involves understanding thermals on a level no consumer can be prepared to look at.

Re:Nice

[Wing_Zero]

Well, if the peak (judging from the last sentence of the summary) is 6w, that is easily in the power envelope of USB. (6w = 5v@1.1a) so, all they mean is a larger Passive cooler. You probably don't need a Heat-sink-fan unit until about 10-15w

Re:Nice

[TubeSteak]

Well, if the peak (judging from the last sentence of the summary) is 6w, that is easily in the power envelope of USB. (6w = 5v@1.1a) so, all they mean is a larger Passive cooler. You probably don't need a Heat-sink-fan unit until about 10-15w

There might not be much room for overclocking, since we know Intel has been having lots of trouble with their 14nm yields.

It's possible that they're rolling out low power dual core chips because that's all they can produce in any significant volume.

Don't get too excited

[Solandri]

This chip has a base clock speed of 1.2GHz, but is burstable through Turbo up to 2.9GHz.

If it's like their first M processor, the turbo boost mode only works when using a single core. i.e. You can run one core at 2.9 GHz, or you can run both cores at 1.2 GHz. That's the price you pay for the extremely low TDP. In contrast, an i5-4250U has a base clock of 1.3 GHz, can turbo boost to 2.3 GHz on two cores, and 2.6 GHz on a single core.

Intel Maths

[rossdee]

"(6w = 5v@1.1a)"

Are Intel still having problems with their floating point maths?

5V x 1.1A = 5.5 W

Re:Not USB powerable

[Khyber]

Uhh, USB 3.0 has a battery charge specification that allows for 1.5A