News - ARM announces 28nm quad-core Cortex-A15 for notebooks

[HEXUS]

Send the design to TSMC and it should work right off the bat.

Read more.

[krazy_olie]

Ah, but will people want a windows rt laptop?
Only time will tell

[shaithis]

There is certainly a market for it. The biggest gripe most laptop users have is battery life.

Now, if only they can make all screens work like the kindle screen....

[Noli]

far exceeds the power of the xbox 360

holy sh1t - I hadn't realised that arm chips had advanced quite that much.
that really puts it into perspective!

It's not x86 of course though (then again is the powerpc?)

[Scribe]

far exceeds the power of the xbox 360

holy sh1t - I hadn't realised that arm chips had advanced quite that much.
that really puts it into perspective!

It's not x86 of course though (then again is the powerpc?)

PowerPC is also RISC-based architecture like ARM which is why I drew the very rough comparison.

[cheesemp]

far exceeds the power of the xbox 360

holy sh1t - I hadn't realised that arm chips had advanced quite that much.
that really puts it into perspective!

It's not x86 of course though (then again is the powerpc?)

You've got to remember the Xbox is ~7 years old now so its really quite slow compared to a modern i7 but yes they have come on a long way but yes give them a few more years and I'm sure they'll begin to match an i5/7 (of course PC's will be on the i777!)...

[miniyazz]

More to the point, how will it compare against a CULV or Atom notebook/netbook. I await benchmarks with interest.

[crossy]

Generating a very rough comparison against another RISK architecture

PowerPC is also RISK-based architecture like ARM which is why I drew the very rough comparison.

That's "RISC" not "RISK"!!!

Good idea, comparing with the PowerPC in the XBox360, although I'm sure that some would have preferred a comparison to the current Intel desktop gear.

(Not me, I'm happy with the XBox comparison.)

Sounds like this processor might make a good engine for a low-cost/size blade server?

[Scribe]

That's "RISC" not "RISK"!!!

Good idea, comparing with the PowerPC in the XBox360, although I'm sure that some would have preferred a comparison to the current Intel desktop gear.

(Not me, I'm happy with the XBox comparison.)

Sounds like this processor might make a good engine for a low-cost/size blade server?

I get a spelling into my head and it wont leave I drew the Xbox comparison because they're both RISC whereas it's quite hard to measure up DMIPS to real performance on CISC because some instructions do a lot more per cycle, also the numbers were pretty close. Will try to slap something together once benchmark figures are floating about, which when we have notebooks all running Windows 8 should tell a fairly acurate story.

[watercooled]

Comparing performance between architectures based on specs is bound to kick up a stink. Yeah you can compare DMIPS but like you say it can be miles off real world performance, not least because it can be heavily dependant on compiler optimisations and cache size of the CPU. And that's before we move on to FPU performance, where for example the Xenon still compares well to even modern desktop CPUs in terns of theoretical performance. Even POWER vs ARM has me pulling funny faces.

[anselhelm]

Given this kind of progress, I shall be one of the many Nintendo users who will be exceptionally displeased if the upcoming Wii U is less powerful than an Xbox 360 or only marginally more powerful.

If an ARM chip can push this kind of power, Nintendo's Wii U had better be a heck of a lot more advanced this coming generation, as a lot of users won't be duped twice by a woefully underpowered console, as good as it was in terms of first party games as usual.

[watercooled]

Well I've heard those rumours but I'd still be quite surprised if it wasn't - they could be started by MS/Sony though don't forget, so take them with a pinch of salt. It's said to use a CPU similar to the one used in IBM's Watson, so it's highly likely to be a POWER7-based chip which is a very capable architecture. GPU is also rumoured to be similar to AMD R770 so anything from 4830-4870 performance really (or possibly more/less depending how they modify/clock it on current manufacturing processes), which would also place it a fair bit ahead of the current console GPUs. But console performance comes down to more than the theoretical performance of the chips, so we'll have to wait and see really.

[Fraz]

Slightly off-topic, but these DMIPS comparisons are pretty interesting. An XBox CPU (blindingly fast back in 2004/2005) can do about 19,200 DMIPS, now trumped by a tiny ARM quad-core in 2012. Today, a high-end PC is about 10x that level. My PC, for instance, now over 2 years old can do about 185,000 DMIPs.

Assuming that graphics capabilities of a modern PC as compared to an Xbox are also 10x greater, it really makes me wonder how hobbled the PC gaming industry is, what with most games being designed to run on consoles. Or, do the tools they use to develop games allow them to properly scale things up nicely?

Also, if consoles are multicore, why are the PC ports often single core? Skyrim, for instance, uses only a single core. Hmmph.

[Scribe]

Slightly off-topic, but these DMIPS comparisons are pretty interesting. An XBox CPU (blindingly fast back in 2004/2005) can do about 19,200 DMIPS, now trumped by a tiny ARM quad-core in 2012. Today, a high-end PC is about 10x that level. My PC, for instance, now over 2 years old can do about 185,000 DMIPs.

Assuming that graphics capabilities of a modern PC as compared to an Xbox are also 10x greater, it really makes me wonder how hobbled the PC gaming industry is, what with most games being designed to run on consoles. Or, do the tools they use to develop games allow them to properly scale things up nicely?

Also, if consoles are multicore, why are the PC ports often single core? Skyrim, for instance, uses only a single core. Hmmph.

I've been wondering the exact same thing. Also, technology is different, consoles are all DX9.0c but we're now on DX11, in theory with tessellation you should be able to push some pretty impressive details even on a low-end device by focusing where detail is concentrated in a scene, however optimisation on this level isn't taking place yet and is appearing only as a way to add high-end detail, which I suspect is a side-effect of supporting consoles.

I can't really answer the multicore question, only that perhaps it's needed on consoles to squeeze extra performance but then why they wouldn't translate that code over to the PC I don't know.