Opinion: Silverthorne fails but PowerVR impresses (+Montalvo trouble)
Wednesday 02nd April 2008, 03:24:00 PM, written by Arun
First, the Silverthorne core. If Intel's management and marketing personnel had bothered to properly analyze the handheld market rather than trying to figure out how to impress the press with empty rhetoric, they'd have concluded they shouldn't even bother. Silverthorne is a good design for UMPCs/MIDs/Ultraportables, but that's about it really. Any mobile phone manufacturer or carrier which is seriously considering making designs based on this architecture should seriously reconsider its strategic planning process.
Intel claims that the 1.6GHz Silverthorne is 4.1-6.5x faster than an ARM11 400MHz core at Internet Browsing. Great - too bad that doesn't seem to be much faster, if it's even faster at all, than 40nm Cortex-A9 implementations coming out in the same timeframe as Moorestown (which is the first chip based on the Silverthorne architecture that Intel will try to sell in the smartphone market). And with a TDP of 2W, it would take 4-6 times more power at peak and as much more on average. If you compare it to slower versions Silverthorne (and presumably Moorestown), Intel's chip will actually be substantially slower and still take more power.
Let's put it this way: I'm not impressed. The fact they went through so much effort on the architectural side of things (going back to CISC etc.) to improve power efficiency is even more telling of the x86 architecture in general. Even the 800MHz variant is very underwhelming in terms of power efficiency. This kind of problem won't stop the Intel marketing department from hyping this to infinity and back though by using the most ridiculous of metrics; their 'average' power consumption is measured under the maximum sleep mode (C6) on 80-90% of the time. Errr, yeah, sure, why not - but that's not very comparable to anything else, now is it?
Shifting gear completely, Intel also revealed much more information on the Poulsbo chipset. There's good news, and then there's bad news. The latter first: it's on 130nm. Yes, you read that right, the memory controller, 3D and video cores are on 130nm. Ugh. But then there's the good news: Intel is using PowerVR IP for the 3D & Video cores, and the video decoding capabilities and power efficiency are incredibly impressive and beat everything else in the market despite the process node. 120mW for H.264 Main Profile @ HD! One day, one day, I'll figure out why so much of the best semiconductor technology is designed in the UK (CSR, PowerVR, Icera, PicoChip...)
This roughly matches PowerVR's claims of 30-50mW @ 90nm (presumably pre-layout) for H.264 High Profile @ HD. Now, given that Moorestown will have the video core in 45nm, it does look like it will ironically have at least one of the lowest-power video decode processes in the industry despite not having a power-competitive processor. So much for benefiting from Intel's traditional strengths! We look forward to seeing how competing SoCs on the 40nm process node will compare. And obviously, we'd like an independent party to be able to verify Intel's claims on this subject.
On a very slightly related note, it looks like x86 start-up Montalvo is in big trouble and may be acquired by Sun - if so, this will make it substantially less likely for Asymmetric x86 to become mainstream. This is positive for GPUs, as it may make GPGPU more attractive for consumer workloads (as opposed to many-core x86) - however, in NVIDIA's specific case, it possibly removes a safety net and puts their x86 integration strategy at the mercy of VIA and a small number of (much leaner) stealth-mode startups.
I am honestly surprised NVIDIA doesn't seem to be bailing out Montalvo (cash infusion, not acquisition); I certainly hope they've thought that through and are sufficiently confident this course of action is in their best interests. It is difficult for us to say whether it is without more real data, but it is a very complex stategic discussion of which I obviously am not part. And in either case, we'll see how this all works out in the next few years - in the mean time, I wish the best of luck to everyone at Montalvo during this difficult period.
UPDATE: Made it clear that I'm referring to the Silverthorne architecture in general and all of its derivatives, not just the specific chip announced today which I know is not aimed at the handheld market. However, future derivatives will be and those are the chips which I claim will not be competitive at all there.
Tagging
Related intel News
RWT: An Updated Look at Intel's Quick Path Interconnect
32nm sixsome over at RealWorldTech
Intel Core i3 and i5 processors launched
Analysis: Intel-TSMC announcement more complex than reported
Intel and TSMC join forces to further Atom
Fudzilla: Intel 45nm Havendale MCM replaced by 32nm+45nm MCM
Intel announce Core i7 processors, reviews show up
Intel's Aaron Coday talks to Develop about Larrabee
Larrabee to also be presented at Hot Chips
