I am also under the impression that Intel is behind the Mac Pro logic board design.
It would hurt Intel to make full systems IMO, as they're then competing against their own market (system vendors). They're an ODM and component maker afterall...
That would benefit everybody by pressuring Apple in providing good value, even if they serve only the high-end of the market. I like the Mac Pros design, Mac OS X, and even its form factor.
MP's are workstations, and the XServes are servers. Both systems are enterprise systems, where appearance has taken a distant backseat to other considerations (noise as well, especially for rackmount servers). Apple is an exception, but others are getting better.
Appearance is in the eye of the beholder (I'd think most would like the MP's look for a desk over other systems available), but none of them use an all aluminum case AFAIK either. The Sun's look fairly decent to me for a workstation, but another could think it's the worst thing to ever sit in front of them.
But from a hardware perspective, there's competing systems that are as good in quality, and better in performance (i.e. capable of using more memory and at a higher clock speed - 1333MHz DDR3 RDIMM/UDIMM sticks, and up to 9 per CPU).
But if I wanted an alternative, it is either a do it myself PC (which I think is the most sensible)...
This is what I actually did, as my needs were specific enough it actually made sense, and was more cost effective (due to the lack of external RAID enclosures by using a full tower case).
Apple may use Foxxconn for doing the grunt work, but at least they keep the profits in the US and employ marketing and sales forces in the US.
Partially, but last I saw, half (50%, possibly more now), is actually flowing overseas (compontents, board work, assembly labor,...).
The funds (not all profits) retained in the US are dispersed amonst share holders, employees, and other things (i.e. insurance, power,...). Not that much goes back into the US economy via employee or shareholder spending, as their small in number (especially in context to the US population - per capita). And this is repeated by most electronics companies based in the US.
...Designed in California. Made in China.
Technically accurate, as Intel's headquartered in California.
Apple (and Sony) put the consistent and heavy marketing behind their products and the marketing leads to the mark up and provides a brand experience.
There's more to "brand" and "user experience" than marketing though. System Integration of the hardware and OS (and apps in Apple's case). That's where you can eliminate things like driver issues, unify the appearance of the OS,... It's much easier to do in a closed system such as the MP, but it's not impossible for an entity like MS either. Just
much harder, even in their own products, as the development cycles aren't in sync (for example, it can result in differing UI's from app to app or apps to OS).
Fix the user experience to something consistent, and the marketing can take that and make a brand out of it, rather than just a developer label (i.e. name slapped on a box or piece of software they didn't develop, or at least not entirely).
If you are talking about the recent news around an Intel 48 core chip, it's a test chip to give software developers to work out better ways to take advantage of multicore.
This one seems less of an experiement to me, unlike the Tera-Scale (Proof of Concept 80 core processor). The Bangalore (48 core) would seem to be on their roadmap, and is aimed at cloud/cluster computing. Given the move to virtualization, and Intel's move to high core counts, it would make sense that we will see this one, or possibly one built from it (seems like an Itanium family candidate to me
). Hard to tell, as Intel is more secretive on that line than the others (far fewer press releases).
If you've spotted a particular article that gave a bit more information than what I've found, I'd appreciate it.
However, if you want to go down that path, the expectations are that Gulftown will deliver more cores at the same or even lower clocks. If your workload can benefit from high-clocks, you may be better off buying a 3.3GHz Bloomfield vs. what will likely be a 2.4GHz Gulftown. Time will tell and at this juncture, I agree, it's worth waiting to find out.
More information is definitely needed, but I do think for the same (approximate anyway) pricing, the clocks will be a little slower. The idea that we'd see a 2.4GHz part instead of 2.66GHz would seem to make sense ATM.
You are right, I think they will also come in around the 2.4GHz speed, give or take a few... A single 3.33 Bloomfield machine will probably be faster at Photoshop (or any app that can't take advantage of multi core well) than the Gulftowns. These will be audio and render powerhouses, not daily task speed demons.
The multicore chips coming out won't be for workstation use. Those are going to be modified consumer models. I.e. ECC may be added to the top end consumer chips, as the Xeon 35xx are i7-9xx parts with ECC added. I don't think this was a fluke/stop-gap measure to fill a need, but the beginning of a trend.
Intel's expecting those needing more cores to go with clusters it seems, given the direction they're taking. Power computer users will now be called "Farmers"... the next wave of power computing.
Edit: I think dual Gulftown systems could be $6000+ ?
I'm thinking the top end models will be in the $6k mark with few upgrades (say graphics only). So definitely not aimed at the enthusiast user.
The 2.4Ghz speeds were an engineering sample, the speeds that can be achieved with the move to 32nm, even with 2 more cores, will be higher than currently available. Probably see a 1 stage multiplier bump over what we currently have and maybe the price bracket move up once stage, with something extra above the $999/$1600 spots taken by the top 4 core processors.
I've a feeling there will be a 2.4 GHz clocked part though. Definitely higher, and possibly a couple slower (L family). No idea on costs, but it's not looking good in terms of reducing the cost of MP's and Xserves.
