It looks like you're new here. If you want to get involved, click one of these buttons!
The cheapest 6-core Gulftown desktop processor was a little under $600. Sandy Bridge-E, a little under $600. Ivy Bridge-E, a little under $600. See a pattern here? Neither do prices drop as the part gets old.
Well, here's a 6-core Xeon E5, which is the same die as Ivy Bridge-E:
http://www.newegg.com/Product/Product.aspx?Item=N82E16819116934
And it's much nearer to $400 than $600. Granted, that's a lower clocked version--but with a lower TDP of 80 W, also. But for an enthusiast part with an unlocked multiplier, does the stock clock speed even matter?
So what makes Haswell-E any different? Intel is bringing 8-core processors to desktops this time, so they're no longer relying on 6-core to be their top of the line. Granted, Intel has sold chips with a lot more than 6 cores for quite some time now. Here's a 12-core Xeon E5, for example:
http://www.newegg.com/Product/Product.aspx?Item=N82E16819116926
If you're willing to look at a different platform entirely, there's also the 15-core Xeon E7:
http://ark.intel.com/products/75258/Intel-Xeon-Processor-E7-8890-v2-37_5M-Cache-2_80-GHz
So it's not that Intel can't make chips with more than six cores. It's that they choose not to bring them to desktops. But Haswell-E is reportedly bumping the limit to 8 cores in a desktop, not 6. Hence my question: does that drop prices on Intel 6-core CPUs? One rumor says it does.
Comments
Right now, I'm betting no.
Sure, AMD has more "cores" but they aren't serious competition in the desktop arena. That and there is no pressing need on the desktop front for more cores - not like there is the Server end anyway.
Most software still runs great on dual core, why marginalize the margin on your upper-tier parts unless you have some motive (like trying to sell a multicore aware compiler or libraries something).
Once AMD starts to get a bit more serious competition - once they start to push 10 and 12 core Desktop variants that are compelling at attractive price points, then we'll see Intel start to push their prices down and core counts up seriously.
There's also the case that Intel gets "stuck" - Haswell wasn't much faster than Ivy that wasn't much faster than Sandy... if that trend gets worse with Broadwell, then the only way Intel can really show "improvement" is by upping the core count (that's the boat AMD has more or less been in since Bulldozer), and then we may see prices come down as Intel tries to sell more cores as an alternative to faster cores (like they did with their big push to quad core back with Conroe - which was right after the infamous Prescott, which was where Netburst hit the end of the road). If Haswell was more or less the end of the road for the Core architecture, we could see another branch out into core counts with Broadwell to get increasing performance.
Would you believe that the intended use of that part is to put 8 of them into a single server? That's Intel's top of the line: an eight socket server at 15 cores per socket and potentially enormous amounts of memory. It's a very small market, and one that AMD abandoned years ago, but with a six figure price tag per server, Intel does make money off of it. And if you really need top of the line hardware, you pay what it costs.
That's why there's been a push toward more distributed computing: if you can get the same performance out of twenty separate servers that cost $2000 each, you don't want to pay well into six figures for Intel's top of the line. But whether you can get that same performance split among more servers depends on what you're doing, as communication over ethernet, infiniband, or fibre channel is much slower (both in terms of latency and bandwidth) than inter-socket communication within a single server.
I really don't know much other than very general stuff. I did know that they were multi chip often enough. That prior to multi core cpu's people made multi chip desktops a lot more. I have seen a small server farm when I was in high school. A friend was running his own local isp by 11th grade. Don't know if I would call him a prodigy or not but let's just say he had many talents by that age that made him a lot of money considering the tech bubbles ect. Backed by his dad's money but still the knowledge was all his. Just never saw a price tag for the chips alone and caught me by surprise.