there are other reasons for buying a mp besides true
n core multi-threading. In the case of what apple offers, it's the only system that can allow for easily upgraded hdd's and pcie cards (not just upgrading the gpu, but allow for things like raid, fc, 10g ethernet, or even infiniband).
Now i'm not saying that users don't want or need true n core multi-threading, but the software a particular person is using may not provide it for all applications within the suite (most in fact, from my observations).
As it happens, creative suites fall in this category (i.e. Some parts of cs 5.5 are single threaded or for limited core counts, such as ps only using 2 cores, though there may be many more), as do some engineering and scientific applications (i.e. Based on ancient code).
Rare case. Seriously. I can't recall how many times commercially available software isn't capable of this across the entire suite.
Privately developed software (not commercially available), is where i see the exceptions. Which is expensive to do, and why it's rare in the grand scheme of things.
Actually, it's not. Cs 5.5 is a good example.
Specific applications are, but
not the entire suite. Call and speak to one of the design engineers of the suite in question, and you'll discover this for yourself.
Do not presume you remotely know what i do or what i require for a system.
As it happens, i no longer use a mp (got one in 2008, and returned it within the 14 day return due to the poor roi once the hardware upgrades and software used were taken into account).
I ended up building a custom solution that was cheaper and was better suited to my specific needs.
As it happens, not all of my software is true n core multi-threaded either, and i'm using engineering software (
electronic design automation). Given the type of software, a xeon is needed for the ecc memory (recursion = cannot afford a bit error in memory).
As it happens, about 50% of my engineering software is true
n core multi-threaded (which is higher than most professional suites i've researched), but i don't spend all of my time in those portions of the suites. Less than half actually, particularly when considering everything (email, research via a web browser,...) that use common, single threaded applications that aren't part of any engineering suite. So when i account for time spent where, a dp wasn't the best way to go (hex core xeon = better cost-performance ratio).
It's due to intel increasing core counts, as they can no longer push clock frequencies.
When the cores are leveraged, they do produce a faster result. But there are 2 parts to this, which comprise of hardware + software. Intel can only control the former. The software is up to the particular vendor, and intel cannot force them to change their code, assuming it can even benefit from true
n core multi-threading in the first place (not all can).
They're not all simultaneously executed due to how resources are used. They're "active", but in reality, most are stored in memory, and are actually idle (not processed in consecutive clock cycles).
The real key however to differentiating between threading in the sense you're using it and multi-core systems, is the
n core aspect, which denotes
multi-core systems. It's this latter case i'm talking about, and expected you realized that.
