Cores vs # of Processors

"Xeon" is a brand name that's been applied to a huge variety of chips. The latter machine is using a newer chip that's significantly faster (marketed for desktop machines as "Core i7"). That said, I doubt that it's enough faster to be worth an extra $1k.

 

Definitely not. The bottom one is probably the best if it's a newer Xeon, but you need to tell us the specs of the other Xeon too.

 

2x2 vs 1x4 of the same core would perform very very similarly.

 

even if it did work like that, there are two processors in the first one so it would add up to 12 not 6. The ghz are not directly comparable between different families of processors (in this case new vs old models). My advice is to try the cheaper one out if thats possible and see if it is fast enough for you. No need to spend the extra money for the newer one if the old one meets your requirements.

 

1. It makes very little difference.
2. You might consider changing your username.

 

The newer system will be 20-30% faster for most things, games barely benefit (but will from faster memory) and some specialist things can get much bigger improvements. The 2.66GHz 2009 Mac Pro is also faster than the older model if $1,000 is a big step up and would be a better choice unless you need more than 8GB of memory.

 

+10 points


back to the original question...you should be able to get a Harpertown for 2k US which will as fast, if not faster (due to the faster front-side bus) for most tasks.

and if you could, go for the 2.93GHz Nehalem...it'll last longer...

 

what some people are trying to say is that the two processors you described in your original post are not the same family of processors.

the newer quad cores are significantly better than the previous quad cores and dual cores.

most of the advantage from the quad core would not be coming from being a quad core versus the two dual cores, it is the newer architecture.

 

Cruel.

Funny, but cruel.

Ok, mostly funny.

 

That is what I thought to

 

It's all down to how efficient the processor is with its caches with respect to its cores. Now go one level up, and it's also a matter of how efficient the memory controller (on-chip and/or off-chip) is at feeding those caches.

Once that is all cleared up, its all about processor throughput.

With all that in mind, the second processor the OP listed is the better performer.

 

But is likely that the other solution gives more bang for the buck. So it still depends of his preferences.

 

Isn't the whole point of putting all the cores onto one processor is to improve performance over having them on separate processors?

According to Cinebench scores, the 2.93 Quad is approx 25% better on single core and 50% better on multi core performance over the 3.0 Quad. Double the multicore performance increase compared to single core performance increase tells me it's better to have them all on one chip.

2.93 Quad - 4,037 (single core) and 15,182 (multiple core)
3.00 Quad - 3,162 (single core) and 10,527 (multiple core)

Plus you get cooler graphics card possibilities being a newer machine. Is that worth an extra cool thousand to you?

 

The part your missing is the first machine may be 2x2, but the bottom one is more like 1x8. The newer xeon uses hyper threading on each core, presenting 2 executable threads for every core. Its a much bigger power house then the 2x2. machine.

That said, both machines are excellent, get what ever fits the budget better.

 

No, it's primarily a cost reduction measure. It has some benefits for inter-core communication, but also downsides in terms of heat dissipation and connectivity to off-chip things.

This is very likely because the 2.93 has hyperthreading, although I suppose it's possible that cinebench has a lot of inter-core chatter. (seems unlikely to me though)

A quad with HT really isn't equivalent to a 1x8 machine though; virtual cores only go so far.

 

I was under the impression that Pro Tools didn't like Nehalem's virtual cores. I'll edit in a link to the discussion when I find it (soon).

EDIT - Here's the Link. In the ends, its officially supported, and one can effectively "disable" hyperthreading to make it run nicer then with it on, if hyperthreading is a problem.

 

There are major architectural differences between the two machines that are far more significant than the arrangement of cores.

The 2x2 is based on Core2 technology which relies on two dual core dies connected by the now obsolete front-side-bus architecture for processor to processor communications. It's terribly old technology that originated with the original Pentium processor over 10 years ago. It's bandwidth is limited and the two dies don't share a common cache meaning there is a lot of data duplication and inter-processor communications going on... definitely not a very efficient architecture. In addition, FSB bandwidth is shared between the north-bridge dual channel memory controller and both CPU's. Certainly not ideal.

The new Nehalem quad's are perhaps the most significant update to Intel's x84 architecture since the Pentium and use a monolithic quad (with hyperthreading) on a single die that share a common L3 inclusive cache. Add to this, the Nehalem tri-channel on-die memory controller and there is simply no comparison.

More info on Nehalem... http://nehalem-news.blogspot.com/2008/04/nehalem-faq.html

 

My mistake, 3000 to 4000 is a 33% increase not 25%. But that's on single core performance only. The multicore performance increase is 50%.

Purely arithmetic, you'd be buying 50% more multicore performance for 50% more money with the prices you mentioned, so both machines are actually priced according to their relative performance. Pretty much basically one bang per buck, so it just depends then how much bang you want, $1900 worth or $2,990 worth.