Can we go for Dual Quad Core Processors around 2.0GHZ for Trading Technologies Gateway Server

To me performance is foremost but purchasing is not willing to spend that much so I have to find a middle way where I can keep up with both Performance and Purchasing Department.

How about if you get a single powerful quad core desktop cpu like the i7 and install a server OS on it?  It should be cheaper than server hardware, yet is very high performance.  You can opt for a high-end disk system to go with it.

That's strange, that cpubenchmark.net page has the E5410 higher than the E5520 yet Intel says the 5500 series is much faster than the 5400 series. I would take results on that site with a pinch of salt, the results are from PassMark Software results "easy PC benchmarking" - emphasis on PC. I think you'd be far better offf using results from spec.org or similar industry standard tests.

The benchmarks on that site are real-world results from people who have uploaded their runs.  If it looks non-intuitive, maybe some benchmarks from other sites would be in order.

A Xeon cpu does not differ much in performance from a desktop Core2Duo - the primary difference is multiple socket capability, but at the same clock, the same FSB and the same cache, they perform the same.  The Passmark benchmarks for cpus are much simpler than disk benchmarks, so there's no reason why spec.org would give you better results.

With SPEC each manufacturer submits one or two of each server/CPU and tunes it to perform well in the test. With the above you've probably got hundreds of tests done on some CPUs and onlt one badly tuned test done on others. Do they take an average or do they pick the best? The most expensive CPUs have probably only been submitted once by someone who didn't have a very good setup so the top CPUs are marked down through statistical error. There isn't an X5570 in the list for example whereas to prove their box is the best all the manufacturers submit server benchmarks with a pair of X5570s. Look at the VMMark results for example, sudden jump when the 5500 series came out but no similar jump on your user-submitted result site.

The site takes an average - if you hover the cursor over the model, you'll see how many samples were submitted.  The more samples, the better, of course, but the lack of an X5570 only means no one has yet submitted a run, and that's to be expected with a new cpu.  I think when they start showing up, they will reflect real world performance, and there will be statistical variations due to different motherboard chipsets and BIOS configurations.  My fellow hardware expert garycase finds nothing wrong with the site, either.  If you've found something that shows their testing is wrong, I'd be glad to look at it.  VMMark is an application benchmark and is fine for measuring how well a setup handles virtualization, but that is not necessarily a cpu-only benchmark.

When was "trading software" mentioned and how does the asker know that I don't have specific knowledge of it?  I have developed systems using the ION platform with real-time feeds.

Recommend http:#a24907073 as the answer to the question "Any suggestions if I want to go for may be AMD Dual Quadcore Opteron Dell machine in 1.7GHZ or 2.0GHZ it will make any difference?" - it WILL make a significant difference.