I intend to buy a new computer. Multi-purpose. My old system has an AMD Athlon XP 2800+ CPU and uses Windows XP SP2. I've compared CPUs using http://www.cpubenchmark.net/ According to that, my old CPU scores 446. The i5 750 scores 4206. So the i5 750 would be 9 to 10 times faster. It's a quad core. If I were to divide that by four, a single core would be twice as fast as my old CPU. Taking into account Moore's law that is somewhat disappointing. The new OS would likely be Windows 7 64 bit. How much faster would it really be ? Traditionally, every new OS of Microsoft has been more resource hungry than its predecessor. Perhaps with the exception of Windows 7 ? Other CPUs are also possible, but I can't spend over 1,000 USD for a CPU. From what I gather, no inexpensive and fast 32 nm quad cores are available ? In the old days comparing the speed of CPUs was much easier. From 286 to 386, to 486 to Pentium.
Before I say anything about the cpu, I'd like to suggest one thing. If you really want to see difference in performance then you should really consider spend less for the cpu/other components and get an ssd as a boot drive. Also 4GB of ram I could say is the minimum for a 2010 64bit system. Now about the cpu. The cpu you have chosen is absolutely great. Only 2 considerations: 1. Does not have hyper threading and 2. the fact that has a relatively low clock speed ( take it under consideration only if gaming is really important to you ). For the rest I have read only great things about this cpu and actually many suggest it instead of some i7 models. Great prices too. About the performance difference. It will be hard to get an exact answer. Since you want to compare two completely different systems ( different OS and hardware components ). The sure thing is that the i5 will be a lot faster. With a good video card and a fast ram, hdd ( better ssd ) the performance boost will be important. It also depends if your applications can really take advantage of the multi core system and all the new technologies your new system will have. Sincerely I don't believe that you'll see x10 performance difference.
It depends what you want to do with the cpu. My test computer is also a semprom 2800 and serves that purpose fine. Where you will see a major speed increase is in video encoding but not on every codec. Some codecs like the h264 codec is design for quad core use. Some examples I encountered when encoding a movie using the h264 codec 2-pass mode 1. semprom 2800 x1 core = 3.5 to 4 plus hours 2. intel e6300 x2 core = 2.5 to 3 hours 3. phenom 9750 x4 quad = 1.5 to 2.0 4. phenom 11 630 quad = 1.15 to 1.5 hours For everyday use internet surfing/game playing a quad will be overkill and I've never notice any difference between the lower power cpu's and the quads. I got the phenon 630 on sale (cpu/motherboard combo) for $109.00 dollars total. Excellent deal for great performance. No need to spend big bucks to move into the quad world. I'm satisfied with the encoding speeds I get with the phenom 630, the intels I-7 would be faster but I can't justify spending the extra money for maybe a .5 hour speed increase. If it's a computer used in a business then the price would be justify. http://www.tomshardware.com/charts/...pdate-1/Mainconcept-Reference-1.6.1,1385.html
Thank you for your comments. My new computer should last at least 4, preferably more than 6 years. If I'll need something like a Blue-ray player or faster graphics I can upgrade later, but replacing the mainboard+CPU isn't that easy or cheap. A few questions: I could get a 128 GB SSD drive for about 260 euros. (344 USD) 128 GB is not a lot. It is not cheap either. Maybe it's better to get a conventional HDD and perhaps buy a SSD when prices have dropped a lot, assuming I'll need the faster SSD in a few years ? How important is hyperthreading ? Not just now, but a few years from now. Isn't buying a quad core without hyperthreading self-defeating ? What about a fast 32 nm duo core ? I presumed a 45 nm quad core is faster, but is it ? Of course, cost is an issue. When/why is clock speed important instead of relying on benchmarks ?
