Advice needed: Buying a PC
I am planning to buy a new PC. I am currently writing my own NS Fortran code (3D & structured). I think at the most I would need about 500*200*50 grids, which would probably need about 1GB meomory. But the problem is that the simulation time is very large, so I would need as much speed as I can get.
I would very much appreciate your help about these questions:
1. Does any one have experience with AMD cpus? Are they better than Intel or not?
2. What about CORSAIR 1600 (-Dominator) Ram? It's price is about $100 higher than CORSAIR 1333. Does it influence the efficiency of the PC or not?
3. Which one is better? "Intel Core 2 Duo E8400 - 3.0GHz " or "Intel Core 2 Quad Q9550 - 2.83GHz"
The first one has 6MB cache and the second 12MB. Do you think the Q9550 is worth the price?
Thanks in advance
1. at the moment I would stay away from consumer AMS CPU's, for high end CPU's its another case
2. Every aspect of your I/O will influence your final speed. RAM, harddrive, bus speed etc... all influence your performance. SO dont ruin a perfect system with a slow 5400RPM diskdrive
3.Only go for multicore CPU's when you can solve in parallel (there is no other benefit to this)
lowest system specs: go for a fast intel i5 quad, 4Gb RAM, 12k hard drive and DDR3 1600Mhz
mid-end:xeon 5500 series or i7, solid state HDD, 8Gb RAM, 64bit OS, DDR3 1600+
Thanks for the help Bram.
Could you please help me with these questions too?
1.What do you mean by I/O? Does it mean input/output?
2. Is the performance of the i7 cpus much better than the Intel core2 quads?
3. Does having a 64 bit OS or a 32bit make a difference? Is the 64 faster?
1. yes he means input/output. He is basically saying the chain is only as strong as the weakest link (computer is only as fast as slowest component!)
2. yes, i7s are faster than core2quads. I imagine if you go to a website such as Tom's Hardware you can find lots of benchmark test comparisons
3. It certainly matters if you plan on using more than 3.5 gigs of memory (32-bit operating systems** have a ~3.5 gb limit)
that is where the speed difference between 32-bit/64-bit comes from
** from my understanding, you can apply changes to the linux kernel to "fix" this. I have never really looked into this though
|All times are GMT -4. The time now is 23:34.|