CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Hardware (
-   -   6core AMD 1090T or 4core core i7 930 or i7 870 for Fluent? (

gsg001 August 17, 2010 18:17

6core AMD 1090T or 4core core i7 930 or i7 870 for Fluent?
I am PhD researcher in the process of buying a new computer for Fluent.

What is the performance comparison between the Thuban architecture and Nehalem. With all the benchmarks it looks like the quadcore Nehalems with their 3 channel memory will be faster than hexacore Thubans with dual channel memory. In some bechmarks even the dual channel i7 870 seems to beat the thuban. Is this true?

Has anyone done any comparisons to compare Thuban vs Nehalem 1366 vs Nehalem 1156?
I intend to overclock them to 3.6 Ghz with 4 or 6 GB memory correspondingly.

I feel I could buy 2 Thuban systems for the price of one 1366 Nehalem system.

Any advice will be greatly appreciated. Thanks.

Dinocrack August 18, 2010 13:08

I also want to buy a new computer for my simulations and my doubt is about a xeon, and a Phenom 1099T.

I also saw some benchmark and phenom seem pretty nice performance for the price.

I also looking for advice. :)

CATFLUENT September 7, 2010 15:58

Performance using i7-950
The i7-950 just came down in price. We are purchaing the best 1366 system we could find for a reasonable price )($3200 CDN)

My research has determined that it is worth going for 1366 instead of 1156 because when you want to get a little more performance, then you get into the price/perf issue. It makes sense to go to 1366 instead of staying in the 1156 setup.

If your budget allows, the OC capabilities of the i7 series are awsome and you can always insert a new 980x or Tesla at minimal cost retaining the rest of the system.

Apart from the CPU, you need to decide on a board. We went with the Rampage III Extreme simply because we didn't see the need to go P6T7 WS (ASUS). Believe me, I spent so much time reading everything I could on all 1156 and 1366 systems that, after a while, I had both my eyes in the same socket! (and I needed glasses to boot!)

Seriously, if you're going to run serious models on your machine, you need all the perf you can afford. If oyu have complex or large systems (or both) to analyse. Tt makes sense to invest in faster results since the overall cost is minimal (compared to salaries, etc).

We are also investing in SSD, which is really going to help. Small investment and big bang, or so we think...

I can get back to you with perf results once we get our system running.

All times are GMT -4. The time now is 22:03.