Is I5 9400f processor is sufficient for deep learning, performing models,Monte Carlo simulation and other heavy works on Matlab
조회 수: 8 (최근 30일)
이전 댓글 표시
I want to purchase a processor for Matlab working. I have initially selected i5 9400f processor . I wanted to know that is it sufficient for my PhD work in CIVIL ENGINEERING (including future work) or shall I go for i7 8th gen processor. Or anybody who can suggest me an ideal processor for the same.
댓글 수: 0
채택된 답변
Michal
2019년 8월 3일
편집: Michal
2019년 8월 3일
All MC methods are in general extremely CPU consuming. So, soon or later you always reach limits of your computer. From this point of view I suggest most powerful CPU you can buy by your budget + additional RAM + fast SSD disk. In some cases GPU capabilities (cuda) are very useful.
Minimum config:
- CPU with 8 physical cores
- 32GB RAM
- SSD system system disk 256GB or more
- HDD data disk 1TB or more (7200rpm)
- NVIDIA graphics card - gpu-cuda computing
- Linux OS
댓글 수: 4
Walter Roberson
2020년 8월 17일
Often the question arises about whether to get fewer but faster cores, or more but slower cores.
The answer depends a lot on your computing mix.
Having more but slower cores only wins out if the majority of your work can be done in parallel and your problem size is such that you are still getting notable incremental improvements with adding cores.
There are some kinds of problems where adding more cores is almost always better: problems where the cost of transferring data into and out of the cores is small compared to the work done on the cores.
However most of the time the cost of starting the extra processes and transferring data to them and transferring back, exceeds the cost of doing everything in one process with automatic parallel processing. It needs special attention to the programming to take advantage of the parallel cores in a way that makes them a net benefit.
A lot of the time, you are not running that carefully programmed code, and at the other times, faster CPUs almost always win.
There is the old saying, "Never underestimate the bandwidth of a tractor-trailer hauling a load of magnetic tapes." There are some tasks that 1000 "slow" processors can chew through much more effectively than a couple of overclocked nitrogen cooled CPUs. But most of the time you are not doing those specialized tasks, and 8 hot-rods can be pretty sweet the rest of the time.
추가 답변 (2개)
참고 항목
카테고리
Help Center 및 File Exchange에서 Get Started with Statistics and Machine Learning Toolbox에 대해 자세히 알아보기
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!