why matlab slow on more good spec PC

i change my hardware to

12th Gen Intel(R) Core(TM) i7-12700 2.10 GHz to i9-11900H (2.5GHz)

RAM 16.0GB 32GB

GTX1660TI RTX3060TI

1TB SSD 1TB SSD

but new PC is slower than old PC why it happen???

old pc took 5min to reading data(size 1098000x21) csv format

but

new pc take more than 6min

what should i do? to make it fast

[Matt J]

Maybe your new SSD is slower? We haven't been shown specs on your new drive or bus.

[Rik]

You also went from a 12th gen to an 11th gen. I don't know how the IPC developed between the two. It seems unlike to make the 400MHz difference disappear.

What code are you using? Changes in code tend to be much more effective than hardware improvements.

[JIHOON KIM]

i just use readtable(); to load csv file

Answer 1

[Nathan Hardenberg]

It seems that you went from a desktop PC to a notebook. Your "new" i9 processor is actually about 20% slower than your "older" i7 if you want to believe UserBenchmark.

There are multiple reasons for this:

• As @Rik pointed out, you went back a CPU Generation
• The H-processors are notebook CPUs and draw way less power than a desktop CPU and are therefore slower

The following reasons can contribute but don't necessarily have to:

• Not enough cooling on the notebook
• Notebook not set to best performance (in Windows)
• Notebook not connected to a charger

I understand your confusion though. Customers just look at the front of the processor name (i5, i7, i9) and think a higher number would result in better performance. Which should be like this, but is not. The generation is very important and notbook CPUs and GPUs are always slower than their desktop equivalent.

But you're in luck since intel H-CPUs are the fastest notebook CPUs (from intel). U-CPUs are the slowest and P-CPUs are a midpoint. This is done, because 'slower' also means better battery life.

[Rik]

Good catch, I missed the mobile vs desktop part. That is absolutely the reason. It is a lot worse if you look at GPUs, but with CPUs this performance difference is even smaller than I would expect.