Hardware update - a new PC

Published on 21 August 2024 at 21:18

It was time for a new PC. My old machine was from 2017 - it was not too bad. Here are some data from the PixInsight benchmark:

CPU Identification
CPU vendor ............. AuthenticAMD
CPU model .............. AMD FX(tm)-8320 Eight-Core Processor

System Information
Platform ............... Windows
Operating system ....... Microsoft Windows 10 Home
Core version ........... PixInsight Core 1.8.9-1 (x64)
Logical processors ..... 8
Total memory size ...... 15.967 GiB

Execution Times
Total time ............. 04:31.71
CPU time ............... 01:32.41
Swap time .............. 02:59.25
Swap transfer rate ..... 92.467 MiB/s

Performance Indices
Total performance ...... 1731
CPU performance ........ 4096
Swap performance ....... 512

The graphics Card was an NVIDIA GTX 1050 Ti 4Gb

The bottleneck was obviously the transfer rate to the SSD.

As I started development of AI based on deep learning with Tensorflow, I soon realized that training of my AIs takes much too long with this Methuselah. I saw the need for an upgrade and decided to configure a completely new one:

CPU Identification
CPU vendor ............. AuthenticAMD
CPU model .............. AMD Ryzen 9 7950X 16-Core Processor

System Information
Platform ............... Windows
Operating system ....... Microsoft Windows 11 Home
Core version ........... PixInsight Core 1.8.9-2 (x64)
Logical processors ..... 32
Total memory size ...... 63.217 GiB

Execution Times
Total time ............. 00:11.40
CPU time ............... 00:09.65
Swap time .............. 00:01.74
Swap transfer rate ..... 9537.135 MiB/s

Performance Indices
Total performance ...... 41250
CPU performance ........ 39240
Swap performance ....... 52822

Obviously this is quite an improvement :-)

The swap transfer was boosted by using a RAM disk - not sure yet if this makes sense for PixInsight - might need to find a balance between RAM and swap memory.

The graphics card is now an NVIDIA RTX 4060 Ti with 16Gb DDR6, which speeds up my machine learning with TensorFlow tremendously.

 

Add comment

Comments

There are no comments yet.