A Pi 4 4 GB and a Pi 5 8 GB work well as file servers with the Pi 5 adding the extra power for a media server along side file server and backup server and some other things. I looked for benchmark comparisons. The Pi 5 is faster than network access so there is no practical use for more speed.
I have an old i7-2700K based computer with the highest single thread speed at the time of release 14 years ago. Overall it is faster than networks and faster than disks, including up to PCIe 4 SSDs. I can run monster programs and it will sit there waiting for the disks.
I am thinking about retiring the Intel machine to reduce electricity use for "always on"machines. Out of interest, I started looking at comparative benchmarks. Looking at some common benchmarks, the ones appearing at the top of search engine results, the Pi 5 is not included but the Apple ARM CPUs are included. Plus the M1 is given a miraculous high single thread speed despite having a slow clock speed compared to everything else.
Most of the modern CPUs appear to have many processors but pathetically slow clock speeds limiting their single thread speed. Some strange chips are rated with performance two to three times higher than their clock speed. Does this mean the benchmarks are bent towards artificial instruction selections to favour one chip over another?
Could it be the benchmarks are too simple to test real life speed?
The Pi 5 with PCIe 3 access to a good NVMe disk is fast for things like searching for text within files. This is the type of benchmark I would look at for the overall computer speed. For a CPU speed, I would use something more like a spreadsheet update or a Web page presentation or a LibreOffice document update.
I do not know of a benchmark where they actually run real world tests.
For processors limited by network speed, the Pi 5 is fast enough and I am looking at reducing 24/7 power usage from 160 watts to whatever the Pi 5 power supply delivers.
I have an old i7-2700K based computer with the highest single thread speed at the time of release 14 years ago. Overall it is faster than networks and faster than disks, including up to PCIe 4 SSDs. I can run monster programs and it will sit there waiting for the disks.
I am thinking about retiring the Intel machine to reduce electricity use for "always on"machines. Out of interest, I started looking at comparative benchmarks. Looking at some common benchmarks, the ones appearing at the top of search engine results, the Pi 5 is not included but the Apple ARM CPUs are included. Plus the M1 is given a miraculous high single thread speed despite having a slow clock speed compared to everything else.
Most of the modern CPUs appear to have many processors but pathetically slow clock speeds limiting their single thread speed. Some strange chips are rated with performance two to three times higher than their clock speed. Does this mean the benchmarks are bent towards artificial instruction selections to favour one chip over another?
Could it be the benchmarks are too simple to test real life speed?
The Pi 5 with PCIe 3 access to a good NVMe disk is fast for things like searching for text within files. This is the type of benchmark I would look at for the overall computer speed. For a CPU speed, I would use something more like a spreadsheet update or a Web page presentation or a LibreOffice document update.
I do not know of a benchmark where they actually run real world tests.
For processors limited by network speed, the Pi 5 is fast enough and I am looking at reducing 24/7 power usage from 160 watts to whatever the Pi 5 power supply delivers.
Statistics: Posted by peterlite — Mon May 12, 2025 8:24 am — Replies 2 — Views 94