move /var/log to a RAMDISK

I gave that away sometime in the late 80s.

My Pi Zero W came with 'noatime' configured

root@heating-controller:~# more /etc/fstab proc /proc proc defaults 0 0 PARTUUID=b8c9fbb7-01 /boot vfat defaults 0 2 PARTUUID=b8c9fbb7-02 / ext4 defaults,noatime 0 1 # a swapfile is not a swap partition, no line here # use dphys-swapfile swap[on|off] for that tmpfs /var/www/data/volatile tmpfs nodev,nosuid,noexec,nodiratime,size=1M 0 0 tmpfs /var/ramlog tmpfs nodev,nosuid,noexec,nodiratime,size=25M 0 0

It seems Raspios has pretty decent defaults.

This is good because I will be reading files a * lot*. Probably 16 a second

PS I also managed to find out from the net how to set the PI Zeros USB port to look like an ethernet adapter and service DHCP requests, so it can be a web sever to a PC just by plugging it into the PC's USB port. Great for setting up the wifi! If anyone else wants to know, ask.

IIRC it used to be well known that SD cards (class 10 cards?) that were sold for use in video cameras etc. had: (a) large storage blocks and (b) no wear levelling

In consequence these cards are not generally useful for use as computer filing systems, being designed to hold large (MB to GB) data files that would only be written to sequentially and would usually not be edited in situ, just deleted in order to make room for recording another video. In consequence only file level operations (record a video, copy or delete a complete file) worked at an acceptable speed.

Slackware used to do BogoMIPS at startup (perhaps still does) but then went on to say measuring BogoMIPS was a waste of time.

That's what I do

My cmdline.txt:

console=tty1 root=PARTUUID=72020302-c396-2148-9d13-caf4f68cd18c rootfstype=ext4 fsck.repair=yes rootwait quiet splash plymouth.ignore-serial-consoles net.ifnames=0 biosdevname=0 usbhid.mousepoll=0 dwc_otg.lpm_enable=0

The PARTUUID is that of half a 120Gb SSD (/dev/sda2). /boot is on a

It was (seemed) easier than trying to get USB boot to work (RPi3B).

Oh that brings back a memory from a long time back. On a visit to Altos I saw a machine that one of the engineers there was playing with - it was a XENIX machine with four eight inch floppy drives as storage - the labels stuck over the drives read (/ /usr /etc and /tmp) - He'd had to turn of atime because it was wearing out the tracks holding the inodes.

Agreed. BogoMIPS is useless for showing the cpu speed. It's much faster than I expected, which allowed me to use it for much more than simple testing of aarch64 installs, which is all I bought it for.

Regards, Dave Hodgins

Agreed. I use multiple ssd drives on my main system. The difference in speed between a spinning rust drive and an ssd drive is well worth it. As I do a lot of installs and other testing on that system, the speed is worth it.

I have as my main drive an OCZ-AGILITY4 240GB SSD drive that I've been using since 2011.

I back it up regularly as I've been expecting it to fail anytime for years. :-) smartctl shows ... 5 Reallocated_Sector_Ct 0x0000 100 100 000 Old_age Offline - 1 9 Power_On_Hours 0x0000 100 100 000 Old_age Offline - 89199 12 Power_Cycle_Count 0x0000 100 100 000 Old_age Offline - 489

The Reallocated_Sector_Ct has been at 1 since shortly after I started using it. The spinning rust drive it replaced only lasted about 5 years.

For the rpi 4b, I'd be looking at an external enclosure connected via usb, so it wouldn't see the speed of a sata or pcie ssd, but would be a lot faster than the sd card, but for what I use it for, not worth the price.

Regards, Dave Hodgins

BogoMIPS on ARM is known to show incorrect results. The CPU tricks the routine that Linux uses to calculate it, so at least it's meaningless compared to the number given on architectures like x86, or actual Million Instructions Per Second. Somehow the Linux developers haven't been able to fix this.

"In 2012, ARM contributed a new udelay implementation allowing the system timer built into many ARMv7 CPUs to be used instead of a busy-wait loop. This implementation was released in Version 3.6 of the Linux kernel. Timer-based delays are more robust on systems that use frequency scaling to dynamically adjust the processor's speed at runtime, as loops_per_jiffies values may not necessarily scale linearly. Also, since the timer frequency is known in advance, no calibration is needed at boot time.

One side effect of this change is that the BogoMIPS value will reflect the timer frequency, not the CPU's core frequency. Typically the timer frequency is much lower than the processor's maximum frequency, and some users may be surprised to see an unusually low BogoMIPS value when comparing against systems that use traditional busy-wait loops."

Perhaps it might seem as fast as a 486 running Linux from the mid

I consider it to be a "valuable artifact" :-)

At minimum, it showed how serious and deep they'd go with the early PCs. That's when programming was still painfully "real". Mostly you could not buy what you needed - so you had to make it yourself.

And I *did* find many uses for that info.

That seems a logical best solution for an early Pi. The later ones will do USB boot, I think.

(where on earth did you get a 250MB SD card: I cant find anything much less than gigabytes these days)

Oh, so did I. I was assembly programming 8086s then - writing BIOS extensions or complete BIOSES for them,

Or interfacing with special hardware people had built that needed drivers written. But that line of business dried up, so I ended up doing yet another career shift into running IT businesses.

IBM simply provided the same level of documentation for the PC as they did for their mainframes as a matter of habit I think. That made life a lot easier for the clone makers who popped up like mushrooms almost immediately.

MIPs was a pretty useless concept long before Linux appeared.

Seems the place I got them from is now out-of-stock on all SD cards.

Anecdotally, GeekBench seems reasonable.

AMD Ryzen 5 7600, 2701, 12071 Orange Pi 5, 841, 2911 Intel i5 2500K, 634, 1903 Intel Core i5-750, 433, 1223 Intel Core 2 Quad Q6600, 271, 764 Raspberry Pi 4 B r1.4 253, 617

I was actually surprised at how fast the new Ryzen was. Maybe it is time to buy a new PC.

I haven't done that in a long time. I used to buy parts and make them but these days I buy "refurbished"* ex-corporate machines which are cheap and far better made than the typical consumer PC.

• Read cleaned, OS reinstalled and stripped of asset tags.

Hence the acronym: Meaningless Indication of Processor Speed.