Puppy Linux Discussion Forum

Posted: **Tue Sep 29, 2020 6:17 pm**

Which puppy uses less ram?

Posted: **Tue Sep 29, 2020 6:19 pm**

Older ones.

The newer the Puppy is the more RAM is requested.

Give some specs of you computer, please.

RAM, CPU, Graphics, Mainboard etc.pp.

Use PupSysInfo to sample that information.

Posted: **Tue Sep 29, 2020 8:17 pm**

Most newer Puppies installed as a frugal install or a live install will boot to a working desktop and use between 200MB to 300MB.
If you use boot option pfix=nocopy.
Someplace around 150MB to 200MB.
But that looses some of the speed in starting programs.

To be able to run the latest programs you are going to need to use a newer Puppy version.
The newer the program the more it pushes the need to have newer dependency files/programs to run it.

What program you are trying to use and how much RAM it uses is the question of how much RAM needed.

All software is getting bigger in code, thus needing more ram to load into to run.

Web browsers and full featured word processors keep pushing up how much RAM they use to run.

Posted: **Tue Sep 29, 2020 8:35 pm**

taersh wrote: Tue Sep 29, 2020 6:19 pm Older ones.

The newer the Puppy is the more RAM is requested.

Give some specs of you computer, please.

RAM, CPU, Graphics, Mainboard etc.pp.

Use PupSysInfo to sample that information.

CPU~Single core Intel Atom N270 (-HT-) speed/max~1333/1600 MHz Kernel~4.10.0-38-generic i686 Up~5 min Mem~237.4/986.0MB HDD~120.0GB(5.6% used) Procs~166 Client~Shell inxi~2.2.35

1 GB DDR2 ram 533 Mhz

Posted: **Tue Sep 29, 2020 8:58 pm**

AtomicPup 2020 Topic
AtomicPup 2020 Download

Down on the archive.org page is some more Puppies listed that may fit...

Posted: **Thu Oct 01, 2020 7:08 pm**

Tahrpup32 should work well. Sounds like about the same specs as my HP 2140 netbook. You don't need PAE, so maybe try this:
tahr-6.0.5_noPAE:
https://distro.ibiblio.org/puppylinux/p ... _noPAE.iso

I still use that one.

Posted: **Thu Oct 01, 2020 8:31 pm**

Are you looking for a standalone distro?
Or do you have some purpose for it in mind?

I have a very old PUP I found from @Barry (I think) that was built in 2007 if what I read is correct. Works when I tested it, as a proof of concept.

Posted: **Wed Oct 07, 2020 11:48 am**

vtpup wrote: Thu Oct 01, 2020 7:08 pm Tahrpup32 should work well. Sounds like about the same specs as my HP 2140 netbook. You don't need PAE, so maybe try this:
tahr-6.0.5_noPAE:
https://distro.ibiblio.org/puppylinux/p ... _noPAE.iso

I still use that one.

I second TahrPup. The main problem for a regular user is newer pages that break old browsers incompatible with the os.

I'm using the next edition Slacko which can run somewhat newer browsers but the Slacko interface is more arcane demanding a more user-friendly desktop.

TahrPup rocks if you can keep it alive.

Posted: **Wed Oct 07, 2020 12:59 pm**

I use "Precise Light" which easily runs within 1 GB of RAM. But, I think this focus on RAM is incorrect. I think if you have at least 1 GB of RAM then the ability of your machine to browse modern web pages, play YouTube videos or play 720p videos will be determined by your CPU not RAM usage. I monitor all my Puppies constantly with htop and whenever Puppy is struggling I have noticed the CPU cores at or near 100% with minimal use of swap space. Specifically, I think if you run the following sysbench test and get a score significantly greater than 10 seconds your machine will struggle mightly (i.e., be unusably slow) to do some or all of the previously mentioned activities.

Code: Select all

sysbench --test=cpu --cpu-max-prime=10000 --num-threads=2 run

--num-threads should match the number of cores.

This is my seat-of-the-pants thinking based on very limited testing - I only use 3 machines. I would be very interested in knowing the sysbench score for other 1GB+ machines that can or cannot do the previously mentioned activities. For me a modern web site is typically an American or British news site (except BBC) which typically is overloaded with too many dynamic ads, a YouTube video is a VP9 encoded video (this forces software decoding on older machines as the VP9 codec did not exists when the GPU was manufactured). "Precise Light" on "older machines" does other things quite well, e.g., play 480p videos, browse less dynamic web sites, e.g., this forum site. However, "Precise Light" does not come with a video player and when I installed VLC the audio stopped working in Firefox. My solution is to have a two session save files one with Firefox and one with VLC.

Posted: **Wed Oct 07, 2020 2:42 pm**

m-cuda wrote: Wed Oct 07, 2020 12:59 pm... I think this focus on RAM is incorrect. I think if you have at least 1 GB of RAM then the ability of your machine to browse modern web pages, play YouTube videos or play 720p videos will be determined by your CPU not RAM usage. I monitor all my Puppies constantly with htop and whenever Puppy is struggling I have noticed the CPU cores at or near 100% with minimal use of swap space...

I've yet to get anyone to explain how "cache" and "buffering" fit into the picture; or where they are physically located as in:

Memory Allocation:
Total RAM: 7853 MB
Used RAM: 2358 MB
Free RAM: 5495 MB
Buffers: 88 MB
Cached: 1763 MB
Total Swap: 0 MB
Free Swap: 0 MB

Actual Used RAM: 507 MB Used - (buffers + cached)
Actual Free RAM: 7346 MB Free + (buffers + cached)

But my 'seat of the pants' understanding is that "Used RAM" is that which is currently occupied by those parts of datafiles and applications currently in use. What CPUs do is move files into and out-of RAM, from and to cache, buffers and storage*; storage being your drives and writable optical media. Consequently, the less RAM you have, the more work your CPUs have to do moving files.

* and swap files/partitions and zram if you've got these.

Posted: **Wed Oct 07, 2020 6:19 pm**

@mikeslr

the more work your CPUs have to do moving files

My understanding is the CPU does not actually move data between memory and disk files. It tells the DMA controller to move a block of memory to/from the disk and then continues working. If you have insufficient memory then the CPU is always waiting for the DMA controller to transfer the needed data from the disk so typically what you see is CPU usage drop because the CPU is continuously waiting for the needed data (It also has to make room for this data by first saving some memory to the disk). I think this is called thrashing - pages being continuously swapped between memory and the disk resulting in high disk activity (from the DMA controller not the CPU) and low CPU usage. In other words if you see 100% CPU usage your problem is probably not caused by insufficient RAM.

Posted: **Thu Oct 08, 2020 2:46 am**

JASpup wrote: Wed Oct 07, 2020 11:48 am
vtpup wrote: Thu Oct 01, 2020 7:08 pm Tahrpup32 should work well.
I second TahrPup. The main problem for a regular user is newer pages that break old browsers incompatible with the os.

I'm using the next edition Slacko which can run somewhat newer browsers but the Slacko interface is more arcane demanding a more user-friendly desktop.

TahrPup rocks if you can keep it alive.

I have no problems with new browsers on Tahrpup32. Running Seamonkey 2.49.4 right now writing this and I also run Firefox 81.0.1 to watch DRM protected video subscriptions. Tahrpup32 is still my regular work OS day-to-day. And this is on a 7 year old laptop.

I do think it's critically important to me to use the "No-Script" add on with both of these, and block all scripts other than ones for sites that I add manually. I allow other scripts only temporarily, as needed, when needed. I have both browsers destroy cookies after closing. I have absolutely trouble free fast browsing wherever I want.

Posted: **Thu Oct 08, 2020 3:06 am**

I just did a recent Firefox install on a legacy Puppy, but that was a rare exception. It was in the repositories. Runs fine but noticed sluggishness.

Seamonkey might be worthwhile. I'm cagey about its multi-trick pony status.

So there are exceptions you've found. Try the same with Chromium & derivatives, Vivaldi, you'll be surprised.

I'm a regular cookie crusher. It's almost an idiosyncratic habit.

vtpup wrote: Thu Oct 08, 2020 2:46 am
I have no problems with new browsers on Tahrpup32. Running Seamonkey 2.49.4 right now writing this and I also run Firefox 81.0.1 to watch DRM protected video subscriptions. Tahrpup32 is still my regular work OS day-to-day. And this is on a 7 year old laptop.

I do think it's critically important to me to use the "No-Script" add on with both of these, and block all scripts other than ones for sites that I add manually. I allow other scripts only temporarily, as needed, when needed. I have both browsers destroy cookies after closing. I have absolutely trouble free fast browsing wherever I want.

Posted: **Thu Oct 08, 2020 1:54 pm**

Hi m-cuda,

Your post above pretty much reflected my understanding until I used puli, and williams2 responded to my post regarding the ability to boot Puppies from a USB-Key and then remove the Key, or as he pointed out CD/DVD rather than USB-Key. See, http://www.murga-linux.com/puppy/viewto ... 78#1048378 and william2's response, http://www.murga-linux.com/puppy/viewto ... 96#1048396.

But this thread isn't the place to continue that discussion. Especially as I can't add more to make the situation clearer than to post this question: If the Key/CD a Puppy has booted from has been removed, no hard-drive is mounted and no swap file/partition exists, than everything which happens has to be happening in RAM; what distinguishes "Actual Used RAM" from "cache & buffers"?

Posted: **Thu Oct 08, 2020 7:20 pm**

JASpup wrote: Thu Oct 08, 2020 3:06 am I just did a recent Firefox install on a legacy Puppy, but that was a rare exception. It was in the repositories. Runs fine but noticed sluggishness.

Seamonkey might be worthwhile. I'm cagey about its multi-trick pony status.

So there are exceptions you've found. Try the same with Chromium & derivatives, Vivaldi, you'll be surprised.

Our experiences differ, possibly due to different computers, but not OS's -- since we are both talking about Tahr32 as a recommendation to the OP for limited computers, correct? I've been using Seamonkey on Puppy for a dozen years without problems, so I don't obviously feel cagey about it.

I've tried all the browsers and have always returned to the Puppy standard for both performance reasons and the extensive amount of easy fine-degree user control over privacy and security options, plus the No-script extension. These security and privacy options are all to a greater or lesser extent obfuscated, watered down, or plain missing in the other Mozilla and Google etc. flavors. I find that when the mass of twenty runaway scripts (including Google's) now loaded on half the Internet's pages are blocked from the get-go via the No-Script extension, Seamonkey is the fastest browser of any I've tried, especially on limited hardware. Firefox is more bloated, less S&P adjustable, and yes, not as fast. But it does do acceptable DRM for Netflix Amazon and Hulu, so that's all I use it for. DRM providers reject Seamonkey as a platform out of hand, so I just switch when necessary. Seamonkey is for me, day to day browsing, and 90% of my online browser use.

But browsers are a very personal thing, to each his/her own for personal reasons.

Posted: **Thu Oct 08, 2020 7:31 pm**

@mikeslr wrote:

If the Key/CD a Puppy has booted from has been removed, no hard-drive is mounted and no swap file/partition exists, than everything which happens has to be happening in RAM; what distinguishes "Actual Used RAM" from "cache & buffers"?

Since, I use htop instead of top, I ran top to investigate the buff/cache fields and was surprised to find that htop was reporting memory usage about 25% higher than top. Here is what I learned from reading several articles on the web.

buff/cache are for transient data. They both live entirely in RAM. Cache are exact copies of disk blocks that were read from or to be written to the disks. Buff are for other kinds of buffers, e.g., a packet read from or to be sent to a network interface.

Since, buff/cache have a transient existence top in calculating "used" use the following formula:

Code: Select all

'used' = 'total' - 'free' - 'buff/cache'

However, I think this is optimistic as even though 'buff/cache' is transient they are actually using RAM at least for a while. I think htop is trying to account for this and using some kind of heuristic adds a fraction of the size of 'buff/cache' to its used field. (That is my conjecture - I haven't found it documented anywhere.) Anyway, that is all I can do for now - gotta get real work done.

Since the filesystem in Puppy also lives in RAM the kernel 'cache' would seem to be unnecessary - the kernel could just directly use the blocks from the in-memory filesystem. However, vmstat reports a huge cache - not sure what is going on here. Is it possible that there are two copies of a disk block one in the kernel cache and one in the in-memory filesystem? -seems inefficient but true?

Also, thank-you,

Thanks to you I discovered that top has the field "avail Mem" which I think is the same as "available" in the free command. htop doesn't have this very valuable field. I think I should switch to top. [ADDENDUM: Later, I found that top on "Precise Light" does not have the field "avail Mem" - must have been added sometime between the "Precise Light" and BionicPup64" releases. Bad news for me as I really wanted to measure the performance on "Precise Light".]

Posted: **Thu Oct 08, 2020 10:54 pm**

mikeslr wrote: Thu Oct 08, 2020 1:54 pmwhat distinguishes "Actual Used RAM" from "cache & buffers"?

I'm assuming that "Actual Used RAM" comes from Pup-SysInfo.

Pup-SysInfo says:
Actual Used RAM is Used - (buffers + cached)

where Used Ram seems to be Total Ram - Free Ram

All these numbers are from the kernel:
cat /proc/meminfo

Other programs, like top, htop, free, etc, etc may use different numbers and/or different formulas, which would show different results.

Usually, when a file system is mounted, a block of ram is setup. If a file is accessed in the file system, the file is first copied to the buffer. The original file on the drive device might be fragmented, and it might be compressed. Usually, a drive device is much, much slower than ram. It would be much more efficient to work with the file in the buffer as opposed to working directly with the data on the hardware device.

Every time something is read or written by the operating system, a copy is put in ram memory that is not currently being used. If the system gets something that is still in the cache, it's much faster then getting it from a hardware device hard drive, usb drive, cd/dvd etc). Cache memory is basically ram memory that is not currently being used. If the system needs more ram, it can instantly use ram in the cache area.

Ram in a tmpfs file system shows up as ram in the cache area. So files in /tmp/ and if the save layer is in ram (pfix=ram, or usbflash or ataflash), these files show as taking space in the cache area. These files in the tmpfs file systems can not be cleared from the cache area. Normally, the Puppy sfs files are copied to the tmpfs file system, in my case, BionicPup64, they are in /initrd/mnt/tmpfs/

So used ram is:
ram used by a video card builtin to the mother board (not usable by the system)
the kernel and kernel modules and kernel threads and hardware drivers
executables and library shared objects
data in tmpfs file systems

Here is ram used by BionicPup64 with Xorg, cups, etc killed:

Code: Select all

3536	/initrd/mnt/tmpfs/pup_rw/lib/modules/4.19.23
3556	/initrd/mnt/tmpfs/pup_rw/lib
4956	/initrd/mnt/tmpfs/pup_rw
space used in ram tmpfs in KiB

 Private  +   Shared  =  RAM used	Program

116.0 KiB + 555.0 KiB = 671.0 KiB	busybox (6)
456.0 KiB + 284.5 KiB = 740.5 KiB	dhcpcd
  1.1 MiB + 295.5 KiB =   1.4 MiB	ntfs-3g
  1.3 MiB + 285.5 KiB =   1.6 MiB	dbus-daemon
  1.6 MiB + 272.5 KiB =   1.8 MiB	udevd
  2.0 MiB + 425.5 KiB =   2.4 MiB	sh
  3.0 MiB + 324.5 KiB =   3.3 MiB	wpa_supplicant
---------------------------------
                         11.9 MiB

It's using about 12 MiB of executables, and about 5 MiB in the tmpfs save area (no save file, like pfix=ram)
(using du and https://github.com/pixelb/ps_mem )
https://www.man7.org/linux/man-pages/man1/free.1.html

Posted: **Fri Oct 09, 2020 12:33 am**

I'd be happy if Seamonkey works and don't have great preferences except I believe Vivaldi is best looking and Tor uniquely useful.

The issue is really time-updates.

Since I wrote the post you're reply to, my newer Firefox on Slacko 6.3.2 crashed:

New browsers are slower and more resource hungry and improvement isn't always there.

vtpup wrote: Thu Oct 08, 2020 7:20 pm
Our experiences differ, possibly due to different computers, but not OS's -- since we are both talking about Tahr32 as a recommendation to the OP for limited computers, correct? I've been using Seamonkey on Puppy for a dozen years without problems, so I don't obviously feel cagey about it.

I've tried all the browsers and have always returned to the Puppy standard for both performance reasons and the extensive amount of easy fine-degree user control over privacy and security options, plus the No-script extension. These security and privacy options are all to a greater or lesser extent obfuscated, watered down, or plain missing in the other Mozilla and Google etc. flavors. I find that when the mass of twenty runaway scripts (including Google's) now loaded on half the Internet's pages are blocked from the get-go via the No-Script extension, Seamonkey is the fastest browser of any I've tried, especially on limited hardware. Firefox is more bloated, less S&P adjustable, and yes, not as fast. But it does do acceptable DRM for Netflix Amazon and Hulu, so that's all I use it for. DRM providers reject Seamonkey as a platform out of hand, so I just switch when necessary. Seamonkey is for me, day to day browsing, and 90% of my online browser use.

But browsers are a very personal thing, to each his/her own for personal reasons.

Posted: **Fri Oct 09, 2020 4:00 pm**

Can't speak for Slacko, and Firefox, the subject was Tahr32 as a suggestion to the OP's question, and I recommend its on-board Seamonkey as reliable, and compact, based on long experience with it in that distro. This keeps wandering off topic, and off suggestion.

Puppy Linux Discussion Forum

Which Puppy uses least RAM?

Which Puppy uses least RAM?

Re: Less ram

Re: Less ram

Re: Less ram

Re: Less ram

Re: Which Puppy uses least RAM?

Re: Which Puppy uses least RAM?

Re: Which Puppy uses least RAM?

Re: Which Puppy uses least RAM?

Re: Which Puppy uses least RAM?

Re: Which Puppy uses least RAM?

Re: Which Puppy uses least RAM?

Re: Which Puppy uses least RAM?

Re: Which Puppy uses least RAM?

Re: Which Puppy uses least RAM?

Re: Which Puppy uses least RAM?

Re: Which Puppy uses least RAM?

Re: Which Puppy uses least RAM?

Re: Which Puppy uses least RAM?