I am a long time FSF member, and while I’m not QUITE as fanatical about it as RMS, I do try to be pretty firm about only using Free / Open Source software if at all possible… I’d define myself as about 85% of the way up the “Freedom Ladder”
FreeCAD runs natively on GNU/Linux, and you have a full unlimited license… (GPLv3+ mostly) There are other FLOSS CAD applications that will run on GNU/Linux but FreeCAD is the only one I’ve found that can also generate the toolpaths and G-Code to actually make your parts on a subtractive CNC mill or lathe.
F360 runs on other peoples hardware, and the ‘free’ license is majorly restricted - no tool changes, so you have to treat each op as a separate job… No G-0 moves, everything is G-1, and they can change the rules (and have several times) any time they want - and you can only look at your work in F360 because of the proprietary file format.
I’ve never done anything with OnShape, but again it’s a thing that runs on other peoples hardware (there is NO CLOUD! just other people’s computers) under a license that can be changed out from under you…
ART
1 Like
I’ve been working on PC’s since the ORIGINAL 640K RAM (More than Bill thought you’d ever need) IBM PC and a bit of other hardware before that… So I’m very familiar with the older ISA hardware, but haven’t spent as much time with the newer stuff. I built both of the desktops I’m running now…
The machine I most want to upgrade has an Intel DX79SR mobo that per Intel doesn’t have any upgrade options that would be worth it… I’ve been running it with an SSD and a using bunch of assorted scrounced spinners for backing up. ALL of them were showing “EOL” or “Pre-Fail” in the SMART Drive checks, so I tried upgrading to a couple of new 8TB spinners (I’m amazed at how cheap new drives are…) Even though I hadn’t touched the boot drives, I started having boot issues after I started shuffling the other drives around… I still can’t get it to come up in UEFI mode, and have to use the boot menu to get it to come up in Legacy BIOS mode, but at least it’s sort of usable now…
All the connections are good, and the drives show up in the BIOS. I haven’t checked the PSU as such, but it’s a high end “PC Power & Cooling” unit so I doubt it would be a problem…
I think mostly what I’m needing to figure out is what the “bleeding edge” hardware is that will be hitting the discount shelf because it can’t ‘officially’ be downgraded to W11…
ART
The power supply can look like it is working whereas in fact it is faulty. I have seen that before. They don’t last forever, and they may meet with stability issues even when not failing completely.
It does not mean it is the case for your computer, but you still might want to check. Also were the boot issues already happening before you added new storage devices? I stumbled upon this PSU calculator once, found it handy : https://pc-builds.com/power-supply-calculator
According to this documentation related to your brand/model motherboard : https://data2.manualslib.com/pdf2/48/4733/473251-intel/boxdx79sr.pdf?d3dee6fb890ba39d0522b77c863ae1c0
Pages 16 and 17, this motherboard takes 3rd generation Intel processors. I was curious about the Xeon processors, and looked at this particular one : https://www.cpu-world.com/CPUs/Xeon/Intel-Xeon%20E5-2680.html
it offers CPU cores 8 - The number of threads 16
Here is the full list of possible CPU’s : https://www.cpu-upgrade.com/mb-Intel/DX79SR.html
and as for the RAM, with 4 slots in it, if the doc is right, you can install up to 64GB RAM, specifically:
• 1.35 V DDR3 SDRAM DIMMs (JEDEC Specification)
• Four independent memory channels with interleaved mode support
• Unbuffered, single-sided or double-sided DIMMs with the following restriction:
Double-sided DIMMs with x16 organization are not supported.
The possible PCI Express x16 Graphics presented in page 16 of your motherboard’s doc seem to me pretty high end too, (provided they can be found), for a motherboard of that age. (Intel is what, 12th, 13th generation now?)
When I read what you say:
that per Intel doesn’t have any upgrade options that would be worth it…
I wonder why? Ok, you can’t run Windows 11 on it, so maybe wherever your read that was referring to the Windows world?
1 Like
The machine has 8 RAM slots, not 4… When I got the machine off the Artisans Asylum “cruft shelves” it had 16GB of RAM (4x4GB sticks). I added 4 more 8GB sticks to give me 48GB, which was the best upgrade option w/o throwing out the existing RAM…
The I-7-3930K @ 3.2Ghz already in the machine gives me 6 cores / 12 threads, and was a “High End” CPU in it’s day, so according to the stuff I’ve read, the benefit of going to a Xeon would be minimal, and not worth the cost… The graphics card is an old (2009) Nvidia G92 / GeForce 8800 GT, but it works well enough, I’ve never felt I had video issues with it.
Essentially the machine seems like it would have been pretty much ‘bleeding edge’ when it was built, and isn’t horrible today, but it doesn’t seem like there are many options for any significant upgrades given the existing motherboard… That it won’t run W11 is almost an ADVANTAGE as far as I’m concerned… I have ONE old laptop that runs a bootleg version of W10, which is all I need to run the only THREE programs I have that are only available on Windows, and that I haven’t figured out how to get running under WINE… (and that machine is NEVER allowed to connect to the Internet!)
Getting back to the power supply question, I’ve not had any problems when the system is running, or unexplained boot problems before this. What I’m not sure about is how best to check for stability problems. The load is actually LESS currently than it was as I had been running about 6 spinning drives, and replaced them with two (The tower case has room for about 20…)
ART
Indeed, it is a very good computer, any Linux distribution should work a treat on it, and I guess you are right about CPU and RAM resources, you probably have way enough, even if using some greedy applications (I guess way enough for FreeCAD, probably ok for Blender too if you decided to give it a try).
If you had not such a good CPU, a Xeon such as the E5-2680 could have been a good choice. I see some which does not cost too much at ebay or aliexpress.
To check a power supply, you can either do it with a multimeter, which you can check how on Youtube videos (keywords “testing a power supply with a multimeter”), or with a dedicated tool, such as this one : https://www.ebay.fr/itm/276811636747/ which is handy too. You can find a manual showing the desired values here: https://www.lindy.co.uk/downloads/43058v2.pdf and a picture with the useful info here: https://m.media-amazon.com/images/I/81TySD9ClUL._SL1500_.jpg
1 Like
The tester you point at looks reasonable (though most of my DVM’s go to much higher accuracy) but it looks like just a basic voltage checker, to see if the supply is making the right values at no-load status… What I’ve seen in PSU failures is harder to see as the supply looks fine at no-load, but breaks down under heavier loads, or when it gets hot.
The PC does have a screen in the BIOS that displays all the mobo voltages in case you want to try overclocking (I don’t) and those are all well within the right range (although sitting in the BIOS screens is pretty close to ‘no load’
I am simply not seeing ANY symptoms that make me suspect the PSU as a problem.
ART
1 Like
I agree the small devices I pointed to are basic tools. Testing with a multimeter is better, and I heard of specific tools which can help by putting load on a power supply, which is even better, but apart from people dealing with electronics, and ones who are really into hardware who have them?
I am not an expert, but I am pretty sure the signs one can get on a PSU by looking at what happens in the same computer where the problem lies is not the right solution.
Testing components on a computer is about removing as many internal peripherals as possible, testing one by one, and testing a power supply is about removing it from the computer and testing it apart.
Here is a page which seems serious to me : https://www.rohde-schwarz.com/fi/products/test-and-measurement/essentials-test-equipment/dc-power-supplies/5-essential-tools-on-an-electronics-bench_256910.html
I found that video too : https://www.youtube.com/watch?v=qqzelYnX35U
A very simple test could also be to get another power supply, new preferably, equal or higher in qualité and check if you find similar signs. A good PSU is about stability.
What else : have you tested the RAM modules? (They would have to be tested one by one, with a live CD coming with memtest, for instance).
I haven’t run Memtest86 on the machine lately, but I did a 24 hour run when I first got the machine, no problems… (BTW, no need to test modules separately UNLESS you have a failure and need to isolate which module is bad, however if a machine has a lot of time on it (as opposed to a DOA failure) it is likely that anything that caused one module to fail is going to damage others, so I’d class all as suspect.)
The articles - The test equipment article was reasonable though far higher end than is likely useful in a repair shop. I have fixed lots of different things over the years, and have found that nearly all were clearly good / bad issues that could be traced with a cheap meter… The kind of gear in that article is more what you’d need in a design / development lab where you were creating an item, or testing for things like regulatory compliance… On the repair bench I’m dealing with much more binary questions of good or bad…
The Load testing video is not ‘wrong’ and their process would work with a simple supply, but would NOT work on a computer PSU as there are a lot of other complexities involved in the PSU like the “power good” line that has to come on at the right time, and all the different voltages it puts out.
Replacing the supply with a ‘known good’ one is a test that is simple in theory, but a new supply in the same quality range as what I have is well over $100, and replacing it is a non-trivial task to get to all the connections… Given that a PSU is not going to be returnable, I am not really wanting to spend that sort of money just for a test…
ART
For any RAM test I’d always test them separately. What is the point of the test, if some error occur and you can’t tell which RAM has an issue?
You would have to start over and to test one by one anyway, wouldn’t you?
What about the.PSU connections, why are your’s complicated to unplug and plug back in?
Flip side on the RAM test - whats the point of testing them separately if there is no problem? Since they are already installed, I’d have to pull them to test, which has it’s own risk of causing problems… If there is a problem given the hours on the machine, I’d consider it most likely to have been damaged by a power surge or similar thing that would make all of the RAM suspect…
Also many years back, I actually had an issue with the high-reliability telco equipment I was working with that Memtest86 and several other memory test programs did NOT find even after several days of running their toughest settings, but was definitely a reproducible and verified failure in the RAM. Oddly enough the test that showed the failure fastest was supposedly a torture test for the hard drives! I forget the exact issue but it was a parity chip failure… (RAM mfg by Micron BTW, not cheapo brand X)
In terms of the PSU, it is hard to get to some of the connectors because of other things in the way like the plumbing for the liquid cooling unit, and because I put a lot of work into cable routing when I built the box and I’d have to undo a lot of that to get the PSU unplugged.
ART
I suspected, or I thought I read in one of your posts that this is a machine with a water cooling system. I have never had or used one of those, so I don’t know how they are built. However, in one of your first posts of this discussion, talking about some boot issues, you said you would consider changing all the components, motherboard included.
If it comes to that, once the diverse components tested you could still do something with them, either reuse them or resell them, wouldn’t you?
I could possibly resell the parts I replaced, assuming there is much demand for them…
Basically the case I have is a BIG tower (a CoolerMaster “Cosmos” model). The mobo mounts near the top, with the dual 120mm fan radiator attached to the inside top, so that the fans blow up and out of the case. There are two fairly stiff hoses that come off the radiator to the CPU cooling block that also contains the coolant pump, etc. Between the fans hanging down, and the coolant hoses, plus the graphics card in the highest slot there isn’t a lot of room to get in with my hands… It’s also a bit of an awkward angle to reach from my chair, unless I get the machine up on a table which is not easy (It’s HEAVY!)
ART
It seems you might need a tripod-mounted pulley (I pulled the word from deepl.com, as I didn’t know how to say it in English).
Do you go to the repair cafes around your place sometimes?
I have access to a couple different shop cranes, so its a solvable problem if I have to, but I’m hoping to avoid it…
I don’t generally do much outside the maker-space, and we do get some folks that come to us, particularly in the "Electronics & Robotics shop, which has a weekly free and open to the public “Shop Night”. However I’ve found that it is problematic for me to try to go to the shop on those nights because it gets really difficult to get around the shop in my wheelchair… The space is tight enough that I can barely get through when all the chairs are empty and pushed in, if there are people in, I have to get everyone to stand up and let me by which is a major disruption…
ART
To add my experience. Just upgraded my old pc to elementary.os, as it’s like windows and was recommended. Athlon chip with UEFI/Legacy bios. All went well and then it stopped working, wouldn’t boot up either from pen drive or hard drive. I suspected that it was something to do with the fast boot, which I probably should have turned off before starting, as I could see there were some unexpected boot options. As I couldn’t boot into elementary I couldn’t do the full shutdown process so I did the obvious thing and took the power cable out. I waited a while and it still did not clear the boot list. So I tried booting again and after an enormous amount of time, with nothing happening, 30 minutes or so, elementary got itself started (rather impressive I thought). I did a full power down and now it’s fine. So the message is that some distros don’t like fast boot on uefi/legacy bios and that might not be immediately apparent.
2 Likes
Would you like to add the Bento Openbox Remix distribution? (Remix means “built on Ubuntu”, which is what is required by Canonical, for the Ubuntu derivatives not supported by Canonical).
I have been rebuilding them since 2011, and more than one of my clients use them, as well as myself and probably a bunch of people I don’t know who can get it.
You’ll find detailed descriptions and downloads at https://linuxvillage.org.
At a time, Linux distros were either light, or easy, but not both. I needed both, and when I finally got to the point I did, I learned how to rebuild a distribution, to share the setup in the form of distributable versions.
In the context of performance testing, I currently have 9 DELL Inspiron 15R Enterprise Edition machines (from 2012 and 2013). These devices were upgraded from 4 GB to 16 GB of DDR3L RAM and equipped with 480 GB SSDs to enhance performance. i3-2xxxM, i5-3xxxM & 5xxxM processor.
Observed Boot Times
- Machines that were less frequently used boot Linux Mint in approximately 25 seconds.
- More heavily used machines take over 40 seconds to boot, despite the identical hardware upgrades.
Initial Configuration
- Originally, these laptops ran Windows 7, Windows 8.1 or 2008 or 2012 Server.
- All were upgraded to Windows 10 IOT.
- Used for desktop work with workstation like a desktop classic PC.
Migration to Windows 11 and Linux
- Four of them successfully migrated to Windows 11 without major issues, achieving boot times between 21 and 39 seconds.
- For the others, migration to Windows 11 was problematic, with boot times exceeding 2 minutes.
- Even switching to Linux Mint XFCE did not significantly improve the performance.
Installation and Multiboot Configuration
To test performance and comparisons, I configured a triple boot on each machine:
-
Windows 11:
- One native installation.
- One optimized version (Atlas OS), each on a 30% partition.
-
Linux with GRUB Multiboot, featuring the following distributions:
- Mint XFCE
- Mint
- Zorin
- Zorin Lite
- Winux
- Lubuntu
Objective and Project Context
These machines are intended to be resold as part of the Zero Waste Project, aimed at the educational sector, specifically for children with dyslexia. A detailed comparative table of boot times and OS performance for each setup is being prepared and will be shared as soon as it is ready.
2 Likes
Fonts - Windows vs Linux
I wanted to look into this for the sake of Windows users transitioning to Linux - and bringing across their Word documents - only to find all their hard work now looks like something the cat dragged in.
This is based on my own experience in an office with a mix of Linux & Windows - and people swapping files.
Although there are any number of font collections for Linux like ttf-mscorefonts-installer & fonts-liberation they’re mostly just for the Windows system fonts - or are badly out of date.
So I wanted to focus on the default MS Office font - which used to be Calibri - but is now Aptos.
Surprisingly - Aptos is available from Microsoft as a free download. That leaves Calibri. Google (because of Chromium OS) has done a FOSS drop in replacement called Carlito. That should be in your package manager as fonts-crosextra-carlito.
Then - in LibreOffice go to Tools → Options → LibreOffice → Fonts - and under ‘Replacement Table’ check ‘Apply replacement table’. Then - under Font - Calibri won’t be in the drop down font list - so type it in manually - and then ‘Replace with’ Carlito - which should now appear in the drop down.
There’s also Caladea to replace Cambria - fonts-crosextra-caladea.
As well - the good people at Wine have done FOSS replacements for Tahoma. Tahoma regular and Tahoma Bold.
Of course - all this could just be left as a PDF on the Desktop of any new Linux install to save wear and tear on repair volunteers.
Please comment if I’ve got anything wrong - or if there’s an easier way. It may be that the substitution is handled automatically making the whole ‘Replacement Table’ process redundant.
1 Like
My hunch was right. Font substitution is handled by ‘fontconfig’.
fonts-crosextra-carlito includes an XML script:
<fontconfig>
<!-- Microsoft -->
<alias binding="same">
<family>Calibri</family>
<accept>
<family>Carlito</family>
</accept>
</alias>
<alias binding="same">
<family>Carlito</family>
<default>
<family>Calibri</family>
</default>
</alias>
</fontconfig>
No need to configure LibreOffice.
