I'm planning to buy a new PC. Can I use Trisquel with it?
- Inicie sesión o regístrese para enviar comentarios
Hello, Recently I am using my computer very much because of the COVID situation, especially with online classes and etc. (Although I try not to use Zoom!)
The computer I'm currently using has an Intel's 2nd gen Xeon E5-1650 processor, 16GB DDR3 ECC, some ssd, and nvidia's gtx 670 and so on.
I am planning to buy a 10th gen Intel i5 processor (10400?), and 32gb ram (8*4), and intel internal graphics. Does Trisquel support 10th gen intel processors?
As far as I know trisquel uses a 4.x lts kernel as of now and I'm concerned if it supports the 10th gen processor, and also if my intel internal graphic will run well on trisquel with non-proprietary drivers. (My gtx 670 was detected and working without additional issues and it also has accel.) Thanks!
Your current computer is far far better than most others', and doesn't have major freedom issues. Both chipset and GPU are well supported by free/libre software. I don't see any reason purchasing a new one.
If you want more memory, your CPU can support up to 768 GiB.
My computer is a workstation, it eats up too much energy/resource and is also so hot, so I am trying to buy a new one and use this as a sub pc/ etc things so that I only use this sometimes and not daily. (I'm paying more like $50 dollars for electricity after buying this PC)
To save electricity bill, it's reasonable to obtain a more power-efficient computing device.
You can buy a 10th or 11th generation platform. But if you want 32 GiB memory, it's strongly recommended that you buy one 32-GiB module, rather than four 8-GiB modules. For the motherboard and CPU, 11th generation is strongly recommended. One reason is that most 9th and 10th generation CPUs don't support latest instruction set (AVX512 and some more).
I have been waiting for a CPU supporting AVX512 since 2017. The 9th-gen. Cannon Lake was said to have been released in 2017, but actually it was Coffee Lake Refresh that was released. The 10th-gen. Ice Lake had AVX512 support, but very few of them were produced. Most 10th-gen. processors were Comet Lake (with only AVX2, no better than Haswell). This time it's 11th-gen. Tiger Lake/Rocket Lake. Thankfully they all have AVX512. (Purchasing a new CPU without new instruction sets is pointless.)
One 32 is like 200 dollars here and 4 8gb is like 160 dollars. Is there any benefit for using one 32gb module?
Also, I actually don't know well about what AVX512 would do in normal usage and compilations. After a quick search it says that it makes performance higher, but what other instruction sets does the 10th gen processor not support?
Thanks. I'm just trying to make the best purchase as I can.. :)
If you purchase one 32-GiB module, you can upgrade to a maximum of 128 GiB in the future without having to throw away anything. If you purchase four 8-GiB modules, you'll have to throw away at least one of them when you upgrade in the future.
AVX512 and some other new instruction sets (such as vectorized AES NI) are available in Ice Lake, but not in Comet Lake, although they are both 10th generation. However, these instruction sets are available in either Tiger Lake or Rocket Lake (both 11th generation).
Since new instruction sets are of vital importance for me (when doing scientific calculations and cryptographic operations), I won't purchase new processors without new instruction sets, no matter how cheap they are.
I'm just curious if the new pc's parts will be supported with trisquel.. if else I'll just check out other foss distros for now or try to put mainline kernel in trisquel and come back to trisquel after the parts get supported
I don't know how latest hardware works with an old kernel. I always use latest kernel from a rolling-release distribution. And I have no interest in testing this.
I recently tried to install Trisquel 9 on a VAIO FE14 laptop with a an Intel Core i7-1065G7 10th gen processor (the model is VJFE43F11X-B0611H). Unfortunately, Trisquel's installer could not install GRUB whatever the combination of activation/deactivation of "Fast boot" and "BIOS Legacy mode" (if I remember well) I would choose in the BIOS' configuration. I ended up installing Debian testing (soon-to-be Debian 11 Bullseye), of course without accepting proprietary firmware (the internal Wifi card is not functional: I use a dongle with a AR9271 chipset), disabling the contrib area of the repository (which was enabled by default, as far as I remember) and without enabling the non-free area.
Great news, thanks for your feedback. It means that Q3-2019 Intel integrated graphics still work well without any proprietary blobs.
The wifi thing is always a bit annoying but easy to solve (and anticipate). Having a buggy display on a laptop really is a bummer.
The Intel integrated graphics work well (including on Trisquel's live system) on that laptop, yes. My touchpad does not work, at least on the Debian's X session I am running right now. I believe it works on the Wayland session. At least I have seen it working and I believe it was when I was trying Wayland (that OBS Studio, in Debian's repository, does not support, hence my use of X). I use a USB mouse and should make some test for a definite claim regarding the touchpad. Anyway, I do not recommend the model I chose: Trisquel 9 cannot be installed, the Wifi card does not work with free software (what is unfortunately the norm, as you wrote) and, if like me you use some program that does not support Wayland, you need to occupy an additional USB port for a mouse.
According to my experiences, Wayland is far worse than Xorg on any of my systems (higher power consumption, lower graphics performance). Most of all, I can't use "ssh -X" on Wayland. Even if someday Wayland becomes the default session on MATE, I'd still use Xorg.
i7 1065G7, an Ice Lake processor with all new instruction sets supported. I planned to find such a notebook equipped with this processor. However, there were very few (if any) such products available.
It is still possible to try Trisquel. Just don't use the Live media's installer. The network installer should be similar with Debian Installer.
If you didn't load non-free firmware, the contrib repository for "debian-security" would be enabled, but not for "debian".
Indeed, I have not (and should) have tried the network installer.
Is the contrib area for "debian-security" OK freedom-wise? What is in there that justifies it is enabled by default?
The contrib repository for debian-security is enabled by default after installing a testing branch of Debian using Debian Installer. I don't know the behavior of Debian stable, since I always use testing.
Regarding the installer, I'd like to discuss something in another post.
Changing contrib and or non-free to main would be better freedom-wise, I'd imagine.
"Q: How is security handled for testing?
A: Security for testing benefits from the security efforts of the entire project for unstable. However, there is a minimum two-day migration delay, and sometimes security fixes can be held up by transitions. The Security Team helps to move along those transitions holding back important security uploads, but this is not always possible and delays may occur. Especially in the months after a new stable release, when many new versions are uploaded to unstable, security fixes for testing may lag behind. If you want to have a secure (and stable) server you are strongly encouraged to stay with stable."
-https://www.debian.org/security/faq#testing
Both Think Penguin and System 76 have laptops with i7-1165G7 processors. Are they not suitable?
When I really need AVX512 to run my scientific calculation software, I'd like to consider them.
If AVX2 is enough, I don't need 5th through 9th generation Core processors. Haswell is good enough.
If AVX is enough, then SandyBridge/IvyBridge is enough.
If SSE4.2 is enough, then Nehalem is enough.
If SSE4.1 is enough, then Penryn in enough.
If I don't need any instruction sets newer than SSE3/SSSE3, then Merom (Intel's first mobile 64-bit processor) is enough.
Out of curiosity: what do you actually do that requires close to 32 GB of RAM? Buying more RAM than probably required for the next few years makes little sense, in my humble opinion. Nowadays, I would advise a mere desktop user (maybe not your case) who buys a computer to get 8 GB of RAM, to be at ease. It is nice to have a free emplacement for another stick. In this way, if the Web keeps on requiring more and more RAM (the usual reason for most of the RAM usage on desktop systems, for those who do not disable JavaScript), another 8 GB can be bought in a few years, at a lower price, because the price per GB goes down. Planning further ahead is pointless. Not only because the price per GB goes down but also because most people will be using another computer, accepting faster RAM, in (say) eight years. It is not that I find it OK to not have our computer last more: it is a prediction based on past observation. Of course, components can be reused from a computer to the next one. I actually recently bought a laptop without an SSD, because the SSD in the laptop I replaced (with the battery not even detected by the OS anymore anymore, a dying DC-IN and a dying display) was good enough.
Not everyone is a "mere desktop user". One day someone sent me a biophysical chemical model (as part of that person's Ph.D. dissertation) that required at least 200 GiB of memory to solve. I did it on an old Xeon E5 platform with a maximum of 256 GiB memory support. (And I planned to build such an E5 platform for my lab to solve problems at such scale.)
Besides caching, memory (RAM) has many more usage. For example, with plenty of memory, one can use Zram to eliminate most of disk wearing.
I hope you get the best PC safe and sound, my friend.
Newer Xeon E5 could support 768 GiB or even 1.5 TiB of memory. I could also consider purchasing such a platform but begin with 256 GiB of memory. When necessary, I could upgrade to 1.5 TiB (maybe six 256-GiB or twelve 128-GiB modules) someday.
Scientific calculation is indeed money consuming.
nadebula.1984 considers near-astronomical amounts of RAM, but from my recent experiences with nmap searches in
my Lenovo T420 with 8GB of RAM that forced me to introduce cooling-off periods, what keeps the processor(s) in
such a machine with terabyte RAM banks from overheating ?
- Inicie sesión o regístrese para enviar comentarios