#Tech Lab forum - General Discussions!
1 messages · Page 28 of 1
The problem is, SteamOS is based on Arch, same with its predecessor.
I personally think they should've used Debian.
yes, I know that. I dont get how your response to me is.. a response to what I said.
cant you use the debian build on arch?
Debian can run windows apps
??????
You can't always use Debian builds on Arch no, it's more like a 30/70% chance of working of not working.
yasee, theres an appropriate and meaningful response to what I said
Joy thanks for the info
Anandtech is gonna stop
Click here to see the CPU gamers dreamed of that Pat Gelsinger Cancelled.
[SPON: FlexiSpot Anniversary Sale! Use “24AUG30” for $30 OFF E7 Desks w/ Combo Discounts on C7 Chars: https://bit.ly/46FUDNz (US), https://bit.ly/3YEHamX (CA)]
0:00 Jim Keller hinted at Rentable Units 5 years ago!
1:38 Pat Gelsinger is the one that Killed Royal Core
7:2...
So as a Linux newb the mood launcher I downloaded was a zip that I decompress into it's own folder it contains a binary and 2 compressed .tar.xy folders one labeled control the other data
Do I need to uncompress these folders or do I just run the binary using the terminal?
If you inspect the tarballs is there a readme in them? Are there instructions on the site you downloaded them from?
Like, what led to you downloading this thing?
is moores law loosing brain cells as time goes on?
this is the most crack pot conspiracy thing ive ever heard come out of him
A Intel just layed off 15k employees
using the argument oh a bunch of ex intel engineers are working on risk V so they can make high performance cores is comedic insanity
I havent followed him since he overhyped the heck out of ryzen 2000 and never apologized
"there were plans to fuse E cores and P cores at intel"
LMAO
no there fucking wasnt
I don't know😅
1 channel says this, the other that, hard to know which to follow😐
I just want a good efficient CPU that supports 2 full PCIe 16x slots😅
And have space for at least 4 SSDs
So I thought that story from MLiD was weird so I shared it
"cores that split into cores"
he knows they already do that right single cores arent single processors
they have like 6-12 in them
Everyone makes mistakes. It's how they respond to it that determines if I find them reputable or not.
Someone like Jayz2cents likes to talk out of his ass, but he has been very good at correcting himself. Mlid does not have a history of that, which leads me to believe he is just interested in breaking a story, regardless of whether it's true or not.
I guess it depends on how you define cores. Bulldozer was arguably cores split into cores, though I am not sure that is a desirable comparison
CPU stories are wack lately, I'll wait for the full release and see if I may use it or not, all I hope for is CAMM2 coming to the consumer market.
Doubt it will. Fitting it on itx would be a nightmare, and the one stick limit is not desirable for desktop
Cannot figure out what program you're talking about but GENERALLY SPEAKING:
- Using Linux you need to find out what kind of packages your distribution accepts (.Deb for example) and preferably use these.
- if your Software does not ship as such a package you will probably download the zip
- there-within should be a file without an extension or .AppImage extension.
AppImage is great because it's like a docker image, it contains everything the app needs to run but is therefore bigger.
This file needs permissions to run (chmod +x <file path> but should already have them set normally.
Usually this file is also recognized by your file explorer as an executable and therefore has an image or displayed as something like some gears etc.
Those you need to run then
CAMM2 would be cool for the higher speeds
CAMM2 benefits cooling though.
And CAMM2 for laptops do fit on ITX boards, and from what I know are already pretty fast.
For a new standard to take hold it has to be a upgrade, and have limited interferece with existing stuff
There is a reason 12v only and back connectors are not taking off
Camm2 is a side grade. Better preformance, but more board space and less flexibility
Like seriously, what was the last update to the typical mobo? Probably m.2, and before that pcie
I just hope they'll be a thing, currently on very small builds you're tight to low profile slower RAM, with CAMM2, even if you have to slightly extent the board, you have more flexibility.
And this may sound weird, and it's just an idea, you always can move the SSD where the RAM is and the CAMM2 where currently the M.2 is on an ITX board.
Newest addition is zero-cable solutions I.E. power passthroughs or connectors on the back
I can't wait for that to properly take-off
No you can't, there are electrical reasons for the typica board layout
You would need to modify the cpu
The board layout we have now is what, 20+ years old, it's time we change and/or improve on that.
Actually no - the layout today is the way it is because the CPUs are designed in a way that minimizes trace distances and pcie is at the bottom of the socket while ram is on the right.
You currently don't have CPUs with the necessary socket layouts and you should not put ram and pcie on the same socket side, crossing over eachother
Wishful thinking but it grew like that for a reason lol
It was just an idea😅
(Beat me to it haha)
A nice one, if we would start from scratch but profits and backwards compatibility requirements (aka. The Market) don't allow that :/
Remember BTX?
That had the CPU I believe 45° tilted and in the front, and it worked😅
Damn, I miss the wacky BTX.
If it ain't broke, don't fix it. You would have to have a real good reason to change. Like 5x uplift
Sometimes change has to be forced😉
Never heard but a 45° angle might help (although not with cooler mounting as the rest would be perpendicular and you'd need to make the perpendicular coolers line up with the diagonal sockets
As I said, there needs to be a good reason to change. You shove all the new stuff into a new form factor, what kind of benifit would you expect?
Fun fact: current GPUs actually hang upside-down, with BTX the GPU coolers were on top.
New form factor would be a worthwhile change
Thx for putting that, connection is shit and can't Google lol
BTX had the CPU close to the front so it'll always get fresh air from the outside 🙃
Yes, but you will notice that the ram and pcie are still on opposite sides
Research has shown that order of components in the heat-transfer usually does not affect temps enough to change
Even on laptops that is the case
Hell even stuffing the case full of cables does not change the temps lol
Framework likes a word with you.
They changed the entire motherboard layouts.
Haven't had a good look at their movies tbh
But given the new "there's one more standard" GPU module standard I would probably guess they did
BTX apparently had lower latency and better airflow
Does framework offer a board with dgpu
They offer a board with changeable gpu
Nope, an add-in GPU
There boards look standard to me
IMG pls .-.
And Dell had the Area-51m, which had a whole gap in the motherboard where the GPU usually sits.
Thx
Usually on laptops the RAM is below the CPU, on the FW16 it's on the side of the CPU
Hmm fairly non-Standard... Dimms to the right, pcie to the top and left...
Also still wondering how that GPU connector works - only seen that it's basically probe points on the PCB?
Even on insane motherboards like on the MSI Titan 18HX the RAM is below the CPU(massive grey rectangle)
Wait actually there's pcie and power delivery left AND RIGHT because of the USB-C 4 ports, right?
Yes
More heat pipes than subway lines in a city lol
Yeah, that's not unique, I have seen similar before. The issue is with stacking traces. Pcie and memory cannot be in the same side of the cpu
Tell me about it, I was like WOW when I opened mine😅
They currently can't on ATX, doesn't mean if we tilt the CPU 45° that we still can't 😅
Some motherboards do have an NVMe "riser" beside the RAM, Asus done this before, it looks like a RAM slot, but it's a slot for an add-in card for 2 NVMe SSDs
So, you can trace PCIe under RAM channels, especially if you have something like CAMM2, where you have less traces which are also shorter😅
No, you can run them around the ram, and then you are limited to how many
I forget, were those running off the chips set?
I don't think so
I mean, we're in 2024, not 1995😅
We can adapt new styles way more easy than few years ago, just like we have a "fan" that's more flat than the camera bump on my phone 😅
Checked there first info for windows but not Debian
No read me files either
There is an.xml info file though
Wait, are we in 2024?
I keep getting confused if it's 23 or 24🤣
Actually it was easier to swap standards back then. Things were simpler back then, and there was less technical debt
@balmy flicker
If you tell me your distro and the program I can find the correct thing and probably instructions
But traces and such were way bigger than they're now
Nope, but I seen wacky laptop boards where component placement makes no sense, not looking at you Framework and MSI.
At launcher (for modded mine craft) and steam deck so arch
Thanks
Ouuuhhh the AT launcher might not have one, if you have like half an hour or so I'm at home and can check
Even curse forge Linux support is crap
This is what ram used to look like
Figures I need to head back to work I will have to fiddle with this when I am off
Thanks for the help
Honestly i think CAMM2 is a pretty good standard for desktop its just not viable for Workstation or server CPUs
but for Work station and server it matters less as they are after stability and can exceed well over double the bandwidth CAMM2 could ever do
@jade scaffold have you heard of Lenovo actually wanting to put Oculink on their laptops?
For atx it would work, but itx would be tough. And the cost of splitting the market would be too high
DTX🙃
I love that form factor.
It's ITX with an extra M.2 slot below the GPU.
What about we take ITX, tilt the CPU 45°, and have both the PCIe and CAMM2 under the CPU, and a place for the SSD beside the CPU, you would even have place for 2 Dimm2 slots, mmmmm extra SSDs.
But now most cases are incompatable with your board. Not something you want as a mobo vendor
Well now all the coolers are incompatable
And that's assuming it works electrically
Nah itx would be fine
Why would you need another cooler?
BTX also had so far I know normal Intel coolers, only the CPU was 45° tilted.
Itx supports rear mounted ram
I forgot about that, CAMM2 on the back😅
I have seen ITX with flat laying SO-DIMM slots, which I think would take as much space as CAMM2
I think AsRock did that if I am correct?
Btx was oem only as far as I know. You can fit a stock cooler on them, but even though they are a fairly large board you would not be able to fit a larger square cooler on them
All the flat sodimm boards I am seeing lack pcie
Still, the cooler was on straight while the CPU wasn't.
Fun fact: Mini-BTX also was a thing 😅
They aren't common, I seen them more in larger barebone systems like years ago, been trying to find it.
Wait wait wait, do I see that correctly?
RAM on 2 sides of the CPU on an ITX board.
https://www.asrock.com/MB/Intel/X299E-ITXac/index.asp
Supports Intel Core™ X-Series Processor; Digital PWM, Dr. MOS; Supports 4 SO-DIMMs Quad Channel, DDR4 4000+(OC); 7.1 CH HD Audio (Realtek ALC1220 Audio Codec), Supports Purity Sound™ 4 & DTS Connect; 6 SATA3, 1 Ultra M.2 (PCIe Gen3 x4 & SATA3), 2 Ultra M.2 (PCIe Gen3 x4); 2 USB 3.1 Gen2 10Gb/s (1 Type-A + 1 Type-C), 6 USB 3.1 Gen1 (2 Front, 4 Re...
It's possible, depending on the cold plate. But modern coolers are designed to cover modern cpus. The issue is not with the mounting, it's with making sure the cold plate is actually covering the cpu
Shadow, look at that board, it has 2 SO-DIMM slots between the CPU and PCIe 😅
Yep, that is the first time I have seen that
Possibly, but remember sodimm have lower bandwidth limitations. The whole point of Camm2 is bandwidth
Modern pcie and ram would have higher trace requirements
The point of CAMM2 is it's size and how much space it takes, speed is just an added advantage of having all RAM lanes so close together.
I believe you also need less space for CAMM2 in terms of lanes than RAM
Desktop doesn't care about size. The only reason for Camm2 is preformance, which is dictated by bandwidth
No my board can have 4 top mounted and 2 rear mounted and has a pcie slot
Performance is also dictated by latency
Camm2 is lower latency
Latency is bandwidth in this case
So you get lower latency and higher frequency which gives you vastly higher bandwidth
Isn't DIMM also more complex to run compared to CAMM2?
Like you need longer traces for DIMM?
Well Camm2 is single module
Don't need all the crazy tech that goes into making 4 sticks of ram work together
You need to have the same trace length for all 4 dimms
I think this is CAMM2 on an ATX board, look how close it is to the CPU, DIMM needs extra space or else it blocks the cooler.
That looks about the same
You still don't have to use CAMM2 for desktops, you can save space and use the 1 for laptops on ITX boards.
Then it can fit even more snug
We already seen 2 SO-DIMM on between the CPU and PCIe slot, I believe that would take more space than CAMM2, so it ain't impossible.
What board do you have?
2 ddr3 sodimm
That matters
And CAMM2 isn't much wider than SO-DIMM
The space we are concerned about is the traces
Which should take around the same space.
No? Sodimm have lower bandwidth due to lower trace count
And modern ram and pcie have even higher requirements in that regard
And are more sensitive to interference
But having 2 DIMMs means you need more space as the lanes would pass over each other, something you don't need with CAMM2
Still, look at how much space DIMMs take compared to CAMM2.
I still think CAMM2 is a good thing, even on ITX, where you often are soace constrained, especially with how tall current DIMMs are.
You are contending with multiple things. You will notice that both boards in question lack m.2 (at least as far as I can tell
Nvm, the x299 does have them on the back
So possible? Maybe, but it really depends on how much physical space traces for pcie 5 and Camm2 take. (also number of pcb layers, you might be able to physically build it, but if your entry level itx costs 500 your kinda doa)
I hate typing on my phone
I seen people buy MSI Godlike motherboards, so there's always a market😅
https://atlauncher.com/downloads
Given that steamOS is arch-based you should download either the AUR package (Arch User Repository) or the flathub, as flathub is an app-store-system where all dependencies are also bundled in and it practically works on every platform.
Download ATLauncher today, available for Windows, Linux and OSX. Installation is simple and you can get up and running and playing modpacks quickly.
Halo products don't make up bulk of sales. That's why I specified entirely level
That looks wacky as hell, I love it
Am I blind or is that missing m. 2?
Looks like it had 2 m.2 slots on the bottom , but they were removed because they conflict with the ram
yes
but a normal desktop CPU is a quartar the size of that one
theres easilly room for a single CAM module on top
its not even a debate
Well yes and no, cpu might be smaller but space occupied by the cooler mount is about the same. That said you move power delivery to the top and it would probably fit
The question isn't can we make it fit, it's can we make it fit in a way that we can retire the dimm form factor entirely (for desktop)
Yea easily
It's not a question really cooler mounts for AM5 intel 1700ect are vastly smaller than that of socket dimensions for lga 3647
For desktop cpus CAMM is a great form factor
For workstations not so much
Like how am I fitting 2-6 CAMM modules on a board reasonably without loosing all those fancy advantages
For desktop its less an issue of capacity as it is one of upgradability. Yes, Camm preforms better, but is that better preformance worth giving up the upgrade options that 4 sticks of ram give you? Since most ITX boards only have 2 dimms, you can probably get away with replacing them with a single Camm. But for ATX, you are going to need to be able to fit at least 2, and have the modules be able to mix and match sizes if you want full customer support
An 8 gb ram stick still has value. An 8gb camm module on the other hand would be useless
And a 16 gb module would only be useful to someone who has an 8
theres never been a good reason to have 4 sticks of ram on dual channel boards
It massivly kicks performance in the balls
theres AM5 boards that have dual chipsets and 4 Ram slots and they just cant achive decent speeds
For most people the capacity is far more useful than the speed
with how cheap DRAM currently is i see no real issue with supplying 48gb CAMM2 Modules which will last you throughout the entierty of this generation
I have not looked into it with am5, but it was fairly modest with am4
easy upgradeability is nice yes
but the current DRAM method is basically responsible for the massive ineeficiency in APU GPUs
Ryzen CPUs just love RAM speed generally
mainly because AMDs mem controler in normal Ryzen Desktop chips is turbo ass
Yes, apus are hit a lot harder, but at the same time most users who care about preformance are not running apus
i mean alot of people wanting ITX systems run APUs
or even alot of budget gamers use them
CAMM2 might also allow a further drop in DRAM Prices if they can be standardased as both a laptop and desktop standard
They love a mix of latency and speed, and they tend to have a sweet spot which is below the fastest ram of the generation
High speed high Cas sticks tend to preform comparably to lower speed lower cas
you need both
With Ddr4 3600/22 was very similar to 3200/16
you can calculate overall latency with the cas speed and transfers per second
CAM gives you the ability to have low cas latency and high Speeds
like actually high speeds
currently DDR5 8000+ is only possible by BURNING YOUR CPU
and anything over 9000 i think needs some kind of active cooling
CAM would in theory allows you on DDR5 to hit speeds of 10,000mhz
thats alota mhz
also about VRM placement
i dont think its an issue modern VRMS are vastly too overbuilt as is
you can knock an entire side of VRMS off and still be able to comfortable pass several hundred Watts over a platforms rated power kinda insane
I do agree that going CAM will change mobo design a fair bit maybe add an inconvinience or 2 but i think for the most part it would be for the better especially as next gen DRAM comes along and thats wanting to be even faster i think DDR5 makes it very clear that the current DIMM design isnt really upto the task of handeling DRAMs speed anymore
Thank you that is my bad
And thank you for informing me on flat packs
@jade scaffold from what I hear a lot online, AM5 actually has a huge trouble with 4 DIMMs, I personally never had this problem building systems.
But a single CAMM2 going to 128GB would solve that issue entirely 😅
Yea it does
At high speeds anyway
Even more so on xX70E type boards
Yeah, I never overclock other than using XMP and it always ran fine😅
But in this case CAMM2 would only be beneficial.
It does kinda kill your ability to upgrade as cheaply is the 1 issue with it
And you can only have 1 on a board reasonably speaking
But I think the speed advantages and for the itc ballers out there the extra cooler clearance is worth it also lower power consumption
For me it's the cooler clearances.
More often than not, especially with ITX builds, youre limited to smaller coolers because 80% of the RAM is just too tall, with CAMM2 you can get a much larger cooler and it even cool the CAMM2 as well as it cools the CPU and VRMs.
To me, CAMM2, even harder to upgrade, it is a good upgrade.
Sometimes I love TrueNas Scale, other times I hate it. Going to update to the beta version to see if it fixes the problem I've been having with apps.
Not fond of using a beta as its not listed as stable, but it should be close as the projected release of the stable version is October.
Well, its allowed me to reinstall my apps, where as the previous verson of Dragonfish did not.
Spoke to soon. Pihole isn't showing my ethernet interface/isn't connecting to it for whatever reason. Keep getting an error that the interface doesn't exist.
Its the same interface the server itself is on, so its working and active.
Fantastic. In trying to figure out what broke in my Pihole setup via my TrueNas, I somehow after a reboot into a previous version of TrueNas (that was working just fine) has broken the permissions of the entire server drive. I cannot share any file to the mapped drive (I unmapped and remapped as well). Says permission is denied.
I have 2 pools each with a Share. Both have the exact same permissions. Yet in my main share named after the server I have no permission to make a new folder or transfer a file. Any ideas?
So, I can add/share/create in the VM data set. I cannot in the Abyss Server dataset. Any folder I put in there when I did have permission inside those folders I can add/create/share.
Are you only trying to manage those via the gui?
I'm trying to figure out what I broke that killed my ability to share anything to the Abyss Server datapool
I’d be looking at what perms are in place. There might also be something in /var/log that you can chase down
Everything I can see on the TrueNas side says permissions are the same. But clearly something isn't.
If I go indo windows and check properties of the 2 drives. VM shows 2 users called creator owner and creator group. The abyss-server does not show those permissions in properties.
You might want to go into the cli in Scale and see what that says
Eh cli?, how would I do that?
Either ssh into it or I think there’s a console tab in the systems section of setting
Have you done much Linux?
Very little. I see the cli is on the startup screen for the os. Do I just select it and then look at what it says?
Not exactly. I’m talking about doing some typing to check what permissions are on the mounted directories and marine matching uids from there with the users in the GUI
Okay, well I am in the cli command, what do I type?
ls -larth will show permissions and ownership
root root in that example means root owns it along with the root group
if you type id $user it will give you the uid for the user and and group id (gid) of associated groups
Your going to need to check the path for your mounts to see if they match up with the acls in the gui
Typed in ls -larth was I supposed to hit enter? Cause that pulled up a menue I think, and I can't get out of it.
Ctrl+c is your friend
Not doing anything, just stuck with (END)
Okay, I'm going to need a screenshot or screenshare to understand what you're looking at
I have a list that says
account
app
auth
cluster
directory-service
network
reporting
service
sharing
storage
system
task
then a bunch of tilde key strokes
and lastly (END)
Was I not supposed to be? Thats the first thing that came up when I searched cli for truenas.
No, that's my bad. I forgot it's the default for a fresh install. I disabled that on both of my systems at work because it felt redundant to the GUI
If you click into the console window and hit ESC does it do anything?
It responded, got a summary is less commands that popped up. And says press return for more or q when done
q should be good
And back to the previous list, lol. Sigh, its just not my day.
Okay back to being asble to enter commands.
What are your options?
interesting
Maybe let's try shifting back to the GUI
Are the ACLs the same for both datasets?
and should they be?
Unable to check permissions of the root datasets.
The datasets were existing and you imported them, correct?
No, never changed anything related to datasets. Only was updating TrueNas Scale, then downgrading after a bunch of problems. Ive done that several times before and never had an issue before.
Was trying to figure out why an app wouldn't work anymore (still haven't sorted that out, but thats not important atm). Not sure what else I did, to screw up the permissions.
What did you unmap and remap? The SMB or the Pool?
Neither. I was unmapping and remappingthe drive from windows.
oh
okay
Could you try mapping a new dummy SMB to test if that works from Windows?
So create a new SMB in TrueNas them map it to windows?
Well, shit. Can't create a new pool as I don't have any drives left unused.
Thats going to force me to add to an already existing pool and dataset though no?
Also
An SMB share should just be exposing contents of the dataset it's mapped to
That... is interesting
It will let me add a test smb to the VM share, but not the Abyss-Storage.
I feel like this is coming full circle as I had apps running on the root share cause it worked (and then broke, which was what I was trying to originally fix).
The-Abyss-Server isn't a child of /tank/?
No its a root dataset
oooooohhhhhhhh
I think, this is apparently a problem then?
When iXSystems was walking me through setting up the system at work they made a point to state that /mnt/tank/ was the preferred practice for mounting datasets
so
/mnt/tank/$dataset/
was their recommendation
Yeah, the guide I followed was originally for Core, so I migrated over to Scale back in Feb.
So, what or how do I fix how the dataset is mounted without losing any data?
If it were me with my system I would unmount it and then remount it with the new path
which... I haven't ever done that in BSD
and is definitely console stuff
Well, I definately don't have time for that, hell I should have been asleep hours ago. Gotta get up in less than 4 hours.
I'll be on a staycation in a week, so good time to tackle that then.
But, correct me if I am wrong. I can unmount and remount without losing any data?
Yes, so long as you don't reformat the LUNs or anything
Should be straightforward and it may not immediately correct the perms issue I think you're having
but it should get you on a path to having things working as expected
Blah, I got the app working. Pihole is up and running but ads are not being blocked on wifi...
Scratch that, add blocking doesn't work on my phone, but is working on the laptop.
On the phone you shoud set a custom "private DNS" server and/or in the wifi settings tell it to use pihole as a DNS (it might have the old dns server saved)
Aha, that was it. Wifi settings on the phone had an old DNS stored. Entered the Pihole DNS and its working.
Okay, ready to tackle my TrueNas Scale pool setup. I need to umount and remount? Or import data when creating a new pool? That will not result in data loss correct?
so i tried the flatpack launcher but for some reason steam cant launch it. i tested it on my pc where it steam did launch it. so now i have the arch download but dont know what to do past that there are 2 executables that look like they should build the launcher from the git package but they will not run
Remount and then import as a new data pool
Yes, flatpak is an installer (which requires the flatpak framework to be installed beforehand), not the program itself
You could download the flathub appstore and it will have everything else as a dependency also installed.
You need to download the AUR Bin package, then double click or right click and open with "installer" or similar
I got a working version from the flat pack but it is not compatible with steam (this is an issue as the desktop mode controls are suboptimal for gaming)
I so I down loaded the aur-bin hopeing that it would be an install more akin to what I use on my PC that is comptible with steam
@jade scaffold I just noticed, my AyaNeo Flip DS has something "illegal", it came with an Intel AX210 Wi-Fi and Bluetooth, in an AMD system😅
intel do tbf have the best wifi chips
Well, before I unmount my NAS pool, as it has my Pihole running on it, I bought a separate raspberry pi to run Pihole, to keep my network up and running while redoing the nas. Plus, if I ever have to shut down the server I won't loose my network.
Not a bad idea
Just trying to figure out what ethernet interface is required since it lists 3 different ones, lol.
Found it, 3rd try, lol.
Oof
They do, but AMD system 95% of the time have to do with the bad Mediatek WiFi 😅
Nice and toasty😅
intel has been trying to compete with nvidia for spiciest chip, but nvidia really only wants to compete with themselves
I wonder how long it'll last, Nvidia's monopoly though
The difference is though Nvidia still lead efficiency over amd
I wasnt even meaning compared to their competition, just in terms of highest number
I think Gruzzob was comparing power consumption
I was
ah then its been nvidia for a hot minute h100 could reach 700w iirc
the dual die blackwell chip will do 1000w
in the consumer space though.
intel and amd have both had hedt/workstation cpus that can push near 1000w
I think one of the TR chips could be pushed near 700w or something
Only in turbo boost and overclocked, or else it can stay around 450w in boost only.
Although mine(7970X) doesn't pass 350w.
not by default
@CoolKid369 's original VOD: https://www.youtube.com/watch?v=X3eOsb5DLDQ&t=7s
Music used:
Wh40k: Mechanicus - Children of the Omnissiah
Wh40k: Mechanicus - Caestus Metallican
Wh40k: Mechanicus - Noosphere
Join the Discord: https://discord.gg/unTPjHQ2CR
Buy me a coffee if you want to: https://ko-fi.com/bluprint27
Follow me if you want to:
Twit...
the power cables xD
Holy hell, troubleshooting random iot devices not able to remote access with the new pihole was a chore. Apparently some of them use IPV6 for their connection and they all had to be individually configured.
And now that I am home I confirmed my pihole is blocking my vpn that I use on my phone while at work.
Would a R5 5600X be a good pairing for a 3060 or RX6600 or is there a lot of imbalance risk in that pairing?
3060 is a good GPU, and pairs nicely with the 5600X
That's a pretty common pairing. You might also look at the 5600g or 5700x/g, both chips tend to be pretty close in price
Is the G any better to the X? Sorry I’m a first time builder so I’m kinda swinging into it all
G is slightly worse, but is often cheaper. It also has a igpu which is nice for debugging
I see I’ve seen the X running about 125 bucks so far not looked into the G
If you post a parts list someone here will be happy to go through it for you
I’ll do that.
Though I do have one question related to CPU coolers
I’m a tad confused on the major differences among them
I placed a Thermalight Peerless Assassin on my list but I’m not sure if that’s overkill too little or if there’s better alternatives
Cpu coolers are not something that you can really go overboard with. With pwm it just means it will be quiter.
Overkill is when it has little to no benefit over a less capable cooler
The peerless assassin rates very highly though, can't really go wrong with it
Strictly speaking it's overkill for that class of part, but at that price you are not really missing out
Depends on how tight your budget is
The Ryzen 5000 chips run a bit hot for their power usage, so a better cooler is... better 😆 (up to a limit of course)
Biggest issue with bigger coolers is actually installing them a lot of the time, can be a bit fiddly because of their prodigious bulk
Something like a 212 would be cheaper and preform adaquitely, but are getting more for your money with something like the peerless assain
A d15 or darkrock 4 would be overkill though
Oh no, the hyper212 is a really old design now. You can get better for cheaper or about the same price (don't ask me what though)
Yeah, I don't know what the modern equivalent is ither. I remember when it was all the rage though
It's like the d15, it's not bad.. but it's dated and other stuff has moved past it
It's been around for what? 10 years at least now?
07 for the 212 I think
Ah, yeah that is anchent.
People were sticking them on p4, c2d, and athlons
And it's reputation was very well deserved
So basically a peerless is something that can stick around awhile
A good CPU cooler can see you through multiple systems
Yeah. You got enough headroom to cool any modern consumer chip, though you might be pushing it with an x950 x3d class
Pushing my wallet too lmao
That’s like a 550 buck chip
Also in your guys opinions would you say the RX6600 or the 3060 are better
Ik they are similar cards
Honestly though, for CPU coolers (like cases), my suggestion is to find something you like the look of and then check to make sure it isn't shit. For coolers this of course is only really important of you have a transparent side panel and will be seeing it
I will disagree there. Aios look cool but I would not touch one with a 10 foot pole
The 3060 will be better if you are doing any RT (ray tracing) stuff, but iirc the 6600 out performs it otherwise.
I got one for free and I am still running my stock cooler
Maybe I should have said "something you can live with", which encompasses looks, price, etc
Now another thing I have to inquire about is I notice the 3060 and 4060 are in a similar price range. Why? Are they a copy paste or is there some benefit to the latter card?
Prices are often a mess
The 4060 uses less power, but performs about the same. It was a lazy product
Ahh not worried about power tbh
It's generally not a recommended product
4060 is the more modern card, but it's preformance compared to last Gen is middling
I can safely say if there’s one thing I ensured I won’t need to worry about it’s power plant lol
So basically avoid it got it
Well depends on the price difference
less power means less heat, and also means you dont need as big (ie expensive) a powersupply
From what I can tell 3060 still seems to be ever so lightly cheaper depending on the card
There is no bad products, just poorly priced ones (with some exceptions)
that makes them a bad product, but not a bad device
Fair enough
I chose a 650w power supply on the basis that would get me through a few sets of upgrades over the course of building. So o can’t particularly see a downside to the 4060 but neither for the 3060. But for the right price seems the 4060 may be the way to go for the thermal aspect
650 watt is on the lower end for modern hardware
Personally I would target 850, 650 will work but you have a lot more options with something bigger
RM850e
I thought the 850w was only for the RM fully modular series
the RM line is solid, cant really go wrong with it
Just make sure you connect it to the upper PCIe 4 slot, and not at the bottom
I imagine it’s not good for the system if I do that
Eh, 4060 probably won't care that much
Nah, it's because lower motherboard slots are usually slower, like at max PCIe 8x or 4x, but I seen 1x for the bottom slot as well
Okay so ik I set on the 5600X but there’s a 40 buck difference between the 7 5800X. I only ask because all I really know to base on is price Ik that’s dumb but ig we all start somewhere.
Oh gotcha gotcha
8x is more common, and 4060 could probably be fine on a 4x
But yeah, may as well use the best slot
If pure gaming is all you do, the 5800x is not going to change much
But if you want to do other tasks, the extra two cores won't hurt
Would it benefit someone like me more to have those extra cores when I run other tabs while gaming?
It's fine as long it's PCIe 4 4x or 8x😅
Because the 4060 seried are only PCIe 8x
It will feel more snappy yes, especially if you play DX12 or Vulkan titles
What refresh/res are you targeting?
Ideally 144hz but I know that is indeed a possible ask here
And res?
I hope it went well
You should be able to do 1440p 144Hz.
I change sit up a lot ngl but I am worried about cases atm
Changed*
I am also worried I created imbalance in the force (Video Card and CPU)
I seen worse, I seen Celerons being paired with a 3090😅
Pffft
I have a 7800X3D mated to a 4070 Super
I worry if the peerless can keep it cold
Biggest issue now as I said is trying to find a case that will support it well. I don’t know much about the dimensions of these things tbh
4060 isn't that big, so don't really worry that much about case size.
What's your budget for your case?
Couple hundred at most
Couple hundred, you know that isn't smart to say😅
But I won't go overboard, look at the Fractal Design North Midi tower
Oki and yeah buuuuuut it’s fine
Okay, Hyte Y60
Cooler Master HAF 700 Evo
Fractal Design Torrent
NZXT H9 Flow
@silk jasper what about the Montech King 95 Pro
Thermaltake The Tower 900
I think I like the H9 Flow more
Out of them all but I also like the North Midi
I guess the only other thing is how many additional fans should I buy
As much as you want, but don't buy more than the case supports.
Unless you want to push-pull on a radiator ofcoarse.
I recently replaced all the fans in the Hyte Y70 for Corsair Link fans, it's a costly endeavour on a costly case😅
I’ve always been info the Li Lian’s
I forgot they existed 😅
Oi, still running into network issues. Anyone familiar with pihole on a raspberry pi? My 32GB sd card is full after a day or 2 using pihole.
Well what's on it? Did you configure some kind of network caching?
Usually pihole only needs the OS, the application binaries and space for the block lists (just a couple megs)
So the pihole/var/log file is 22gb somehow
The only thing running on the pi is pihole.
Okay, was able to finally clear the logs, had to install SQLite3 to get the proper commands in the terminal.
Not just to figure out why the logs were so full. Perhaps because I restored a backup from when I was running pihole in Truenas Scale?
Oh fffff. knocked over my old Synology diskstation and now one of the disks is saying its degraded.
Removing it and reinstalling it didn't help, but great news. The other drive after I turned it back on failed. CRashed, and went read only. So, I may have just lost my entire diskstation.... Fuck
Noooo!!!!!
Welp... Im having an adventure. Diskstation held Windows backups from before I built the server this year. Plus some ofther files. I think, I have everything on the Truenas server but I can't be sure.
Turned it off, took disk one out (that one went into read only) to check disk 2 (the one that had degraded, but after a shut down removal and reinstall said could be initialized) and it doesn't has DSM on it. So, I think disk 2 is a total loss. Hopefully can get disk 1 working or at least pull the data off it.
It won't let me re initialize drive 2, and drive one it won't allow a repair because drive 2 failed...
Oh, I have access to Drive 1 and can copy files over. Thats an upside. Downside is its nearly 8TB of data on a 1gbps lan port on the diskstation.
And for Drive folder one its more than a day to transfer over.
Just looked at the specs for this very old diskstation and yeah. Hopefully the remaining disk doesn't die before file transfer is done.
Its a DS218J. And it only has 512mb of ddr3 memory. A 1.3ghz dual core cpu on a 32bit system.
Okay, back the the raspberry pi and pihole. I'm running out of space again. screenshot shows df -h results after I deleted all the stored logs. It freed up 4.5GB from /dev/mmcblk0p2 But its still using 23GB and I have no idea why.
du -sh /* might give you a clue where to look
oh, sudo it
Does pihole perhaps have verbose or debug logging enabled or something?
What commands did you use to delete the var log files?
I wonder if you have an orphaned file handle
I have no clue. Its a stock pihole install. Ive changed no config settings.
Trying to find the command I used. It started with rm I know that.
I dont think I did after words. Let me do that quick.
Yeah, the file might not be visible to tools like du because the file handle is gone but would still be occupying the space because the service has the file open.
This is after a system reboot.
what does the df show?
That did it.
Nice
Now to watch for a day and see what happens.
logrotate is your friend
5.2GB still seems really big. But the log data should clear every 24 hours.
What's clearing the log data?
I think I read, or thought I did that the standard log data should start deleteing after 24 hours.
Ah, found the commands I used to clear the logs, and yeah I had never restarted the system as I had stopped and restarted the FTL service thinking that would be enough.
Rotating normally means it starts a new log and renames the old one. Ideally it compresses the legacy file to help save some room
Yes, after I found more info it rotates the logs at 24 hours and deletes after 6 days.
Nice
Well, ordered a new 12tb drive for the diskstation and doing a repair and rebuild now. Fingers crossed the other drive checks out and I only lost 1 total.
Full day of the pihole running. The /dev/mmcblk0p2 has only increased in size by 3%. So far so good.
So, need more pihole help. My dns query log is being spammed by something. Like tens of thousands of requests every couple minutes. e.f.0.0.0.2.6.2.ip6.arpa has the most hits.
From what I can see its a reverse query for ipv6 addresses. Buts its pamming my dns requests and I'm hitting the rate limiter. And if I turn the limiter off, my pihole starts to lag.
Is that coming from inside your network or outside it?
The ip address is the UDM Pro gateway itself.
It says type is PTR, the client is my UDM Pro, status is okay, reply is N/A.
Does it show up in your unifi logs?
Not sure, does the UDM log that?
I think it might
It shows me traffic, but no actual log that I know of for domains.
Yeah, saw that. Haven't found a solution yet.
And I keep losing internet with my pihole. Ended up setting up a secondary dns to the udm.
Or losing internet connection using the pihole as the dns server.
I'm also not using conditional forwarding in pihole.
I get query spam like this. This goes on for pages and pages.
Starting to wonder if ipv6 is causing a problem and if I should just disable it
Possibly?
Ipv6 always causes problems ;(
Like you have to think in two entirely different network structures.
One being the nested ipv4 sturcture where you translate stuff
and one where every device has like 3 globally unique long ass Addresses
And I get home from work, half my network is down. My zigbee hub is no longer connected to the internet, just my private network so no remote access. Nothing was changed overnight. Just what am I doing wrong here.
A while back I had an issue with a modem, kinda similar like you where it would disconnect and not reconnect, and it was down to thermals, so my curiosity, have you checked the thermals?
Wait a minute, wtf. Just did an ns lookup of the raspberry pi. Its pulling the wrong ip address, the ipv4 address for the previous pihole install. But, ive changed the dns in the unifi udm pro.
Its in a open air rack mount.
Yeah I get a communication error with the pi using nslookup in the terminal for the ipv4 address.
What would be pulling the wrong dns? My router is manually set to the correct one.
A computer or a phone can pull the wrong DNS
I mean, so was my old router😅
So shit can happen.
Yeah but, I'm ssh the raspberry pi and pihole. nslookup is just for the pi no?
Always when I change settings on a modem or router, I reboot every device connected to the network.
Just to prevent stuff like where 2 devices ask for the same IP or any setting getting messed up.
Well, the zigbee hub didn't connect, so the reboot didn't fix that problem.
I mean all connected devices 😅
I'll reboot the zigbee hub first, see if it connects.
Hmm. Client list on the pihole still shows an ip address to a defunk address.
I have a MAC address, but I don't see it listed in the UDM devices list.
Can you remove it?
I don't see it on the UDM pro, I see it listed as a client under the client tab on the pihole gui
Hm, weird, then I don't know, you could try to block the "ghost" address
Unless the pihole just stores orphan clients in a database?
It could, some devices do that
Whats weird is the client list on the pihole gui has a client listed as pihole using the old pihole info.
Okay, still really confused by this.
242 is the pihole. That address is plugged into the UDM pro. That 53 address was the old address of a pihole install.
I see why IT networking people can be paid a lot. Nearly 4 hours and finally got everything working. However, using nslookup on the raspberry pi still shows the above image. Yet, my zigbee hub has an outside connection and my vpn is working again, so I have no clue.
Back to square one with massive dns query requests coming from my udm pro.
Literally turned the per client rate limiter up to 80k per 60 seconds to see what happened. And the udm pro ip address showed 80k were of dns queries in 60 seconds.
Its reverse DNS lookups.
Pihole shows my UDM pro IP address, the pihole ipv6 address and the ipv6 link local address for the udm pro as well as the udm pro mac address.
What I'm struggling with is the PC has its own client listing, and its barely over 2200 requests. So, its not the pc causing the issue.
And internet crashed. Had to set a secondary DNS (ipv4 and ipv6) that pointed away from the pihole to reconnect.
Sounds like you have to reinstall the PiHole
Already done that, multiple times yesterday. It would work for a couple hours then fail.
And I just finally figured it out, I think. Its been 2 hours and I haven't seen the tens of thousands of quries I was seeing before on the pihole. I had the DNS set on the WAN side of the UDM Pro, and auto/manual (I was switching back and forth during diagnosing) the UDM Pro LAN. Ended up putting the UDM Pro WAN on auto, and manually setting everything on the LAN.
SUPPORT OUR REPORTING: Buy a coaster pack, solder mat, or more at https://store.gamersnexus.net/
In this report, we close out the story on what's going on at the tech company that sold over $250,000,000 of products in its time in business. The company has been collapsing from within for years now, silently, but grew louder this year as contrac...
And fuck, looks like both drives from my old Synology diskstation have completely failed. EaseUS data recovery found the raid diks as lost partition 1 and 2, but so far have failed to even pull any data on them.
Yeah, this isn't good.
This is why I never use a raid set-up where you make 1 disk out of 2.
Well, neither recovery tools I have installed worked to get access. So ended up sending a quote in to a local very well respected data recovery company to see if they can do anything. And, if they can't, it costs me nothing.
They also specialize in raid array recovery. So hopefully I can get my data back.
damn
Yeah, can only blame myself. I knocked over the diskstation off the bottom shelf of my server rack. Couple hours later the drives failed.
I got some stuff off of it before the drives straight up crashed. Even purchased a new drive as one hadn't failed completely but needed to be "repaired" according to the diskstation.
Didn't work. Installed the new drive and the DSM software was gone, the diskstation had reset itself, and every single DSM install failed or just froze.
i need to get myself a propper server rack
so far all my stuffs just in normal cases
server rack would make accessing my stuff so much easier when i wanna do upgrades and such
I have a 12u Startech roller rack. It works for what I have. I'm about to add another switch to the empty top slot for 2.5gbe ports. Not sure why my phone decided the blue indicator light on the server needed to be blindingly bright.
Rack will end up with a Dream Machine Pro, a 24 port switch with 16 poe ports, a 2.5gbe switch, and my silverstone 4u server chassis.
yea something like that could work for me
well I did it... 4s lithium ion battery for my power tools. I am afraid to even touch the damn thing https://cdn.discordapp.com/attachments/524188294456672258/1282631211788931144/image.png?ex=66e00ee7&is=66debd67&hm=30ad5f773a4fe5227cdbf13b0d43cd6c20a22f83255115260b367dabd7aa43ea&
I think I did it right but really no idea
You won't feel up to 48v😅
Unless it's like 200 amps
I am not afraid of it shocking me, I am afraid of it it exploding!
actually it did shock me, 14v is more then enough to shock you threw your skin
you can shock yourself with a 9v battery remember
and those are some of the shittest power sources around
not on external skin, 9v is a little low for that, just the mouth
have you stuck a multimeter across the output to verify the voltage looks right?
unless you wet your skin with some salt water
yeah, voltage is right and it makes the drill go Brrrr
nice
its more charging it I am afraid of. I really hope I got the BMS wired up right
I once survived 400v with 12 amps
Wasn't my own doing
All I remember is waking up in the hospital
I have been zapped more then enough times but thankfully never anything that landed me in the hospitle
electrocution isnt a risk with this. it just scares me because these are lithium ions and i wired/welded/soldered all this shit my self
the pack is connected with 0.15x10mm pure nickel strips (they are real, I checked) spot welded to the batteries. then wires soldered to the tabs to connect to the BMS and the output terminals. along with balance leads to each battery for balance charging
I have to build the charger for it next
And, because I am a glutton for punishment, now to fix the server NAS by remounting the drive pools.
Good luck
I'm updating Scale, and then going to rebuild the pool. This I assume is an easy task? As I literally lost my 2nd NAS to complete drive failure just the other day.
Make a backup of your configs and make sure you’ve got something handy for looking up z pool configs
Okay, and a remount did nothing, or at least what I did was wrong. I unmounted the pool. But if I go and create pool, it says it will wipe the current data on the drives. So I imported the pool as I saved the data and config, and I'm back to square one with one caveat. Now my PC cannot connect to the server via network mapping.
Did you update the share?
Current datasets.
How do I do that?
Click edit on the share and update the path
Also, when I go to edit share acl I get this error:
Here?
So Just restart the sha\re first and then check?
Cause the original issue was I had no ability anymore to add to the main mnt The_Abyss-Server as I didn't have permission.
Yeah, but let’s try getting you back to the previous state
To restart the smb I just restart Truenas or?
Ah, found it.
Turned it off then back on.
And trying to map the network drive fails. Windows cannot access \TRUENAS
Hhhmmm
How do I find the actuall path again? I'm wondering if its just not auto detecting the correct name?
Actual path of the mount?
Yeah. Cause nothing I have should be named TRUENAS. The share name is Abyss-Storage, and the path is /mnt/The-Abyss-Storage
I remember something about needing the windows username is the path if I recall
Do you have a dns entry for the device?
Or you can try going to \\192.168.you.rip in file explorer
Yes, it points to the pihole I use as a DNS. I just changed that in network config as I noticed it was pointing to an OpenDNS dns address.
Didn't work.
NSlookup TRUENAs give a completely different address.
Is the ip static or dynamically assigned?
Static. I assume the ip is the same as the truenas web gui
It should be, yeah
Yeah, same, no response.
It still on hitting browse poulls up TRUENAS, just won't connect.
SMB service is definitely running?
space marine 2 is out and as per usual the 5700xt is having a fucking stroke over it
Thanks AMD
Hhhmmm I wonder if it would persist through a reboot
please revisit your previous GPUs and Drivers rather than just ignoring everything released before 6000
Scale itself? or the PC?
The PC. Or is Scale in like a vm/Container?
Scale its on its own computer.
Okay, then yeah, I would try a reboot
Rebooting now.
No change
Could this be the user issue I was having earlier breaking the ability to connect after I remounted the pool?
Checked pihole, found the server ip that Truenas uses and nothing is blocked there.
That actually would make sense
So I need to fix the user permission issue on TrueNas then, the original problem, lol.
So, now to look at users in the smb share?
That would be my next step
Okay, if I go under credentials and Users, I see my username. But under roles it says N/A, whereas the root usewrname has listed Has Allow List, Full Admin
As far as users for the smb share would it be under share acl or filesystem acl?
Yes? I’m still puzzling that out myself
Some of the stuff is more responsive to the SMB part and some is more responsive to the Linux part
The SMB is going to need access no matter what. I’d start there to figure out access from the network users and then switch to the file system
Uhm, could I not just give every user full permissions to keep this from happening again?
You can
So, both data sets show the same permissions. Back when I could access the NAS I could edit one dataset but not the other, if I go into share acl and just hit save changing nothing I get this long error message:
I can’t 😬
Does it do the same thing if you create a secondary test share with the same permissions and path?
I don't have any more drives available to create a share unfortunately.
You don’t need a drive to create an smb share
It can path to the same ones you already have mapped
Share, not pool
Pools need drives
Share’s are like overcharged shortcut icons
Ah, gotcha, created a new share called test.
Filesystem shows test user permissions as:
"VM" share shows
And the abyss server shows:
Checking the share acl the test share shows:
VM: top user that I couldn't fit on the screenshot is everyone.
The abyss server has the same permission set.
It looks like it’s using uid and guid for some groups
Good, bad?
Unsure. Different and maybe odd
I'm wondering if I should just change it to everyone full.
You can definitely try it
Changed VM and Abyss Server to everyone and my username.
🤞
Okay, checking filesystem acl if I add my username I get this message: Good, bad?
It’s a reminder of the risks. It can be fixed via the cli if it goes horribly wrong
Changing share acl gave no change. Should I delete my username and just try everyone?
Filesystem acl gives me this window when I first open it by the way:
Up to you on what to do there. Go ahead and add the everyone perm
Btw, don’t type what it actually is. That’s a nasty timeout
Okay, tried to add user to filesystem share for VM:
Whaaaa——-
I don't like the sound of that.
Not sure how to do that?
When setting up the SMB you should be able to go one dir deeper than the mount point you’re trying to point at
Okay, also tried using the ip address to link as I discovered not only do you need the ip address but the name of the share as well. I got a new error doing that:
Progress!
This is before I added any share.
Okay. Add the share with everyone and see what it does
Also, btw, are you using some kind of ldap or AD?
? That sure what those mean.
Trying to add a share in just /mnt called test gives me this:
User authentication. LDAP is Linux. Ad is Active Directory
It does allow me to create a share in /mnt/The_Abyss-Server or /mnt/VM however.
Yeah it won’t like that
Test is what I was talking
So now you need to figure out how to connect as the user that has access permissions
I just remembered they maintain their own discord
They might be able to get you sorted faster than I can
SO, looking at my user permission does this all look normal?
Oh, forgive me. Not my area of expertise.
HOWEVER, @surreal moss you wanna discuss what you did? xD
This illustrates what I was trying to send
I tried to break it up with a \ and discord thought I was doing something different
But again… why should that even work!?!?
For tagging everyone?
I was trying to break it up. Not escape it
It doesn’t actually tag it, it is just that the AutoBot caught you on “attempting” to do it
Why is it show "op" mode for Jack?
Because I made this thread lol
Ahh, "Original poster" not "server operator"
AutoBot and I now have a vendetta
Discord too for whatever shoddy programming would have let that trigger an at-everyone
So the user looks normal then?
The Samba auth. might be hurting you if you don’t have an AD connection configured
YEah, not familair with AD configuration meaning.
Its linked with my windows login.
My user doesn't have a home directory it looks like.
That’s normal if you didn’t tell it to create one
Okay, I didn't.
As far as permissions
Also, nslookup won't connect the the truenas server.
The perms seem alright
Fucking halleluiah... WTF I'm happy as fuck but at the same time I have no damn clue what I did to get access.
I still can't add a folder or anything. But I have access again.
I will take any, any victory at this point.
And more progress. I added a new dataset inside the main pool and have read/write access. And it looks like after a long time reading that when I setup all this I was running Core, and then migrated over to Scale. Well in Core you are able to use the root dataset to store individual folders and such. In Scale permissions were removed for the root folder.
So now I need to figure out how, via shell move all the contents in the root folder into the new dataset.
Okay, more progress. Shell shows the issue for /mnt/VM why I can't even access it. As well as why I have write access to the Storage folder.
Okay, 2 questions. First related to the shell screenshot above. How to I change the VM dataset ownership group and other to match what is listed under the abyss-server datatset. r-x to rwx for ownership basically.
2nd question. I need to move all the files and folders I created in the root dataset the abyss server while in windows to a new Storage folder dataset inside the abyss server root folder.
chmod 777 /mnt/The-Abyss-Server/
chmod -R 777 if you want it to be recursive
I would use a Data Protection job to either sync or rsync
That changes permissions to rwx?
Scratch that, yes.
Would I need it to be recursive though? Not familair with what does does.
THAT FUCKING FIXED EVERYTHING.
I have full access. The fuck. Literal days of pulling my hair out and searching the web. I'm so very pissed and happy at the same time.
Thank you so very much @surreal moss
I have full access. The fuck. Literal days of pulling my hair out and searching the web. I'm so very pissed and happy at the same time.
Welcome to Linux!
Thanks, I hate it. Nothing intuative at all, or simple, lol.
Okay, so new problem. While I have access. Anytime I try and move a file/folder in windows it tells me I need permission from TrueNas/Username to make changes. What do I change in shell to fix that?
Going to try editing the local username auxiliary groups by adding in the buillt-in admin group.
That didn't work.
It needs to match the group of the user trying to make the change
but honestly you've entered unknown territory for me now. The one's I manage are joined to Active Directory and can associate with the Windows user so things are shared correctly.
But, I'm in windows, where I am the username accessing the network drive I thought. I'm confused as to why its asking for permission. Unless, unless its cause I have guest access on.
Maybe?
... Fuck.
Turned off guest access and lost access to both network drives. Gonna try that command again with guest access off and see if that fixes it.
Nope, didn't fix it. wtf
I'd have to pull up the manual but my guess is that it needs to have guest access enabled and, because it doesn't know the user connecting, it requires you to authenticate as a user on the system it does know each time you need to execute an action
Well, I disconeccted the VM drive to reconnect it. Its asking me for username and password. But its telling me what I enter is wrong?
Is that with guest on or off?
Guest off.
But with guest on I bet it would work
Hmm. Under aux groups there is a builtin_guests. Gonna add that and see.
Okay, got access back to the abyss server in windows via ip address, still no access to VM share.
Alright, I think, fingers crossed I got the abyss storage dataset straightned out. I can now read write in windows via mapped network when gues access is turned off.
Still cannot map the VM drive though.
Only difference I can see is the VM dataset doesn't have the everyone listed on the access control list. And it won't let me add it either, not even listed.
What I can see in the difference between the smb shares is the sbyss storage is listed as smb/nfsv4 as the acl type and the mode as passthrough. The VM is listed as possix with acl mode as discard.
Also noticing I still don't have permission on random files to copy/delete/move in the abyss storage smb in windows. Not sure what that is about as I set permissions to everyone full access.
I'll take what I can though. I am absolutely hammering the server copying and moving 3.5tb of data. And the old 8700k isn't even breaking a sweat. So thats nice at least.
Fucking Finally
Guys since when do laptops with 40GB RAM have an i3, 2TB SSD, Win Pro and cost only 670€??
Am I crazy or what?
Devils number
It looks like a Lenovo from the business line-up
But without seeing the exact model, I almost think it's a 2nd hand
Or refurb
I’m not superstitious but I am a little-stitious I think I’ll avoid that one
There's like an HP one like that aswell??
I'm very stitious because like what? It's too good to be true!?
Windows 11 Pro is like 250 bucks alone, a 2TB SSD costs 120 AT LEAST, then there's 40GB (there's no RAM module size that would account for that!?) which would be like 200 bucks in sodimm.
There's like not nearly enough profit for this to be true, right?
A 32 and an 8 would account for the 40gb ram
Which, why are the oddball ram sizes suddenly a thing?
Well 24G modules are a thing now (which is fairly sensical in my eyes)
It hurts my 8 bit brain
Since Asus started doing it in some of their Zephyrus laptops
It’s not like Asus would have invented it though
But they made it more popular 😅
Then Lenovo copied it on some of their Ideapads
Now HP is doing it as well
I wonder who's next
I’m not talking about pop computing though. I’m wondering who said “we need 12gb modules” and why
Corsair probably, but they sell so far I know 24GB, 48GB and 96GB kits.
But I haven't seen it with other companies yet
Crucial also sells them apparently 🙃
96GB isn't new. I've seen several servers with that configuration
It's new on consumer side
I think that system has 8gb soldered ram
Which is where you get the 24 and 40gb ram sizes
Soldered+expansion
Yes, all my new networking equipment was just delievered.
New dedicated 2.5ge switch plus each switch and gateway now linked with 10gbe fiber via sfp ports. Also ordered 3 flex switches for the garage, attic, and basement.
WTF Biostar, that card goes EoL soon
https://videocardz.com/newz/biostar-unveils-white-edition-radeon-rx-580-2048sp-six-years-after-original-launch
Biostar intros Radeon RX 580 2048SP, 6 years too late Believe it or not, Biostar has just announced the release of a new Radeon RX 580 graphics card. Yes, the RX 580 from 2017 (actually from 2018). The RX 580 graphics card, powered by the Polaris 20 XT and featuring 2304 Stream Processors, was one […]
Uhhh
Strix Halo added to AMD ROCm The next-gen APU design could turn everyday laptops into ‘small supercomputers’. The upcoming Strix Halo APU support was discovered in AMD’s open compute platform, ROCm. This finding was made by David Huang (known for the first Zen5 review), who noticed that Strix Halo, based on RDNA 3.5 architecture, is […]
ROCm is interesting
its good that its going to be going onto more consumeresque parts because its AMDs Answer to CUDA
I think it's really good development, because more and more companies would like to see alternatives to Nvidia.
yea sadly ROCm is very early on and has a long way to go
The more companies join, the faster it'll go
And seeing how much companies already shown interest in it, it can take off like a space shuttle without rocket boosters
not really
AMD scaled back the developement of it rather substantially
completely ditching the CUDA Emulation part of it
I have the feeling Nvidia has something to do with this🤔
I mean, ZLUDA was going to be the CUDA translation layer as well, and AMD told those devs to stop.
Emulating CUDA only really helps them fufill neiche tasks CUDA can do that ROCM cant yet do more expensive than its worth doing for AMD
because they were using AMDs Code
ROCm is good just gonna be a while till it can do any of the niche things cuda can do but thats not a particularly huge market
main market for CUDA and Tensor ATM is AI so AMD are focusing on that
Talking about AI, FSR4 might not run on older cards if they're gonna use AI🤔
lunar lake is getting close to properly launching really excited to see how it does i hope it does well
I wonder too, but seeing no reviews, does make me worried about Intel.
its not out just yet but its not far off i think its end of this month the first devices with it launch
I mean, usually we already see "previews" by the likes of Jay and Linus, but we haven't seen anything yet
yea but usually only a few days before release
I am still worried about Intel after the entire 13th anf 14th gen shinanigans
Just realized the new chipsets for AMD and Intel are going to cause a lot of confusion. X870 and Z890.
yea xD
Windows revokes kernel access from anti cheat
FUNK YEAAAHH
Microsoft has officially announced its intent to move security measures out of the kernel, following the Crowdstrike disaster a few short months ago. The removal of kernel access for security solutions would likely revolutionise running Windows games on the Steam Deck and other Linux systems.
Wow, that would make Linux an actual legitimate OS for a gaming pc.
So, is this just going to brick all games that use kernel level anti cheat unless they get an update?
Glad I'm not the only one that had that thought
This is amazing, especially for Linux users.
And even security wise.
Too many Kernel Level Anti-cheat can be exploited for something bad, and if idiots do that they have near full access to the system.
Damn. I think I just lost a drive 😦
Nah, Linux has their way in running games that do have a Kernel Level Anti-cheat, but it's a huge pain in the ass.
We're talking about windows games that use an anti cheat system that's dependent upon kernel access
Games that use one of those and aren't maintained anymore might become bricked once that update rolls out
Basically almost any Ubisoft title to date
Not just Ubisoft
I am curious about how they handle it and if they have an alternative to it, like a fake kernel for example.
That would require an update of the software though
I mean, there are devs who seem to love kernel level anti-cheats to hell and back.
I remember the day I got banned from Genshin Impact for playing on Linux and a fake Kernel.
wanna know something funny
windows itself can just act as an anti cheat anyway
Windows has the capability to rather accurately mind you detect and prevent memory manipulation of a program, Might be a windows server only thing not sure if it requires any of the fancy security things Server CPUs have but it can do it so you wouldn't really even need a super invasive anti cheat
not that an anti cheat can do anything against a cheat acting as a keyboard or something anyway
Wouldn't modern AMD CPUs be able to do it with their fancy security chip?
everyones chip has some kind of inbuilt security
not sure what features arent and are enabled and what is and isnt needed
i know standard ryzen just flat out has half of its possible security features dissabled
Lets enable them😃
unlikely
🥲
on Ryzen they are fused off
The AM5 EPYCs have the ones ryzen doesnt but still lacks all the ones propper SP5 and SP6 CPUs have
😭
It would be awesome if they had it though, or not (Remembers the days of Intel Minix)
infact Epyc 40004 supports only like half of the infini guard spec
no SEV Encryption State or Encryption for virtual machines or Nested Paging.
just gets basic security boot, Does get standard memory encryption tho and Shadow Stack
Now I just want Coreboot instead of all the security feature
Just simple
No bullcrap
Lenovo just posted a vid of a laptop with the new intel chip achieving 23+hours of battery life
Now I wonder how much of that is actual use or the average idle time
It was a YouTube test
In comparison with the M2 and M3 macs of comparable size they get 18.5 hours
That's pretty impressive
especially if you consider how much more light weight Mac OS is compared to Windows 11
Not really sure what M3 that is but its got 24GB RAM.
Also to note is the Chip used by lenovo is an ES chip so its likely actually clocked slightly lower than the final versions which will be properly tuned
and sorry it wasnt YT playback it was local video still though scoreing higher than the M3 mac book is regarless impressive
Christ intel may have totally bungled 13th and 14th gen but christ what a tiny lil beast this thing is
The question always stays, have they fixed the older issues
yea the oxidation issues wont be an issue
otherwise every AMD NVIDIA Qualcom and Apple chip would have the same issue
I mean the Harakiri issues
Yea no that won't be
Chips far to low power for it to at any point request a high enough voltage for the micro code to overvolt the CPU
They finally decided that a power limit is good😅?
well also the origional issue was just the CPU requesting more voltage than it actually thought it was
oh wow intel have actually ditched hyper threadding on their standard desktop chips too
thats interesting
I wonder how programs will work with it.
same way I pressume lunarlake does where it just treats the E cores as a Hyperthread for programs that arent automatically programmed to pour over into cores
Except the big.LITTLE were a mess for some programs as Windows put the programs on e-cores only which lowered the performance.
in theory using E cores as a Hyperthread is faster, Hyperthreads only bump up performance by some 30% or so
thats because of the thread director on die
and some windows tomfoolery which ALSO affected AMDs chips
I know on Star Citizen, Windows launched the game on the E-cores and it halfed the performance, or even more than half.
On some PCs people were forced to turn off the e-cores all together, especially people that still use Windows 10.
yea that was fixed like a week after launch
same as AMDs issue
the thread director on 12th gen isnt great nor is 13th/14ths all be it theirs are a lil better but they never treated E-Cores as Hyperthreads they were just other cores
Big.LITTLE can be done right, but I think you should make all cores hyper threaded or not hyper threaded.
hyperthreads are inefficient
Star Citizen benefits from it currently though😅
if you are trying to go for efficiency hyperthreads will allways hurt that efficiency
Wait, you say if I want to increase my AyaNeo's battery life I should disable HT? 🙃
you get a substantial additional power draw then very little additional performance because hyperthreading is only an extra data lane into the CPU its nothing but that
inactive sillicon still consumes power
data interconnects are now so fast that hyper threadings benefits are just kinda rapidly diminishing
sure in a super data intense workload hyper threads might help but thats only gonna be for big simulations games arent gonna really care that much
infact quite alot of games run faster without hyperthreading as they are only built to scale across say 4 threads and the OS doesnt know the difference between a thread and a Core so it often will throw 2 tasks in effect onto 1 core
Currently Star Citizen can only utilise 4 threads, but if you disable HT, it goes down to 2 cores only, losing that little boost.
They're busy changing it with Vulkan so it can use the entire CPU.
My tests were around 3.12, we're now at 3.24😅
They must've made it better then😅
Although, I had the i7-6700K back then
My last ever Intel CPU
Now I wonder if that'll also work on the 8840U 😅
Before you ask, why I am no longer on Intel, it has to do with them removing functionality from the CPU on the mobile side
Intels modern chips before lunar lake are all also dogshit
so like theres that as a reason
they stagnated at 6th gen only just about still matching AMD at 9th and 10th gen then lost 11th and onwards
I mean, Intel did remove graphical options on the mobile chips, where it's now impossible to make custom resolution, which I use to use a lot.
Or on some chips they even remove the option to disable turbo or the wattage step-down options.
And I remember the days where laptops have an HDMI, but were unable to actually use them.
And now Intel has Harakiri CPUs, I hope they fixed these issues on the newer ones, I am kinda tired of their shinanigans.
Battery life only doesn't mean much to me.
they never used to have mobile chips
they were physically identical to the desktop ones
the new Mobile chips are a separate architecture
With mobile chips I mean the ones on laptops 😅
i mean thoes graphics were just display outputs really
you could still change resolution in windows
tho why would you
Because gaming on a laptop with a 3240x2160 on an intel GPU is near damn impossible 😅
But it was possible on lower 3:2 resolutions
Also, the basic setting for scaling is on Stretch to Full-screen, which you only can change when making a custom resolution.
can still change the in game or windows resolution
and you can tell windows to do 3:2 fine
The only resolutions that were in Windows and games were either 16:9 or 16:10, which were all stretched, and the only 3:2 was 3240x2160.
So I was forced to make custom resolution
Which in early Intel drivers was possible, in later drivers your got thrown at you that is wasn't possible.
Iv'e made the mistake of looking at used server equipment...
Found a Dell PowerEdge R730 16SFF 2U server with Xeon E5-2680 cpu, 128gb of ram for $200.
Just use custom resolution utility. It's very compatible with everything (even most really obscure graphics drivers/devices!)
I know, but I just find it sad that Intel simply removed the function without warning
And then pretend it was never possible
I tend to prefer CRU anyways instead of using the graphics driver.
Intel do be goofy like that sometimes haha
They did it multiple times with 1 CPU 😅
Most likely cases of "the developer made this feature cuz he really liked having it but now it's incompatible and we chose to not pay anyone to apply a small fix"
On the same CPU it was possible to make custom resolutions and to even lower the wattage of the CPU and even disable Turbo, and they removed that 2 years after launch
Now I have to rely on 3rd party tools for custom resolutions.
Lunar lake is pulling some insane gaming performance
Gonna be a good year for handhelds this year me thinks
I wonder how the drivers are, if they're good then they could compete with AMD in the handheld market
I was almost gonna click "purchase" during the steam deck sale but it wasn't for the OLED and that price was just a little too much for me
They seem decent probably still not perfect but they are alot better than when they first did a proper graphics architecture
But yea lunar lake is trouncing last gens 7840 from amd not too sure how it's holding up against the Ryzen HX370 but they are very different category of part
Nice, finally some competition, I just hope that the software is good for Intel.
There's still some games that are barely playable on Intel, mainly older games.
And I hope Intel's AFMF alternative can't come soon enough
Yea it's because Intel's last architecture to save costs doesn't PHYSICALLY support DX11 or older so has to emulated them
Which for old games wasn't much of an issue unless you were on the weaker cards or the Xe integrated gpus which didn't have the raw performance to do that translation efficiently
It's also like 1800% faster at running ai upscaling than the last gen according to intel really random thing to include but with FSR4 on the horizon that could really help
Will help reduce some of that upscaling latency as well as do it more efficiently leaving more power room for the CPU and GPU to do their thing
If FSR4 will work on non-AMD GPUs and not use AMD's own AI cores


