#off-topic-tech
1 messages · Page 11 of 1
as long as it can run netflix and some other tabs idc
not gonna be where i play games
I mean. That arm battery life is nice..
But the shenanigans that come with it.. I'll pass and hold on to my x86 machines that I actually feel like I own for as long as possible.
i’ll prolly get a m2 13inch macbook air if i ever get one
Pretty soon your pc will probably be arm and have a nivida cpu in it anyway.
And they'll probably be no difference between a windows machine or apple.. the only computers you'll be able to actually own will be raspberry pi
Lol fr
Your cpu dies, your memory dies, or your ssd wears out/dies. Throw it away and buy a new pc because it's all soldered on, and all firmware paired together.
any hints at what's going to be on fire sale for black friday?
Idiot resurrected old thread and tells that if anyone has temperature problems with Intel CPU to just "Disable the Turbo boost system completely!": #1163142914109603850 message
Instead of tuning it to work with the cooler...
If you actually clicked on that, and then looked at the top right corner, it would show you have 32GB installed, with 31.9GB usable, because of various UEFI reservations for RAM needed by the firmware and hardware.
And only having 0.1GB reserved is very small reservation, normal is 0.2-0.4GB, if there isn't IGP taking another chunk that would usually be 2GB from 8+GB of RAM.
Anyways, planning to get Reolink RLC-520A IP security camera for inside the apartment vestibule:
reolink.com/us/product/rlc-520a/
Will only be recording very small area inside of the outside door.
Any recommendations for something else instead?
Same form factor instead of the long tube kind, smaller the better.
Physical Ethernet a must.
Preferrably without WLAN.
POE power strongly preferred.
Must be able to record to local Linux LAN server in some way.
Must be able to work without Internet connection.
Cheapish.
@jagged snow Just completely bonkers... #1163142914109603850 message
Probably drives and peripherals.
Maybe coolers, GPUs, RAM and cases.
Unlikely PSUs, CPUs, mainboards.
Also remember:
Just because something is on sale doesn't mean it's a good deal.
Never look at a product and its price in isolation - always compare it to the competition.
A $72 SSD with a 25% discount ($18) still is a worse deal than a $50 SSD at regular pricing
I would bet there's probably going to be plenty of good Gen 4 SSDs for sale. Those have already been going on and off sale all year.
just like with avocados
Today I learned almonds are like the evil opposite of avocados, where you throw away the fruit and eat the pit.
😈 🥑
Prices are rising now.
The low-cost market in the US is a barren landscape
Also, good Gen 3 > mediocre Gen 4
and even then you have almonds you can't eat. So super evil avocados
I'd pick a solid Gen 3 SSD over NV2 any day of the week
btw have you seen cashews???
And yes, I concede.
Both the original Steamdeck SOC and the new one are AMD Semi-Customs, so when all the subparts in the SOC already had both 7nm and 6nm versions, just doing the "same" thing as die shrink was cheap.
And the differences between the TSMC 7nm and 6nm are minimal anyways.
they look so dumb
Worse I accidentally mistook some sort of imitation cheddar made from cashews for the real thing when I was hastily shopping on vacation.
who put the damn mushrooms up the tree i swear
But anyways, 122e for camera, POE injector and 256GB Samsung Pro Endurance MicroSD card.
And later about 100e for Pi 4 with airflow plastic case, passive heatsinks and PSU.
To act as hidden inhouse FTP server for the camera.
If someone in the next 6 hours doesn't give me better idea.
oh god, chrome updated and what the hell did they do to it
every damned standard icon they have now in the titlebar area, bookmark bar, etc... are 'blury' icons
Get anti-aliased idiot 
the apprentice casts gaussian blur and I just put my wizard glasses on
imagine using Chrome, smh
its apparently even affecting every "SVG" in websits (but not other images)
get unvectored
you only want one camera ?
Yes?
I live in apartment building, and I don't want to record anything in the real living areas.
Only the entrance door.
ok
because if you planned to upgrade I think poe switch may be worth it
and it may be possible to power the pi with it
No, don't need full POE switch, the active AF (15W/12W) injector costs 11e.
And not planned for the PI, that will just normal USB-C charger.
The pi will be with WiFi once it is added, so no network cable leading to where it is hidden.
Current plan is to put it behind the fridge, that is inset into the cabinets.
Taking power from the same outlet.
i wonder how managing all my google profiles using firefox profiles (since it doesnt link at all) would be a pain to get used too ... cause i see chrome going to shit when they remove the experimental flag to go back to the old UI and not this new 2023 bullshit they created and shat out their ass without vetting
seriously whomever stamped this as ok to ship needs to be fired. now
so, depending what happens in a few months when they retire the experimental flag to use the old ui ... i may be joining you team firefox guys 😦
lol its even blurry on my ultra res macbookpro screen ... holy crap
... installs firefox to mess around in and see about switching too ...
nice reveal of complete gibberish lol that actually makes fieforx seem less
I'm free of youtube ads or their blocker so far
ive not run into it myself, and im still on chrome
firefox + ublock and everything auto updates
but then i dont watch youtubes a LOT so maybe there is something to it
thankfully im not the only one complaining about the new 119 chrome build .... for once, other people are bringing up just how bad something is that i agree too 😄
im not alone anymore! (so thats what this feels like) feels great!
Maybe the Amcrest 5MP Turret POE? As long as the camera supports onvif you're good.. what software do you plan on using?
The manufacturer doesn't seem to have any European presence.
And like I said, that specific cam can do just dump FTP.
ONVIF mostly seems to need purpose build recorders.
There isn't open source or like software for it.
I've been using shinobi.. but it's kind of junk, and it has a nag about subscribing.. also doesn't allow fastforward through timeline with out subscribing.. I really wouldn't have an issue with it.. but the software is feels very beta
Onvif/rstp streams can be pulled/viewed with pretty much any software, vlc included
That specific manufacturers cameras can do phone app, windows program or just direct http.
And can just send the video to FTP server.
Yes, but the point is where do you store them?
Either on the camera, or you need specialist device or storage software.
🤣 peak comedy to wake up to
Took quick look if there was linux text mode program to act as the storage server.
But mostly talk was just about the software to access the video for viewing.
And then that for example Synology can act as the recorder, or specialist devices.
Yeah. Shinobi half-ass works.. there's frigate nvr, which supports using a Google coral for object detection.. I haven't gotten to mess with it yet but it's on my list
Which start from 200e or so for the recorder device IIRC.
I've heard synology is good.. you've also got unifi protect.. but I think that's only for x86 machines
I need simple and cheap single camera solution that doesn't use the cloud and can store to hidden small server.
But Synology NAS would still cost a ton, and are big.
Yeah.
I'm not doing "full" system with multiple cameras etc.
I just want fire and forget single camera local recording for cheap.
Shinobi might work for you for a recorder/viewer as much as I hate recommending it
I've only got 2 cameras on it. They're both wifi. Ones a cheap China cam and the other is tplink tapco. The tplink one is better, but it's wifi
Yes, I could go with ONVIF it seems, but basically overkill for my current needs.
Even when that could run on Pi.
When just FTP server with logrotate like solution to delete old videos to keep space open and with VLAN separation and firewalling will work.
When I don't have need to access the videos ever, except to confirm that they still are being recorded and work, and in case there has been a breakin.
Needs of "proper" system are different from mine.
what's worrying me is how to do space efficient encoding on pi
The camera is doing that already?
ok
Yes, I know.
But I don't need the functionality.
I don't WANT to give easy access to the videos.
And you can just use copy, so there's no encoding
Yes, I know what the protocol is.
if you can get the flow in the pi, all that online stuff you can diy with nginx and ffmpeg on pi probably
Just much more complex than needed in this case.
And the camera can do it too.
But like said, the Pi is only for storage, even if I was using that specialist protocol.
The camera already encodes the video to H264 with selectable settings.
And just then streams that to the storage device using one of several different protocols.
And the storage just stores it, and with more advanced solutions, allows you to easily look at it too.
But as there will be only single camera, and that will just be recording small part of the interior of the apartment, there isn't need to watch the video.
But time to look at the exact software etc. is when I'm going to be after buying the Pi.
Just going to go first with recording to camera local card, for budget reasons.
Even when that causes it to be easily stolen, if there is break-in, and if the person thinks to grab the camera from the ceiling with them.
I checked earlier about pi poe, and it's now fully supported, that's nice
Yes, but again, the Pi will not use POE, it will not have hardline network connection.
Reason I'm going with Pi 4 to get AX dual-band WIFI.
And will just have USB-C PD wall wart.
The camera will use POE.
I was not talking about your setup specifically, it's just that I like that, and I talked about it earlier, so I think to share that it supports it well
And yes, I know that the Blue Iris support for Reolink cameras doesn't actually work, because of encoding settings used.
Blue Iris wants pretty specific ones in relation to keyframes etc.
And Reolink cameras don't fit the bill currently.
Blueiris is windows only too.
Yes. But as point that I know that the Reolink cameras have problems with some of the common "use this" NVR softwares.
I think zoneminder will run on the pi. I'm pretty sure all of that is outside of what you want though.. if you just want to simply record with no live view, I don't know of any software for that
Just that it is the inexpensive option that isn't WLAN only with cloud storage.
Like I said, the Reolink cameras can just use any FTP server as the destination.
Including FTPS (still not completely secure, FTP itself is all plain-text, including logins).
What's the best noise cancelling + surround sound headphones under 500?
Just so weird...
But as the guy has bought too good CPU for the case and cooling, and then just gimped it...
But anyways, that Reolink camera support ONVIF, Dump FTP, and Reolinks own access protocol.
- just plain browser access for setting it up.
And RTSP too it seems.
And the other manufacturer mentioned doesn't seem to be available in EU.
I dont know if there are bare sftp or even tls cameras with good encoding and poe ethernet with ethernet embeding (I mean physically 'sealing' the ethernet inside the camera stand, so it can't be easily disconnected)
Of course the person breaking in can trash the camera.
Point is that it will record to hidden small server.
The moment the person breaks the door.
And that it can in this case be done with just FTP server software, instead of full ONVIF one that are much more complex.
But I was asking about camera alternative recommendations, not about the storage side.
Most of the time, a camera working or not will deter shit like that. If it doesn't then they don't care and they're going to do it regardless
I can't believe he kept arguing the point
Not about deterrance.
Will not even know the camera is there before the only entry door is broken.
Just hoping to get good enough pic of the perp for the police, to have higher chance to get the stuff back.
When the problem is someone living in the same building.
Damn.
AFAIK only has broken TWICE into same other apartment, in one of the floors above me.
So most likely lives on that same floor.
I paid for the whole computer, but I'm only going to use half of the computer
btw one could say the same thing about my embeding stand, it can still be forced out, but the point is :
- the time it gets to be forced out, the pics are already uploaded, the guys are screwed already
an easier to disconnect camera can be disconnected fast enough the pics are not recorded, and easily enough the guys can still be hiding mostly
Yeah, you need something that's recording. I think having a hidden pi is a good idea.
If that's becoming an issue I'm surprised the apartment manager doesn't do something. I don't think that's illegal for them to setup cameras in the hallway or any common area
recording outside can very quickly become a legal headache if the managing entity does not own every apartements around it
The entrances to the building and the common storages already have cameras.
And the outside doors have electric locks with AFAIK logging.
Same for the common area locked doors.

But adding cameras to the upper hallways isn't legal without lot of hoops that would probably fail.
Like I said, the chance of breakin is very minor, but after the second time same door has been broken (if the ramblings of the specific dweller are to be believed), I'm going to use some money to add that camera.
Steel core door would be the real fix, but that would cost double to quadruple.
And cannot be easily gotten, as mostly those are being sold to cases where all the doors in complex are replaced at same time.
But anyways, going to put the order in for the starter pack with camera, injector and card.
Couldn't find anything else prosumer segment.
Just the WiFi & Cloud consumer crap, then Realinks, and then expensive professional stuff.
That's pretty much it. I'm familiar with the reolink name, so they're probably alright. Sadly the market is saturated with cloudsumer crap.
You wouldn't think finding a simple ip camera would be difficult, but it kind of is.
Because "Simple" is "Uses Cloud" these days.
And that is in theory good option, as it stores the video away from the robbers.
But for privacy reasons I don't want that, even when the video would only contain audio from the apartment most of the time.
Anyways, buying 256GB version of this, the largest write endurance SD card on the market, by large margin, AFAIK.
semiconductor.samsung.com/consumer-storage/memory-card/micro-sd-pro-endurance/
the very fashion that market as been so much pushed is hinting bad intents
140160 hours of 3.25MB/s recording as the spec.
That is my concern
"It's just an added feature, trust me, but now we will charge you double if you dont adopt it soon"

that's like 48 years of common recording throughput
(1MB/s)
26 Mbps for comparison to next one:
Kingston High Endurance: 26900 hours at 13 Mbps.
SanDisk High Endurance 256GB: 20000 hours at unspecified bitrate for fullHD (like the other two).
Not so little difference in the datasheet spec for endurance.
Any sd card will probably last long enough to not matter
No, most fail in less than year for dashcam or security camera like use.
Often in less than half a year of constant full-hd loop recording.
And the warranty terms in all but those three explicitly rule out use in video recorder type things from warranty coverage.
If you go read the small print in the SD card retail package.
My old SanDisk dashcam is still using the same 64gb card. It's been in there for about a year. It's nothing special and I'm rewriting the entire card every 16 hours I think
I will some day learn how pi are booting
smh my "some day" to-do list is insanely long
Yes, some old cards could take it.
Some couldn't.
But that is getting worse and worse with new models.
There were/are threads etc. in dashcam forums listing the suitable card SKUs.
I don't drive my car much anyway. My point is if you want to research the best sd card you can, but with 1 camera at 1080p any modern Samsung or SanDisk card will likely last years. Unless the write endurance of modern cards is lower
Yes?
And that 256GB Samsung Pro Endurance cost me 37e.
So there is price premium, but it isn't that big.
56e in my local store.
Cheapest 256GB card would be 20e for special deal on Kioxia Exceria Plus 256GB.
And 21e for cheapest otherwise.
So something like this won't last?
AFAIK not anymore, usually.
Haven't checked the specialist places in few years.
Absolutely no endurance spec in current Sandisk Extreme datasheet for example.
Ah, SanDisk also has "Max Endurance" line,
apparently reddit says to avoid them like the plague
120000 hours at full HD, 15 year warranty, again no bitrate spec.
Some of the old ones worked fine.
Today the non-specific ones are really bad idea mostly.
Samsung seems to be highly recommended.
Like I said, that Pro Endurance is "only" twice the price of the cheapest other options mostly, and has insane high endurance ratings in the official spec.
So why take a chance.
got the 512GB of the Kioxia one for 40€, great deal tbh
but honestly, I didn't bother checking anything in regard to speed or longevity
so it could just as well be the shittiest of microSD cards known to man
Mostly the normal cards are just fine.
But continous recording needs specific ones, and not all are ok.
And the pool of ok ones outside the specialist cards has been going down with the time.
And there can be changes in the FLASH used without warning, usually the exact SKU code changes, but the branding doesn't.
As long as it works. When it dies, replace it. With the pi you can image your initial install and settings so it'll be easy to swap the cards out after a year or two if you wanted
I think they use Broadcoms CFE
Problem is the time between the card starting to fail or fail, and when you notice.
If it has been corrupting the data for X time before you check because you need it?
Yes, they are specialist cards, that cost some extra.
But the extra is so small in percentage and absolute terms that didn't see need to skimp.
Today I learned that the 3rd video in youtube list is always a low viewed video.
The YT algorithm is written by kids.
just had every 2nd video in my shorts feed be a video I had already watched
The algorithm feels less and less natural. Or like what I am used of an algorithm. It's so scripted. Second video: a video arleady watched. 3rd video: a video with low views. For a few years I am dreaming for some competition against YT. In that sense I am happy we have TikTok. Yet I never used TikTok before 🤣
Hey, does anyone here know their way around Three js and SLGL? I have a uni assignment that's kicking my ass, and I am failing to find answers or apply any of the examples I find around...
Hmm...
Linpack Extreme 30GB stress test now makes the 7800X3D pull about 100W?
And the from wall power is now at 150W.
But the VCore current and the VCore voltage don't support that...
Some kind of reading or firmware error in the chain?
PPT percentage matches the wattage numbers...
Still only at 75C CPU hotspot
How do you measure wall?
Ok, it was real.
Because the normal full load from wall measurement was and still is 123W.
Linpack just causes some abnormal power usage in this case in the CPU.
Getting all eight cores to permanent almost 7W usage.
So 72W for cores, then 12W for SOC.
No idea where the rest is coming from, as the MISC etc. don't have specific readouts.
Even the one that is Core+Soc+Misc is only 89W.
With cheap inline meter.
This kind of thing, from Biltema in this case.
Oh, I have one that looks exactly the same
I trust them more than readouts from software tbh.
Yeah, something some Chinese company produces and sells for various brands.
Just don't measure the same thing.
mine is directly reported from the PSU 🤷♂️
I highly doubt it's the meter. Always possible you are unlucky and have a faulty one.
The only way to be sure is to have a second reading at the wall.
No, I used the external meter as sanity check.
It showed the same increase as the software.
You don't have a multi meter lying around by any chance?
That was the point.
That somehow the Linpack is using tens of watts more power than previous other tests.
In specific part of the test.
Inline with norms in others.
Just that the frequency, voltage and temperature aren't increased during that higher power usage part.
sorry, just wanted to share... I feel old
I had to use a 8bit ISA card for my serial mouse
Do you miss the screws? 😄
not really, but I always used them
There was a satisfying feeling to have screws. Don't know why. Maybe the fact you were sure the cable couldn't snatch out?
but then again, if someone is stepping over cables in their house and they trip, it could prevent the case of landing on the floor if the cable just snaps out.
Yeah, secured it. But not really needed
I think the main reason they don't do it is to reduce costs in production 😛
parallel printer cables were horrible stiff sometimes and really needed to be locked down
cost savings are always a concern
another thing I noted with those old cables is that they can get more dirty than modern ones. More hooks and crooks. The had a color. Were bigger etc.
you need a 🪥 to clean those 🤣
and the pin are now reversed female/male so it get less stuff inside cable connectors
i.e cable connectors are more protected than pc connectors now,
makes more sense that the old otherway around
one could arg pc connectors cost more to replace then, but its also cheaper to make bare connector bulkier in a pc case than on a cable
Wow, wire nuts for VGA signal? Never thought I would see them used for that 😁
I mean :
stole image from google, no clue what they are doing. I assume reanimating the cable or just messing around with it to see how it works.
and yes I hand mouse made that
still something satisfying about the chunkiness of old connectors
like driving an old car where it's mechanical connections everywhere
It looks like Alder Lake/Lovelace laptops might be going on clearance. Gigabyte G5 with a 12650H and 4060 spotted for $800. I have no idea whether that machine is any good, but I'm sure some decent machines will be available.
that is probably the worst way ive ever seen to splice a data cable
that's the max, but yeah that's not ideal
That's even slightly outside the range you want your Hotspot ag
At
You normally want your Hotspot to be likr delta 20c of the global temp
hmm its stock air cooler though
D:
i think i could push my 6600XT up to 108C on the hotspot
but that'd also freak me out a bit & i'd calm it down
The temps are fiiiiiine
86 global and 112 Hotspot is okay
It is the max and you're fully pushing it but it's fine
Linpack crashed the computer last night...
And then second time with CO offsets removed...
So seems that the RAM timings with IF OC isn't stable when RAM ECC is on...
Time for more testing. Set the RAM ECC to off and running the same linpack settings again.
And if that works for several hours, time to set IF to lower frequency and ECC on, and test again.
Rebooted without ECC, and that means it probably is IF uncorrectable ECC fault.
So lower the FCLK, enable ECC, enable CO, run again.
Yup, seems to be that.
So need to set the FCLK to "just" 2100MHz for now, and wait for AMD to update AGESA to support reporting the detected IF errors.
What case?
Extra fans or no?
3 intake 3 exhaust (2 top 1 rear)
All 120mm
Uhhh only gpu sounded noisy
Idk if intakes ramped
Get Fancontrol software and use the GPU temp for case fans
What fans are you using for intake and exhaust?
Uhh some Corsair 120mm
Needed a y-splitter to connect all to the mobo, only 3pin
Hmm I'll see if I can play around tmw with getting the fans controlled. I'm using gigabyte gaming x ax
So might need their stupid bloatware
Download Fancontrol
Side note why's there no open source non garbage rgb controller
OpenRGB works decently well, unless you have Corsair
Because Corsair are assholes
I'll check these out tmw. 🙏
Fancontrol is open source freeware
Lol
That is a big yikes.
no way.
https://videocardz.com/newz/asus-premium-700-rog-maximus-z790-hero-evangelion-motherboard-has-a-typo
ASUS ROG EVANGE-N-LION The board has a misprint that hasn’t been fixed for over a month. The ASUS ROG Maximus Z790 Hero motherboard, a flagship component within the EVA-02 series inspired by the ‘Neon Genesis Evangelion’ anime, may present a small unexpected feature. There is a misprint on the I/O cover, where “Evangenlion” is erroneously […]
Yes way 🤣 - Not fixed for over a month 👍 So happy I ditched Asus.
I am not a big fan of this case. It's an eye catcher for sure.
Yeah, I was kinda interested in this thing even if Intel only
If you are a fan I understand.
Evangelion was my entry drug to anime
I am open minded, but that doesn't seem like an anime I would like 😅
but might be totally wrong. Just seen some awfull animes with girls in spandexes 🤣
I think it is a great story, Spandex is not everything. yeah, some cringe moments about those coming of age things are there, but not many. and the animation is really great for my taste
This is 2-3K.
Guess whose employer was also attacked but had their IT systems up to date so it wasn't successful
Btw, the attack was on Halloween - a holiday in Northern Germany where the vast majority of our IT department sits (and the infosec responsible who was contacted by BSI about it).
Day later (November 1st), Northrhine-Westphalia was on holiday - including BSI, so nobody answered our infosec rep's calls.
i’ve seen several of these heavily-themed evangelion builds, and they all look so similar
Either you make an eva build or you don't. Mostly Asuka and/or Rei themed, but I've seen Shinji, Engel, Seele or other chars
my guess is that many have a crush on Asuka
Eva 01 was a thing too, just last generation
there they could write it correctly
Ubuntu didn't like that I had only 64MB RAM dedicated for the IGP...
the H6 looks nice at first glance. Hyte Y70 should have way better thermals than the Y60
for 110usd for the h6 flow vs 240 for the h9 flow, i’ll take the h6 flow any day
The GPU driver kept crashing or something when trying to load the graphical user interface.
Works fine when I upped that to what auto would have been.
Lmao
i added the h6 flow to my planned build, 130usd decrease in price for one less fan and a bit smaller of a case when i already wasn’t a huge fan on the size, i love it
Decent bit, yeah
Also IO on the bottom, ideal for use on the desk - shitty when on the floor tho.
yeah i have thermalright fans
it only comes with three fans anyways
wish they had an option to not get fans for a reduced price
Just sell them, some brandhead might even pay you more money for them than you spent on Thermalright
if it was even 5usd off, that’s 5 usd i could spend on something else
not worth the effort and shipping cost
one less fan in the h6 actually works better, now i dont have to get 3 3-packs, plus a random 1 pack, i can just get the 3 3-packs
Asus thinking about Evangelion...
Eh-fun-geh-lee-uhn
interesting
english pronunciation can be hard... especially as a non native I mix my teachers 'english', british, american and australian slang.
honestly there not bad i kept one for the fun of it and its the quietest and the smoothest overall i think they leveled up there fan game
They move air, but pretty much any other proper fan out there beats them
If they are quiet, they don't move air.
If they move air, they are loud.
Any other decent quality fan out there beats it in either of those metrics (or straight up noise-normalized performance)
So in conclusion, they are bad
You are better off paying $15 for a three pack of TL-C12C
thats probably why the AW fans sound like rocket ships
don't forget the scousers!
AW? Alienware? I think those might be even worse offenders - also, Alienware prebuilds tend to have practically no air access meaning higher rpm required for similar performance
well yea you have a point there but there not bad also the rads are quite good for there size
checked myn and its like 3.5 inches thick
they have air access but no fans forcing air in... only out
In comparison to literally any reputable third party fan they are bad
And Alienware PCs are a meme
Just check GN dismantling one of their Aurora PCs - both physically and figuratively
That wasn't nearly as hard as steve roasted 14th gen intel
I haven't laughed that hard at a review ever
☕
at best really its 13.5 Gen
Yeah, all just marketing
not seen Steves review, but I can imagine, lol
It really should've been the 13x50 series
And then there was the "Waste of sand" 11th gen.
isn't that odd? you can't get a machine to boot without graphics.
if you can, let me know how
Just couldn't handle running on just 64MB of VRAM, and didn't know how to use the dynamic part.
Windows can.
the bios won't let it
So the driver worked, just ran out of VRAM when the GUI booted and tried to run.
And the GUI crashed and burned.
I've got an am3 board that I was going to use for experimenting. no igp, won't post
I think you can setup windows server without a gui.. not sure tbh
The IGP worked fine.
Just that the Xorg etc. couldn't run with just 64MB of VRAM.
And the driver/Xorg didn't know how to use RAM as VRAM extension.
So text mode linux worked fine, and even safe-mode GUI worked fine.
Windows worked fine.
IGP:s have variable allocation size for "Dedicated" RAM as VRAM.
I had set that to minimum setting of 64MB, to be able to test as much of the RAM with Memtest etc., which cannot test that VRAM allocation.
And then the IGP in Windows can use up to half of the RAM as Dynamic RAM as VRAM.
But the Linux driver or Xorg or something else whichever should do the same in Linux couldn't do it.
And the 64MB wasn't enough for whatever layers over the Xorg current Ubuntu uses.
apparently people are indeed killing sdcards with the pi lol
Yes, by default Linux often does excessive logging for SDcard endurance.
Or even SSD endurance.
I was reading stuff about truenas and whatnot.. learned that ZFS's pools without ECC memory is a horrible idea
Not really.
Not more than any other system without ECC.
Just thing that runs around in that specific community.
You sure?
For "never get any data corruption ever, in any case", you need the ECC RAM too.
But that is pretty specific niche, just that those people are very active in various datahoarder places.
This was someone who seemed knowledgeable. and apparently if there's a bit flip for whatever reason and it's written to the pool it's very bad.
Not any more bad than with any other file system.
So ECC is good, but ZFS doesn't care any more than any other file system.
ECC should be everywhere, but the specific thing in the ZFS/Datahoarder/TrueNAS community is way overblown.
this is related to zfs pools, so many drives in raid. it was enough to steer me away from it.
If you get RAM bit flip in just right place, you can write the corrupted checksum to the drive.
And you then lose whatever that specific checksum was checksumming.
Same as with any other filesystem where data or metadata gets bit error and is written to the disk.
The chance to get that with ZFS in just the right place is just so miniscule...
Less than with non-checksumming file systems.
It is just that the specific community is extremely paranoid about absolutely any amount of chance for data corruption.
And of course you cannot get as close to zero without ECC RAM, as you can get with ECC RAM.
But if the RAM isn't actually faulty and it is only about random bitflips from cosmic rays etc, then the chance is miniscule, and should only affect individual files even if it happens.
And the same chance for random data corruption is there on any file system.
I can understand the ECC enthusiastic people in the zfs community. Why go to the lengths of using zfs and then not go all the way.
can stay at ext4 or btrfs if 💯 % data integrity is not one of your major concerns
is it needed for personal use? usually not
It is good idea, but personal system can do without.
And in large scale storage deployment, ECC is pretty must no matter what file system you use.
Damn you Intel and your segmentation!
what segmentation?
As in Intel disables the ECC support in their consumer CPUs based on chipset.
To not have those that absolutely need the ECC buying the cheap systems, but need to get the more expensive chipsets.
All Intel CPUs support ECC RAM.
They just don't allow it to work in Firmware if you don't have "Xeon" chipset.
Have for tens of years.
yeah, old problem. I remember that some i3 processors had ECC enabled
Just segmentation, so that consumer platforms don't steal sales from the higher tiers.
And then have FUDed it to "Consumers don't need ECC".
In marketing.
And then you have chicken-egg problem in ECC RAM.
High prices because of low sales.
And low sales because of the high prices.
Combined with that "You need much more expensive MB to support ECC, which won't have consumer focused features".
So the Intel CPUs support ECC RAM, MBs could just add the 8 extra traces.
But as the firmware doesn't allow that to work when the MB has one of the "Consumer" chipsets, what else is that but pure segmentation?
When the chipset has nothing to do with the RAM?
i3-12100E CPU supports ECC, but I have no idea if the board has to support it too
Has to.
And that is only in the C-series "Xeon" chipsets.
AMD does the same with their desktop APUs.
The consumer APUs have the ECC support disabled in the CPU, even when the IMC could support it.
And only the Pro APUs have it still enabled.
But no chipset difference.
Desktop non-APU, non-Pro CPUs have it enabled, but not officially supported.
Both APU and non-APU Pro:s have it officially supported.
And the MB has to have those extra traces.
Asus and ASRock have them.
Gigabyte and MSI don't.
For AM4 and AM5.
kinda sad that since DDR 5 ECC is not real ECC anymore but many use it for marketing
The mandatory On-Chip ECC isn't real ECC.
You can still have real ECC too.
Just that often the bad marketing drops that "On-Chip" part.
Real End-To-End ECC I mean.
The On-Chip ECC does fix bit errors during storage.
But it isn't meant to be there for the user, but for the RAM manufacturer to be able to use worse chips.
yield increasing ECC
Yup.
You can sell chips that have RAM cells that are stuck to either position or are flaky.
If they are split so that no single row has too many of them to be correctable.
And the On-Chip ECC can correct multi-bit errors if they are spread out suitably.
Micron implementation splits the data into four areas that are separately corrected at least, and can do 1+ bit correction on each of those four, independent of each other.
IIRC from the datasheet.
When trying to find out if the On-Chip ECC can notify the IMC about errors or not.
It cannot easily.
There was some kind of error counter that the IMC can read.
error correction reporting? what's that?
they have to leave some room for improvement for DDR 6
Well, if the user got reports from the RAM that contained knowingly flaky cells from the start?
Oh, they would complain
define X corrected ECC errors per Y timeframe as good enough?
Not really workable when you can then just do reads to the specific row you found out to contain failty cell?
mark cell x as bad and never touch it again? best done at the check right after assembly
a bit like wear leveling on a SSD
Got the pcie to m.2 adapter in as well as its power. Went fairly well.
You can see it coming in between the pcie x16 riser
and card works ?
This isnt for the graphics card.
I know I'm talking about the htc connector card
Oh. Dont have it yet. I got the adapter only as i wasnt sure if it would fit. It might fit but would the card work normally? Thats my only question now and i dont want to spend $350 on something that doesn’t work
what adapter did you get then ?
also you never answered what was the card PCIe version rating
I dont know what the rating was of the htc card but this is a pcie 3.0 x4 adapter so it should be fine. I got simple m.2 to pcie adapter
the adapter is rated pcie 3x4 that's what you said ?
now you must look at your motherboard documentation to know if you have a pcie 3.0 x4 M.2 here
and htc card what version it uses
I can't know in place of your own eyes, I can't know what card you want to buy exactly
Telco nightmare
if its green theres no problem 
Someone is going to spend a lot of time cleaning those connectors 
Hypothetically
If I remove the compressed attribute from the many windows system folders that have it, will the pc boot/run ant faster?
I assume the answer is that it would be a negligible impact
highly depends on your disk speed and processing power
are my sensors fault
either that or it's getting ready for ignition
Known as Post Package Repair, Hard version causes that to be permanent, Soft version just until next boot.
Feature of ECC RAM sticks.
IMC can order the chips to do it if it wants.
Just limitations amount of those that can be done.
Damn it's easy to be sick now in germany. Usually I had to go to my doctor for the little "I'm sick" paper and send that paper to my HR as proof of my debilitating illness so I can get paid sick days. Now all that shit goes through the insurance provider electronically so I just have to call the Doc and just be like "uhouhouho I'm sick" and they just do their do and I don't have to do shit but to notify work
Okay does anyone here have experience with bypassing SSL in Java
Dev server I have to use has broken Certificate, and http requests redirect to https lol. Asked few people already and none can figure out how to force Java to not even try parsing SSL Certificate 
You can't?
Because that would be bad...
You define at startup what certs it will accept. If the cert is broken good luck, but you need to re deploy the service with a new cert or ask whomever is responsible to update the certificates in that case.
Or deploy a new one with certs disabled.. not recommended unless it's on a local network, and ofc only temporary.
It's local dev server, so that's why I want to bypass SSL instead of fixing the issue... and the fact I don't have access to the server configuration to fix the certificate myself :P
Idk then.
Would need to override the functions doing the checking in the libraries.
There are some guides I could find for that.
already did that, as this is known workaround. The issue is that Java libraries are responsible for parsing, and that's where the issue is.
And overriding java libraries isn't fun 
Java isnt fun 
true.
Well.. gonna talk to few people that are responsible for the server to fix the certificate, and I guess I have a day off as I can't do anything without the acess to the server and the API
Is it broken, or just self-signed?
When in the second case you just need to add the self-signing authoritys public key to the computer.
Second one is pretty common for internal systems.
It's.. both.
It's self-signed with some information missing completely, causing Java to fail parsing of the certificate.
some poor guy gonna go to to the server then hihi
Still around AFAIK?
It is just the number key type that is broken.
The button press works fine AFAIK?
Way to do WiFi connections without the full password entry manually.
One is 8 number code, that is completely broken, because you can try it as two separate 4 number codes.
And there isn't limit how fast you can try.
Other is to press a button on the router, and next connection attempt is accepted automatically if the other side supports it.
why are printers such a pain to have and use
Entering the WiFi password on the buttons of the printer.
ink is expensive, they like to kill themselves periodically, horrible connection
the upcharge on ink should absolutely be illegal
they already made it so you can’t use third party ink
time to go back to the printing press
Laser printers \o/
it’s about money too
costs them literal cents to produce an ink cartridge, just to sell it for an absurd amount
my printer is balanced precariously between two desks, with a cabinet underneath it that sits about a half an inch below the printer
it also weighs as much as 4 toddlers
good thing i never print anything
Laser printers are by far superior to any other tech unless you absolutely need color
i just have a usb connection between the printer and the family pc
Unless you need photo quality color.
it refuses to print without a wired connection
Color lasers are fine for almost anything else.
That's true
i don’t like that emoji
why that emoji
you’re making me nervous maxeek
the family computer is a poor thing with most of its io and motherboard connections filled
was lucky enough to get ethernet tho
said computer, and printer
there are good color laser too
but they are very very expensive for entreprise stuff
and it sucks lot of power too
i did dust it out, but it was struggling before hand
i did take it outside to dust
Got the DNI version of that thing as personal printer.
otherwise i would’ve gotten popcorn lung from the dust alone
Yes, it is big.
But printing is pretty cheap, even including the image drum replacement costs etc. into the per page costs.
Something like 3.7 cents per black standard coverage page, including the paper IIRC?
At retail paper prices in small quantities.
dust annoys me to see, so my main computer gets dusted weekly
And the high power usage is only during the printing, as the heating element pulls power.
usually whenever i can visibly see dust in the dust filter
isn’t monopoly a critique of capitalism?
350W during print, 51W while keeping the heating element near-hot (standby), 1.1W while not doing that based on specs?
the one I was talking about get up about 750W when printing 150W when waiting and 1.5W when standby iirc
Yes, the old ones were much more power hungry, and same for the BIG machines probably.
it's very stable printing though
that's what I like with laser
it needs hundreds of W to print
environment does not have hundreds of W to shit on the paper printing
so it naturally is very stable
you don't get crapped paper when very cold or very hot environment
That slot is wired up for up to pcie 4.0 x4
ok nice, remains the htc card pcie
did you double check the different wiring depending on used cpus ?
Yes.
So what MB? What CPU? What slot?
24 usable lanes, 16 for gpu, 4 for os drive and 4 left over
Meant the Callies case.
Callie: So can you answer these three?
without restarting ? you mean if you restart it detects if you dont it does not detect it ?
seems a OS issue
Nzxt h1, asus rog strix b650e-i gaming wifi, ryzen 5 7600x
Using m.2 slot on back of mb
And what SSD?
no ssd
For the os? That is on the front. Its a 870 evo i think
2 m.2 slots. One on front, one on back
Back one has the adapter
Brb. Gotta go do stuff
I think it's baldur is not aware of the target setup
Ah, what is going on then?
Misunderstood?
thank you to finally provide the info we can help with, especially motherboard wiring
may be power problem, is the icy box powering up fine or does it is just waiting when not restarting ?
but yea if its only for backups then restarting is fine I guess
But what is Callie trying to connect to the other M.2 slot then?
Convert it for something else that uses PCIe?
an htc vive wiring card, but they never shared what card exactly
its not wireless M.2 card
Ah, again my mistake.
Probably trying to plug the WiGig wireless adapter card?
Because the options seem to be normally to use wired connection.
And then there is upsell to that PCIe wireless card for 60GHz wireless connection.
So would need any m.2 to PCIe conversion cable, that in theory do exist, but when you try to find one you just get returns for the PCIe to M.2 cards.
But would probably need active conversion cable in this case, because of the lenght of the run.
For 4.0, few centimeters can be too much.
And I expected it needed to be pulled to below the Mini-ITX MB.
I dont think it's pcie 4 you checked it ?
For the card, probably not, didn't think again.
Brain not working today.
Ok, at least older versions were just 2.x
HTC doesn't seem to publish the spec.
thats nasty then "hey, we designed a card you can use .. But we will not tell you how to use it Kappa"
And just 1x based on pics.
Yeah, for FAQ entry about what length PCIe slot you need they tell the card dimensions instead...
that's why version matters, because it can't be too slow
Sabrent finally got back with me... after speaking with their tech lab, they are going to send out a replacement ...
I guess we'll see what I get from them directly ...
And reports that it is only linking as that 2.x, even on 3.0 slots etc.
WiGig is in theory up to 7Gbit/s (less in reality), PCIe 2 x1 is about 5Gbit/s.
I am back
Htc vive wireless adapter
There is only one version of the card.
Yeah, so M.2 to PCIe slot converter.
And you might need to put it on the front M.2 and then use a PCIe riser.
If you cannot find one that is combined one.
PCIe 2.0 or 3.0 with 1x slot is enough based what I found.
So something like this:
www.delock.com/produkt/64131/merkmale.html
I can’t imagine it will fit on the front m.2 slot due to where its positioned
So one of the M-key versions, or similar from other manufacturers.
This is the one i got
So you got the parts, but it doesn't work?
I got the parts and it fits in the rear slot, but i have nothing on me atm to determine if it works as i would like it to.
I dont have the wireless adapter yet
Yes, should work.
And cannot you just test with any PCIe card?
Including GPUs that don't pull too much power from the slot.
Dont have any regular pcie x4 or x1 card and my gpu wont fit cuz its not an open back connector on the adapter
Ok, looked in the pics that it had open back.
Only question is the signal integrity.
But for 2.0 it very probably works, when the slot is 4.0 and the adapter is 3.0.
But plugging in 4.0 device will not work, unless you limit in the BIOS first.
Because the end points see both are 4.0 or better, but don't know about the 3.0 part in between.
You could go to the BIOS, limit the M.2 slot to PCIe 2.0, and try?
Just if you get picture and get to BIOS with the GPU in that.
Will try it later
lol
Cortana can be UNINSTALLED ?!??!?!? wow
now only if apple will follow suit and let us get rid of Siri
constant prompts "Finish setting up your mac! Turn on Siri." .... and im like ... no
the negotiation is normally testing the line and fallingback in the specs
No!
though I agree it can still fails
Doesn't always work right.
Sometimes seems to just try to go with what the end points support.
"A.I. Are People Too" 🇦🇮
hmm... that flag choice is wrong ... should be a skynet logo
ive still not run the windows update .... kinda hesitant
always something changes for the worse
They already could be uninstalled
You had to use the deployment command
So I guess they're just making it easier to do now?
deployment command didn't do it clean. Left over garbage sometimes and references etc.
makes sense to me... seems a logical solution
@twin dew #satisfactory message
...

Good thing throttling works to well today, otherwise he would probably have dead CPU.
Now a days, its hard to kill a cpu
new factory dedi server uses so much less memory. 8-10GB to like 2-4GB wow
i have achieved eternal battery life my battery wear level went from 38% to 37% 
i can plug a usb-c to usb-c cable into my mac and it begins to charge itself. I have unlocked perpetual energy!
*corporation wants to know your location
and then you simply have to accept defeat....
pretty sure it's not supposed to do that
No specific check to see if both are connected to itself.
"Do you wish to turn on your wi-fi?" in-app purchase
"Incoming bluetooth shared file, pay .49 cents please to accept"
"File Sharing Tax"
corpo want a piece of every bit of bandwidth
"Set up a home network! Only $499.99"
"Unlimited local bandwidth (but you will still get ads)"
its like the days of PAYING for ads delivered to your FAX machine because they were eating up your paper and ink
heh
my ISP applies a "Call before you dig recovery tax" ... which is a service provided by the city where you call before you dig and they come out and mark the land where all the wires/pipes are.... thing is... the city already pays for that (from our generic property taxes).... so for my ISP to 'charge a recovery fee' for that is actually false and robbery
its also like since the ISP charges a seperate "equipment rental fee" ... they can also charge "sales tax" on that equipment fee because its not the digital service provided, its a physical good ...
waaay back 12 years ago, the service fee included all equipment, so there was no sales tax on your service bill
not sure who to blame for that change ... cause 'sales tax' is supposed to be reported and turned into the gov ...
all the sales tax WE collect from texas residents, we have to send in to the texas gov each year
which is strange, given we dont sell physical goods ... yet we are required to charge sales tax
haha ... corpo no like small biz
yes
they dont call it 'bribery' ...
but is the same
RT looks amazing in satis
like... they dont want to call it "con job" cause its degrading, they would rather be tried on "anti-trust" as it sounds more civil
Now i see why people obsess over cable management so much
Lumen Theres a few tweaks needed to denoise some of the lighting
But my gpu can mange the game at 1080p MAXIMUM settings at 60 fps
uncapped framerate is 67fps
would there be any point in investing in going from a 5600X to a 5800X3D for gaming stuff now that AM4 has been officially dead for a while
just spent 20 minutes cable managing stuff at my desk with some cable clips i ordered and holy it's so much nicer
no more dropping cables into the cable abyss behind my desk
Budget upgrade Ig
mmm cable abyss
doesn't feel like enough of an uplift to justify if ryzen 8000 is out in less than a year along with more stable AM5 mobos
"unplug charging cable from phone... cable slides off back of desk... DAMNIT"
if you saw behind my desk you'd know why i call it a cable abyss
if a cable falls back there you bet i'll be looking for it for the next 15 minutes
what's even better is when it's the entire damn usb hub
i need motherboard makers to add more USB slots
satisfactory lags ALOT wile leading new terrain sometimes
hehe... i have my desk pulled 2 feet from all walls (it sits near center of room) ... got so sick of trying to reach back and fish cables
.8174782423 meters ?
i play vr in my office, that isn't an option for me
holy does it get toasty in the summer playing vr games in the same office as two gaming pcs
oh, i guess its actually .6096 meters
or... .69 meters to give it an extra bit of 'length' bowchickabowwow
though one of them sits idle a vast majority of the time
poor alienware pc with a dying psu or gpu
i have a box on the back of my desk with a 6 inch duct, which runs through to the closet into the other room where it has a inline duct fan sucking all the hot air out 😄
I am so jealous
i would kill to have a duct that takes all the hot air from the pc out of my room
yeah ..... i did it this year cause the new pc generates a LOT of heat
its fine in winter 😉 just not summer... spring.... fall ......
when i upgrade to a 7900xt and 7800x3d i'm gonna be sweating
since the A/C controller is out in the hall, it would never really read the heat coming out into the office ... so trying to turn the a/c down cooler didnt work right
i recently realized pcpartpicker has a dark mode, this is so much better
ha
oh yea my pc is right by the air return, so that definitely saves me from cooking
my office is tangential from the rest of the house, and has a big ole window
in the summer i’m baking
my google chrome is recommending me the U8 EA release video, but i never searched anything related to it or consumed an U8 content online other than discord, creepy
Its because you play satisfactory, are in a satisfactory discord, and U8 was released
Discord says in their TOS your messages are used for advertising
Discord = DataMiningChatSystem 😄
That's pretty much the go to
I got a Poco F5 recently, also Redmi Note series is great for value
so i got the U8, booted it saw the driver update window, updated the driver, restart the machine, open the game and still says need update? what do?
did that, copied the link, (reset nvidia password lol) got that all happy as far as i can tell, it still gives prompt on game start?
weird
well, it hasnt crashed after a few minutes...maybe its fine? just doesnt 'know' it yet? give it a day?
do any of u know how to blacklist websites on my search engine?
want to exclude fandom.com
both terraria and satisfactory have a wiki hosted there and for both games there is another superior version
the magic of induction.
nvm found something, ublacklist
There's also an extension that will automatically redirect all fandom links to wiki.gg
oh wow, whats it called?
a few months ago my family was talking about doing a LotR marathon this xmas for the 20th anniversary, & we talked about making a xmas tree look like mount doom. our phones were clearly listening to us because we all started getting Mountain Dew ads
@jagged snow with U8 maxing out settings with TSR at native. Makes it really hard to run at any framerate above 60. Aka uncapping I get 67 fps. And I consume basically 230w Vs consuming 150w and getting 60fps
That and uncapped temps run away again
I need to get my GPU professionally serviced. As it used to be able to handle 230w loads fine
Now I can barely pull 180w
Without going into the spicy zone
My F5 has it. Works not quite as well as on my Note 9, but once I'm used to it, nothing that's hindering me.
All other Xiaomi phones that I've seen fail were damaged by user error
Try DDU
Basically full driver reinstall
So you repasted? With what results?
There's not much a PC shop can do extra other than what you can do yourself.
Pretty sure there's more phones with Android One
Wikipedia says Android One is still being supported
There are phones with stock Android, but the Android One branding has been killed.
So finding them is much harder.
In HMD Global (Nokia branding) we trust?
It is the Nokia mobile business in everything but name.
Bough back from MS by Nokia as separate business.
I just got this and im scared to do anything. What should i do?
Update the GPU driver?
So just click yes?
do you have nvidia geforce experience installed?
No, that just continues to the game.
No?
Visit the url listed and download latest drivers.
Then do what Baldur said.
Ok.
If you have time, look into experience. Can keep your drivers up to date.
The version you have is now almost exactly one year old.
Nah, that just prompts you to select a driver.
It is not automatic update.
That gets the latest installed on your system, or in Windows Update catalog in theory.
Not the latest.
Do i just click "download now?
Yes, and then you run and install it.
Huh, funny. Where is windows update now with their forced drivers 🤣
I have to jump through hoops to get it disabled. And people who need it, mcmiko05, don't get it.
Honestly mcmiko05, when you have time. Look into geforce experience. It will notify you of driver updates. And download and install them for you.
It also has some other features you might like or not.
The login requirement is just asine.
Yes?
Not the best, but good enough.
And best of the free ones.
The Yes? was because the answer was so obvious in my head.
probably always will be
Not limited to old people.
Pretty minor part of corporate hackings are actual remote hacks.
Most start with some employee doing something stupid with phishing or like.
And still even cases of plugging USB "drives" or like they found outside the work into their work computers happen.
Always funny when you forget a usb stick in the kitchen and an hour later the @all goes out in teams
3 photos with the worst lighting showcasing your dirty and broken usb stick attached
i mean it should do unless your like really really trying to get a bunch of viruses
Bad infosec training is the most common vulnerability
https://youtube.com/shorts/VAWwtjtRM98
Watch the stream here:
https://piratesoftware.live
#Shorts #Blizzard #Hacking
im going to assume this is not a good firestrike score
i would say 'some basic restraint' protects you from getting a virus... but thats about as useless as saying 'use common sense'
common sense is incredibly rare
No idea, how does it compare to PCs with similar spec?
dont really run firestrike cuz im ussually not working whit DX11 only
no comparison info so idk would have to search deeper and tbh dont have time rn
Thing is with CS: would you rather risk helping a hacker or not helping a customer?
helping a hacker could prevent you from helping any customers, but it’s more likely you’d not help a customer than you would encounter a hacker
but common sense greatly depends on your location and community you live in... what is 'common' sense is not equal across the globe
so it always has bugged me when people just say "use common sense"
train your employees properly and hope for the best?
our customer service team gets frauders who write in trying to game the system ... and also phish attempt tickets and the like. they gotta be on their toes
Also, common sense sometimes goes against training and infosec "oh, he seemed nice and I didn't reveal any meaningful info, just the name of one of our department leaders, you could probably figure that out yourself".
Next thing, CEO fraud
Pretty much, yes
Also: it's best if the one hacker they do fall for is you
So try pentesting your employees
cause CS is generally a good vector to breaching into a company
hehe, i did that recently (unintentionally), i was testing the system on spotting phishing by sending some stuff through an odd entry point, and the team was like 'wtf is this crap?'
heh
im paranoid at work
white hat hacker type thing?
even if they get through, helps you determine what and where to improve stuff
true that... some holes can just be plugged before they go anywhere
like, for the loving pete.... any system that just forwards on html in a message should be shot dead
please don't pentest your employees
why not? HR can go screwtest themselves 😉
HR is too busy hiring actual monkeys
haha, enough of them and things actually get done, right ?
right….?
Yup.
I recommend this talk:
https://youtu.be/NNqBwb0DC_Q
https://media.ccc.de/v/36c3-11175-hirne_hacken
Menschliche Faktoren der IT-Sicherheit
Die überwältigende Mehrheit der erfolgreichen Hacks in freier Wildbahn setzen auf menschliche Faktoren. Wie können wir Systeme und Interfaces gestalten, um diese Schwachstellen zu mindern?
Ob Ransomware oder Phishing, APT-Angriffe oder Stalking: Die am häufi...
Better than not doing it
"we never had any reports of security incidents before we hired an infosec responsible"
That doesn't sound normal, I play at 1440 high/ultra with lumen on and get about 90-100 fps on a new save
That's all I pull on a regular basis anyways
I'll pull above 185 in some games
Love these comics btw
gotta love bot farms ... Steam stats change over last month LOL
Temps are stable up until 75c GPU temp. Then it runs away. Delta more then 25c
Used to be able to handle much higher wattages up to 230w
Did you repaste?
Yes using MX4
I may just get it professionally serviced.
Maybe something doesn't have proper thermal contact after reassembly?
Or did the issues arise before switching?
And MX4 is AFAIK very bad on GPUs and fails pretty quickly?
But might be remembering wrong too?
That it worked on previous GPU doesn't mean it works on current.
Tho people have used mx4 on a 3090 iirc
Should I use what I have for the hydronaut on the GPU die
I may need to buy some more thermal paste
Depends much if the GPU die and the cooler are straight or not, and how tight the mounting is.
Many pastes fail if the mounting when tightened is too loose, like often on GPUs and laptops vs. CPUs.
Yeah, MX-4 is 1-3 months on GPUs.
It's been about half a month on the gpu
Just get Cryopad?
MX-4 is working on being very very thin and staying very thin with no air
so if it's not tight mounting it's probably shit past
GPU mountings are usually pretty loose, and lot of thermal changes for lot of pump out.
And the straightness is usually pretty hit and miss too, causing even more pump out.
So most CPU thermal pastes just don't last in GPUs or laptops.
What size should I use
What was the GPU?
6700XT
Basically you want die sized.
why is it that way though, wouldnt gpu also benefit from tight mounting ?
I have my GPU tight. Without stripping screws level tight
Danger of cracking the die.
It's easier to not go tight when you're cooling direct die
Don't do that... tightening the screws doesn't make the cooler any tighter 
Yeah, it is determined by the spring, not the screws.
Damn
So I'm kinda screwed aren't i
No pun intended
And prolly can't rma the GPU at this point
Plus I don't have a spare one
Ehh no complaints to amd this is on me
you probably need to carefully clean it
and put appropriate pads in there
I'm using stock thermal pads on all parts
aren't kryosheet too thin for that ?
they are designed for cpu tightening too
what
Like a cleaning brush eseentally
why brushing it
@sharp oasis it makes me wonder if you have a thermal pad bunched up that's lifting the coldplate
0.2mm thick before being compressed, so completely fine.
I have make sure to lay them flat
If you're using isopropyl it's not typically a concern
The plastic brushes without carbon added are huge ESD generators.
25x25mm for RX6700XT:
www.thermal-grizzly.com/images/downloads/SizeGuide_Kryosheet_TGU20231005.pdf
90% iso alc specifically for cleaning electronics
It's quite litterally simply alc and purified water
No added skin care stuff
It's pretty unlikely that you killed the card, the small water content of the isopropyl aids a lot in reducing esd danger
Yeah, just bad idea.
That was the point.
It is done, hopefully didn't cause anything, but bad idea in future.
So I can use what to clean my gpu
When semi-conductive brushes don't cost a lot.
I normally use like a cotton pad and the brush
Or just use toilet or kitchen paper?
Microfiber paper towel with isopropyl for the die, conductive brush with isopropyl for the surroundings
Kk
Alight sorry bout that
tissus are a lot better at catching and retaining small paste parts
brush will merely spread it
I really wouldn't worry about it, chances are high that the card is fine
My GPU isint dead as it's still working in my system. But I'll take this as advise for later if I do repaste / repad my gpu
Yeah, was just for future.
I may get some new therm pads for the vram / chokes
Reapply all thermal pads and then do thermal paste again
the only bad thing about tissus is left over fibres on the pcb metal parts, which you should check for, but it usually go out with blowin air on it if it's clean
Hence why I use the cotton pad
It doesn't fall apart as easily
👍
So probably 18.31mm x 18.27mm die for Navi 22?
I know it's 335 mm2
And then that 25x25mm is the smallest that will cover it.
24x12 will not, and would need to cut it down anyways.
Or protect the small components.
And I'm pretty sure it's a square die from when I repasted my card
Yeah, that is the area, but most places don't list more than that.
The source for the dimensions were leaks from before the actual launch, so that is why cannot be completely sure.
And yes, the die is square, so that would match.
But need to do some very careful snipping to get the oversized pad down in size.
(Putting it in freezer will help AFAIK?)
And then use tiny dab of paste to get it to stick in place when installing the cooler.
But the freezer effect in this case will be minor, and the pads are very easy to tear
KryoSheets I mean.
As most of the carbon fibers are in height direction, for better heat transfer.
ohh, tomorrow is my steam accounts birthday, big one next year
older than a lot of the people I game with.. lol
I joined somewhat late, in summer of 2004.
Just checked mine:
mine was Half Life 2
I joined with the Half-Life Generation set
damn, you had to be in the very early days
I remember the days of manually patching Half-Life and Team Fortress Classic which was before Steam
Nah, that is the 2004 timeframe, or very late 2003.
Ah, sorry, Steam was only launched at fall of 2003, another part seemed to say it was in 2002.
So my registering was probably to register my Half-Life 1 key, before keygen people did.
lol
There were major problems with people running keygens for Valve games back then and trying to register the keys in Steam.
Until they found one that worked.
But then the legit owner couldn't register it anymore.
I had a few friends run into those problems.... but Valve were quite good at sorting it out quickly
@twin dew if I am to use a thermal paste again what do you reccomend for my gpu
I have hydronaut on my card at the moment
No idea really.
Haven't looked closely enough.
The paste is not to blame for your temps
It can easily be, if the temps are first fine after repaste and then start to climb again.
Pulling 200w for 15m at a room temp of 18 I reach a maximum edge temp of 68 and a maximum hotspot of 84
I meant the hydronaut that he had on his card before
Ok, that.
Because previously he tightened completely wrong.
And MX-4 in turn just doesn't last in GPU.
MX-4 is also a thicker paste if I remember correctly, so it's very possible that the mounting pressure is not sufficient to properly distribute the paste and it is keeping the coldplate off the die
You don't need that much paste, in the previous pics you had over 10 times too much.
It was nice remembering the good ol' days of half life 2
nice
im only at 14 years .... i joined it after they kinda turned into a store for other games 😄
Mine is "longer" 😅😅😅😅
im only clossing in on 3 bruh
@jagged snow i unironically reccomneded a friend to get a threadripper.
For context. He loves to do renders in his spare time
He already has a 7900X and that's not enough
For cpu. I said to him. Either go Intel or go threadripper at this point
And Boi just get yourself like a massive ass GPU or so lmfao
Yeah, they just released a new generation
Even the 5000 Pro:s were only 1 year Lenovo exclusive.
And the new ones didn't get any exclusivity.
But you cannot buy one yet, just announced, not released.
Ah, I had mixed that up
I know, its already thursday.
Genuinly i said Whataver fits his budget
Preferbly new threadripper
Oh fr
wait
Not for This lad
Bro actually needs the threadripper
Hes addicted to blender and Runs like Multiple renders on his machine at a time
He has a 7900X and even that aint enough