Nexxen wrote on 2025-07-29, 22:32:

Yep, the PCIe lanes are the ones of the slots connected to the cpu. I'm not concerned, it isn't affecting the results one bit. […]
Show full quote

chrismeyer6 wrote on 2025-07-29, 21:43:

Nexxen wrote on 2025-07-29, 21:23:

I bought a 3rd graphics card from Sapphire, RX 5600XT Pulse. I alredy have 2 same model. This one was cheap and I made a windows […]
Show full quote

I bought a 3rd graphics card from Sapphire, RX 5600XT Pulse. I alredy have 2 same model. This one was cheap and I made a windows only gaming rig.
Even this 3rd card instead of PCIe 3x16, it show 8x. Not a problem, they output the same results with usual statistical variations (all in line with general results).

I'm starting to think that something is wrong with the 2 mobos I tried them on, X370 + 3600X and B450 + 2600.
Both cpus are capable of PCIe 3 16x.

I tried multiple OSs and programs, all showing the same 8x.

Other cards show PCIe 3 16x. I am truly puzzled. Makes no difference, but on an old i5-6xxx they are reported 16x.
Old i7-3820, 16x.

Is intel beating Amd on its turf? 🤣

Is the slot on your motherboard actually an x16 slot? A lot of times you can have a full length slot and it only have pins for 8x, 4x , or even 1x. Check the slot and see how many pins are in it.

Yep, the PCIe lanes are the ones of the slots connected to the cpu.
I'm not concerned, it isn't affecting the results one bit.
IIRC a RTX 2080 isn't maxing out the bandwidth of a PCIe 2.0 16x. Thus 3.0 8x is totally fine for this card (like a 2060 performance). It's 6GB GDDR6 192 bit.

I understand that it's not a performance issue I'm just more curious if that 16x slot is actually wired for the full 16 lanes. More just morbid curiosity

chrismeyer6 wrote on 2025-07-29, 22:35:

Nexxen wrote on 2025-07-29, 22:32:

Yep, the PCIe lanes are the ones of the slots connected to the cpu. I'm not concerned, it isn't affecting the results one bit. […]
Show full quote

chrismeyer6 wrote on 2025-07-29, 21:43:

Is the slot on your motherboard actually an x16 slot? A lot of times you can have a full length slot and it only have pins for 8x, 4x , or even 1x. Check the slot and see how many pins are in it.

Yep, the PCIe lanes are the ones of the slots connected to the cpu.
I'm not concerned, it isn't affecting the results one bit.
IIRC a RTX 2080 isn't maxing out the bandwidth of a PCIe 2.0 16x. Thus 3.0 8x is totally fine for this card (like a 2060 performance). It's 6GB GDDR6 192 bit.

I understand that it's not a performance issue I'm just more curious if that 16x slot is actually wired for the full 16 lanes. More just morbid curiosity

I'll have to check the cpu socket.
If it's not an error of some sort on the BIOS-Cpu-program, probably a broken lane. And that can be, absolutely.
Tomorrow I'll pester my dad to test it 😀 What are dads for??? 🤣

I'll also check the second motherboard.
It makes non sense. Also it could be a dead demux.

Not the CPU socket you need to check the pci Express slot itself. Not all full size slots have all the pins populated. Asus has some mid range Intel b850 boards with full size 16x slots but only wired for 1x and the rest of the slot has no pins.
Would you mind posting a picture on the 16x slot?

Ordered a pack of 600 colored jumpers ... set for life! 🤣

Sorted some of my parts shelves. Much better than before.

Ordered a 16GB eMMC module for my Le Potato. Not 100% sure what I'm going to use it for now.... Portable SDR station? Dedicated DOSBox machine? OctoPie server for one of my printers? Either way, eMMC is faster than an SD card. I do like the Le Potato, similar to Pi 3B+, but less power draw and 2GB DDR3 RAM vs the Pi 3B+ 1GB DDR2, same basic footprint.

Released ByteAether Ulid v1.3.0.

Blog post: https://byteaether.github.io/2025/byteaetheru … l-and-security/

Did quite an overhaul on the generation/creation part of the ULID. Better than ever before! 😀

chrismeyer6 wrote on 2025-07-30, 00:59:

Not the CPU socket you need to check the pci Express slot itself. Not all full size slots have all the pins populated. Asus has some mid range Intel b850 boards with full size 16x slots but only wired for 1x and the rest of the slot has no pins.
Would you mind posting a picture on the 16x slot?

I tested one GPU with Z75 (3rd gen i-5) and Z87 (4th gen i5), both PCIe 3.0 x16 lanes.

GPUz detects PCIe 4.0 @ x16 running PCIe 3.0 @ x16 ---> values that never change, not at idle or 100% under stress. With 2n x16 slots it is detected running @ max speed allowed (both PCIe 2.0 @ x4) and results are lower than expected. But consistent between them.
I tested the GPU with 3DMark 11 and results are consistent with other cards' results. This is probably just a glitch and I should probably stop wondering.

HW monitor is the same.
IDK how GPUz and HWMonitor work but probably they use the same system to check lanes? If I'm wrong correct me, I'm clueless on this subject.

My guess spans from m/b bios to gpu bios.
There have been many updates and who knows if by flashing the latest version I just introduced a bug.
Who knows! - Prof. Farnsworth

Converted Smart Preview for Pale Moon forum, a XUL bootstrapped extension, to a user script.

I tried some of those gaming keyboards when I was in Best Buy on Sunday. I wasn't expecting the different switch types (at least the ones on display) to have similar noise levels or tactile feel. They do have different noise characters, but the "clicky" sound was a different type of obnoxious high-pitched metal-on-plastic tapping noise than I expected. The MX Red clone switches have a better-sounding click, but they still seemed loud. At least they feel smoother than the random cheap rubber dome keyboards I've been using (full-size ones, not the ultra-thin ones).

Yes, I've seriously never used any kind of "mechanical" keyboard (never mind that rubber domes are mechanical parts) since Alps switches were still common.

lti wrote on 2025-07-28, 21:53:

UCyborg wrote on 2025-07-28, 17:04:

Dealing with bugs and staring at the code for hours with nothing to show for it. Is it still better than dealing with idiots and annoying people? I don't really know.

You're still working with idiots, but now they're in upper management. I had a health scare after hearing the VP in one of those all-employee meetings last spring.

I tried to take a few days off, but nothing really sounded interesting, especially with the extreme heat that is now normal (and even the most extreme climate change activists deny how much temperatures have actually increased). Then I saw a weird bug at work (the kind that makes even less sense as you keep testing and leads you to some ridiculous potential causes), and I want to see what happens.

Small companies seem to be their own kind of hell. I recently notified the developer of the issue with labels not being associated with checkboxes on one form in the internal application we use. He fixed it, then reverted it because the moron from another department told him to (the developer himself thinks the association makes sense). WTF? Have you seen a program where clicking the label associated with checkbox doesn't trigger the checkbox? It's one of the first accessibility guidelines, makes it easier to click since you don't have to strictly target that square. And it helps assistive technologies. Would a sane person go out of their way against it?

This is a big company, but I shouldn't go into too much detail when I'm talking about upper management online.

The weird stuff I saw at work really was the ridiculous answer. A bug in the development software resulted in hardware damage.

My work computers are both broken, and my personal phone is giving me an Android update alert in Spanish. Then I'm waiting for the rainforest idiots to ship one of those shitty cheap power meters so I can see if my big Ryzen is really heating the room or if I'm full of shit. I ordered it a week ago, and when it finally does ship, it will probably be put in either an envelope or a huge box with no packaging material.

I took my computer outside to clean it, and even the outside of the case is warm around the "extremely efficient" CPU and motherboard VRM. Something is seriously wrong. I think MSI is fucking with power limits and overvolting, but everyone is pretending that that's an Intel-specific problem. Now what?

lti wrote on 2025-08-17, 02:20:

I took my computer outside to clean it, and even the outside of the case is warm around the "extremely efficient" CPU and motherboard VRM. Something is seriously wrong. I think MSI is fucking with power limits and overvolting, but everyone is pretending that that's an Intel-specific problem. Now what?

Have you tried updating to the latest bios and fiddling around with BIOS settings? I know part of Asrock boards (and others) frying 7000x3d and 9000x3d chips has been an entire debacle. Wouldn't surprise me if some stupid default bios setting is causing VRM issues

I don't know what the BIOS settings do. It's a lot of low-level stuff instead of the simple "optimization" switch that Intel motherboards had (what was initially blamed for the last batch of CPU failures before the oxidation problem was found). I do have PBO and XMP/EXPO turned off, and the switchable TDP is set to 65W. One weird thing I noticed is that the rear case fan is set to a fixed speed (I think it was 60%) by default. That fixed speed is pretty low, so I'll eventually change the fan profile to vary the speed, but make its current fixed speed the minimum (it's pretty slow with the fan I'm currently using). There are some BIOS updates that I didn't know about, so I'll try that.

I'm seeing 54W at idle, but my old computer (i5-8500 with integrated graphics) draws 22W at idle (with a less efficient power supply). Even though the Ryzen is about twice as fast in the applications I've tested, it uses over double the power compared to the i5. Unless the BIOS is set to "bigger bar better" settings by default, it's another example of efficiency dropping in more recent products. If the BIOS is really set for optimal power consumption, then I think I'm going to sell it to someone with more air conditioning. For all I know, the "inferior" Intel parts could have lower power consumption in more basic office/Internet tasks (including installing updates - even the old i5 spikes up to 85W when Windows is updating, but that's still better than 135W in my Windows VM on the Ryzen).

Playing around with Windows 11 LTSC IoT on my Ryzen 7 build. Seems to be working fine so far. No Recall/Copilot/whatever, no pre-installed bloatware, and no intrusive data collection, at least as far as I can tell. The only trace of AI that I can see is in the Edge browser, which I don't use.

I restored the Microsoft Store (it wasn't available by default) and then installed the Xbox App and Game Bar. So far, games seem to be working fine, but I still have more testing to do.

Swapping out my original Ender 3 with a newer, faster, and the overall "better" Ender 3 V3 SE.

Got the V3 for my wife last year, but, she's not using it. So, it's mine now. She didn't like the manual filament swapping for color changes. Going to have to get her a multi-color unit.

My old Ender 3 served me well, have a ton of upgrades on it, but this V3 is so much nicer. Going to make my 3D printing life easier. 😀

Decided to cut my losses (time and mental headroom as well as money and space) and recycle my Ryzen 7 mobo/cpu/ram/gpu. Case has already gone along with its sticky rubber. Rest goes today.

Have moved to Windows 11 on a mini PC. Haven't tried any games on the mini PC yet, but external GPU is an option. Things all work so far - the usual peripherals, plus such software as was immediately critical.

Contemplating whether to try to get my old Windows 10 install converted to a VM to run on the Windows 11 machine, in order to retain the software I can't reactivate (because the companies have either withdrawn the software, or no longer exist). Need to look at clonezilla.

I got curious about forcing newer versions of DLSS in older games via Nvidia Profile Inspector, so I tried it with Rise of the Tomb Raider.

That game used an old version of DLSS by default (not sure which one) and it looked kinda crappy to me. So I forced the latest one, and also made sure that DLSS is only used as DLAA (different setting). This resulted in a very nice image, possibly even cleaner than the game's own 4xSSAA.

I spent most of my weekend reorganizing files on my NAS as a background task.

In the process, I lost about 80 GB of non-essential data because I mistakenly assumed that rclone works the same as rsync. I learned the hard way that rclone sync deletes files in the destination that aren't present in the source. What I needed was rclone copy.

On a brighter note, in my files I rediscovered an old and hilarious Duke Nukem 3D RTS file that my friends and I thought was lost forever.

mtest001 wrote on 2025-09-01, 07:18:

In the process, I lost about 80 GB of non-essential data because I mistakenly assumed that rclone works the same as rsync. I learned the hard way that rclone sync deletes files in the destination that aren't present in the source. What I needed was rclone copy.

We’ve all done it… that’s why now I always run robocopy with /L (list rather than copy) to do a dry run first, and never use /MIR to mirror unless I’m 100% sure that’s what I want to do.

Yep, I did a rookie mistake here. I was moving tons of files and it was taking too long and I got impatient. Rclone has the "--dry-run" option so I have no excuse.