BitWrangler wrote on 2024-09-06, 12:38:

I think even if you sort out the errors at 17, you get a new set of problems trying to get past 20/21

I don't have the abilities. It would be a total waste of time.
You sound familiar with it though 😉 Hope it didn't ask for a human sacrifice 🤣

Ensign Nemo wrote on 2024-09-06, 15:38:

BitWrangler wrote on 2024-09-06, 12:38:

I think even if you sort out the errors at 17, you get a new set of problems trying to get past 20/21

This sounds like good life advice.

I immediately let go. As soon as I see red letters during its process and halting to command line... I know I am über screwed.

It was fun for a bit though. 😀

Today testing the upscaling difference from a modern good brand DVD player a Sony DVP-SR760H DVD device against a low end good LG BP-250 BlueRay Player using as test a modern movie in its commercial DVD original version running on both players in a 1080p calibrated TFT monitor; I was expecting the BlueRay player obviously having a better result but I was shocked to see how MUCH the LG BlueRay player (most probably using some Mediatek SoC into it) surpass the DVD player hardware/software upscaling using the same MPEG2 DVD disc.

But anyway the Sony DVD player seems to have a good drive silent and solid but the main feature the upscaling and also colors quality looks cheap. Also the the DVD player default colors seems to be off, with green colors heavy and missing dark and bright areas details and mostly useless brightness/contrast setting while some other setting seems to help improving the final result. Of course I expect a BD player having a much powerful SoCs to do a better job but I wasn't expecting such difference. I might still suggest the Sony DVD player as option for DVDs for features like AV latency control, good drive, good remote and also the coaxial digital audio the LG BD player had not BUT for better video quality of old cinema discs I'd totally suggest a good BD player like the LG above.

(EDIT: grammar corrections)

You get some 90s DVDs and maybe early noughts budget releases that are horrible whatever, think they only used an 8-12bit digitiser on them or something, see the rings around every light source.... Maybe it was a first gen satellite broadcast Mpeg2 converter or something.

I bought a GPDWIN4, and have been playing around with it.

I got mine used off FleaBay for a vastly reduced price, but it works just fine. It's the 2023 model. (ooh, a year old, Big woop.)

I can say the following definitively about it:

HoloISO "works" on it, but not properly. The version of Radeon Mobile video hardware inside it does not play properly with their version of the RADV vulkan driver. Some games run at 1fps and 100% CPU-- others have randomly vanishing textures, and more yet have strange spaghetti texture stretching, to the point of it being a crazy ball of multicolored fuzz when trying to play. (And others still, WORK JUST FINE.)

Bazzite works much better-- with actual accelerometer support Out of the Box-- and so far, without any of the above strange graphical bologna.

Testing continues.

UMPC doohickey? Can you boot steamOS off a card or anything?

It has a microSD card slot, and 2 USB type C connectors.

Internally, it has an NVME 2TB SSD.

I have dropped a 512gb microSD card into the slot, and will (soon[tm]) set it up as a bootable recovery volume.

Both HoloISO and Bazzite are "SteamOS-Alike" distros for such handheld PCs. Boot directly into steam in game mode, just like a steamdeck.

Replaced Magisk 21. something with latest version of Magisk Delta on my phone. Now looking for the damn ZIP files of the modules I had installed before and re-granting permissions for root apps.

Financial app devs and everyone around them are a bunch of scumbags.

That was a disaster, replacing Magisk 21.1 broke another app (can't access my online bank without it...), still broken after removing Magisk. Works when restored from backup with old Magisk. There went my evening.

Also, funny output of adb after restoring data partition:

1>adb push data_before_delta.img /dev/block/mmcblk0p24
2data_before_delta.img: 1 file pushed. -1.5 MB/s (1879031296 bytes in -1188.994s)

Haven't updated Android Platform Tools since 2019...that and the phone's flash memory must be getting tired after 10 years.

Yeah I hate having anything "important" on my phone, I was pissed when I had to add my cell number to my credit card account .... this ain't adding security, it's increasing attack surface.

It seems most banks love to tie their services to a smartphone and employ activation procedures that make users jump through hoops and have no way to transfer it to another phone without contacting the bank. 10 years ago, you could have a certificate on a computer and things were simple(r). Now certificate is replaced by a smartphone...

I somehow got both apps working in the end, now I don't even know if replacing Magisk was necessary because in the end, I had to force basic attestation through MagiskHide Props Config for the new app to work surely. Not sure why the new app launched at first without it. I noticed first app doesn't lunch at all when completely removing Magisk.

if I knew to look into MagiskHide Props Config first, which I had installed, just didn't actually use (figured it might come in handy some day)...probably needlessly restored partitions few times because I assumed the process broke something permanently...

In the past, I was restoring partitions on the phone using Linux on desktop PC and used a chain employing netcat and ADB's TCP port functionality. Didn't know I could just use adb push, pulling partition from the phone works when referring to partition's symbolic link, but not the other way around. Pushing works as long as you refer to partition's path directly, eg. /dev/block/mmcblk0p24 for data partition on my model. Path to symbolic link is longer and descriptive, you know exactly which partition you're targeting. But when you know the numbers, referring to them directly shouldn't bring any surprises.

Flash memory is still holding, it's just that adb pushing on Windows was slow on my PC. Seems to be noticeably more performant on same machine on Linux. I'd have to try again to be absolutely certain about exact numbers, but in best scenario, it's something like 3 MB/s on Windows and 6 MB/s on Linux. No idea how apart ADB versions were, my Linux has whatever version came in the Ubuntu's repos in 2020 if I'm not mistaken.

One interesting difference that I noticed, that ADB on Linux I have always does "push install" for app installations via adb install command. The old ADB on Windows I have does "streamed install". But brand new ADB downloaded recently straight from https://developer.android.com/tools/releases/platform-tools on another computer running Windows also does "push install". I think "push install" sends the whole APK file to temporary folder before install while "streamed install" doesn't.

These days I'm going to install the new M2 HAT board on a Raspberry Pi 5 8GB board I have so I'll replace the Micro SD card which is the best I can find with a NVME 2230 disk that should increase really a lot disk speed in a o.s. real world scenario. It connects to the board with a PCI-EX connection which seems really a nice thing to have here. Only bad side the active cooler being already hot with the closed case, I suppose it will be even worse but I'll see.

I installed the latest Kdenlive on my laptop (I was previously running 20.08.1) and discovered that videos created by the Windows built-in Camera program have a variable frame rate. I thought it was a bug in Kdenlive (or more likely MLT), but MediaInfo confirmed it. The old version of Kdenlive didn't give me any warnings. I'm a little disappointed that Intel Quick Sync encoding is not available in Windows (NVENC and AMF are available - not that you want to use AMF because AMD never seemed to care about any video features that don't relate to 3D rendering for gaming), and NVENC appears to take slightly longer to render than x264 on my laptop (i7-9750H and Quadro T1000). The new version seems to use different render settings that are much faster, but I don't remember how the old version performed on the laptop.

Also, when you update Kdenlive in Windows, you apparently have to fully uninstall the old Kdenlive version and manually delete the directory out of Program Files before installing the new version. If you just install the new version over the old one (or just uninstall normally, which leaves a lot of stuff behind), it claims to crash at the very end of every render job, but the video plays fine.

My desktop is still resisting Linux. I downloaded the latest Lubuntu (since that's what I had a good experience with when I couldn't afford anything that wasn't sub-Dorito class), and it still has screen tearing (even though that was supposed to be fixed) along with introducing the video stutter that was in Mint. Lubuntu 22.04 and MX Linux didn't have the horrible stutter, but MX Linux was totally broken in other ways, as I think I mentioned in another post.

Lubuntu worked for me for such a long time, and now this video stutter exists in multiple distros. I don't know what's going on.

I also noticed that videos from the Windows Camera program are even worse because they're using a constant bit rate as well as a variable frame rate. From my limited video knowledge, that looks like the worst possible option. I'm not surprised because it's Microsoft software. I'm trying to switch to Linux because Microsoft is totally incompetent and still tries to force us to use stuff that's so buggy that it's completely unusable (like OneDrive).

On my work computer, OneDrive failed in a new way. I suddenly got a message that said that syncing failed for some files, and it was going to create duplicates for each computer that I've signed into OneDrive from. Then it started duplicating files that I've only ever accessed from that one computer. It has been claiming to be downloading files for months now (which is weird because they shouldn't have been edited on any other computers), and those downloads never completed.

lti wrote on 2024-08-25, 02:11:

The desktop (i5-8500) took 34 minutes to render a 4-minute 1080p video with Kdenlive

I installed the latest Kdenlive on my desktop now, and the same project rendered in 3:03. It uses different x264 settings (the "veryfast" preset instead of "faster" and not setting the "bf" parameter in FFmpeg anymore), but that isn't enough to explain the huge performance difference (and the CPU was loaded to 100% in both versions).

I didn't notice that the Camera program was using a variable frame rate because the minimum frame rate isn't taken to ridiculous extremes like Handbrake (I have seen it drop to 5fps - Handbrake defaults are "what not to do," but people seem to like it).

BitWrangler wrote on 2024-09-15, 14:13:

Typical old fart linux rant ...

Linux pisses me off, keeps going in cycles where "Big new fresh release...plus 50 bugs, next 10 updates fix 4 bugs each, introduce 1 new feature, with bugs, last 10 original bugs pronounced insoluble under this paradigm... junk everything, Big new fresh release, desupport 1/3 of hardware/software, old bugs gone, 50 new bugs, repeat cycle... " I want one, just one, that's all the way working, for at least a 5 years spread of hardware and software. You get on a distro, and it's hopeful, and it seems to improve and improve and then "Aaaaarrgh you maniacs, you blew it up" because yeah you wanted to spend the next year chasing answers, and getting given solutions on linux forums that only used to work 3, 5, 10, or 15 years ago. It's getting to seem to me that you have to have single purpose linux installs, have to get distros tuned to the one thing you want to do, and trying to put stuff on that allows it to do other things has big roadblocks.

Then sometimes you get a distro that seems okay but then you discover it's someone's foibles and peccadilloes cast into code, all other ways are wrong, there is only their way, and they've only considered the 15% of use cases that are important to them. So you get these artificial blocks, built in quite deep, to doing anything else any other way. It's particularly annoying if you mesh with the original 15% of use cases, think the whole thing is perfectly designed, get real comfortable, settle in, then a year later try to branch out and try something annnnnd *bang* that's your head against the brick wall you didn't realise was there. Raise forum questions in the other ppl comfortable in that 15% and it's cult like denial, they haven't banged their head yet. The people who experienced it, aren't around, they're on another distro now. Not solely a linux thing that, I guess it happens where ppl repackage windows too, it's clearer cut though, a thing doesn't work because you've got a "damaged" windows install.

it 'funny' how Linux on the desktop, 25 years of it, has certain characteristics that stay the same - the proliferation of distro's, the dying off and forking of the distro's, the unfulfilled promises, the next big thing

that said, certain mainstream distros are ok and drivers etc work much better. ironically, for the perpetual claim of versatility and 'do it yourself' i now find linux just fine for use on older hardware to do some simple things and going online - leaving the complex stuff out because its a labyrinth of forums, guesswork and an endless 'choice' of methods

Consulted the Thinkpad T480 Hardware Maintenance Manual to open up my friend's laptop and exchange his failed internal battery with a new third-party one. The battery was visibly bloated. It's a rather annoying thing as batteries become internal - their physical deformations are not visible until they start pushing against other parts, sometimes breaking them.

I installed the M2 HAT for the Raspberry Pi but found out that the original case while it can have the HAT installed which seems designed for it, it'd work without the top cover at the end. So I went for a cheap metal case designed for both the single board computer and the SSD disk option. Too bad the top glass cover isn't really glass and doesn't have the power on/off button. On the speed side the Kingston NVME 2230 board runs at around 400MB/s on write and 230MB/s on read with very high random IOPS numbers on the PCIEX Gen2 x1 bridged connection. It can unofficially support Gen3 speed too.

On the real world usage I can see good differences in heavy apps like Chromium based web browsers but still imho need to improve the GUI latencies higher than a low end x64 system even when technically slower than the RPi.

NVME test:

Sequential write speed 388937 KB/sec (target 10000) - PASS
Random write speed 93622 IOPS (target 500) - PASS
Random read speed 100054 IOPS (target 1500) - PASS

The best previous very expensive fast Samsung Micro SD I had gave like 20x lower numbers. It doesn't seems to increase too much power usage at the wall just few watts but still I've to test in specific heavy benchmarks.

386SX wrote on 2024-09-16, 20:44:

I installed the M2 HAT for the Raspberry Pi but found out that the original case while it can have the HAT installed which seems designed for it, it'd work without the top cover at the end. So I went for a cheap metal case designed for both the single board computer and the SSD disk option. Too bad the top glass cover isn't really glass and doesn't have the power on/off button. On the speed side the Kingston NVME 2230 board runs at around 400MB/s on write and 230MB/s on read with very high random IOPS numbers on the PCIEX Gen2 x1 bridged connection. It can unofficially support Gen3 speed too.

The RPi5 is a cool gadget. I wish I could connect both an NVMe drive and a PCIe card to it. I suppose one would need to use their PCIe HAT and some sort of additional PCIe-PCIe bridge card and then a PCIe-M.2 adapter. Cumbersome.

dr_st wrote on 2024-09-17, 08:57:

386SX wrote on 2024-09-16, 20:44:

I installed the M2 HAT for the Raspberry Pi but found out that the original case while it can have the HAT installed which seems designed for it, it'd work without the top cover at the end. So I went for a cheap metal case designed for both the single board computer and the SSD disk option. Too bad the top glass cover isn't really glass and doesn't have the power on/off button. On the speed side the Kingston NVME 2230 board runs at around 400MB/s on write and 230MB/s on read with very high random IOPS numbers on the PCIEX Gen2 x1 bridged connection. It can unofficially support Gen3 speed too.

The RPi5 is a cool gadget. I wish I could connect both an NVMe drive and a PCIe card to it. I suppose one would need to use their PCIe HAT and some sort of additional PCIe-PCIe bridge card and then a PCIe-M.2 adapter. Cumbersome.

Really a cool board. I had all the versions and still have the RPi1 512MB, the 4 8GB and the RPi5 8GB version. The new CPU is really powerful and having some specific acceleration that make it much faster in some area even against some early i7. But the problem imho is the speed of GPU tasks not necessary in OpenGL apps but latencies seems higher than any old systems even much slower. Still I like to use these as main machine and the CPU can easily work with intensive apps, actually faster than I thought. I also wish to try these with a Intel Arc or AMD video card with some PCI-EX GPU and to see if things can get better for the GUIs.