Leaderboard

Salut ; since you are the party interested, i.e. the one in possession of an SSE-only CPU, you can "try" yourself whether the following route leads to somewhere: The CPython (py3.9) implementation used by nicolaasjan is a fork and recompile of a py3.9.x release by @cmalex , targeting his SSE-only processor; do note that this compile has special configuration that deviates slightly from the PSF's one, as I had discovered in the past myself when messing with it (e.g. pip wouldn't successfully install some modules, while the installation would be fine with the PSF compile). You don't need a yt-dlp_x86.exe binary per se; you can try to install yt-dlp as a Python module python -m pip install yt-dlp using the py3.9-win32-sse library; the above command will attempt to co-install all necessary dependencies; however, some of these contain binary parts (C/C++ extensions, they're the files with ".pyd" extension) that have been precompiled as wheel packages by their authors, not specifically targeting < SSE2 CPUs... As cmalex has done himself with some python modules he was after, these had to be also (forked/)recompiled locally on his target machine, to ensure no SSE2 instructions sneak in the compiled code; this only complicates things further (i.e. when python modules have to be compiled from source rather than installed via wheels); a similar approach would be to "try" and run yt-dlp directly as a python script, using, of course, the aforementioned py3.9-win32-sse compilation; again, the "requirements" must have been successfully installed, which is the real question here... (rant mode ON) Python coders of this era have no consideration at all for backwards compatibility; they'd like to only code in Python 3.20, if they could get their hands on it now, supposedly because it makes their lives "easier"; with the official PSF stance being "we only support what Microsoft supports", this practically means that Python projects will soon ONLY run under Win[10/]11, on H/W compatible with these OSes only ; to add insult to injury, with Win11 being 64-bit only, I have trouble already finding 32-bit compiles for many of the (open source) programs I want ... If it weren't for dirkf, who decided to go against the tide and tries single-handedly to keep youtube-dl code compatible with older Python versions (thus able to run on older Windows, too), things would've been already dead for XP/Vista[/Win7] yt-dl users... (rant mode OFF)

I also tested there my coming release uBlock Origin Legacy 1.16.4.31. No ads in either of the two YouTube videos. And this release also works flawlessly and without any issues in New Moon 28, Serpent 52 and Pale Moon 32.5.2. Under both versions 1.16.4.31b2-1.54.0 and 1.16.4.31, I don't see any ads in the Ad Blocker Test Video. It stays empty and black. Here is a screenshot with 76 ads blocked:

I know Youtube Downloader HD very well and use it for many years, too. YouTube in web browsers on old hardware is no longer a good idea. This site has gotten worse and worse from year to year. If I nevertheless want to watch a YT video directly in New Moon 28, I use the extension uTube or the service Invidious.

@nicolaasjan I did further tests. This time, I tested uBlock Origin Legacy 1.16.4.31b2-1.54.0 in the most recent version 32.5.2 of Pale Moon under Windows 7 on a notebook. I cleaned the profile completely. I deleted all cookies, website settings, cache and so on. I removed all remnants of previous uBlock Origin Legacy installations. I can't see any ads in either of the two YouTube videos. Please, check if you use a completely clean profile and install my mod again! In any case, delete all cookies and settings related to YouTube! Especially the file ublock0.sqlite in the subfolder extension-data.

Thanks for reporting and your kind feedback! Very appreciated! Did you also test the two YouTube videos from the quoted post above in terms of embedded ads? I need urgently further confirmations.

And about this "test" : these are the weaker cards , much weaker than GTX 980. So PCI 1.1 is enough in this case. GTX980 is like 80% faster than GTX680. Don't forget about the onboard VRAM . 8GB vs 2GB . See the screenshot by asdf2345. https://gpu.userbenchmark.com/Compare/Nvidia-GTX-980-vs-Nvidia-GTX-680/2576vs3148

First off , I'm not the only one who is "mistaken" , scroll back and you will see the same issue has been reported by uCyborg and asdf2345 . Their Nvidia CPL also shows PCI-e ver. 1.0. Second : why would you even mention Kepler cards ? It is off-topic . This is only about 900 and 1000 series, asdf2345 and I have tested on GTX 980. Forget about my Titan , it was used only to confirm the issue with XP64 not detecting PCI-e 2.0, besides , asdf2345 has tested on a laptop. As you can see , it has nothing to do with my mobo.

For testing purpose, I opened a small video with low resolution on Imgur. Nearly 100% CPU usage and tons of the same JavaScript error in the Web Console. One more bloated website and service which does not run bearably in New Moon 28 and Serpent 52 on Windows XP. Now, I know why I have always avoided this service, especially as an image hoster.

Why would they ever be there?

I've just read this on TheRegister (about a firefixed flaw in Fireferret=). Does it mean anything for our UXP browsers? And by the way - why does imgur.com perform so badly in Serpent 52.9 (in recent releases) - videos/images just start to load very slowly - is it the RAM? It used to perform smoothly not so long ago.

What do you think? (It's a rhetoric question, no need to answer)

Greetings; yes, you're posting in the Supermium thread, so you must be using the Supermium browser ; but it would've helped greatly if you mentioned already from the beginning the specific version of Supermium used and the OS you're on ... As for Win XP (SP3?) x86, there have been several pending issues in the official Supermium tracker already: https://github.com/win32ss/supermium/issues/940 https://github.com/win32ss/supermium/issues/867 https://github.com/win32ss/supermium/issues/797 https://github.com/win32ss/supermium/issues/444 This appears to be an XP-exclusive issue... As for sourceforge, Supermium versions 121-r2, 122-r6, 124-r2 and 126-r3 (all 32-bit) DO NOT trigger here the cloudflare anti-bot protection ; all are "dirty" profiles, different to each other, with several extensions installed; OS is Vista SP2 x86... KafanMiniBrowser (Cr87) and 360EEv13.x (Cr86) BOTH trigger the CF "protection" and I can tell you it's practically IMPOSSIBLE to pass the challenge on those two browsers; in desperation, I tried a fresh new profile of KMB (with no extensions) and I eventually managed to pass the test only once after maybe twenty successive attempts; but it gets worse, because access to sourceforge.net is shortlived (despite the cf_clearance cookie expiring a year after ); once the browser session terminates, you're being blocked anew ... Given that recent Chromium versions on Win10 don't get the door, I used this extension https://chromewebstore.google.com/detail/custom-useragent-string/lejiafennghcpgmbpiodgofeklkpahoe (v0.2.1 is the last MV2 one compatible with Cr86/87) to set a SSUAO for SF: and I can now access SF unobstructed ... NB: Both KMB and 360EEv13.x DO NOT send out CH by default (you have to enable #enable-experimental-web-platform-features for that first; and even then, this is an immature/alpha edition of the CH API). Can the link to this file of yours be made public, please? I suspect some Supermium default security setting might interfere there, if that "file" isn't palatable to Google's notion of "security" ...

@VistaLover If you don't try, you don't know... @nicolaasjan OK, thank you for your efforts. Don't spend too much time on this. Regards

I'm using - Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/122.0.0.0 Safari/537.36 This matches the string used by Thorium.

Due to the certificate issues all 360Chrome browsers suffer from time and time again, I would suggest to try ProxHTTPSProxy TLS 1.3.

I can well imagine that. More than 50% of the world's population is male, and such porn websites are highly frequented. And once again you're completely wrong. I hate Microsoft. Especially for the restricted, disempowering and bloated (newer) operating systems, websites, programmes and services. One of the reasons why I use Windows XP, from a time when Microsoft still produced sensible things. Moreover, your beloved porn website is totally crap as you have to register first. Apart from that, such filthy websites should be discussed elsewhere and not here since there are of course underage members and, in addition, those who consider such pages to be offensive. And as I already said, I consider requests for members to register on such a questionable website to verify login problems to be highly dubious and a no-go.

So, you prefer to have a unique ID number of the tiny Supermium userbase, or an empty string? Ungoogled userbase isn't huge either.

See also my build instructions here: https://github.com/ytdl-org/youtube-dl/issues/32076#issuecomment-1518535677

I won't make a SSE version of yt-dlp, because the PyInstaller building module produces exe's which require SSE2+. The alternative would be a crippled py2exe build. See here.

ytdl-patched is built automatically each day from an outdated master tree, so each build is the same size... E.g. release youtube-dl (Unix executable) from 2024.08.01 (on page 7) is 1923305 bytes, same size as their latest youtube-dl from 2024.10.07. The maintainer Lesmiscore seems gone. That's why I pointed you to this pull request there. When you want to build from source, use the master branch from https://github.com/ytdl-org/youtube-dl. (but it won't fix your issue yet)

I had already looked at these resources before @UCyborg mentioned them. But in any case, a very interesting contribution. The file scriptlets.js is only present in the more recent webextensions and can't be transplanted without code changes if at all. The legacy extensions provides functions via the file resources.txt which was missing in my previous releases and had to be restored in my latest one. From now on, I won't merge legacy extension and webextension any longer in terms of uBlock Origin. The webextensions of uBlock Origin are moving further and further away from the code base of the legacy extensions. Too much incompatibilities.

We will see what YouTube will come up with next. Personally, I hate all these constant changes and perfidious mechanisms that are supposed to block ad blockers. In any case, thank you for your persistent reporting and testing! And one more request: keep testing to make sure everything runs smoothly! It's a good thing to do when you're surfing anyway. I personally think this version should now do its job.

Due to the problems in terms of filtering and blocking ads in YouTube videos, I had to correct my release uBlock Origin Legacy 1.16.4.31b2-1.54.0 once again. I added some (necessary) code and two filter lists which were not present in uBlock Origin 1.54.0 but in uBlock Origin Legacy 1.16.4.31b2. The two lists are uBlock filters - Legacy and uBlock filters – Resource abuse. Here is a screenshot taken from the setting page: Additionally, I updated all filter lists once again. The new download link and all information can be found in this post: Please re-download this corrected release from the linked post above especially if you already downloaded the old one! Please, test this special version and report here! Any feedback is of course welcome! Cheers, AstroSkipper

Here is your PSY Gangnam Style YT video https://www.youtube.com/watch?v=9bZkp7q19f0 filtered and adblocked by my new mod of uBlock Origin Legacy 1.16.4.31b2-1.54.0 which will be released very soon:

I don't see any ads in either of them. And both videos are tested by me with default lists in New Moon 28, too. As far as I can see, my fixed release uBlock Origin Legacy 1.16.4.31b2-1.54.0 works fine here. But we need further tests and results from other users for confirmation purpose. As for PEBCAK, I wouldn't say you are a problem.

Can you please provide the YouTube link and a screenshot of the ads (maybe, you can outline all ads in red) you see by using uBlock Origin Legacy 1.16.4.31b2-1.54.0 which is on the other hand successfully blocked by uBlock Origin Legacy 1.16.4.31b2? This would be very helpful for reproducing here. I personally had problems to find ads in your last provided YouTube link.

Thanks for your detailed feedback! The original idea for creating this build was to provide all filter lists from the webextension uBlock Origin in the extension uBlock Origin Legacy. That there are filter lists which are no longer fully compatible was expected. I just wanted to correct the version uBlock Origin Legacy 1.16.4.31b2-1.54.0. Any list which is not compatible shouldn't be enabled or should be disabled if not working properly. My next release will be an update version and mod of uBlock Origin Legacy 1.16.4.31b2 itself with all filter lists that have already been their plus the list "uBlock filters – Quick fixes". In general, it is no problem to implement any filter lists into the code or to remove them from the code.

I use https://curbengh.github.io/malware-filter/urlhaus-filter-online.txt and have removed the obsolete link from rhe code of the internal list. BTW, from now on, I try to avoid GitLab links if possible. GitLab causes errors (see the uBlock Logger) as they force the user to login into their account.

I spent my vacation at a nice hotel located on a charmingly beautiful island Sint Maarten. The islands in the Dutch Caribbean, part of the colonies, now Constituent country of the Kingdom of the Netherlands. https://en.wikipedia.org/wiki/Dutch_Caribbean Sint Maarten is a constituent country of the Kingdom of the Netherlands, and as such the Monarch of the Netherlands is head of state, represented locally by a governor. https://en.wikipedia.org/wiki/Sint_Maarten I've been there several times before, and always liked it! So far, I visited many, but my ultimate goal would be to visit all islands in the Dutch Caribbean.

1- I'm pretty familiar with 90's style , no . 2 - I didn't change any settings at all . I think it happened after I reinstalled Nvidia drivers. And the nice bug happened when I was using RTM. I can't replicate this. Got some time to test again , I'm pretty convinced SP1 (RTM) runs faster , overall better . I can clearly see the desktop is crystal clear . SP2 looks more like Win-7 , washed out , not clean , it is still better than win-7 , of course. I think I'll have a dual boot with Vista x64 SP1 and XP64 (RTM) with GTX980/Siemens.

Thanks for the explanation . The absence of DWM is a big downside , that's for sure . I don't know what I did , it was so many attempts and testing procedures (trying various drivers , reinstalling again and again , etc) , but somehow I've managed to switch to "cartoonishly moving mode" instead of tearing . I do not know how to name it properly . It's like when I move a folder , it draws a new one (where I place it) , while keeping the folder in the previous position intact . In this "mode" the games that I run in windowed mode ran without any tearing , still slow , but without tearing . And now it's gone (the "mode"), after reboot . And I do not know how to get it back . Funny.

Got some time to test on my other system with Haswell : Nvidia CPL still says "PCI-e" with XP and "PCI-e gen 3.0" with Vista (3.0 because this is a newer mobo). The games still feel much slower on XP. Also , there is a terrible screen tearing when I move any folder across the screen , the same with all of the games that I try to run in windowed mode. Is it supposed to be like that on XP ? V-sync and tripple buffering aren't any help. The same is with Win7 when aero is turned off . By the way , quality wise Win7 is on the same level with XP. The picture is flat and washed out . Games are cut off by several mm when launched in windowed mode , but this is not the case when running them on Vista. This is my first experience with XP . I skipped XP and went from 2000 to Vista in 2007.

Where did you see such config ? Anyway , I'm not familiar with modern laptops , sorry. And my notebook from 2013 already had PCI-e 2.0. On one of my desktops (where I test XP64 dual-boot with Vista) I can observe some games can easily reach and stay at 86% for some time on PCI-e 2.0 with Vista and GTX 980, but with XP64 it's like around 30% only. For my precise config just scroll back and replace GTX Titan with 980 . I still think that even if GPU-z detects PCI-e 2.0 with XP64 , it seems like Nvidia drivers just can't take any advantage of it on XP. Otherwise why would Nvidia CPL say "PCI-e" with XP and "PCI-e gen2.0" with Vista ?

Wolfenstein: The New Order has rather simple demands . The recommended graphics: AMD Radeon HD 6850 or NVIDIA GeForce GTX 460 with VRAM 1GB , so there's no way it can saturate PCI-e 3.0 or even 2.0. I'm guessing this is the reason it feels smooth on XP. OK , GPU-Z shows , but nVidia CPL not , why ? Seems quite strange , the same result with several different people on absolutely different hardware. Yet again , if I reboot to Vista the CPL will show the highest version of PCI-e. This is on the same drivers version . I've tested with almost all known drivers . By the way , 368.81 is the weakest among them , asdf2345 uses 347.25 for his XP and it's the wise choice , I would also recommend 347.09 and 344.60 for 900-series. But yet again , you've just confirmed my discoveries . XP is slower even with DX9 games. Why do I need this ? It's just out of curiosity , I've never used XP and decided to try it. I went from Windows 2000 to Vista in 2007 (skipped XP completely).

Yes , I know , thanks cap. My GTX 980 is 4GB , yours is shown as 8GB. And the dude is talking about 600-series ancient cards in the wrong topic.

I'm not sure how GTX750ti is relevant to this discussion , being lower end and from another series. Yeah , I know it's the first gen. of Maxwell , but still another series . Btw, artifacts in Unique Heaven oftenly tell that somethig's wrong with the videocard itself . No such issue for me with high end 900 series and 344 drivers. What about the other issue with displays ? Besides , what version of XP you were using ? 32 or 64 bit ?

Well , that's at least 3 by you and a couple by me . I can confirm all mentioned by you and add these to the list. 1 - Significantly decreased performance starting with 35.x 2 - Problems with idle modes on GTX980/970 , GTX780 (new revision, reissue with another chip) , Titan Black. Though, the first GTX 780 (from May 2013) and the original Titan did not behave oddly during idle. Mind you , this happened on both , Vista and XP , so XP has nothing to do with it , it's the drivers themselves. My opinion : 344 series are the best for XP. If someone wants newer , 364 drivers seem to be OK , but not as fast 344.11 and they need to be modded to work with 9xx series. For some raging publishers : I didn't test with Titan X , so no comment. EDIT : I did not test with lower end cards .