What tests do you do to validate the changes you make to preference values and/or introduce new user prefs?
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
foxPEP and TenFourFoxPEP - For A Faster Firefox / TenFourFox
- Thread starter z970
- Start date
- Sort by reaction score
What tests do you do to validate the changes you make to preference values and/or introduce new user prefs?
Because specific loading times and other benchmarked number values can often vary between tests due to factors outside PowerUOC's control, testing is mainly conducted on the basis of simply feeling a difference. If a change does not noticeably improve the all-around experience, or impacts it in a negative way, it does not get implemented.
This ensures that the end user will benefit, as it's already been confirmed prior to release that there is a noticeable, self-evident difference between revisions in realistic scenarios (oftentimes confirmed on multiple machines running different OSes).
Usually, it's a four-part process that goes as follows:
1. Apply change, save, restart session.
2. Stress test with various resource-heavy sites like Daily Mail, CNET, CNN, or many high-resolution images onscreen at once via a search engine.
3. Pay close attention to site loading speeds, text display immediacy, image rendering times, scrolling smoothness... etc.
4. If there is even a small positive improvement, the value gets logged and added to the master template.
Take this FWIW, from someone who hasn't yet tried PowerUOC (sadly, my daily driving of late requires more modern equipment
), but... well, that's all very subjective.As such, it seems to explain the differing experiences of those who've tried it. Of course, subjectivity is perfectly valid - no one would care to apply any set of tweaks that didn't make a meaningful difference for them. Back when @eyoungren's tweaks were the only game in town - and when I was using my PPC Macs regularly - I gave them a shot. They subjectively, considerably, improved my TenFourFox experience. I didn't require benchmarks (supposedly objective, though not always relevant) to prove to myself that there was a difference; I could see and "feel" it. That was good enough for me, and it's probably good enough for those who've benefited from your TFF tweaks and all the other things you've produced.
But, when you brand your efforts with hyperbolic terms like "ultimate", etc.; and introduce them with countdowns and other sensationalism, people rightly have high expectations. And, when the product doesn't meet those subjective expectations, they rightly look for objective means to quantify your claims. But there aren't any that can satisfy, because everyone's experience and needs are subjective, just like the standards by which you gauge the product.
I'm happy that it works so well for you, and clearly, for many others who've tried it. I look forward to trying it myself, because new stuff for our old hardware is always cool, whether it makes a difference for me personally or not. So, thanks for what you're doing, I really appreciate your efforts and others should as well. It's just that the "marketing" - for lack of a better term - is a bit over the top.
Again, FWIW.
Marketing and promotion will always have a net negative and detrimental outcome if the product falls way short of the promise. I simply don't understand the need for branding and self promotion in a community like this - a contribution can stand by it's own merits and normally respect is derived on those grounds.
Conversely there is a fun, tongue in cheek element to creating a buzz and giving something you've created an identity.
Take this FWIW, from someone who hasn't yet tried PowerUOC (sadly, my daily driving of late requires more modern equipment
As such, it seems to explain the differing experiences of those who've tried it. Of course, subjectivity is perfectly valid - no one would care to apply any set of tweaks that didn't make a meaningful difference for them. Back when @eyoungren's tweaks were the only game in town - and when I was using my PPC Macs regularly - I gave them a shot. They subjectively, considerably, improved my TenFourFox experience. I didn't require benchmarks (supposedly objective, though not always relevant) to prove to myself that there was a difference; I could see and "feel" it. That was good enough for me, and it's probably good enough for those who've benefited from your TFF tweaks and all the other things you've produced.
But, when you brand your efforts with hyperbolic terms like "ultimate", etc.; and introduce them with countdowns and other sensationalism, people rightly have high expectations. And, when the product doesn't meet those subjective expectations, they rightly look for objective means to quantify your claims. But there aren't any that can satisfy, because everyone's experience and needs are subjective, just like the standards by which you gauge the product.
I'm happy that it works so well for you, and clearly, for many others who've tried it. I look forward to trying it myself, because new stuff for our old hardware is always cool, whether it makes a difference for me personally or not. So, thanks for what you're doing, I really appreciate your efforts and others should as well. It's just that the "marketing" - for lack of a better term - is a bit over the top.
Again, FWIW.
Thanks for the input, RD. Much appreciated.
It was only ever branded as "ultimate" and whatnot on the relative basis that I really couldn't imagine any other way to improve stock TFF (aside from a different browser) so exponentially, at the very least specifically for individual configurations of the user's. As in, this is as good as it will get (for this browser on this OS on this machine). - It seems that I initially tried it, got overly excited at the results, and shared it with the same shortsighted assumption that the experience would duplicate itself 1:1 on all configurations...
But now that opposing experiences are coming to light, I can take steps to dial everything down to less obnoxious levels, to what it's really worth, so that perhaps instead of the user feeling disappointed, they may feel pleasantly surprised.
A chance to improve is always valuable. And for that, I'm thankful.
To be fair, that's easier to achieve when the contribution's thread is stickied and is assured to be seen by people, otherwise it just falls down into the abyss and is ignored, then forgotten, which has happened before. On the flipside, I thought this was such a benefit to have, that couldn't have been let happen... So perhaps it was overcompensated, in that sense.
This is true.
Last edited:
In good branding language (my first career before university was in this area), it should be abundantly clear what novel initialism or acronym represents.
At this point, and after a few searches here and elsewhere, I’m only certain that the “U” in “UOC” is “Ultimate”. I’ve no idea the rest.
"Ultra One Core", to symbolize the power of single-core CPUs.
...Hey, I'm not the one that came up with it. Besides "somethingAccelerator", I didn't have much better, either.
Hell, only well into development was I ever made aware just what it did stand for in the first place...
Thanks.
Yeah, I literally thought it was something like “Ultimate Operating C…” — where nothing was coming to with for the “C”.
But that's just the nature of the forum - I like to think I've made useful contributions over the years - things that I've spent time on but they all slide down the page into obscurity. If you stick around long enough you'll witness new members join and maybe at some point they will post their new whizzbang prefs file for TFF and you will have gone full circle...
Gave it another go on my 1.33 Powerbook - first trying the Antutu HTML5 benchmark which gives rather bizarre results with UOC coming bottom below stock and my own tweaks.
However, I can verify now the latest UOC definitely improves Youtube playback without a doubt and although I wouldn't dream of playing HTML5 Youtube video on a portable with the CPU blasting away at 100% it will obviously reap benefits on multi G4/G5s.
UOC
Stock
Personal
However, I can verify now the latest UOC definitely improves Youtube playback without a doubt and although I wouldn't dream of playing HTML5 Youtube video on a portable with the CPU blasting away at 100% it will obviously reap benefits on multi G4/G5s.
UOC
Stock
Personal
That's the thing... Useful contributions should never be forgotten. I simply don't conform to the concept.
That's where you come in, remember?
[automerge]1575306391[/automerge]
PowerUOC V is out, in correspondence with the release of TFF FPR17.
Alongside the usual improvements, PowerUOC now officially supports vanilla Firefox (including Quantum), Arctic Fox*, and Pale Moon on Windows, Linux, and higher OS X. Though, support for 10.8+ and Windows 10 is offered as is.
Scrolling should be slightly faster now, too...
It is also recommended to delete your profile folder at each release if possible, as sizeable speed boosts have been previously evidenced with fresh folders.
That aside, the PowerUOC Project will be taking a quick break for the remainder of the year. Afterward, version VI will be a new type of release further optimizing the patch mainly by removing components instead of adding more in, which is so far expected to exhibit a much greater boost than previously seen.
See you next decade!
*The lines user_pref("layers.tile-height", 1); and user_pref("layers.tile-width", 1); must first be removed.
Last edited:
At a glance, media.webm.enabled, false forces the video source to stream hardware-accelerated MP4 video instead of the software-rendered VP9.
media.gpu-process-decoder, true
media.hardware-video-decoding.failed, false
media.hardware-video-decoding.force-enabled, true
...And all of the above forces the acquired video stream to be offloaded to the GPU for playback.
Those are probably the heavy hitters.
media.gpu-process-decoder, true
media.hardware-video-decoding.failed, false
media.hardware-video-decoding.force-enabled, true
...And all of the above forces the acquired video stream to be offloaded to the GPU for playback.
Those are probably the heavy hitters.
I don't think it can be these in isolation, half of these I have in my own tweaks and I tried the other two to no effect. Also, in theory there should be no hardware acceleration available - certainly not in regard to h264 decoding.
I need to do some tests with the mp4 enabler turned off - maybe that's getting massaged by the prefs somehow?
Do you have improvement with any of them enabled, the ones in your prefs?
...In any case, I'm afraid this will require further investigation.
I stand corrected, it does appear to be the media.webm.enabled (false) toggle that gives the speed up, despite the mp4 enabler already taking care of video.
Turning the mp4 enabler off and reverting to VP9 certainly degrades performance.
@Dronecatcher Just to clarify, can you confirm that media.hardware-video-decoding.failed, false and media.hardware-video-decoding.force-enabled, true do not bring any benefit on their own when playing video?
Thanks.
Thanks.
Tested on my 1.33 Powerbook and the DLSD I can't see any difference with values reversed.
A Celeron N4000 is actually faster than a C2D E8400. Also, I have a HP Core 2 Duo PC laying around will run circles around a $50 G5 when it comes to daily usage, or even any iMac before Late 2009. I got all that for $0. It was free.
Rock solid or not, it's outdated.
Guys, when you say "G5", please be more specific.
Also, i3 has all the Intel ME troubles, don't forget! Even Core 2 Duo, which (often? always?) has removable Intel ME (unlike any Intel processor that came later, like the i3), it doesn't work with a lot of OSes once it's removed.
On PowerUOC and the prefs file, I can't comment yet. But I can already say the important thing here is that these efforts are taking place at all, and that's what matters. Eyoungren's efforts, UOC efforts, UOC PPC efforts etc.... It's good that all this is coming and being discussed.
If you're worried about that kinda stuff but you still want a PC you can use every day, get a AMD A10-6800K. It's the best chip they made without PSP and it is faster than anything in the LGA775 socket.Fair enough - a 3 GHz Core 2 Quad Extreme on a 945 chipset has no ME and still packs a punch. But let's go back on topic
Good to see someone getting my point.
Good point, however, the problem with the A10-6800K is that it's based on the Bulldozer architecture which isn't exactly known for efficiency or outstanding performance.
Compared to PowerPC anything, it is quite efficient relative to its compute power.