You and I have the same reasons for running these things. For me, it's just about using interesting hardware that you can't really find elsewhere. Plus running code on big endian CPUs helps expose bugs I wouldn't find if I was Intel only.
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.
You Can/Can't Still Use PowerPC Macs in 2020!
- Thread starter Dronecatcher
- Start date
- Sort by reaction score
- Status
You and I have the same reasons for running these things. For me, it's just about using interesting hardware that you can't really find elsewhere. Plus running code on big endian CPUs helps expose bugs I wouldn't find if I was Intel only.
Wanted to circle back on my progress with the Mini I purchased. AS a refresher it is an early 2009 2.0GHz Core 2 Duo model, 1GB of RAM, 120GB SATA HD, and GeForce 9400M GPU. It came with the power adapter and mini DVI to VGA adapter. Total cost, including shipping, was $60.00. I purchased this unit for a couple of reasons:
Handbrake transcode of Scooby Doo Season One DVD, Handbrake 0.9.4, normal preset, PowerMac was running OS X 10.5.8, the Mini was running OS X 10.6.8 (I wanted to use Leopard but the Mini requires 10.5.6 to install and all I have on hand is 10.5.0). To transcode the entire DVD to MP4:
For the web browsing test I decided to install El Capitan. Unfortunately El Capitan requires 2GB of RAM and the Mini had only 1GB. Since I had a spare 1GB module laying around I decided to install it. The process was rather straightforward but not as easy as previous Mini's nor the PowerMac. Some might note that the initial $60.00 did not include this upgrade and they would be right. To which I'll respond that I purchase 4GB of RAM for $6.00, including shipping. So let's adjust the cost of the Mini to $66.00, still quite reasonable.
After (finally) getting El Capitan installed I have been using the Mini to browse the Internet. First with Safari and then with Firefox (latest version). All I can say is there is no comparison between the two. Even though this is a subjective measure for which I cannot provide metrics all I can say is the difference is night and day. No longer does the system hang when I scroll down the page. Pages load quickly with little delay. Browsing on the Mini is slightly slower than using a current Mac. Slight delays can be observed but browsing is not an exercise in patience and frustration as it is with the PowerMac. With the Mini I just started the browser and began browsing. No optimizations were needed.
So what purpose do I have for posting all of this? I can tell you what it is not: It is not to try and convince anyone here that PPC systems are unusable. Nor is it intended to convince anyone here their decision to use one is in error and they're a fool for doing so. This is a PPC forum and everyone here has decided, for whatever reasons they have, to use one.
Now that I've explained what the purpose was for not posting here's why I did. The YouTuber in the first video of the OP stated that a low cost Intel system would be a preferable system to a PPC based system. This statement was intended for the average user and not for hobbyists, enthusiasts, or those with specific needs / constraints which would prohibit using an Intel based system.
Some final thoughts:
- It can run OS X El Capitan where as earlier Intel models were limited to OS X Lion
- It was the lowest cost, complete system I found on Ebay at the time. I could have purchased just the Mini for lower cost but not having a power adapter was a non-starter as I'd have had to buy that too negating the lower cost.
Handbrake transcode of Scooby Doo Season One DVD, Handbrake 0.9.4, normal preset, PowerMac was running OS X 10.5.8, the Mini was running OS X 10.6.8 (I wanted to use Leopard but the Mini requires 10.5.6 to install and all I have on hand is 10.5.0). To transcode the entire DVD to MP4:
- PowerMac: 14 hours and 11 minutes at an average frame rate of 5.4 fps
- Mini: 2 hours and 52 minutes at an average frame rate of 26.9 fps
For the web browsing test I decided to install El Capitan. Unfortunately El Capitan requires 2GB of RAM and the Mini had only 1GB. Since I had a spare 1GB module laying around I decided to install it. The process was rather straightforward but not as easy as previous Mini's nor the PowerMac. Some might note that the initial $60.00 did not include this upgrade and they would be right. To which I'll respond that I purchase 4GB of RAM for $6.00, including shipping. So let's adjust the cost of the Mini to $66.00, still quite reasonable.
After (finally) getting El Capitan installed I have been using the Mini to browse the Internet. First with Safari and then with Firefox (latest version). All I can say is there is no comparison between the two. Even though this is a subjective measure for which I cannot provide metrics all I can say is the difference is night and day. No longer does the system hang when I scroll down the page. Pages load quickly with little delay. Browsing on the Mini is slightly slower than using a current Mac. Slight delays can be observed but browsing is not an exercise in patience and frustration as it is with the PowerMac. With the Mini I just started the browser and began browsing. No optimizations were needed.
So what purpose do I have for posting all of this? I can tell you what it is not: It is not to try and convince anyone here that PPC systems are unusable. Nor is it intended to convince anyone here their decision to use one is in error and they're a fool for doing so. This is a PPC forum and everyone here has decided, for whatever reasons they have, to use one.
Now that I've explained what the purpose was for not posting here's why I did. The YouTuber in the first video of the OP stated that a low cost Intel system would be a preferable system to a PPC based system. This statement was intended for the average user and not for hobbyists, enthusiasts, or those with specific needs / constraints which would prohibit using an Intel based system.
Some final thoughts:
- This should not be taken to mean there are no benefits to PPC systems. Especially the higher end G5 systems.
- The only optimization made were to use Eric's preferences file for TenFourFox (no cache entry). Without these optimizations then the browsing experience on the PowerMac is even worse.
- I used TenFourFox on the PowerMac as it seems to be the most compatible / supported browser.
- I made no attempt to optimize the Handbrake transcode as my train of thought was to do what the "average" user would do. If someone has recommendations to decrease the PowerMac Handbrake numbers I'd be happy to make them. However they need to be done in the context of what can be done with version 0.9.4 which is the final version that supports PPC.
- There is a notable difference in performance between Snow Leopard and El Capitan on the Mini. This might be due to the 2GB RAM installation (which is why I purchased the 4GB upgrade).
- Installing El Capitan requires one set the date back to some time in 2017 in order to install it. This is due to an expired certificate.
- Oddly the processor usage on the PowerMac is at a constant 100% when browsing the web. On the Mini the threads routinely went to idle. I can't imagine the Core 2 Duo is so much more efficient than the G5 that it idles most of the time. This leads me to suspect software is the cause. Specifically TenFourFox. Anyone have any thoughts on this?
CPU usage was probably higher on the Powermac while web browsing, because everything on screen is rendered via software rendering. The Intel machine is probably leveraging the GPU.
Is it possible to utilize GPU rendering on the PowerMac? If I could address this issue I would be really happy.
Z970's prefs.js has the proper about:config bools toggled to enable HW accel in TFF, but I don't know if you'll notice much of a difference with the GeForce 5200.
Layers.accelerate is the most important one, I believe. There are several others that I can't remember verbatim off of the top of my head.. There is a stickied post with a download link for his prefs file. Just use that.
Just my 2 cents, but TFF is not a good browser to compare to anything. Under ppc Linux I'm using what's comparable to Firefox 68 on a Mac mini g4 and it runs circles around TFF on osx on the same machine. I can't explain it, but it is what it is.
Last edited:
Are you using uMatrix or NoScript (one or the other, no need for both - although my preference is uMatrix if not using a DNS address for systemwide ad blocking)? Even with my 1Ghz 17" PB I can browse 'reasonably' as long as those are installed. It's near impossible without them.
Last edited:
If a machine's OS or graphics card is explicitly unsupported by Firefox, no additional preference will be able to change the state of hardware acceleration. On TFF, foxPEP will improve the experience by a marked amount (as every preference included has been vetted for nothing less than maximum potency), but it does not do this by enabling hardware acceleration, as it is unable to. Usually, this can be solved by using another browser that does not have any problems with the GPU drivers and will happily leverage what is already there. But for TFF's case, Tiger and Leopard are the only OSes that it will run on, and unluckily for it... Either both operating systems do not support GPU rendering for Web browsers, or TenFourFox was not compiled with support for hardware acceleration.
I do not know exactly if WebKit leverages the GPU or not, but I suspect it's the former as @wicknix's IceWeasel and Arctic Fox do not support hardware acceleration either, which were both individually compiled browsers running on the same operating systems. I also suspect OS X (10.4, 10.5) does not support Web hardware acceleration because TFF blacklists parts of WebRTC, WebGL, hardware video decoding, and others via preference (thanks to the way it was compiled), but it does not blacklist WebGL or hardware acceleration on their own, which heavily suggests that the operating system's lack of support for them does the rest all by itself, as they are clearly disabled as per about:support.
Because of this, TenFourFox is unfortunately only really half-supported by foxPEP, as the browser for several reasons isn't able to follow all of the script's directions, which is yet another reason for Linux on PowerPC. These machines are gimped out of the box, and people then unfairly judge that they are slow because they are never given the chance to glimpse their true power on Web 3.0.
A frustrating catch 22 situation, indeed...
Last edited:
I am not. I have limited optimizations to utilizing your preferences file. I might give this a try to see what impact it has.
[automerge]1584101430[/automerge]
As previously mentioned in my post I am using TenFourFox on the PPC because:
I used TenFourFox on the PowerMac as it seems to be the most compatible / supported browser.
I tried LWK and found its performance acceptable. However its stability, or lack of, was not. Because I was using a Firefox variant on the PPC I thought it appropriate to try Firefox on the Mini. The idea being to use the same general code base on both platforms (unfortunately TenFourFox is not available for Intel systems). The performance of Firefox on the Mini easily bested Firefox on the G5. It wasn't a slight difference, it was very obvious.
Last edited:
@defjam I did a similar video transcoding benchmark here - others chipped in with Intel comparisons:
forums.macrumors.com
DIY Benchmark
A recent thread discussed the accuracy of Geekbench and methods by which to establish the performance differences between machines. With that in mind, I put together this test: A 5 second 1080P MPEG4 video clip, converted in MPEG Streamclip to a 640X360 h.264 file. All files are bundled...
forums.macrumors.com
Last edited:
Try https://sourceforge.net/projects/tenfourfox/files/unstable/contrib/ - I use TenFourFox-FPR14-i386-Leopard on my Snow Leopard machines
It's regarded as old being FPR14 but still more up to date than old Firefox.
Cheers
Hugh
I assume you'd like me to perform this benchmark with the Mini? I did so and it completed the transcode in 44 seconds.@defjam I did a similar video transcoding benchmark here - others chipped in with Intel comparisons:
DIY Benchmark
A recent thread discussed the accuracy of Geekbench and methods by which to establish the performance differences between machines. With that in mind, I put together this test: A 5 second 1080P MPEG4 video clip, converted in MPEG Streamclip to a 640X360 h.264 file. All files are bundled...
[automerge]1584109298[/automerge]
I appreciate the information, as well as that provided by other members in this forum. However I am attempting to keep things simple. I want to attempt to replicate a situation of what an "average" user would be willing to do in order to use their system. There have been a lot of good suggestions on how to optimize a PPC system to improve performance but the majority of them are, IMO, beyond what an average user would be willing to do. Contrast this to the experience on the Mini which requires no optimization in order to have what is, IMO, an acceptable browsing experience. I'm not sure why browsing on the PPC is so slow as I cannot see the Core 2 Duo used in a 2009 Mini as being so much faster than the G5 as to explain the difference.Try https://sourceforge.net/projects/tenfourfox/files/unstable/contrib/ - I use TenFourFox-FPR14-i386-Leopard on my Snow Leopard machines
It's regarded as old being FPR14 but still more up to date than old Firefox.
Cheers
Last edited:
Interesting, that's half the time of my Mac Pro despite your Mini having roughly 50% of the CPU horsepower - you did use the bundled MPEG Streamclip? Just goes to show how much impact OS/software versions have.
Regarding web browsing, I've found, inaddition to using a tweaked config file, the greatest improvement is loading the oldest useragent you can't utilise without breaking the site - the speed difference is immense when you roll back to a more 'basic' page (I tend to use IE 8 UA.)
You can hack Mountain Lion onto GMA 950/X3100 machines but that's it.
El Capitan definitely needs at least 4 GB RAM and an SSD to be enjoyable. Case in point: I have a white MacBook with 2 GB RAM and a "normal" HDD. The 2.4 GHz CPU underclocks to 1 GHz because the battery is gone. Performance is acceptable on Snow Leopard despite all this, but Lion and later versions are much more sluggish. Yosemite is just painful, and that's with a Radeon 6870 bolted on.
Last edited:
I did. I downloaded the ZIP file and followed the instructions contained within. Here's the log output:
Code:
File: test.mp4
Friday, 08:14:01: Export Stream started
Friday, 08:14:08: Encoding
Friday, 08:14:16: Encoding
Friday, 08:14:42: Encoding
Data rate 4.79 Mbps (4.67 video, 0.13 audio)
Friday, 08:14:45: Movie completedUPDATE: Here are the Mac Pro results (2010 Mac Pro, single 6-core, 12-thread 3.46GHz Xeon processor [X5690], 32GB RAM, Radeon HD 5770 GPU): 17 seconds.
Code:
File: test.mp4
Friday, 08:50:02: Export Stream started
Friday, 08:50:05: Encoding
Friday, 08:50:08: Encoding
Friday, 08:50:18: Encoding
Data rate 4.79 Mbps (4.67 video, 0.13 audio)
Friday, 08:50:19: Movie completedUPDATE 2: Here is the benchmark on the PowerMac G5: 7 minutes and 3 seconds
Code:
File: test.mp4
Friday, 10:52:03: Export Stream started
Friday, 10:52:17: Encoding
Friday, 10:52:37: Encoding
Friday, 10:53:16: Encoding
Friday, 10:53:59: Encoding
Friday, 10:54:42: Encoding
Friday, 10:55:25: Encoding
Friday, 10:56:08: Encoding
Friday, 10:56:51: Encoding
Friday, 10:57:34: Encoding
Friday, 10:58:17: Encoding
Friday, 10:59:00: Encoding
Data rate 4.79 Mbps (4.66 video, 0.13 audio)
Friday, 10:59:06: Movie completed
Last edited:
As has been stated, the fact that Firefox can leverage the GPU on the Mini, whereas TenFourFox cannot on the G5, probably explains this difference to a substantial extent. Even my 2010 11in MacBook Air with its 1.4 GHz Core 2 Duo still just about manages to deliver an acceptable browsing experience (El Capitan + Safari + uBlock) thanks to the GeForce 320M.
I understand this and it seems like a very likely reason. However, z970mp has a post which talks about GPU acceleration. I don't know what the reason is, all I can say is that when it comes to browsing the Internet the Mini is hands down the winner wrt performance. If GPU acceleration is available on my G5 by upgrading the GPU I'd consider doing that if the card were inexpensive and that it is passively cooled. If someone has a recommendation I'd be willing to give it a try.
Assuming this is correct, the only way to have a GPU-accelerated Firefox (derivative) on a G5 is to... switch to Linux.
@wicknix' experience corroborates this theory:
That's interesting because whilst I certainly agree with you on the RAM requirements (I ended up installing 6GB), El Capitan has been very enjoyable for me on my MacBook 5,2 with a spinner.
[automerge]1584112242[/automerge]
Which distro(s) would you recommend for a G5?
That's interesting because whilst I certainly agree with you on the RAM requirements (I ended up installing 6GB), El Capitan has been very enjoyable for me on my MacBook 5,2 with a spinner.
Thanks for the data - perhaps I'm just hopelessly spoiled by the SSD experience as I've been using them as OS drives since 2010.
Although I haven't tried them yet (shame on me), @wicknix' Lubuntu remixes sound just great.
My guess is it's down to the javascript that OSX Tiger/Leopard gets bogged down with rather than GPU assisted acceleration, hence my observation about older user agents speeding things up by presumably not loading new script driven features.
I've yet to see any GPU acceleration in Linux browsing on PPC - and I've been trying for years, admittedly on Powerbooks not desktops. And when I've done benchmarks to establish any merit to config file acceleration tweaks I found none.
Javascript is the main issue that brings PowerPC to it's knees during web browsing. A lot of JS is used by third party analytics and advertising. Even MR, which has ads, is heavy on JS despite it's minimal theme.
Using either of these addons allows you to stop that JS cold from places you don't want it coming from. Which gives the CPU time to process all the stuff you do want it to process.
You can use just my prefs but without stopping the Javascript by using either addon, it doesn't really matter.
And it would explain why you have so many issues with T4Fx.
- Status