I web browse with a IBM Thinkpad 760 (the original 233mmx version) with 232mb of ram (or something like that) running 2k and the other day played YouTube on a P4 524 w/2gb ddr2-400 and a pci Geforce 6200 running under xp. 360p worked great and 480p-720p stuttered using the intel 900 graphics. Upgrading to the 6200 made 480p work fine and 720p then worked with minor stuttering. I guess IMHO with "fast" hardware so plentiful I just do not see the real reason to wait around for 2-5min for a web page to load.

m1so wrote:

You don't need Flash, HD video, latest Chrome or any such crap to consider a PC web capable. Anything that is actually useful ca […]
Show full quote

You don't need Flash, HD video, latest Chrome or any such crap to consider a PC web capable. Anything that is actually useful can be viewed with such things turned off.

I think also that "CPU matters more" because a.) people are spoiled and consider minute delays "unusable", b.) people are noticing the effect of CPU speed because cities and rich countries are getting 100 Mbps and faster connections. I remember browsing the net in 2006 with a 1 Ghz Celeron and a 128 kbps EDGE modem, and the modem had a far, far greater effect on the speed than the relatively slow even then CPU.

Core 2 Duo "struggles" with modern webpages? Seriously? My parents use a "slow" C2D (late 2006 iMac) for daily use and it works perfectly well.

Up until half a year ago we used an old P4 3.2 Ghz to work with an online database. It was a bit slow, but usable, even through it is a piece of badly programmed crap that is not really quick even on an i7. But it was usable on a P4. Too many people make the fallacy of extrapolating "oh, I have to wait a second on my i7, that surely means I'd wait a trillion seconds on an old PC!". A 5x faster PC does not mean 5x faster software.

When we're talking web on a K6, we're not talking about autoplay videos, "HD" etc. I remember regular webpages in 2004 worked okay with JPGs even on my uncle's 75 Mhz Pentium and 2004 top CPUs vs a P75 was a FAR larger gap than Athlon XP or even K6 vs a modern CPU.

Agreed. I actually have a P4EE running Vista that handles "the web" just fine - it doesn't like HD YouTube, but that's an inefficiency on YouTube's part (I still remember watching 720p video content on an AthlonXP many years ago, and while it did heavily load the CPU, it was playable). It really depends on what you're doing online - primarily text-based sites (e.g. Vogons, Project Guttenberg) will still be "light" and still work on hardware that could run them ten years ago. But modern "social media everywhere" where there's 30 or 40 trackers, a non-stop barrage of Flash/Silverlight/HTML5/whatever multimedia content, tons of huge graphics, etc web will grind without hardware acceleration. Blocking that stuff can improve page load times even on very robust hardware, because it's just less junk to deal with.

To the original question though, I'd still go with the AthlonXP 3200+ if you already have it. If not, the XP 2000+. That's just me though - I preference using the best of what I have available when possible.

If you're looking for something exotic and K7, why not a dual-socket machine? 😀

devius wrote:

Because more and more websites are owned by big companies that want to gather as much information about their users as possible, […]
Show full quote

mr_bigmouth_502 wrote:

Modern websites are horrendously inefficient, and it makes me wonder why they are coded in such a way with the proliferation of lower-power devices like tablets and smartphones.

Because more and more websites are owned by big companies that want to gather as much information about their users as possible, and to make as much money as possible, which leads to websites being infested with tracking code, animated ads and ads that also do tracking on their own. And I'm not talking about 1 or 2 ads, but tens or even hundreds of HTTP requests to various tracking services and data gathering code that all runs in your browser, using up memory and CPU power like there's no tomorrow. All of this when you're trying to view a single page.

mr_bigmouth_502 wrote:

If anything, I think web developers should be trying harder to make their sites work on low-power devices, rather than relying on the brute strength of modern x86 desktop CPUs.

It's not like they don't try, but there's just so much that can be done when the higher-ups demand all this tracking and ad code. Seriously, try some of the ad and script blocker add-ons that exist and you'll be amazed by just how much unnecessary crap is being thrown at us. Of course there's always the poorly coded and ineficient website that uses a ton of javascript libraries for even the simplest of things, but even then it's nothing compared to the rest of the stuff I mentioned.

For an old computer browsing the web it's vital that these ad and script blockers be installed. It will make for a perfectly good browsing experience in most sites.

PS: And forget about Flash and streaming video. That is really taxing on its own.

I block most of that crap whenever I can on my main PC, but there aren't any mobile browsers that support all the addons I use. 🙁

devius wrote:

Because more and more websites are owned by big companies that want to gather as much information about their users as possible, […]
Show full quote

mr_bigmouth_502 wrote:

Modern websites are horrendously inefficient, and it makes me wonder why they are coded in such a way with the proliferation of lower-power devices like tablets and smartphones.

Because more and more websites are owned by big companies that want to gather as much information about their users as possible, and to make as much money as possible, which leads to websites being infested with tracking code, animated ads and ads that also do tracking on their own. And I'm not talking about 1 or 2 ads, but tens or even hundreds of HTTP requests to various tracking services and data gathering code that all runs in your browser, using up memory and CPU power like there's no tomorrow. All of this when you're trying to view a single page.

mr_bigmouth_502 wrote:

If anything, I think web developers should be trying harder to make their sites work on low-power devices, rather than relying on the brute strength of modern x86 desktop CPUs.

It's not like they don't try, but there's just so much that can be done when the higher-ups demand all this tracking and ad code. Seriously, try some of the ad and script blocker add-ons that exist and you'll be amazed by just how much unnecessary crap is being thrown at us. Of course there's always the poorly coded and ineficient website that uses a ton of javascript libraries for even the simplest of things, but even then it's nothing compared to the rest of the stuff I mentioned.

For an old computer browsing the web it's vital that these ad and script blockers be installed. It will make for a perfectly good browsing experience in most sites.

PS: And forget about Flash and streaming video. That is really taxing on its own.

This post didn't get enough love. I am one of the web developers that program for a big company, and if another inefficient or (in my eyes) unnecessary addition gets requested from me, there's not much I can do about it.

Luckily, with the ongoing trend to mobile devices, and therefore making web pages responsive, i.e. working on every screen size, we will see lighter web pages in the future.

I'm not the biggest fan of ad blockers, because a lot of sites rely on the ad revenue, but for an older PC I can see why they are necessary.

oerk wrote:

devius wrote:

This post didn't get enough love. I am one of the web developers that program for a big company, and if another inefficient or (in my eyes) unnecessary addition gets requested from me, there's not much I can do about it.

Luckily, with the ongoing trend to mobile devices, and therefore making web pages responsive, i.e. working on every screen size, we will see lighter web pages in the future.

I'm not the biggest fan of ad blockers, because a lot of sites rely on the ad revenue, but for an older PC I can see why they are necessary.

There are a few tests arround that claim that these AdBlock (due to the way they work) actually increases the browser's memory usage. Using a "hosts" file should be more eficiente. And yeah, I find the concept of blocking advertisement to be unfair to websites, so I only use the hosts file on my really retro PCs, as browsing is almost impossible without it.

obobskivich wrote:

I still remember watching 720p video content on an AthlonXP many years ago, and while it did heavily load the CPU, it was playable

Uninstall newest flash player, then install version 10.0.45 and it will be playable again. From this archive: http://fpdownload.macromedia.com/get/flashpla … p10_archive.zip
And config firefox in about:config page to do not notify about outdated plugins.

Meanwhile I am using another system based on the 3.2v 233Mhz K6 with 256Mb. If sites are like this not necessary a bad browsing experience but obviously even Opera 9.x is quiet heavy for this.

zyga64 wrote:

Uninstall newest flash player, then install version 10.0.45 and it will be playable again. From this archive: http://fpdownload.macromedia.com/get/flashpla … p10_archive.zip
And config firefox in about:config page to do not notify about outdated plugins.

Gotta try this with my Athlon 64. Do you know if I should I dial back to Firefox 3.x or would this work fine on the latest versions of Firefox just as well?

Works for sure with Firefox 12 (last compatible with windows 2000). If I remember correctly, no problems also with version 3.6.28.

With YouTube running on HTML5 now, you don't even really need Flash IMO. If you do need Flash and want hardware acceleration though, an older computer can get a HUGE boost by giving it an GeForce 8xxx series GPU, which is the earliest GPU to support modern Flash hardware acceleration.

fyy wrote:

With YouTube running on HTML5 now, you don't even really need Flash IMO. If you do need Flash and want hardware acceleration though, an older computer can get a HUGE boost by giving it an GeForce 8xxx series GPU, which is the earliest GPU to support modern Flash hardware acceleration.

Just to point out, the original GeForce 8800 (G80 based) do not have complete acceleration features. The GeForce 8600 was the first member series to implement complete support, followed by the 8400/8500, and then the G9x based cards (which includes the revised GeForce 8800GS/GT/GTS). IMO I'd go with the 8600 or some other mid-range (or newer entry-level, like GT 610) card if the whole point is just modern multimedia support - no point in the extra power/heat/noise of a high-end gaming card if it's never going to game. 😊

fyy wrote:

With YouTube running on HTML5 now, you don't even really need Flash IMO. If you do need Flash and want hardware acceleration though, an older computer can get a HUGE boost by giving it an GeForce 8xxx series GPU, which is the earliest GPU to support modern Flash hardware acceleration.

On my Athlon 64 HTML5 is just as bad as Flash, if not worse.

obobskivich wrote:

fyy wrote:

With YouTube running on HTML5 now, you don't even really need Flash IMO. If you do need Flash and want hardware acceleration though, an older computer can get a HUGE boost by giving it an GeForce 8xxx series GPU, which is the earliest GPU to support modern Flash hardware acceleration.

Just to point out, the original GeForce 8800 (G80 based) do not have complete acceleration features. The GeForce 8600 was the first member series to implement complete support, followed by the 8400/8500, and then the G9x based cards (which includes the revised GeForce 8800GS/GT/GTS). IMO I'd go with the 8600 or some other mid-range (or newer entry-level, like GT 610) card if the whole point is just modern multimedia support - no point in the extra power/heat/noise of a high-end gaming card if it's never going to game. 😊

Good to know. It's really a matter of which version of PureVideo the card itself has for Nvidia and which version of UVD the card has for AMD cards. UVD2 and higher is where you get hardware acceleration for modern Flash, which is Radeon HD 4xxxx series or higher and it's PureVideo 2 and higher for GeForce cards, which the newer GeForce 8xxx cards have you said.

This issue annoyed the shit outta me for the longest time wondering why some video cards had full acceleration in Flash while others only had partial.

So even a Pentium 4 with an 8400gs for example should be able to watch 1080p Flash videos quite easily in theory.

fyy wrote:

Good to know. It's really a matter of which version of PureVideo the card itself has for Nvidia and which version of UVD the card has for AMD cards. UVD2 and higher is where you get hardware acceleration for modern Flash, which is Radeon HD 4xxxx series or higher and it's PureVideo 2 and higher for GeForce cards, which the newer GeForce 8xxx cards have you said.

This issue annoyed the shit outta me for the longest time wondering why some video cards had full acceleration in Flash while others only had partial.

So even a Pentium 4 with an 8400gs for example should be able to watch 1080p Flash videos quite easily in theory.

Worth also noting: on the "bridged" GeForce 6/7 cards, video accel is generally disabled/unsupported per nVidia specs. I'm not sure if the 8400GS PCI would have a similar limitation or not. PCIe of course the 8400-8600 or G9x+ will be solid (and note that GeForce 8 has both G8x and G9x parts - the G9x parts have the improved decode features like the 8600). Regarding 1080p, my Quadro FX 1700 (GF8600 based) has no troubles with YT 1080p. 😀

Give this a look as far as performance goes:
http://www.anandtech.com/show/2220

Note the 8800GTX really doesn't fare much differently than the 7950GT, despite being considerably faster in terms of gaming performance. Also note the lack of VC-1 support even for the 8600; G98 and beyond include full VC-1 decoding. If you just need to support Flash and more common formats, the 8400/8600 would be fine, but if you're also wanting full Blu-ray support and whatnot, I'd probably go with a newer card like GT 610.

ATi seems to be much simpler afaik - HD 4000/R700 series and above should do everything (including VC-1). 😀

Radeon HD 3200 & HD 3300 IGPs (AMD 780/790 chipsets) support UVD 2.

obobskivich wrote:

Also note the lack of VC-1 support even for the 8600; G98 and beyond include full VC-1 decoding. If you just need to support Flash and more common formats, the 8400/8600 would be fine, but if you're also wanting full Blu-ray support and whatnot, I'd probably go with a newer card like GT 610.

ATi seems to be much simpler afaik - HD 4000/R700 series and above should do everything (including VC-1). 😀

Even though it's not fully supported, the 8600GT does give VC-1 a considerable boost, at least when used with LAV Video Decoder built into MPC-HC. Win7's built-in VC-1 decoder completely ignores the 8600's unofficial VC-1 support and falls back to software.

With the 8600GT and MPC-HC, a 28 mb/s VC-1 stream ripped from a BD only uses around 20-30% of a S939 X2-3800. With a G80 based 8800, that same VC-1 stream uses 80-95% of the CPU and the video occasionally falls behind the audio.

VC-1 seems to be harder to decode than AVC/H.264. Or perhaps the decoders just aren't as refined, since VC-1 never really caught on outside of Blu-Ray (new BD releases don't even use VC-1 anymore). In full software mode, the X2-3800 has no problem decoding 30 MB/s H.264 streams.

Standard Def Steve wrote:

Even though it's not fully supported, the 8600GT does give VC-1 a considerable boost, at least when used with LAV Video Decoder built into MPC-HC. Win7's built-in VC-1 decoder completely ignores the 8600's unofficial VC-1 support and falls back to software.

With the 8600GT and MPC-HC, a 28 mb/s VC-1 stream ripped from a BD only uses around 20-30% of a S939 X2-3800. With a G80 based 8800, that same VC-1 stream uses 80-95% of the CPU and the video occasionally falls behind the audio.

VC-1 seems to be harder to decode than AVC/H.264. Or perhaps the decoders just aren't as refined, since VC-1 never really caught on outside of Blu-Ray (new BD releases don't even use VC-1 anymore). In full software mode, the X2-3800 has no problem decoding 30 MB/s H.264 streams.

Just to note, the 8600 does not have "unofficial VC-1 support" - it has partial VC-1 support in hardware, just like the GeForce 7 and 8800 (and it actually has somewhat better functionality than the 7/8800 - it can do IDCT acceleration). It's not surprising that various applications completely ignore the feature however; VC-1 was more heavily used with HD-DVD than Blu-ray, and is probably most commonly found with Xbox360 content (as VC-1 is the Xbox360's primary video format). The huge performance delta you're seeing is probably as much to do with the load VC-1 places on a system as the lack of optimization/refinement in the player software.

Also note that GeForce GT 610 will completely handle any of these formats, and is very affordable even brand-new - if the system has PCIe and you need to handle modern HD content, I'd just go that route versus an older card like the 8600.

alexanrs wrote:

oerk wrote:

devius wrote:

This post didn't get enough love. I am one of the web developers that program for a big company, and if another inefficient or (in my eyes) unnecessary addition gets requested from me, there's not much I can do about it.

Luckily, with the ongoing trend to mobile devices, and therefore making web pages responsive, i.e. working on every screen size, we will see lighter web pages in the future.

I'm not the biggest fan of ad blockers, because a lot of sites rely on the ad revenue, but for an older PC I can see why they are necessary.

There are a few tests arround that claim that these AdBlock (due to the way they work) actually increases the browser's memory usage. Using a "hosts" file should be more eficiente. And yeah, I find the concept of blocking advertisement to be unfair to websites, so I only use the hosts file on my really retro PCs, as browsing is almost impossible without it.

This is what I've heard as well. I haven't looked into hosts file-based blocking on my PC, but I have a utility for it on my phone called AdAway. Found it on F-Droid, but you need a rooted phone to use it.

mr_bigmouth_502 wrote:

This is what I've heard as well. I haven't looked into hosts file-based blocking on my PC, but I have a utility for it on my phone called AdAway. Found it on F-Droid, but you need a rooted phone to use it.

http://winhelp2002.mvps.org/hosts.htm

No substantial performance hit overall, and on many junk-heavy pages a performance improvement is noticeable, because it isn't loading nearly as much junk (it's much less resources intensive for it to resolve a bunch of 127.0.0.1's than a bunch of images/graphics/banner ads/whatever). Running that coupled with Ghostery (which blocks basically all social media junk, among other things) and not allowing Flash to run by default makes a lot of media-nightmare sites run much smoother, even on nicer hardware, as well as making them easier to navigate. Blocking multimedia junk locally will generally always improve performance, but if it's handled through a proxy or other external service there can be a performance hit depending on the performance of the proxy. Some sites also behave really weirdly when you don't let them drop supercookies, a million Flash applets, a never-ending stream of Javascript trash, etc onto you. 🤣

If you don't believe me, try it out for yourself - all of this stuff is very easy to implement and very easy to unload if you aren't happy for whatever reason. 😀

Do note: I've avoided the actual "AdBlock" software because it can cause compatibility problems, and has instituted a pay-to-play program where anonymous advertisers can buy their way out of blocking (and at that point, what's the point?).

Do note: I've avoided the actual "AdBlock" software because it can cause compatibility problems, and has instituted a pay-to-play program where anonymous advertisers can buy their way out of blocking (and at that point, what's the point?).

You can opt out of that. And besides their program never lets flash or video ads through.
BUT adblock+ DOES seem to leak memory.

Also, when blocking things with hosts, I find 0.0.0.0 seems to have less performance hit than 127.0.0.1 for some reason