Skybuck's PC for 2015/2016. (inquiry)

Skybuck Flying wrote:
Hmmmmm I have some bad news for you guys and for me too.

Firefox is not multi threaded... at least not on my system... Firefox
sucks balls.

So does Internet Explorer.

Bad day for the net ahead.

Not sure why Fire**** does not multi thread properly... maybe it's flash
holding it back.. but I don't think so.

I ll try and see in options if Fire**** has some multi threading options
or something.

Needless to say... this sux balls.

If you believe I am in the wrong point me towards links screenshots of
multi threading action in firefox... multiple performance windows/cores
etc task manager etc.

I ll try a google later.

Bye,
Skybuck.

Use Sysinternals.com copy of Process Explorer to list the
thread count. The thread count is over 20 or so (not that
this is important).

Parallelism in browsers is not fine-grained. There is no
intention to split the execution of Firefox over four
cores, such that each core has exactly identical computing
load.

The parallelism is course-grained, and is a side-effect of the
architecture. If the browser uses a sandbox for something,
the execution may end up scheduled separately by the OS
scheduler.

Look at the 3D games you play for inspiration. The "boss"
thread in a game, runs a core at 100%. The "helper" threads
run at 30% utilization or less. That is asymmetric, and
you would argue, not making the best usage of the CPU.

You'd be right, but the software development environment
simply doesn't allow transparently chopping things any finer.

To put a finer point on it, the latest versions of Firefox
are a pig, and even with spiffy hardware, there is
hardly a reason to celebrate. Firefox has video card driven
hardware acceleration, but one wonders why they even bothered.
The software just keeps getting bigger and bigger and bigger...

Using the latest Firefox, I had Firefox use 1GB of RAM to
hold a single web site tab on the screen. How is that for
performance ? And the CPU usage kinda "oscillated", for
reasons unknown. There were bursts of CPU activity, after
the web page was viewable, and it seems to be internal
garbage collection running inside Firefox.

A threading model, is the least of Firefox problems.
It's just too big as a design.

Paul

On Tue, 16 Jun 2015 21:55:02 -0400, Paul Gave us:


Use Sysinternals.com copy of Process Explorer to list the
thread count. The thread count is over 20 or so (not that
this is important).

Parallelism in browsers is not fine-grained. There is no
intention to split the execution of Firefox over four
cores, such that each core has exactly identical computing
load.

The parallelism is course-grained, and is a side-effect of the
architecture. If the browser uses a sandbox for something,
the execution may end up scheduled separately by the OS
scheduler.

Look at the 3D games you play for inspiration. The "boss"
thread in a game, runs a core at 100%. The "helper" threads
run at 30% utilization or less. That is asymmetric, and
you would argue, not making the best usage of the CPU.

You'd be right, but the software development environment
simply doesn't allow transparently chopping things any finer.

To put a finer point on it, the latest versions of Firefox
are a pig, and even with spiffy hardware, there is
hardly a reason to celebrate. Firefox has video card driven
hardware acceleration, but one wonders why they even bothered.
The software just keeps getting bigger and bigger and bigger...

Using the latest Firefox, I had Firefox use 1GB of RAM to
hold a single web site tab on the screen. How is that for
performance ? And the CPU usage kinda "oscillated", for
reasons unknown. There were bursts of CPU activity, after
the web page was viewable, and it seems to be internal
garbage collection running inside Firefox.

A threading model, is the least of Firefox problems.
It's just too big as a design.


I am reasonably certain that every last bit of that went right over
his granulated, four active neuron brain.

One core/thread per tab would already be an improved me thinks.

Right now it seems very ****ty.

Perhaps locking contention/issues.

Perhaps GPU is actually bottleneck... who knows...

I might have to dive into it's source code if available to take a closer
look...

Bye,
Skybuck.

On Wed, 17 Jun 2015 18:03:02 +0200, "Skybuck Flying"
Gave us:

I might have to dive into it's source code if available to take a closer
look...

Bye,
Skybuck.

Trust me... that task is far beyond your capacity to understand.

There is also something wrong with FireFox's audio system.

It starts to crackle after playing a couple of hours of videos.

Shutting firefox down completely and restarting it seems to solve it.

I can't believe this software is so bad !

Bye,
Skybuck.

Skybuck Flying wrote:
There is also something wrong with FireFox's audio system.

It starts to crackle after playing a couple of hours of videos.

Shutting firefox down completely and restarting it seems to solve it.

I can't believe this software is so bad !

Have you ruled out that the problem lies in your computer?


Bye,
Skybuck.

Skybuck Flying wrote:
There is also something wrong with FireFox's audio system.

It starts to crackle after playing a couple of hours of videos.

Shutting firefox down completely and restarting it seems to solve it.

I can't believe this software is so bad !

Bye,
Skybuck.

You can find sites filled with "conspiracy theories".

https://forums.adobe.com/message/4377590

Notice how all manner of factors (not just Firefox)
seem to be implicated. It's more likely
to be a plugin or codec issue, than Firefox itself.

Paul

On Fri, 19 Jun 2015 22:10:43 -0400, Bill Gave
us:

Have you ruled out that the problem lies in your computer?

No... it DEFINITELY lies with the operator thereof.

Yeah could be some flash virtual machine corruption or timing issues or some
other code bug.

Perhaps a slight buffer overrun or something.

Could be a security issue though.

Bye,
Skybuck.