On 22/06/2011 04:44 PM, Darren New wrote:
> On 6/22/2011 1:35, Invisible wrote:
>> Sure. It's not the whole OS. It's just the entire GUI. Subtle
>> difference there.
>
> Not the entire GUI. One program running the file manager.
>
> Your email client still works. Your background processes still work.
> Programs trying to open files might get stuck.

The Start menu stops working, all open Explorer windows stop working, 
the task bar stops working, the system tray stops working, windows stop 
repainting, Alt+Tab stops working, and even Ctrl+Alt+Del becomes very 
slow. And for what? Because of a piffling hardware event which I might 
not even care about? Great...

>> No reason, just poor design.
>
> More like "insufficient engineering time given to edge cases to make
> them efficient."

I'm not sure "the entire PC stops working every time you touch the CD 
drive" is what I'd call an "edge case", but whatever...

>> Then again, SSDs are here now, and gradually becoming actually popular. I
>> guess history repeats itself, eh?
>
> Yeah. Sadly, it has to actually *look* like a disk, so you still get a
> lot of the overhead.

I did see a flash device that's a PCI card. It requires special device 
drivers and stuff, but apparently it goes stupidly fast.

Then again, the advantage of a vanilla SSD is that it works with 
everything. A bit like the way 30 years after the original IBM PC was 
released, almost every desktop computer on Earth still pretends to have 
a 20-bit address space until you twiddle one of the control registers on 
the keyboard controller (!?!) to enable access to all of the installed RAM.

>> Warp's original comment was that partitioning an arbitrary problem
>> such that
>> you gain parallel speedup is probably impossible "in practise".
>
> I wouldn't be surprised if it was mathematically impossible in theory,
> either.

Well, no, the search space is presumably finite (although absurdly 
vast), so theoretically it's trivial.

>> The other thing I've thought about is having multiple heap areas and
>> tracking pointers between them. If you could arrange it so that all the
>> garbage is in one heap chunk, you can just drop the whole chunk rather
>> than
>> doing complex processing over it. However, that's not easy to achieve in
>> general.
>
> That's called semi-space garbage collection.

Wikipedia's description of "semi-space GC" seems to be "2-space GC".

>>> Yep. That's why NUMA is getting more popular and such.
>>
>> Only for supercomputers.
>
> The Opteron is a supercomputer?

How many desktop PCs contain an Opteron?

>> I wonder how long it will take for desktops to catch up?
>
> You're probably already running it.

Desktop PCs have had non-uniform memory access times for decades now. 
But currently, they all *pretend* to have uniform memory. NUMA is about 
explicitly admitting that there are different pools of memory, with 
different properties.

> Comic Saaaaaaaaaaans!

Yeah, well, I didn't write it...

>> The interesting bit is slides #3 and #4. I've often wondered why all the
>> books about supercomputers talk a lot about parallel processing, but
>> [until recently] it's never seen in normal computers. Now I know, I guess...
>
> Because until recently (like, the last 10 years), putting multiple CPUs
> in something was very expensive.

Putting multiple chip in is *still* very expensive. ;-)

-- 
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*

Post a reply to this message

On 6/22/2011 14:09, Orchid XP v8 wrote:
> windows stop repainting,

All of them? Or just explorer windows?  Because none of that stuff happens 
with me.  Maybe you have something installed that goes compute-bound or 
something when you mount a disk.

> I'm not sure "the entire PC stops working every time you touch the CD drive"
> is what I'd call an "edge case", but whatever...

Clearly it doesn't "stop working."  Try opening the task manager and 
watching the chart scroll when you put something in the CD drive.

> Then again, the advantage of a vanilla SSD is that it works with everything.

Exactly, yes.

>>> Warp's original comment was that partitioning an arbitrary problem
>>> such that
>>> you gain parallel speedup is probably impossible "in practise".
>>
>> I wouldn't be surprised if it was mathematically impossible in theory,
>> either.
>
> Well, no, the search space is presumably finite (although absurdly vast), so
> theoretically it's trivial.

Hmmm. Maybe, yeah, if you tried every possible rearrangement of opcodes sort 
of thing. But that assumes you can tell that the rearranged opcodes will 
have the same effect, which again is impossible.

I.e., the number of ways you can chop up the problem is finite, but the 
number of *wrong* ways to chop up the problem is non-zero.

>>> The other thing I've thought about is having multiple heap areas and
>>> tracking pointers between them. If you could arrange it so that all the
>>> garbage is in one heap chunk, you can just drop the whole chunk rather
>>> than
>>> doing complex processing over it. However, that's not easy to achieve in
>>> general.
>>
>> That's called semi-space garbage collection.
>
> Wikipedia's description of "semi-space GC" seems to be "2-space GC".

Well, you're either talking about generational heaps, semi-space heaps, or 
separate heaps that might have roots in other heaps, in which case it's the 
same as a generational heap technologically-speaking.

>
>>>> Yep. That's why NUMA is getting more popular and such.
>>>
>>> Only for supercomputers.
>>
>> The Opteron is a supercomputer?
>
> How many desktop PCs contain an Opteron?

http://sites.amd.com/us/business/products/desktop/Pages/hp.aspx

Come on now. That wasn't even difficult.

> Desktop PCs have had non-uniform memory access times for decades now. But
> currently, they all *pretend* to have uniform memory.

Well, except to the OS.  Applications pretend they all have contiguous 
memory addresses, too.

-- 
Darren New, San Diego CA, USA (PST)
   "Coding without comments is like
    driving without turn signals."

Post a reply to this message