On Wednesday December 24 2014 11:11:59 Thiago Macieira wrote:

> The problem is that they must be tuning the I/O scheduler for SSDs these 
> days, 
> so if you have a regular HD ("spinning rust"), you'll suffer.

Possible, though the parser remains so long on a single file sometimes that I 
have a hard time believing it could only be the I/O scheduler...
I don't think disk benchmarks have gone down significantly when I'm also doing 
other work.

> I also can't really explain this. When building something big on my Mac Mini, 
> the first few minutes the total system usage is around 20% and varies a lot; 
> then after that it shoots up to 100% and stays there.

I've seen clang do really weird things, though that was more on 10.6 . It could 
use loads of memory for a short while, and then after releasing it remain 
almost idle for a very long time.
There must be some serious changes in memory management, including memory 
compression, which I suspect to be involved. I find it very suspicious that I'v 
never yet had more than 2 VM files (= a bit over 1Gb) while under 10.6.8 I 
routinely found myself at several times that when doing serious work. (Couple 
browsers open, each with numerous windows & tabs, 1 or 2 IDE sessions, the 
occasional VM session etc)

> You can call it a conspiracy to make you buy newer hardware...

Amen to that - just like those Gb of useless stuff they so often add to an 
app's payload.


>       goto fail;

:)

R
_______________________________________________
Interest mailing list
[email protected]
http://lists.qt-project.org/mailman/listinfo/interest

Reply via email to