Mesa 17.1 will use a set 1GB size for the shader cache

Not sure if GOL really needs to cover that kind of low level implementation details of graphics drivers.
I agree that the scope of driver functions is very interesting but the actual size of particular driver cache is a little too deep from my point of view.

Not sure if GOL really needs to cover that kind of low level implementation details of graphics drivers.
I agree that the scope of driver functions is very interesting but the actual size of particular driver cache is a little too deep from my point of view.

In principle I agree with you, but...

I am thinking that probably as long as this kind of news are not too great of percentage of total news, it is more simple and better to not doing such filter so _really_ interesting news will not go unnoticed. On the other hand if this is getting out of hand I would say news should be filtered more.

Not sure if GOL really needs to cover that kind of low level implementation details of graphics drivers.
I agree that the scope of driver functions is very interesting but the actual size of particular driver cache is a little too deep from my point of view.

Not sure if GOL really needs to cover that kind of low level implementation details of graphics drivers.
I agree that the scope of driver functions is very interesting but the actual size of particular driver cache is a little too deep from my point of view.

In principle I agree with you, but...

I am thinking that probably as long as this kind of news are not too great of percentage of total news, it is more simple and better to not doing such filter so _really_ interesting news will not go unnoticed. On the other hand if this is getting out of hand I would say news should be filtered more.

Then perhaps the deeper more technial stuff should go on https://www.lifeonlinux.com/

?

Not sure if GOL really needs to cover that kind of low level implementation details of graphics drivers.
I agree that the scope of driver functions is very interesting but the actual size of particular driver cache is a little too deep from my point of view.

I personally really like having these tid bits. I do agree that I wouldn't want these types of news to be the main thing I read about, but I think Liam does a good job of sprinkling them in without it becoming overwhelming.

I kind of like the idea of the two sites merging and when you click on technical news it sort of swipes to LOL and Gaming news swipes to GOL ( but im being overly optimistic here and probably just day dreaming )

Not sure if GOL really needs to cover that kind of low level implementation details of graphics drivers.
I agree that the scope of driver functions is very interesting but the actual size of particular driver cache is a little too deep from my point of view.

We have covered stuff like this plenty of times in the past and you're the first to make a complaint about it. Usually, people appreciate it.

I appreciate it as well. That's actually a topic we're discussing in my lectures, because it's quite tough to find a fast and easy (regarding the gate expense) way of realising the cleaning of a cache. Our mentor said, that usually use 3 different ways of doing that:
1. Random
2. Deleting at the end (like a stack)
3. (And most common) by timetable, where the last use of the address is stored in an extra set of bits. If the cache is full, the oldest address will be removed. Or if a predefined time of last use has been expired.

As you may see, this is a really interesting topic for some. Especially because Linux users are usually power users as well. If you are not interested, you don't need to read that. Only thing I could criticise is the lack of an explanation. But as this is a hobby project (afaik), I don't bother.

Thanks to Liam for that article, made me sign up to GOL ;)

2. Deleting at the end (like a stack)
3. (And most common) by timetable, where the last use of the address is stored in an extra set of bits. If the cache is full, the oldest address will be removed. Or if a predefined time of last use has been expired.

I almost always use a combination of 2 and 3. I usually have a separate list that references the files and stores the epoch time of the last access to it. If a cache file is either added or "touched" I move it to the top of the list. This way there's no need to recurse each file and check for oldest dates. I get it all immediately from a fairly small file. The files at the very bottom of the list will be the least recently used and the ones at the top will be the most recently used. So if the cache is full I just start deleting off the bottom of the list. I can also check the timestamp and if the last access has exceeded a period of time I can just nuke it.

Are you aware of any instance where the random method would be the ideal for deleting cached files? I'm trying to think of something but am drawing a blank.

I kind of like the idea of the two sites merging and when you click on technical news it sort of swipes to LOL and Gaming news swipes to GOL ( but im being overly optimistic here and probably just day dreaming )

I had not actually assumed Life on Linux was supposed to be a technically-oriented site. One reason I don't bother looking at Linuxtoday much any more is that it's practically all about business-technical stuff which is of no interest to a desktop user. I was kind of hoping that, since it's split off from a gaming site, and Linux gamers are mostly Linux desktop users, Life on Linux would be a bit more people-oriented and desktop-oriented than some of what's out there.
If anything, I would think the technical underpinnings of graphics-related stuff would be more appropriate to a gaming site than a general-interest Linux site.

Right now I'm not sure there's enough game-only news on Linux, but as time goes on, if the amount of game news increases significantly, maybe it would be nice to have these technical news on LifeOnLinux...

LifeOnLinux is like the newer, better Phoronix (or at least I wish that it is what it would become ;) ), and this is the kind of news that Phoronix covers.

But all in all, on GoL or on LoL, I don't really care, I'm following both (though I need more and more a Google+ page for LoL :/ ) and as long as the news is on one of these websites, I'm happy ^^

Not sure if GOL really needs to cover that kind of low level implementation details of graphics drivers.
I agree that the scope of driver functions is very interesting but the actual size of particular driver cache is a little too deep from my point of view.

We have covered stuff like this plenty of times in the past and you're the first to make a complaint about it. Usually, people appreciate it.

May be read my post again. That was not my point.
I was complaining about a news that covers a low level implementation detail that concerns more a developer. As a gamer it has no big impact whether the cache size is a relative value of 10%, 5% or an absolut value of 1 gb. It's just a small configuration issue. It works either way.
On the other hand I appreciate news about Mesa implementing new features (like the news in the past) because these increase an actual feature set which has a much higher impact.

Not sure if GOL really needs to cover that kind of low level implementation details of graphics drivers.
I agree that the scope of driver functions is very interesting but the actual size of particular driver cache is a little too deep from my point of view.

We have covered stuff like this plenty of times in the past and you're the first to make a complaint about it. Usually, people appreciate it.

May be read my post again. That was not my point.
I was complaining about a news that covers a low level implementation detail that concerns more a developer. As a gamer it has no big impact whether the cache size is a relative value of 10%, 5% or an absolut value of 1 gb. It's just a small configuration issue. It works either way.
On the other hand I appreciate news about Mesa implementing new features (like the news in the past) because these increase an actual feature set which has a much higher impact.

Like I said, we have covered details like this plenty of times in the past.

I understand what you're saying, but GOL has a big reach, currently LifeOnLinux (or "LOL) does not, so for now it's still suited for here.

The shader cache is an important feature us AMD owners have been waiting for a long time on mesa. Cutting loading times and smoothing out hitches. Its important, although not all games need it as much as others.