Four Years After Launch, AMD Kaveri Sees Huge Performance Boost On Linux

[ElectricPrism]

This is why open drivers rocks, instead of companies trying to push for EOL and planned obsoleteness the software can be improved on if there is room for improvement.

This type of real world benefit builds brand loyalty -- 9 times out of 10 I now make buying decisions based on the long term quality of a product, the most comparable thing that comes to mind is solid wood furniture -- if it gets damaged usually you can sand and refinish it no problem and have it last decades.

Open drivers easily double a hardware's lifespan and really concrete brand loyalty after consumers have a good experience.

[dungeon]

Open drivers easily double a hardware's lifespan...

Lifespan deserve and should be prolonged because initially open drivers were and are in not so ideal shape.

So, problem is that open drivers were and still are usually and sometimes severely borked in their prime time That is a main joke bucause that is also period of time when hardware have most sense to be used

If open have perfect execution and if countinosly in line with hardware launches, that would be also perfect killer... but in reallity it is not like that.

Instead to serve prime time, they are powered by prime time Just model of development is different.

[johanb]

Yet another example of "AMD FineWine Technology"

Not sure if it's a good thing or a bad thing though...

[haplo602]

Yeah, where were you guys till now.

Waiting on performance ... still waiting on proper Ryzen Mobile drivers on either Windows or Linux ....

[chithanh]

So will this also speed up my Carrizo (a10 8700p) laptop since it has similar graphics?

No, Carrizo (and Bristol Ridge) are 3rd generation GCN and have always used amdgpu. The workaround for radeon problems was not necessary there in the first place.

[TemplarGR]

This is why open drivers rocks, instead of companies trying to push for EOL and planned obsoleteness the software can be improved on if there is room for improvement.

This type of real world benefit builds brand loyalty -- 9 times out of 10 I now make buying decisions based on the long term quality of a product, the most comparable thing that comes to mind is solid wood furniture -- if it gets damaged usually you can sand and refinish it no problem and have it last decades.

Open drivers easily double a hardware's lifespan and really concrete brand loyalty after consumers have a good experience.

While i agree in general there are a couple of counter-arguments:

1) Those opensource improvements we get for old hardware seem good only because initially opensource drivers sucked hard. They were behind in OpenGL support and in some ways still are, not to mention compatibility and performance issues, and other missing features. Sure it is nice that we are getting all the support now, but this hardware is already obsolete. For people who bought Kaveri when new, they had to use it for 4 years with half the performance. Now that they are approaching the time to upgrade they unlock the performance, not exactly ideal... But, the good thing is that for current hardware drivers are already in a better shape so i don't expect this Kaveri situation to repeat itself, open drivers are much more mature now.

2) Supporting ancient hardware is not always a good thing. There is a reason blobs drop (mainline) support after some years and relegate drivers to legacy status (for bug fixing and kernel/xorg support only). After some time, any performance you could squeeze out of hardware has already been squeezed, features are all 100% supported, and the hardware itself has dropped in performance so it is not relevant anymore. For example, why keep updating drivers of early 2000s era gpus these days? Assuming said gpus still work which is rare, their performance will be attrocius by modern standards so they can't really support modern distros unless they use very barebone DEs or the console. They don't play modern games, they don't have video accel support, nothing. So why bother? Even the cheapest apu is miles ahead these days.

[bridgman]

On Linux, but on Windows is not like that. There is no such artifical separation there. Or If that is not true then you are not sure so better talk to your marketing team, as i don't see any VS there

VS ?

So you have decided how Linux user of PRO APUs, does not need to run workstation workloads

Better say 'might not need' or the best just learn to say clear 'No'. 'No' is easy, everbody understand that

Nope, there's nothing Linux-specific there. If you check the release notes for the corresponding Windows drivers the supported hardware list is also dGPU-only. Every so often a workstation APU SKU will be produced, and in those cases the business folks ask us to provide support in the workstation drivers.

Just curious, other than the natural human desire to pour sand in gears, what made you say this was a Linux-only decision ?

BTW, what are Radeon Pro drivers now?

My understanding is that Radeon Pro covers a slightly broader range of target users than FirePro did, with more focus on things like 3D rendering.

[Mez']

The 560 is probably still 2x as fast as the Kaveri, even with the latest updates. My guess would be a bit over 2x due to faster/more VRAM.

Supposedly, I agree.
In practice, that was around the difference between both when I added the RX560. With these improvements, the gap should be quite close now.

[paregistrase]

Hi people.

I have a AMD® A10-7860k radeon r7, 12 compute cores 4c+8g × 4 (AMD® Kaveri).

I try to test the amdgpu using instruction that I read in other post:

Adding "modprobe.blacklist=radeon amdgpu.cik_support=1 radeon.cik_support=0" to GRUB_CMDLINE_LINUX_DEFAULT="quiet splash" line in /etc/default/grub

But the second monitor (plugged via VGA) is not recognized.

Is this a bug or I'm doing something wrong?

[paregistrase]

Hi people!

I'm trying to test the amdgpu in my AMD® A10-7860k (AMD® Kaveri) adding "modprobe.blacklist=radeon amdgpu.cik_support=1 radeon.cik_support=0" to GRUB_CMDLINE_LINUX_DEFAULT line in /etc/default/grub and the second monitor (a tv plugged via VGA adapter) isn't recognized.

It is a bug or I'm doing something wrong?