Early Open-Source Linux Benchmarks Of The AMD Radeon RX 470

[bridgman]

I'm sad to see that the 290 doesn't get any love anymore...

I don't think it's "not getting any love" as much as "no developer being able to reproduce the rumored performance regression yet". As of yesterday (or day before, I forget) we have a bug ticket and can hopefully start accumulating enough info for a developer to be able to repro.

[Kendji]

And don't forget that the AMD drivers are open source.
So when you factor that in...

This is a big factor! Open-source performing this well is impressive.

[Azpegath]

amdgpu.ko kernel module in linux-git gained user clock control for Hawaii chips about a week ago. /sys/class/drm/card?/device/pp_dpm_{m,s}clk

Oh, I meant in Michael's tests. He doesn't seem to include it in testing anymore, at least not frequently. I know there was a big regression earlier, I hope that's been fixed but haven't seen anything about it.

[nomadewolf]

Why can't you draw your own conclusions based on the benchmark data? Michael is busy as it is...

Michael presents the data, and we draw the conclusions

[nomadewolf]

Good enough to buy over nVidia? Currently running an gtx 960, in certain games it outperforms the rx 480. Would like to buy AMD, considering the companies financial troubles unlike nVidia.

My old ati card had sufficient performance, but had problems rendering 3d textures.

The answer is:
If you can find an AMD card which has enough performance for you TODAY, go for AMD as it is slower than NVidia, but is more future proof.
If you can't fullfill your current performance needs with AMD or want a low power media center pc for your living room, you're better off with NVidia.

[nomadewolf]

Those drivers for AMD is starting to look decent compared to nVidia

And don't forget that the AMD drivers are open source.
So when you factor that in...

[Xelix]

Why can't you draw your own conclusions based on the benchmark data? Michael is busy as it is...

Because I have little interest in going through 5 pages of bar plots. Typically, I'd like to read the first page to know the test setup, and a conclusion to get a high level summary. Then if something catches my attention/curiosity I can go through the data as needed.

And I get and respect that he is busy, but I won't buy that writing a 2-3 sentences summary will increase the workload. There are some articles where he does it, such as this one: http://www.phoronix.com/scan.php?pag...-anv-wsl&num=5 .

[Azpegath]

I'm sad to see that the 290 doesn't get any love anymore...

[Azpegath]

Why no Vulkan tests? For Doom I've heard AMD to be much faster.

How about actually reading the article in the future, or at least do a ctrl+f search on the page for what you're about to ask, before asking it?

[joh22n]

Michael, I've mentioned that on a previous thread, but could you please start writing conclusions to your article?

Why can't you draw your own conclusions based on the benchmark data? Michael is busy as it is...