R9 Nano Reviews Tip Up, But Will Be Not Too Useful For Linux Gamers Right Now

[Passso]

I'd call this bad journalism. If you want to be a serious HW reviewer go test ist - then write something about it. (Regardless what past experiences we had with the free stack supporting ultra-new-HW or experiences with fglrx.)
And eventually even suggesting the 3.5 GB (etc. etc.) GTX 970. Wow.

This is the same hardware as Fury, already benchmarked. And Michael bough it himself.

No need to buy twice a product to talk about it.

[bug77]

This is the same hardware as Fury, already benchmarked. And Michael bough it himself.

No need to buy twice a product to talk about it.

Actually, it isn't. See here: http://www.techpowerup.com/215776/am...y-tpu-not.html

More exactly, this part:

We can't emulate an R9 Nano on an R9 Fury X. The Nano appears to have a unique power/temperature based throttling algorithm that we can't copy.

That aside, you can safely look at Nano as a 15-20% lower clocked X. Since X doesn't run at 1GHz all the time either, 15-20% is the maximum gap you'll see between these two.

Edit: It's sad AMD chose not to supply a card to TPU, they're the only ones that I know of that include a relative performance graph at various resolutions and card power consumption in different scenarios (idle, multi-monitor idle, playing a bluray and so on). Then again, maybe that's what AMD wants to keep away from the public </speculation>

[artivision]

No wafers are bigger now too so more dies per wafer. But it's more complicated than that. The fabs cost billions of dollars and they need to be constantly updated which costs billions more. That's Intel's biggest advantage. People like to claim that AMD going fabless saved them all that money, but it's not true. It's actually costing them a whole lot more. Global foundries does not have a sustainable business model. They've only made it this long because of investors. Actually I was kind of shocked that they got enough investment to adopt a 14nm process. I didn't think there would be anybody willing to do it.

But lithography investments was also there before, when the best GPU where sub 200 bucks. I think that they have choose a Apple like model with idiotic premium standards that are out of computers philosophy. For example AMD sells the PS4 Apu to Sony for 100 bucks and that Apu has half the Fury transistors. Why i can't get the same Apu at the same price?

[dungeon]

This is the same hardware as Fury, already benchmarked. And Michael bough it himself.

No need to buy twice a product to talk about it.

No one said he needs to buy it, but they are not the same even picture say it all. Michael big card is in the middle and Nano is well Nano

And is not like a Fury, but squized down in size and power needed from Fury X effort.

[boffo]

Very nice card, the price is high but do to the low production volume it will not be a problem.
It's nice to see what the future will look like.

[Adarion]

That's pretty much what was said in the article was to wait, but then he and I were commenting on how Adarion took issue with what I wrote.

Yes, I took issue over that sentence. Michael, you are the official institution for Linux Hardware testing, so there is a certain expectation. You built that, but now you got to live up to your standards.

The sentence was too bland. Regardless how similar the architecture might be (Fury vs. Nano), I'd say this little monster is still different enough. And not to be treated like "ah, the grapes are sour anyway!".
Also, as you wrote later ( http://www.phoronix.com/scan.php?pag...-radeonsi-1315 ) the free stack does make good progress.

It is completely okay, though, if you complain about the lack of review samples.
You know, if I had the money I'd be back to premium (kind of tips for buying samples) and beer (on occasions), but I hit the official poverty line years ago and now I finally dove even under it (likely until I get that darn PhD finished). :/

[hoohoo]

What it is sad is that you're trolling or completely obfuscated... AMDGPU driver is in a very active development, it will support newest chips in an open source flavour. It is not production ready yet, but you can't say they don't care, expecially if you take a look to the RadeonSI driver which is performing almost the same as catalysts.

I think there is something great going on in AMD driver development, and we will see the result in some months.

I did start my comment by saying I'm not trolling or flaming. I guess you can't read very well.

As for "obfuscated", that word is an adjective meaning it's subject is obscured or hidden. Well, since I post using an alias I suppose I am obfuscated. What of it?

---

I'll try to spell my argument out more simply so you might find it easier to understand...

Certainly there is good work being done in the AMDGPU driver.

1) There is no AMD driver, proprietary or open source, *at* *this* *time* that is worth using unless all one does is two-D work.
2) *At* *this* *time*, AMD Fury GPUs on Linux suck in comparison to the competition.
3) Benchmarks of AMD Fury GPUs on Linux are an incentive to buy Nvidia GPUs for use on Linux.
4) *Therefore* AMD has no incentive to give Phoronix a Fury card to benchmark.
5) *In* *the* *future*, if AMD Fury drivers (either proprietary or open source) offer comparable performance to NV's products, *then* AMD will have an incentive to give Phoronix cards to test.

Is that a simple enough chain or reasoning for you?

[Kano]

What i really hate about AMD hardware is the complete ignorance of HEVC Main 10. Not even DXVA2 supports that. You can get HDMI 2.0 support via the first DP adapters out there but without HDCP 2.2. They should at least write a hybid driver for that. I still remember the introduction of the new HEVC accelleration by AMD - they implied it is better than Broadwell because of no hybrid appoach. But they did not mention it would be 8 bit only... Buying AMD hardware for future 4k use is very stupid...

[blackshard]

I did start my comment by saying I'm not trolling or flaming. I guess you can't read very well.

As for "obfuscated", that word is an adjective meaning it's subject is obscured or hidden. Well, since I post using an alias I suppose I am obfuscated. What of it?

The fact that you state you're not trolling doesn't mean that you're not doing it actually.

---

I'll try to spell my argument out more simply so you might find it easier to understand...

Certainly there is good work being done in the AMDGPU driver.

1) There is no AMD driver, proprietary or open source, *at* *this* *time* that is worth using unless all one does is two-D work.
2) *At* *this* *time*, AMD Fury GPUs on Linux suck in comparison to the competition.
3) Benchmarks of AMD Fury GPUs on Linux are an incentive to buy Nvidia GPUs for use on Linux.
4) *Therefore* AMD has no incentive to give Phoronix a Fury card to benchmark.
5) *In* *the* *future*, if AMD Fury drivers (either proprietary or open source) offer comparable performance to NV's products, *then* AMD will have an incentive to give Phoronix cards to test.

Is that a simple enough chain or reasoning for you?

You forgot to say that *at* *this* *time* Linux is not a gaming platform, so spending so much resources to improve benchmarks is quite useless for AMD.