AMD Radeon R9 Nano Review by TPU...Not

[ampre]

asyncronus compute is not an amd Feature it's a Feature for the future and the next Logical step for make the gpu more parallel for better feed the shader.

I'm not an expert, but i can understand what experts are saying!

 

[arbiter]

asyncronus compute is not an amd Feature it's a Feature for the future and the next Logical step for make the gpu more parallel for better feed the shader.

if it wasn't an AMD feature then why is it in 7000 series cards? how does fact they had it in cards from 4 years ago make it not AMD tech?

If PhysX was put as a standard in DX12, AMD fans would gone ape crap over it. AMD would also be whining up a storm over it.

 

[ampre]

asyncronus compute is not a bling bling Display Feature like a Smoking Effekt. It's a Feature to handle data more effektive and in parallel which gives you more power on the same chipsize. I'ts the next Logical step in gpu evolution. So why Pascal should get it like mixed precision? Nvidia is realy far behind gpu development!

 

[the54thvoid]

Only a real kid knows a other kid!

Go home and tell your mom!

Tell me where is the article about Nvidia is missing dx12 key Features in the Hardware?

You cant answer the54thvoid because you are a brain washed Person!

Maybe you think you have a watercooling System it makes you an gpu architecture expert?
I love this selfmade heroes, which think they are the greatest and have no Argument!

This will be my last response to you. I'll make it short.

Goodbye.

Love that ignore feature. Wish it was pre-emptive though - I'm sure that's Sony or his embryonic twin.

 

[HumanSmoke]

Because it's the first introducion of DirectX 12 and leaded to this monstrous thread\discussion:

So what? TPU didn't bench alpha builds of DX10 or DX11 software either.
You expect this site to benchmark an alpha build based on a discussion at another site - who also didn't benchmark AofS ?
If you are pining for that discussion, feel free to go to OCN and close the door on the way out.

Only a real kid knows a other kid!

That sounds exactly what a kid in the preoperational stage of cognitive development would say. Once you get older, you'll probably realize that there are adults who understand children (or "know kids" as you term it) - parents, teachers, care givers, psychologists etc. Something to look forward to!

 

[Pill Monster]

you mean missing a key AMD made feature? That probably was locked out to nvidia til after maxwell was finalized.

What makes you a gpu arch expert?

1st off Async isn't required part of DX12, 2nd off don't know how many dev's will even use it outside ones AMD pays off to use it.

It's already been established nobody will have full DX12 support, features vary dependent on the specific GPU and vendor. I don't know the specifics on that yet,

I did see a pretty scathing article on Guru regarding Nvidia's DX12 claims...and sidestepping questions at TechEd conference.
Actually come to think of it a similar article got published after the GTX970 3.5GB fiasco.
Nvidia finally made a statement and released it through PcPer, When Hilbert found out he was super pissed because they had snubbed Guru.
In his view Nvidia were trying to avoid PR damage by releasing statements through cherry picked sites. I guess I'm saying he felt PcPer was sympathetic to Nvidia.

Tbh the whole affair was way way overblown in my view.


For the guy asking about Ashes of Singularity, (Ampere) come back when DX12 games are available.
Benchmarking an API no games even support is benching for the sake of benching.....which smells a bit like hype.
Reminds me of the 3DMark drawcall BS that was flying around the web a while ago.

Who really cares about benchmarks anyway?

 

[lilhasselhoffer]

One question TechpowerUP where are your ashes of the singularity Benchmarks?

ahh nvidia was not good enough so you didn't try!

If you're going to make that request, how about a few others.

1) Where's your benchmark showing DX13 compatibility?
2) Where's you benchmark showing Black Ops 4?
3) Where's your benchmark showing 16k resolution?


Asking for what may eventually be viable, but is currently not testable, is foolish. As yet, DX12 is a largely unexplored and even more largely a software driven pursuit.

 

[Phildo Gaggins]

If TPU doesn't have a review of something that I'm interested in buying, I don't buy it. This is hands down the best PC hardware review site on the web.

 

[xenocide]

Tell me where is the article about Nvidia is missing dx12 key Features in the Hardware?

It's literally right below this one in the News section. You're trolling--poorly.

If TPU doesn't have a review of something that I'm interested in buying, I don't buy it. This is hands down the best PC hardware review site on the web.

I wouldn't go that far, but I definitely weigh their opinion more than most other sites.

 

[arbiter]

It's literally right below this one in the News section. You're trolling--poorly.

async is like adaptive sync in 1.2a, its optional not required to have to be compliant for dx12.

 

[xenocide]

async is like adaptive sync in 1.2a, its option not required to have to be compliant for dx12.

There are a lot of features in DX12 that aren't required, but if you don't support any of the features it's hard to justify saying you're DX12 compliant. The guys who work on Unreal Engine put out a statement saying Asynchronous Compute was risky to implement, and shouldn't be leaned on too much, so I imagine Nvidia didn't count on the first batch of DX12 games using it to extensively. Just happens to be AoS was originally a Mantle title and heavily marketed as such.

 

[FordGT90Concept]

I believe it is part of the Compute Shader/DirectCompute but all Microsoft literature on the subject appears to end at DirectX11. I think it may be something that has been available but developers haven't used.

I'm literally finding nothing that links async shaders to DirectX 12. It's all DirectCompute which was launched with DirectX 11 so, in theory, all you need is a DirectX 11 card.

 

[15th Warlock]

Seems I'm really late to this party, just returned from a week long vacation with zero internet access

To me AMD killed the Nano when they decided to price it at $649, but limiting the number of cards available for review? Just wow AMD 2015 will probably be remembered as one of the worst years in terms of PR management for the company

On a side note, like many others have stated, it wasn't until I realized I had a few trolls on my ignore list that the thread started making more sense, thank goodness for that! LMAO!

 

[nem]

Async Shaders: AMD native support vs NVIDIA emulated...

This quite clear the situation with the new AMD DX12., nvidia does not look good that would be emulating the Async computer (shaders asynchronous), and is becoming more obvious that DirectX 12 is based on Mantle and hence the advantage of having asynchronous hardware shaders, NVIDIA also provided two new effects the raster order Rasterization conservative views and that these two effects are what make the extra support to generate the DirectX 12.1, but this does not actually add much if we do remenber the words of NVIDIA said in previous versions of DirecX that decial something like We are not interested in implementing the partial revisions why this dont add nothing important. So it was for this reason that NVIDIA was only DX10 and DX11 only then, but at the last moment also use DX11.1, but in short what matters in DirectX is the main features and that these are for hardware support.

 

[FordGT90Concept]

That appears to be from Wikipedia and note there is no source cited for "Async Shaders." A bit of Wiki trolling methinks.

All DX11 cards support the instructions for async shaders; AMD is the only the manufacturer that took the time to make it work as Microsoft intended (asynchronously).

 

[xenocide]

I believe it is part of the Compute Shader/DirectCompute but all Microsoft literature on the subject appears to end at DirectX11. I think it may be something that has been available but developers haven't used.

I'm literally finding nothing that links async shaders to DirectX 12. It's all DirectCompute which was launched with DirectX 11 so, in theory, all you need is a DirectX 11 card.

Pretty sure it's just Compute Units working as Shaders. Compute is something Nvidia started to ignore during DX11 because it hadn't matured enough to be useful. Now it is useful (apparently) so I expect Pascal will see a return to more Compute-friendly setups. It was essentially wasted die space during Fermi and Kepler, but apparently AMD had faith it would payoff. Good planning on their part.

 

[arbiter]

That appears to be from Wikipedia and note there is no source cited for "Async Shaders." A bit of Wiki trolling methinks.
All DX11 cards support the instructions for async shaders; AMD is the only the manufacturer that took the time to make it work as Microsoft intended (asynchronously).

Async Shaders: AMD native support vs NVIDIA emulated...

They used code for async from AMD hence why nvidia don't support it fully. Anyone that doesn't even see that with that table is complete blind fool. Funny how everyone loves to point out how AMD supports it natively when it was their own feature set that was used.

It was essentially wasted die space during Fermi and Kepler, but apparently AMD had faith it would payoff. Good planning on their part.

good planning or did they pester MS so much that MS caved in and added it?

 

[HumanSmoke]

Async Shaders: AMD native support vs NVIDIA emulated...

I doubt you could stay on topic if your life depended upon it. This thread is supposed to be concerned with TPU not getting a review sample and exploring the possibilities of why this should be so and possible ramifications.

So of course you and the rest of Sunnyvale cheerleaders launch into async shader implemantation (we have threads for that already), the GTX 970 is-it-4GB-or-not-quite (we have about 6 threads for that already), and crying that an unreleased game at alpha isn't the focus on the site - but also has it's own threads.
You guys don't want to blow your entire wad prematurely on thread derailing content before the big reveal, you'll have nothing left for when the reviews go live and the inevitable questions of availability.

 

[geon2k2]

LOL 380 comments :B

pd. i cant way for nano.¬¬

Made my day, really funny, and so true, this shows how much value the nano brings, even if performance will not be that much better.

 

[Athlonite]

I'd ask Hilbert over at Guru3D if you can borrow their review model I'm sure he'd be more n happy to oblige

 

[FordGT90Concept]

The agreement he signed to get it likely forbids transferring it without AMD's consent.

 

[Mussels]

420 comments... you could say its blazin

 

[ironcerealbox]

I'm not an expert, but i can understand what experts are saying!

Where have I heard that before? Oh yeah, the many college students I've had to teach over the years...

 

[Assimilator]

I love the number of new members, or members with low postcounts, jumping into this thread to defend AMD. Sockpuppets anyone?

I'm not an expert, but i can understand what experts are saying!

Son, what you are is a n00b. Shut your mouth, sit your ass down, and be quiet while the adults are talking.

Love that ignore feature. Wish it was pre-emptive though - I'm sure that's Sony or his embryonic twin.

I think you misspelled "retarded".

 

[john_]

http://imgur.com/a/D915a?gallery#0