Why Doesn't Anyone Get to the Bottom of the Aging of Kepler and GCN?

[Silverforce11]

Its so great and forward thinking we already see GCN 1.0 and 1.2 either stagnate or regress in DX12.

Not true, unless you like to compare it to Gears of Wars before it was fixed (it's all fine now btw, 380>960 by a lot, tonga works), or RoTR with a BETA implementation as even said by Nixxes.

In Ashes, all GCN gains in DX12.

 

[ShintaiDK]

Not true, unless you like to compare it to Gears of Wars before it was fixed (it's all fine now btw, 380>960 by a lot, tonga works), or RoTR with a BETA implementation as even said by Nixxes.

In Ashes, all GCN gains in DX12.

Is Hitman also bugged since it regresses there?

 

[Silverforce11]

Is Hitman also bugged since it regresses there?

It is actually, DX12 crashes a lot. That game is basically Early Access, not ready for prime time.

It hasn't looked good for DX12 in released games so far actually. Ashes is still the only one where all GPUs improve, including NVs (once AC is disabled).

 

[ShintaiDK]

It is actually, DX12 crashes a lot. That game is basically Early Access, not ready for prime time.

It hasn't looked good for DX12 in released games so far actually. Ashes is still the only one where all GPUs improve, including NVs (once AC is disabled).

Can AOTS run on Intel DX12 IGPs? ()

 

[Silverforce11]

Can AOTS run on Intel DX12 IGPs? ()

Ask Intel to work with Oxide, try harder and all, who knows, maybe Intel's fail-GPU drivers are the cause, are you privy to real info to suggest it's Oxide's fault?

ps. I am still waiting for real proof of Intel's FL12_1 claims btw, find any legit benchmark sources yet or are we just to take their word for it ala NV's 3.5GB and DX12 Async..

 

[ShintaiDK]

Ask Intel to work with Oxide, try harder and all, who knows, maybe Intel's fail-GPU drivers are the cause, are you privy to real info to suggest it's Oxide's fault?

ps. I am still waiting for real proof of Intel's FL12_1 claims btw, find any legit benchmark sources yet or are we just to take their word for it ala NV's 3.5GB and DX12 Async..

With DX12 the burden is on Oxide. Welcome to the DX12 reality.

Other DX12 games works on Intel DX12 IGP.

 

[Silverforce11]

With DX12 the burden is on Oxide. Welcome to the DX12 reality.

Other DX12 games works on Intel DX12 IGP.

More burden on developers, sure. But they do need functional drivers to do it.

I'm not privy to info to suggest who is at fault, neither are you. Feel free to tweet to Dan Baker and ask. He's pretty responsive.

Btw, do you have legit sources of Intel running DX12 in Hitman, GOW:U or Rise of Tomb Raider? That would be interesting to take a look at. I'm googling but it's not coming up.

 

[prtskg]

With DX12 the burden is on Oxide. Welcome to the DX12 reality.

Other DX12 games works on Intel DX12 IGP.

Oxide has already said they have worked together with AMD, Nvidia, Intel and Microsoft and shared code with them to get the best dx12 implementation. So I doubt the fault is from Oxide's side.

 

[cytg111]

Its so great and forward thinking we already see GCN 1.0 and 1.2 either stagnate or regress in DX12. Unless its a Fury card. And even then it depends on the game.

We are down to specific SKU type optimization. And Tonga users for example are the big losers.

Damn, could you just stay away from anything remotly AMD centric! Please?

@*Anyone* else, I was mentally comitted to upcoming Pascal for my GPU/VR needs, cause, while not perfect, better the devil you know and all that, however this AC and DX12 gives time to pause and consider the competition.
There is ONE thing that really bothers me with team green. The next 14nm card is likely going to last a LONG time, right? So here is my thinking, if AMD gets bought up, belly up, or other derivatives hereof, what about driver support? If AMD tanks in a year or two, then what? Are we left stranded with a card with no future? If its a real possibility, it should factor in when putting your money down..

 

[Qwertilot]

Likely to be at 14nm for a while, yes.

NV seem fairly likely to iterate into Volta on 14nm at decent speed like they did with Kepler/Maxwell. Not sure about AMD, but they'll probably be at least planning for it unlike this time round, so we'll maybe see something.

I wouldn't worry that much about 2+ years ahead - even if you don't really want to, if it does go all wrong then you'll generally be able to replace with something of similar performance for much less than you paid in the first place.

 

[3DVagabond]

Damn, could you just stay away from anything remotly AMD centric! Please?

@*Anyone* else, I was mentally comitted to upcoming Pascal for my GPU/VR needs, cause, while not perfect, better the devil you know and all that, however this AC and DX12 gives time to pause and consider the competition.
There is ONE thing that really bothers me with team green. The next 14nm card is likely going to last a LONG time, right? So here is my thinking, if AMD gets bought up, belly up, or other derivatives hereof, what about driver support? If AMD tanks in a year or two, then what? Are we left stranded with a card with no future? If its a real possibility, it should factor in when putting your money down..

AMD isn't going anywhere any time soon. Certainly they will last longer than Kepler driver optimizations did if that's what you are truly concerned with.

 

[prtskg]

If AMD gets bought up, belly up, or other derivatives hereof, what about driver support? If AMD tanks in a year or two, then what? Are we left stranded with a card with no future? If its a real possibility, it should factor in when putting your money down..

Say Boltzmann Initiative fails and AMD can't make enough inroads in gpgpu market and on top of that zen is another bulldozer, AMD will have to sell the Radeon group to someone to keep to keep the company afloat. The said company can be Intel, Microsoft, Apple, Samsung, Sony or whatever. The thing is, it'll be a big company and I doubt your purchase will go down the drain.
Not that I see any of those happening. Opencl is gaining ground ( no one can ignore a monster like xeon phi, right) and recently AMD got a 1petaflop gpgpu deal. Also zen looks nothing like bulldozer at the moment. I believe Lisa that its IPC improvement over excavator should be roughly 40%. She hasn't yet told a lie. Though only time will tell, how good or bad it is.

 

[cytg111]

Thanks.. I am not convinced, however, willing to take the risc when/if the time comes.

 

[Piroko]

There is ONE thing that really bothers me with team green. The next 14nm card is likely going to last a LONG time, right?

Well, depending on how Pascal competes they could pull in Volta to perhaps limited launch in 2h 2017 or put in an intermediary GPU branch with parts of Voltas features (think Maxwell 1.0). Nvidia has proven before that they can weather a storm (Fermi launch).
edit: That is, if they have a problem at all, which is yet to be seen.

So here is my thinking, if AMD gets bought up, belly up, or other derivatives hereof, what about driver support? If AMD tanks in a year or two, then what? Are we left stranded with a card with no future? If its a real possibility, it should factor in when putting your money down..

A couple of months ago I took a look at a desktop from a friend with very outdated drivers (13.12 iirc) and he had no immediate issues.
You're free to factor in a driver support drop with an arbitrary value, it is a possibility after all. But I doubt the video card itself will stop being useful the day support ends - or even a year after that. DX12 doesn't change that.

 

[Final8ty]

Say Boltzmann Initiative fails and AMD can't make enough inroads in gpgpu market and on top of that zen is another bulldozer, AMD will have to sell the Radeon group to someone to keep to keep the company afloat. The said company can be Intel, Microsoft, Apple, Samsung, Sony or whatever. The thing is, it'll be a big company and I doubt your purchase will go down the drain.
Not that I see any of those happening. Opencl is gaining ground ( no one can ignore a monster like xeon phi, right) and recently AMD got a 1petaflop gpgpu deal. Also zen looks nothing like bulldozer at the moment. I believe Lisa that its IPC improvement over excavator should be roughly 40%. She hasn't yet told a lie. Though only time will tell, how good or bad it is.

Why would you sell that part that's makes money to keep the part that does not make money going, that part is going to fail again, sell that part that is not making money and keep the part that is making money.

 

[Headfoot]

@*Anyone* else, I was mentally comitted to upcoming Pascal for my GPU/VR needs, cause, while not perfect, better the devil you know and all that, however this AC and DX12 gives time to pause and consider the competition.
There is ONE thing that really bothers me with team green. The next 14nm card is likely going to last a LONG time, right? So here is my thinking, if AMD gets bought up, belly up, or other derivatives hereof, what about driver support? If AMD tanks in a year or two, then what? Are we left stranded with a card with no future? If its a real possibility, it should factor in when putting your money down..

? Huh? Why would you be committed to any undisclosed, unbenchmarked, unknown thing? The easiest way to see who meets your VR/GPU needs is to wait for the card and VR headset to be benchmarked by neutral 3rd party press. Way too early to commit to either side if you're buying objectively...

 

[coercitiv]

If AMD tanks in a year or two, then what? Are we left stranded with a card with no future?

But I doubt the video card itself will stop being useful the day support ends - or even a year after that.

Once the red queen dies, all radeons will go silent.

 

[tential]

? Huh? Why would you be committed to any undisclosed, unbenchmarked, unknown thing? The easiest way to see who meets your VR/GPU needs is to wait for the card and VR headset to be benchmarked by neutral 3rd party press. Way too early to commit to either side if you're buying objectively...

lol.... you poor poor child.

 

[tential]

Are we left stranded with a card with no future? If its a real possibility, it should factor in when putting your money down..

Do Nvidia cards have futures? It's not like driver updates help you if you own a previous generation card lol....

They're specifically stating new features are for current generation owners only now. What more is needed to get the message through to people

 

[moonbogg]

Yes, if AMD goes out of business a year after you buy their card, that's the same as Nvidia doing very well a year after you buy one of their cards. Either way, you're screwed.

AMD situation = no driver support
Nvidia situation = no driver support

 

[poofyhairguy]

Once the red queen dies, all radeons will go silent.

Nah, the consoles plus Apple design wins (because Nvidia pissed them off) is enough to make RTG worth buying and keeping afloat.

 

[Silverforce11]

Nah, the consoles plus Apple design wins (because Nvidia pissed them off) is enough to make RTG worth buying and keeping afloat.

2016 is when AMD's $1B contract with MediaTek for custom SOC design starts, so that's extra revenue for them. Nintendo NX later this year is also another win. Polaris perf/w will get notebook wins, and the next round of Macs.

AMD will be fine. 2016 is going to be a great year from them, at the end of it, Zen arrives and attack the server & HPC segment with 32 real cores, along with enthusiast consumers.