DooKey
Golden Member
- Nov 9, 2005
- 1,811
- 458
- 136
Yeah, that wishful thinking thing.......video card company fans are great at it aren't they?
Yeah, that wishful thinking thing.......video card company fans are great at it aren't they?
Azix
Golden Member
- Apr 18, 2014
- 1,438
- 67
- 91
LTC8K6
Lifer
- Mar 10, 2004
- 28,520
- 1,576
- 126
Odd that I keep getting into trouble for wanting a full Tonga card, huh? :biggrin:
Magee_MC
Senior member
- Jan 18, 2010
- 217
- 13
- 81
It seems very relevant. AMD looks to have played a long game and whether or not all of the pieces were intentionally put into play, they do all seem to be coming together for AMD in 2016. This is the picture I've been able to piece together from all of the different discussions on these subjects.
GCN architecture with hardware ACEs - faster AC/more efficient parallel compute processing
Getting GCN into the consoles - weaker processors - more need to take advantage of low level API abilities like AC in order to maximize efficiency
Mantle/Vulkan/Metal/D3D12/XBOne/PS4 - All of the next gen APIs are derived from Mantle or extremely Mantle like, and play to the advantages of the GCN architecture.
Devs are now becoming accustomed to coding their games to take advantage of the strengths of the low level APIs in the console, which will help them to use DX12 to extract maximum performance from AMD's GCN DGPUs.
With AMD_Robert on Reddit saying,
https://www.reddit.com/r/AdvancedMi...ide_games_made_a_post_discussing_dx12/cuom7cc
it seems that AMD may be coming into a GW like position to help devs extract the maximum efficiency from GCN and the new APIs.
It also looks like NVIDIA does support AC on Maxwell, but with an implementation using software/drivers that while functional, is less efficient, has increased latency and uses CPU resources to function, while by comparison AMD's implementation of AC using hardware in the GPU is faster, has less latency, but is more power intensive than NV's software implementation.
This makes me wonder how much of the efficiency between GCN and Maxwell is due to AMD's use of hardware (ACEs) to implement AC as opposed to NV's use of software/drivers.
AMD also has an advantage in the coming VR systems from it's lower latency AC with GCN and its Liquid VR software which is Mantle in a different form.
Another consideration is whether AMD will be able to leverage their APUs to assist their DGPUs in the future.
All of these point to AMD being in a very strong position to benefit from their strengths; GCN in consoles/DGPUs/APUs, low level APIs designed for GCN/GCN type architectures, hardware implementation of AC with lower latency, which leads to better gaming and VR experiences.
On the other side of the picture, NV has several significant challenges coming up with regards its upcoming architectures. Pascal will be a challenge with not only with a node shrink, but also NV having their first crack at implementing HBM.
To add to this, either Pascal has moved to a hardware implementation of AC, which will be another new technology NV needs to get right, or they stayed with the software/driver method in Maxwell and Pascal is more or less an improved smaller version of Maxwell.
If it's the first, then NV could experience growing pains with any or all of the new technologies and possibly by moving to a more GCN like architecture they may become less power efficient compared to GCN. If they haven't incorporated hardware AC then they will need Volta with hardware AC as soon as possible.
The final piece of the picture is that Pascal will be competing with AMDs Arctic Islands which will probably incorporate improvements in all of these technologies based upon AMD's experiences while NV will be working through many of these for the first time.
/RS mode :biggrin:
@Magee_MC
It's why I posted awhile ago, that I was excited about Zen HBM2 APU + DX12 multi-adapter.
Can you imagine a powerful APU with Artic Island iGPU on it, parallel execution of rendering with the dGPU?
There's been a lot of doom & gloom for the past year but I saw beyond that, I saw their potential coming together with next-gen API, HBM powered GPUs and APUs.. even the Nintendo NX win is a big deal because it keeps them afloat while their long-term strategy fall into place.
It's why I posted awhile ago, that I was excited about Zen HBM2 APU + DX12 multi-adapter.
Can you imagine a powerful APU with Artic Island iGPU on it, parallel execution of rendering with the dGPU?
There's been a lot of doom & gloom for the past year but I saw beyond that, I saw their potential coming together with next-gen API, HBM powered GPUs and APUs.. even the Nintendo NX win is a big deal because it keeps them afloat while their long-term strategy fall into place.
hopefully the nx is more like the wii less like the wii u for both nintendo and amd's sake.
Azix
Golden Member
- Apr 18, 2014
- 1,438
- 67
- 91
If nvidia has to put in a hardware scheduler to properly support dx12, their power consumption might not be pretty at all. Considering they don't have this in big maxwell and STILL end up near GCN power consumption at the high end. Major hardware changes might not be pretty at all. AMD hasn't been compromising so they might have a smoother transition. They have the hardware features already and even experience with HBM.
Next year will be fun.
Next year will be fun.
http://www.overclock.net/t/1569897/...ingularity-dx12-benchmarks/2130#post_24379702
Looks like Oxide & AMD's FUD and blatant lies blew up in their own faces.
Looks like Oxide & AMD's FUD and blatant lies blew up in their own faces.
It's funny some of you attack Oxide's credibility when they pointed our the flaws, then now you quote them to justify NV's crap drivers which don't even function for a feature they claim support, which turns out to be a software solution. Classy.
Goatsecks
Senior member
- May 7, 2012
- 210
- 7
- 76
I am sure all the, `men of science and citations', will work quickly to show that this is nonsense or, at least, thier previous conclusions are still sound.
The first DX12 game, in alpha, looks like its benchmark results could be unreliable? Who would have thought! :whiste:
Glo.
Diamond Member
- Apr 25, 2015
- 5,930
- 4,991
- 136
I rather wait for real results, in the first place.trust me nvidia will eat their own words on this one
Black96ws6
Member
- Mar 16, 2011
- 140
- 0
- 0
Spanners
Senior member
- Mar 16, 2014
- 325
- 1
- 0
Sorry are we reading different links?
Do you have first hand knowledge of Nvidia's Async Compute capabilities on a hardware level? If not then until Nvidia comes out with something official to clarify this you're just choosing the second hand information that suits your narrative. Quite hilariously from the same people you just called blatant liars and peddlers of FUD.
- Jun 10, 2004
- 14,600
- 6,084
- 136
Arachnotronic
Lifer
- Mar 10, 2006
- 11,715
- 2,012
- 126
Keysplayr
Elite Member
- Jan 16, 2003
- 21,219
- 54
- 91
Why won't you people settle down? Silverforce, I warned you about blindly following the hype and propagating the hype. What if Oxide's statements are true? What if they are false? Would it not just be wiser to say NOTHING at this point and wait for REAL data?
You chastise a member for not taking oxide's word when it suits them. Fine, but don't you see you are doing the exact same? Just in the other direction?
When will this stop.... I wonder.
Also, yeah, nvgpu your comment does not reflect in the link you gave. All it shows is that oxide has talked with Nvidia and Nvidia states they are in the process of fully implementing Async. Datz it.
You chastise a member for not taking oxide's word when it suits them. Fine, but don't you see you are doing the exact same? Just in the other direction?
When will this stop.... I wonder.
Also, yeah, nvgpu your comment does not reflect in the link you gave. All it shows is that oxide has talked with Nvidia and Nvidia states they are in the process of fully implementing Async. Datz it.
Last edited:
Mate, nowhere does Oxide say anything about hardware async compute in that quote, so I corrected nvgpu, he's trolling with a source he has no understanding of or deliberately misleading.
I have never misrepresented Oxide or AMD when I link them as a source, it's in plain english.
The current issue is quite clear, NV software emulates part of the process and their drivers are borked (if you believe Oxide). So now the ball is in NV's court.
Easy. Nvidia should just make a statement one way or the other instead of letting this snowball like this. Nvidia can stop this by just releasing a statement. But, noooo.... They have to remain silent about the whole thing. Therefore, the community have no choice but to put pieces together and try to figure it out. That's what's happening right now.
It's actually not wiser to say nothing because it would do the community injustice. Instead of allowing this issue to be swept under the rug, the community is exposing it one way or the other; just like the 3.5GB crap. If there's no push back from the community, Nvidia might not be inclined to give us a straight answer.
Like I said, Nvidia can easily put an end to this. But, right now, they're not telling us anything.
You still didn't answer my question about market share so now would be a good time to repeat it I guess.
Do you have a breakdown of Nvidia's sales by graphics card? I'm sure that most sales are slow pre-Maxwell cards that are sold in OEM machines from Dell and HP but still count as a discrete card. So even if Nvidia has 80% discrete market share that number means nothing if 80% of the cards they sell are not faster than APUs.
What difference does it make at this point? Are there any DX12 games available right now? AMD had a head start on NVidia driver wise because of Mantle, which is similar to DX12.
Also, the only thing being done in software is the scheduling, which is actually more power efficient than having hardware schedulers like AMD.
NVidia have some of, if not the best driver engineers in the World and so they have the confidence to implement such a thing in software, rather than using transistors for it. We'll see how effective it is when the driver is released.
As for AMD, they have invested heavily in AC to the point of having multiple hardware schedulers, so they're hoping it will pay big dividends in the long run.
It may be more power efficient, but it also (potentially) comes with significant latency penalties, hence the "potentially catastrophic" remark from Oculus.
more efficient if measuring the gpu power only and not total system power. If his is true then it just shifted power draw from the gpu to the cpu and in the process became less flexible. Doesn't sound like a good trade off to me but they made bank on it, so maybe I don't know anything.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 22K
-
News NVIDIA and Intel to Develop AI Infrastructure and Personal Computing Products- Started by poke01
- Replies: 397
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
Facebook
Twitter