It's chicken and egg scenario. If no one makes the hardware, no developer is going to use it. At least AMD are giving the devs the tools, Nvidia just sits back and tells everyone how useless a tech is until someone else makes it, then they copy it and say it's the best thing ever.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Poll: ** The AMD VEGA Thread **
- Thread starter TNA
- Start date
- Status
It's chicken and egg scenario. If no one makes the hardware, no developer is going to use it. At least AMD are giving the devs the tools, Nvidia just sits back and tells everyone how useless a tech is until someone else makes it, then they copy it and say it's the best thing ever.
Soldato
- Joined
- 18 May 2010
- Posts
- 23,645
- Location
- London
That's the point in Open Source tho, developers can just pick it up and run with it and the community is on hand to provide support.
I thought we were talking about True Audio and it's lack of take up, Nothing to do with Mantle/DX12/Vulkan. The point is if AMD want software houses to take it up they need to push it by helping them, there is no insinuating anything.
What is True Audio FFS?
Associate
- Joined
- 27 Dec 2014
- Posts
- 1,686
- Location
- Southampton
I thought that was Apple ...
Soldato
- Joined
- 28 May 2007
- Posts
- 10,222
I don't know why you quoted me. The post you quoted was just me having a laugh. A few posts above I said Vega is very different to Fiji. I never went in to detail about all the changes as I have read enough to know a bit. Won't be going through them but the fact that Vega is a 500mm2 chip tells you that Vega has a lot extra compared to Fiji as Fury X would be somewhere around 300-350mm2 on 14nm. What everyone is wondering is where the performance has gone.
Last edited:
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 31,192
Nah, they have nowhere near the manpower anymore to do something like that. Hence them struggling with Fury and now Vega drivers imo. They always need about 6 months after launch to sort it.
Caporegime
- Joined
- 9 Nov 2009
- Posts
- 25,754
- Location
- Planet Earth
AFAIK,the PS4 has a similar audio setup,so its most likely Trueaudio was something developed for the PS4 which AMD decided to add to their cards.
It is not binaural audio that you want, what you want is proper convolution reverb that takes environmental features into account, by using wave tracing and first order reflections. Which is what Aureal did back in the late 90's, which is still the best audio system to date. but it is computationally expensive, which is why you don't see it running on software audio stacks.
Now TrueAudio Next does the above, not the first True Audio, but the version that came with Polaris and now Vega. it uses some of the CU's to perform first order reflections and convolution reverb, taking environmental features and materials into account.
AMD were bigging up TrueAudio next for VR, but even in non VR it makes a massive difference to immersion. Audio has been one of the most under developed parts of gaming for the past fifteen years to the point where it is no better than audio we had back in the early 90's.
I wish they'd hurry up and release this mess so we can move onto a new/different hype train
Yeah, we need to make way for the Navi hype train.
i was having a joke. i think if i had to read this thread from the beginning i would consider throwing myself off a building
the performance is there, for me it is simply a combination of a) global foundries producing chips that clock at 1600 rather than 2000mhz b) AMD driver team not being ready for gaming as the focus is on compute perf for deep learning.
the big question for me is whether RX vega using the same silicon or bios as FE with compute nerfed.
if compute is nerfed it will bring the clock speed up and the thermals down. i bit like AVX on Intel.
the fact that the RX card is being shown against a 1080 puts a lid on expectations however, i don't see this as a failure of AMD more a failure of globalfoundry.
Architecture plays a major parts in the clockspeed a part can achieve, considering GP107 in the 1050TI is produced on 14nm LPP as well and it can hit 2GHz.
Also we don't know the limits of Vega clocking yet, since it seems to be hitting a power wall before a frequency one. Since it can hit just over 1700 at 1.2V but hits power limits. it takes 1.25-1.3V to get polaris to only 1500 in comparisson, when vega can run 1600 fine at 1.075V
Soldato
- Joined
- 28 May 2007
- Posts
- 10,222
i was having a joke. i think if i had to read this thread from the beginning i would consider throwing myself off a building
the performance is there, for me it is simply a combination of a) global foundries producing chips that clock at 1600 rather than 2000mhz b) AMD driver team not being ready for gaming as the focus is on compute perf for deep learning.
the big question for me is whether RX vega using the same silicon or bios as FE with compute nerfed.
if compute is nerfed it will bring the clock speed up and the thermals down. i bit like AVX on Intel.
the fact that the RX card is being shown against a 1080 puts a lid on expectations however, i don't see this as a failure of AMD more a failure of globalfoundry.
My humour detector must have been off
. Yea it looking more like a gtx1080 competitor.
AFAIK for headphone users rendering with binaural impulses is still needed to fully take advantage of convolution reverb?
Soldato
- Joined
- 18 May 2010
- Posts
- 23,645
- Location
- London
Why do we have no PCB leaks, box pictures.... Performance leaks etc?
Screams of paper launch!
Screams of paper launch!
Yeah, if there are no actual performance (+power efficiency) improvements in architecture over GCN how it can take this long to have new card?
While sales volumes are elsewhere having proper high end cards available is good for maintaining PR.
If architecture was same old GCN with new paint and wax job (like Polaris) should have been easy to have it ready for release year ago.
That's because sound is best calculated from that original truly 3D sound environment/model of game instead of some discrete channel rendering. (flattened to plane)
Like what Aureal did first, and then Creative in X-Fi cards.
But then came Windows Vista...
The reason why hardware accelerated 3D audio was killed off was all down to Creative. Microsoft were ready to standardise a 3D audio pipeline, but Creative essentially killed that off when they destroyed Aureal through lawsuits. After Creative destroyed Aureal, they bought them up and did nothing with the tech, which was far superior to creatives EAX at the time, since Aureal performed True 3D first order reflections and convolution reverb based upon a games assets and environments.
X-Fi cards don't perform 3D environmental audio acceleration, all of their 3D audio is just clever surround mixing and predefined reverb effects. It has nothing to do with the games actual environment.
Caporegime
- Joined
- 8 Jul 2003
- Posts
- 30,063
- Location
- In a house
Ive said it before, concentrate on the here and Now like Nvidia, and also, get your performance there on release day, not a good 6/8+ months later, when you've already lost the sales, as that won't get you market share, and thats what you need first, before anything else.
Doesn't help when you are cash strapped, Nvidia also had nearly a year to perfect their Maxwell drivers since they had Maxwell 1 on the market before launching Maxwell 2. Then the transition to Pascal was a relatively small step.
Considering they had no smaller part to perfect the drivers on before release and not a lot of time with actual hardware, getting the drivers perfect when you need to launch a part can be difficult.
- Status