Ask Me Anything - Official AMD Radeon Representatives

[sumludus]

Anandtech has taken to referencing your 7790/260X and 290X/290 as "GCN 1.1".

Are you at liberty to say if the 20nm shrink will be more of an Intel-esque tick and give us GCN 1.2, or if there are major reworkings at hand and we can expect a full fledged GCN 2.0?

 

[Skysnake]

My favorite is the new implementation of PowerTune on the 290X and 290. There's a lot of doom and gloom around the 95C temperature, because people are used to a world where the product is designed to run as cold as possible... but that's not the world we're living in with these units. The doom and gloom is based on an old viewpoint.

Ok so far, when you design the card to run as hot as it is, than the RMA should be no problem for the customers but.

95C is the optimal temperature that allows the board to convert its power consumption into meaningful performance for the user. Every single component on the board is designed to run at that temperature throughout the lifetime of the product.

Are you really shure?

The GF100, GF110, Tahiti and i think Hawaii will also show a very strong powerconsumptio-temperature correlation.

my tests of a HD7970 shows, that there is a coupling of round about 1.5-3 W per °C at full load. So colder card means less power consumption for the cards with the chips a named.
So i cant follow your answer.

Mantle can be used to increase raw FPS, or to increase image quality while maintaining the same FPS.

It is a complete and standalone graphics API, meaning Mantle must be capable of coding and rendering all the in-game effects you see today. But Mantle also needs to be sufficiently extensible so that we can collaborate with game studios to create the effects of
tomorrow—and it is extensible to do so!

Yes, Mantle is a complete graphics API capable of doing anything you see today. And more. The primary advantage, however, is that it can speak directly to the hardware in a unique way that other APIs cannot.

means this, that that mantle eliminate the OS traps for the GPU driver?

Has AMD considered about making Downsampling software for high end cards? As you guys have stated before, the 290 is meant for 4k monitors but currently the standard for monitors and TVs is 1080p. It has been proven and tested that downsampling improves image quality. (I think it will be a great feature for SteamOS)

We are not considering any such software.

To poor. For the german picturequality-"Nazis", DS is a very hot and important thing, that gives nVidia a big advantage over AMDs GPUs. DS is a important thing, when you have Crossfire systems. They are so fast, that everything else is a waiste.

 

[Thracks]

What is Team Red or is it Red Team? Can I join? Lol

#AMDRedTeam!

My colleagues Heather, Stella and I came up with the idea to give people a chance to share their GPUs, rigs, favorite games and more. Sometimes people need a "spark" of opportunity, and we hoped that the AMDRedTeam idea for Twitter would give people that chance to share their systems. So far the response has been incredible.

If you want to join, Tweet @AMDRadeon and ask to join!

 

[Zalen]

Hello!

Firstly let me thank you for your incredible products.

Secondly, and although I suspect a direct answer won't be possible, it is an ask me ANYTHING!

Any word on whether Battlefield 4 will be included in Never Settle?

 

[mythr1l]

Can you give us any information on the never settle program? A lot of people are wondering what future titles it will contain and currently the titles on offer seem rather dated.

Thanks.

 

[smeezekitty]

Where can I get AMD Radeon case stickers

Any estimate for crossfire stutter to be completely fixed?

Improvements to Linux drivers?

 

[manitoublack]

I would suggest a second AMD AMA for the CPU/APU side of things. Lots of questions here on the subject(s).

Agreed.

Also to AMD, what's the hold up on Mobile Kavari APU's? I'm hearing March2014 or later for first retail products, that's 5 months from now. Or in another time scale, 4 months from the first appearence of a 28nm/GCN APU in a retail product (XB1/PS4)

Has focus on the XB1 and PS4 pulled resources from getting Kavari to the channel?

Kavari looks like a great product, that I want to support, just not sure if I'll be able to wait that long...

regards
Jordan

 

[Thracks]

Hi, thanks for coming here and answering questions, my question is sort of cpu related but not about the next gen ones that arent released so hopefully I get my answer.
Anyway what I would like to know is will mantle remove/help the cpu bottleneck eg if I play mantle supported games will it be worth my while buying a high end r9 290 card to pair with my trinity 4 core cpu due to the draw call increase or will the cpu still require a big upgrade to get enough out of the card to warrant buying it over say a 7870?

The draw call improvements of Mantle does help alleviate cases where the CPU is the bottleneck. Mantle is very good at parallelization. Beyond that, it's too early to say.

 

[Thracks]

Where can I get AMD Radeon case stickers

Any estimate for crossfire stutter to be completely fixed?

Improvements to Linux drivers?

1) See here.

2) Of course. We've already been making massive commits to the Linux 3.11 and 3.12 kernels to substantially improve performance and features for Radeon on *NIX. Steam Machines only encourage that process.

 

[con635]

Thanks for your time thracks exciting times for gaming!

 

[Thracks]

I would suggest a second AMD AMA for the CPU/APU side of things. Lots of questions here on the subject(s).

Agreed.

Also to AMD, what's the hold up on Mobile Kavari APU's? I'm hearing March2014 or later for first retail products, that's 5 months from now. Or in another time scale, 4 months from the first appearence of a 28nm/GCN APU in a retail product (XB1/PS4)

Has focus on the XB1 and PS4 pulled resources from getting Kavari to the channel?

Kavari looks like a great product, that I want to support, just not sure if I'll be able to wait that long...

regards
Jordan

I can't answer any of these questions directly, because the truth is that I don't know. But I hope some of them will be answered for you at the AMD Developer Conference next week. Lots of information to be had on HSA and the like there.

 

[monstercameron]

what can true audio do outside of gaming?

 

[manitoublack]

I can't answer any of these questions directly, because the truth is that I don't know. But I hope some of them will be answered for you at the AMD Developer Conference next week. Lots of information to be had on HSA and the like there.

Appreciate the reply, will be tuned into the press event. There's demand for these products. But the delays on original promises are hurting.

Regards
Jordan

 

[decimad]

Are you planning to raise the Fan PWM Target when the GPU sound solution detects loud sound/music?

 

[Thracks]

what can true audio do outside of gaming?

I love this question.

First and foremost, TrueAudio is a programmable DSP. Its most obvious use is signal processing for gaming audio, but you could conceivably program it to do audio filtering, voice control, biometrics or anything else a powerful DSP is capable of.

 

[Thracks]

Are you planning to raise the Fan PWM Target when the GPU sound solution detects loud sound/music?

No, but we do plan to switch to direct RPM control vs. PWM very shortly. PWM control does not always yield the exact RPM you're looking. Converting an electrical pulse to a mechanical rotation isn't an exact science! It's subject to the design variances of the fan and the PWM module, so we're going to make it the exact science it deserves to be.

Not a literal answer to your question, but a fun fact I wanted to share.

 

[decimad]

Are you planning to raise the Fan PWM Target when the GPU sound solution detects loud sound/music?

No, but we do plan to switch to direct RPM control vs. PWM very shortly. PWM control does not always yield the exact RPM you're looking. Converting an electrical pulse to a mechanical rotation isn't an exact science! It's subject to the design variances of the fan and the PWM module, so we're going to make it the exact science it deserves to be.

Not a literal answer to your question, but a fun fact I wanted to share.

Is the necessary equipment to measure the Back-EMF or an encoder for detecting the speed already on the current boards? That makes me wonder why it wasn't feedback-controlled in first place. Or are these plans for the future hardware revisions?

 

[Fredrik Aldhagen]

How do you convince developers to use Mantle when other APIs are vendor neutral?

For existing games, how hard or easy is it to "port" it to use Mantle?

What's your position on Mantle and OS X? (currently Apple use nVidia GPUs, so it may not make much sense, but without Mantle AMD GPUs might be a hard sell for future Macs)

 

[Cataclysm_ZA]

I have a couple of questions, hopefully they're not too far-reaching:

For Hallock: PR management for AMD in South Africa is non-existent. One guy seems to be doing the job on his own and there's quite a lot of mindshare being taken over by Nvidia here (Intel as well, but as Thrash said, no CPU guys are here). Why isn't AMD doing any marketing in my country (which warmly welcomed an APU stand at a major games convention in 2012)? And If you're able to answer, what is AMD doing for South Africa in particular to make sure prices stay reasonable and keep your products affordable? Case in point, currently an A10-6800K costs the same as a Core i5-4330. On the GPU front you're still mostly price-competitive.

For Corpus: Mantle has been talked about and mentioned in the same name as some big AAA studios. What can Mantle offer to indie developers and would the API be able to use some of the same optimisations on lower-class products? (e.g. the G-series chips with GCN graphics)

For Nekechuk: Since the R-200 refresh we've gone from having several variations on a theme for GCN down to just seven SKUs (R7 240, 250, 260X; R9 270X, 280X, 290, 290X). Given that there's space inside your lineup for an R7 260 and R9 270, 280, would it be feasible for AMD to release cards for those model numbers, or can we expect a simplified lineup of just seven cards in every generation from now on? (I'm personally hoping for the latter)

For Nalasco: Tom Pieterson is everywhere on the internet talking about Geforce products. I hardly see you anywhere on the net or Youtube. Will this change?

For Parfitt: GSync got tongues wagging in spite of the fact that people wouldn't be able to see the difference on a compressed Youtube video. Is AMD considering a similar solution, or working towards one that's more open than GSync? I was also disappointed that I'm not able to run an Eyefinity setup with monitors of different sizes and resolutions, despite this being one of the features teased in the Catalyst 12.2 beta and was called Eyefinity Bezel compensation (http://bit.ly/1cEGAdh). Lastly, AMD still does not support PLP (portrait-landscape-portrait) monitor setups for gamers - will that ever change?

 

[blackkstar]

I have a few questions!

1. When can we see an updated control panel for fglrx? Is there any chance of Catalyst Control Center switching to a different framework from .net to something more portable like QT or GTK?

2. Will you be bringing Mantle support to other desktop OSes like Linux?

3. Will we be seeing a higher end gaming platform to power Hawaii based GPUs come out of AMD's other divisions soon?

4. Are you working with Valve to bring Mantle to Source game engine as well as with SteamBoxes?

Really curious about Linux in general. I've been running Gentoo and my FX 8350 is much faster in Gentoo than in Windows in a lot of important tasks, and I'd like to avoid Windows as much as possible (plus Aero and Metro are tacky and hideous compared to what you can do with KDE)

Thanks.

 

[Thracks]

Is the necessary equipment to measure the Back-EMF or an encoder for detecting the speed already on the current boards? That makes me wonder why it wasn't feedback-controlled in first place. Or are these plans for the future hardware revisions?

We can already monitor the speed and adjust accordingly, but direct RPM control is simply smarter.

How do you convince developers to use Mantle when other APIs are vendor neutral?

For existing games, how hard or easy is it to "port" it to use Mantle?

What's your position on Mantle and OS X? (currently Apple use nVidia GPUs, so it may not make much sense, but without Mantle AMD GPUs might be a hard sell for future Macs)

We haven't had to convince them. Every single one of them has come to us and asked for it without prompting!

Right now we are concentrating on the PC platform for Mantle, so that's our position on OS X.

 

[smeezekitty]

what can true audio do outside of gaming?

I love this question.

First and foremost, TrueAudio is a programmable DSP. Its most obvious use is signal processing for gaming audio, but you could conceivably program it to do audio filtering, voice control, biometrics or anything else a powerful DSP is capable of.

That makes be curious for other applications such as software defined radio (assuming software support)

 

[decimad]

To make the three complete and then shut up: The 95°C temperature target fits nicely with the theory that you're trying to get the same heat flow through a smaller footprint in relation to nVidias Titan at its 80°C (considering the die size ratio and an ambient temperature of max 30°C). Was the reduced (and more affordable) die size the rationale to reach out closer to absolute maximum of what chips can handle before melting?

 

[SPLWF]

Dear AMD

I wanted to know if you have plans for an 8 core APU?

I also want to know why aren't your partners pushing M-ITX, or better designed mATX motherboards for your 8 Core FX chips?

Nowadays people want small form factor

 

[Skysnake]

I am not a developer, but I will answer as much as I can.

Thx. I know my questions are always not so easy to answer.

1) 4 blocks with 11 CUs each is the configuration of the R9 290X.

Yeah i know, but what is with the real physical implementation on the DIE?

When i think about crossbars, codings and so on, 11 is a REALLY ugly number. As a student of computerachitekturs it is really hard to belief, that there are really only 11 CUs per block. It is really ugly, and i would never design something ugly, when there are no really good reasons, and die size is no really good reaseon. There is enough space for 4 more CUs.

2) GDS via GWS should still be in the architecture.

Yeah, but there is no documentation, and the bit of documentation is gone...

The ISA documentation is no big help, when you don't know the limitations and so on. And nearly nobody would write Assembler code... For SIMD it is possible, but it is notheing you want to do without a really good documentation.

3) Yes, the R9 290 and 290X are 1:8 DP. As to "why" I do not know, except to say that these are primarily gaming graphics cards, so it would make sense that the architectural implementations are optimized for these types of workloads.

Yeah, but the Radeon card were really good development cards and also for students for casestudys. I think there are a lot of people who are really sad, that AMD begin also to cripple there Consumercards. AMD was there a real model student in the past.

4) I'm not sure.

I hope really, this is a feature the GPGPU folks wait for years, and it is a REAL gamechanger. It make complet new algorithms on a GPU possible. To bad, that i am waiting since over a year on this feature ;(

5) I do not know what the FirePro team plans.

6) Not sure!

Is there a possibility to get more informations about this? To poor that i am not able to go to the SC....

RDMA shouldn't be such a big thing. More software and a bit of MMU work than everything else. I know what i say, i am porting a linux kerneldriver for a HPC networkcard.

7) Yes!

8) 290 and 290X support system unified addressing.

Yeah i know, but the important part was the "coherent"

Oh i have also some more questions.

9. nVidia made some marketing stunts with there FCAT. Why did AMD no real deep analysis of FCAT, and the problem of divergency between Ingametime and output-/real-time? Framepasing is not the golden eag. It is just a patch, but it do not eliminate the reason for µstuttering. To poor, that AMD not blame nVidia for this.

Is there a chance, that Mantle give the Game-developer enough knowledge about the rendering and the output, so he is able to sync the gametime with the output-/real-time?

10. Is Hawaii 28nm HPM or 28nm HP process from TSMC?

11. I belong to the GPGPU/HPC folk. So FASTRA was a really interesting thing for me since years. How is the situation with FirePro and Radeons, when i want to put 6+ GPUs on a single Motherboard. Will it boot?

I know somebody who let 8 GPUs run under linux, but under Windows the machine don't boot. Is there a driver fix yet, or will it there in there future? GPGPU Development with VirtualStudio and your reall! good performanceanalyser plugin for it makes you want to develop under windows.

EDIT:
Oh and there is one more thing. I try to get some connections since over 2 years to AMD for informations how to read the GPU registers directly.

Why i want this? I want to write my own "MSI Afterburner" for windows and for linux! Unther linux i know no program that allows you to change the voltage, clocks and fan during you monitore the temperatures. linux becomes more important, and so i think there a more than enough people out there who would be happy to have such a program.