[Sweclockers] AMD opens up about Freesync

[Mondozei]

An interesting interview with Robert Hallock from AMD about Freesync, done by Sweclockers. Some stuff we already know, some stuff is explained in further detail.

Sweclockers have had the good sense to translate the interview into an English version as well. It's a brisk read:
Source

 

[BrightCandle]

Still not technical details about how it works, only further trumpeting that it does but without saying how its actually different (beyond again saying it requires no special hardware which has already been determined as absolutely untrue).

Its also kind of unbalanced when it comes to comparing the benefits. Sure Freesync as a standard in theory should be cheaper and have no licencing. But it fails to mention that the Gsync module also supports low persistence mode and 3D as part of its package of tricks in addition to gsync, which is a benefit and is what a balanced report would have mentioned.

I hope it works as well as they say it does and that its as close as AMD says it is. But having already cut through a lot of the lies and watched AMD back trace repeatedly on what it is and the mistakes they have made in explaining it and further having not even seen it demoed at any point I remain skeptical until more details are revealed, and this doesn't reveal anything we hadn't already heard.

 

[Flapdrol1337]

I'm just glad there's supposedly no extra latency. Was worried about that.

I cringe at the "freesync predates gsync" and the "communication overhead" parts though.

 

[Mand]

Ugh. This is just painful to read. The level of spin is just staggering. Let's go point by point.

The engineering timeline of the Radeon R9 and R7 Series, which feature Project FreeSync-compatible display controllers, establishes that FreeSync predates G-Sync by a healthy margin. Both technologies aim to deliver similar user benefits, but Project FreeSync will accomplish its goals with open industry standards that don’t require any licensing fees or contracts from participating parties. History has more or less proven that this strategy enables technologies to proliferate faster and cost less, and we think that’s the right way for everyone.

Yes, the R9 and R7 series were in development for a long time. So was Kepler. So their claim that because R9 and R7 came first, and happened to include things that not even AMD had ever mentioned of being available for variable refresh is flatly false. R9 and R7 may have happened before G-Sync was announced, but that doesn't tell the actual story involved.

As far as open standards, yes they can provide benefits. But that doesn't change one fact: G-Sync has been out for five months, and FreeSync is a long way from being in a product and has not been demonstrated. That fact is something AMD would like you to overlook by claiming, inappropriately, that just because R9 and R7 development started long ago that they were actually working on variable refresh. They have not done anything that demonstrates that development, or even mentioned it ever, until the FreeSync demo at CES, a good two months after the G-Sync announcement and the same week that an actual product was available for actual purchase. So this entire claim is completely bogus.

Make no mistake, providing dynamic refresh rates to users still takes a lot of ‘secret sauce’ from the hardware and software ends of our products, including the correct display controllers in the hardware and the right algorithms in AMD Catalyst.

Quoting this for emphasis. Nice of them to mention this, one of the few pieces of legitimate information, especially since they tried to claim this wasn't the case in prior statements. I'm very glad they said this, as now I can use it as proof to the contrary when people come in and say that FreeSync/A-Sync won't require new hardware and that it won't take any development and that a VESA spec update for an optionally-supported feature means all DP 1.2a displays will support FreeSync.

What is the basic functionality of FreeSync. What does it do and what are the advantages?

-snip-

Nice discussion of what the technology is supposed to do. This is all accurate. Variable refresh is a good thing for all of the things discussed here, no matter whose implementation it is, FreeSync or G-Sync.

Project FreeSync does not require extensive buffering, because the need for such techniques is virtually eliminated when you can match the display timings to the framerate of the GPU on the fly. Within the range of refresh rates supported by a compatible display, Project FreeSync will allow the absolute minimum input lag with single- or double-buffered rendering.

Good. Previous explanations have indicated that there would end up being buffering involved, which would be bad, but it seems that is not the case. I'd like to know how, though.

There are three key advantages Project FreeSync holds over G-Sync: no licensing fees for adoption, no expensive or proprietary hardware modules, and no communication overhead.

1) There has been no evidence that Nvidia is charging licensing fees for adoption, only that they haven't decided to also give AMD access to G-Sync. In Nvidia's words, they don't want to "do the work for everyone." That doesn't mean that they go from there to charging display manufacturers for including G-Sync. Yes, proprietary tech can be used to extract licensing fees, but that is not guaranteed. They could very well just be looking to use G-Sync to drive GTX sales, which strikes me as incredibly likely. They're a GPU company, not a display controller company. Licensing fees to drive away monitor partners would be counterproductive to their overall goal as a company. The only people they'd charge licensing fees for is AMD, I'm pretty sure they'd rather keep unique tech as a differentiator rather than whatever AMD is willing to pay them for it. Just as I'm sure AMD isn't going to hand them the FreeSync code even though it's an "open" technology.

2) Lie. Contradicted by the very same interview:

Make no mistake, providing dynamic refresh rates to users still takes a lot of ‘secret sauce’ from the hardware and software ends of our products, including the correct display controllers in the hardware and the right algorithms in AMD Catalyst.

So they know it's false but they say it anyway. That's called a lie, folks. Just because the hardware won't be proprietary, that doesn't mean it won't be expensive, particularly if it's taking a lot of "secret sauce" in order to make it work. Lying through their teeth. This goes way, way beyond spin.

3) The communication overhead is in reference to the polling that G-Sync has to do, to see if the GPU is ready for a new frame. The overhead is a real, measured, and characterized thing, about a 2-3% drop in framerate over disabling G-Sync. I would love for AMD to substantiate their claim that there is "no communication overhead." I see none, so far. And given that they were willing to outright lie earlier in the sentence, I'm not too keen on taking statements like this on faith. For their part, Nvidia has said that the polling isn't absolutely required for G-Sync, only a product of their initial attempts, and that they're working on eliminating it.

Compatibility

Pretty standard stuff. Both of the current gen architectures from each company will support their version. One thing I'd love for them to say, though, is that they're going to make sure that Nvidia cards will have access to FreeSync. If they're going on and on about how awesome it is for it to be open and wonderful, is it the case that you need an AMD card to use it? And if you need an AMD card to use it, how is it not proprietary? I get that they can't really comment on what Nvidia may or may not do, but even the attempt at indicating openness of the technology that they presented with Mantle is absent in this case. I find that highly suspect.

There are several industry partners working on Project FreeSync, but it would not be right for me to announce a partner’s plans on their behalf. Rest assured, when the time is right we will tell the world!

Why the slow-play? Nvidia announced four major display manufacturers in their original presentation of G-Sync. Nothing I've heard from any display manufacturer, including ones other than those four, has indicated anything other than overwhelming support for the concept of variable refresh. So this begs the question: why are none of them coming out to support FreeSync like they did for G-Sync?

We expect Project FreeSync-ready monitors to be available in retail within 6-12 months, and prototypical monitors suitable for tradeshows or press demonstrations to be ready within 4-10 months.

That's good, I guess. What about this whole "we totally thought of this first" and "we have several industry partners" if you're ending up 11-17 months behind your competitor? Hard to claim you did it first when they beat you to market so incredibly badly. Let's just recall for a moment that the first we heard of variable refresh from AMD was the very same week that an actual G-Sync product hit the market.

Thank you for your time and we wish you best of luck with Project FreeSync!

I wish them the best of luck too. From everything they've shown me so far - which is not much - they're going to need it.

 

[KingFatty]

It's one thing to deconstruct a statement and poke holes in it.

But it's quite another if all you are doing is refuting one assertion with your own assertion, that is unsupported.

For example, regarding the statement about the timeline establishing the pre-dating of G-sync - if you want to argue against that statement, it would be helpful to support your counter-points with some evidence. Otherwise, it's just your word against theirs, and is not moving the debate one way or the other. he said she said.

As for the "secret sauce" in their hardware, aren't they referring to the R9 and R7 series cards that apparently have the secret sauce already baked in? I fail to understand how you are so certain that it's a false statement they are making, can you elaborate with support?

 

[piesquared]

Excellent interview, well done Robert! I'd recommend everyone read the actual interview instead of out of context, spinning bullet points.

So I think it's clear that A-Sync makes G Sync pretty much irrelevant. NV and their press outlets have been bloviating about how G sync will take over the world for several months now, yet actual hardware is as rare as hens teeth. My guess is that monitor manufacturers are just going to skip it and adopt the much preferred A-Sync VESA standard, as standards always win. It'll be standard across all monitors going forward with no vendor lock in. AMD's Project FreeSync took the wind right out of NV's proprietary sails. G Sync is dead in the water.

When NV had a look at those confidential documents stolen by employees that went to work for them, they probably started rushing something together. AMD would have been waiting for DP 1.2a do be finalized before they planned to implement FreeSync.

 

[Stuka87]

So they know it's false but they say it anyway. That's called a lie, folks. Just because the hardware won't be proprietary, that doesn't mean it won't be expensive, particularly if it's taking a lot of "secret sauce" in order to make it work. Lying through their teeth. This goes way, way beyond spin.

You better have some facts to back this up. You have made it your mission to speak badly about A-Sync ever since it became a standard, but you have not put out one fact to back it up. If you are going to call some a liar, you dang well better back it up with facts.

So lets see them.

 

[Grooveriding]

Not reading that wall, but no doubt it's another diatribe with no substantiation.

No idea why anyone could be so upset the gsync apple cart has been toppled by an open standard coming from VESA ? Unless you are an nvidia employee or shareholder ? If so, okay, otherwise wow...

I'm looking forward to this big time because it will give actual selection in screens that support it with it using part of the DP 1.2a standard which will come in all new monitors, not just one crappy one like gsync currently supports.

Open standards always trump proprietary and are the best solution for end-users.

 

[Hitman928]

Not reading that wall, but no doubt it's another diatribe with no substantiation.

No idea why anyone could be so upset the gsync apple cart has been toppled by an open standard coming from VESA ? Unless you are an nvidia employee or shareholder ? If so, okay, otherwise wow...

I'm looking forward to this big time because it will give actual selection in screens that support it with it using part of the DP 1.2a standard which will come in all new monitors, not just one crappy one like gsync currently supports.

Open standards always trump proprietary and are the best solution for end-users.

The A-sync portion of DP 1.2a is an optional portion of the standard. Don't expect all new monitors to have it available, though if it becomes popular enough, any gaming oriented series of monitors will incorporate it eventually.

 

[Cookie Monster]

Not reading that wall, but no doubt it's another diatribe with no substantiation.

No idea why anyone could be so upset the gsync apple cart has been toppled by an open standard coming from VESA ? Unless you are an nvidia employee or shareholder ? If so, okay, otherwise wow...

I'm looking forward to this big time because it will give actual selection in screens that support it with it using part of the DP 1.2a standard which will come in all new monitors, not just one crappy one like gsync currently supports.

Open standards always trump proprietary and are the best solution for end-users.

You should be careful Groove when you are somewhat calling out other members especially when you fail to realize that its an optional and unproven feature that may not be so trivial to accomplish especially when its taken this long even introduce such a feature. Basically its a very ignorant post on your part and I dont see why every topic has to have ____ shareholder/employee accusations.

And its for another thread entirely but open standards/source rarely trump proprietary ones or else we would be all playing OpenGL based games on a non-window OS using something else than x86 based CPUs.

 

[Grooveriding]

You should be careful Groove when you are somewhat calling out other members especially when you fail to realize that its an optional and unproven feature that may not be so trivial to accomplish especially when its taken this long even introduce such a feature. Basically its a very ignorant post on your part and I dont see why every topic has to have ____ shareholder/employee accusations.

And its for another thread entirely but open standards/source rarely trump proprietary ones or else we would be all playing OpenGL based games on a non-window OS using something else than x86 based CPUs.

No call-outs there. Simple question as to not understanding why there would be persistent and unsubstantiated attacks in multiple threads on something like this ? If you can think of another reason for it other than the ones I gave ? I couldn't.

The nature of what a-sync or g-sync or whatever sync does is a good thing and a net positive. Is VESA now in on the vast AMD conspiracy of misinformation in coming out and announcing support in the DP 1.2a spec ? The level of animosity towards it is definitely suspect and questionable. Especially considering all the claims being made in the attacks using opinion as fact.

I think using Directx vs OpenGL is a poor example for proprietary as well. Think more in terms of something like Thunderbolt vs USB 3 etc.

 

[Spanners]

2) Lie. Contradicted by the very same interview:

So they know it's false but they say it anyway. That's called a lie, folks. Just because the hardware won't be proprietary, that doesn't mean it won't be expensive, particularly if it's taking a lot of "secret sauce" in order to make it work. Lying through their teeth. This goes way, way beyond spin.

In your desperation to discredit this you seem to be reaching a bit far. Just because it contains "secret sauce" doesn't mean it will be expensive or proprietary either, so calling them liars isn't backed up by any information you've presented.

It's one thing to deconstruct a statement and poke holes in it.

But it's quite another if all you are doing is refuting one assertion with your own assertion, that is unsupported.

Agreed.

 

[Silverforce11]

All of this talk is just that. Lets see real hardware we can buy that have it functional.

This level of preempting and spin is getting stupid from both sides.

In the good old days, they would work on a new tech and keep it hidden until its ready for reviews and mass market (re: Eyeinfinity), now its just talk and spin for months (heck, might even be a year?) before the product is ready.

 

[sirroman]

Freesync is probably closely related to AMD's heartbeat (which according to pcper reviews I saw at the time, seems to never drop a frame now, something that even nvidia can't do).

2) Lie. Contradicted by the very same interview:

So they know it's false but they say it anyway. That's called a lie, folks. Just because the hardware won't be proprietary, that doesn't mean it won't be expensive, particularly if it's taking a lot of "secret sauce" in order to make it work. Lying through their teeth. This goes way, way beyond spin.

Slowdown, man.

You were right when you said "Just because the hardware won't be proprietary, that doesn't mean it won't be expensive". It doesn't mean that it will be expensive either.

Provide one source or proof that Gsync is (or will be) cheaper than A-Sync. In time: "it needs R&D" isn't one.

You are correct about time-to-market, announced industry support etc, but going beyond that discredits you.

 

[Wall Street]

I think that a lot of the suspicion over free sync is that AMD has only shown the 'windmill' demo which had a locked frame rate. If this is for variable refresh rates, it would be nice if they actually had a real world gaming fluctuating frame rate demo. Some people have been speculating that the displayport standard requires you to indicate how long the v-blank will be at the start of the v-blank period, which does allow for variable refresh rates, but in a gaming situation it is impossible to predict frame time. If AMD has been developing this for years, they should have a better demo than the windmills. They should just show freestnc with an FPS game, even on a laptop panel. It would quiet a lot of people if they did, but because they are not willing to do that, it is valid to question if they can.

 

[sirroman]

As for the "secret sauce" in their hardware, aren't they referring to the R9 and R7 series cards that apparently have the secret sauce already baked in?

Yes, they are.

Make no mistake, providing dynamic refresh rates to users still takes a lot of ‘secret sauce’ from the hardware and software ends of our products, including the correct display controllers in the hardware and the right algorithms in AMD Catalyst.

 

[Mand]

If you want proof, evidence, and cited sources, please read this thread:

http://forums.anandtech.com/showthread.php?t=2382628

Everything I mentioned is contained there. Real, sourced evidence, most of them quotes from AMD itself.

But, because they say things that certain people here don't like, they feel the need to harass me and claim I have some kind of untoward agenda.

On calling them out on their lie, not only did they say otherwise in this very article, which you seem to be ignoring, they've said the same thing as in that quote in other interviews. Repeating the lie, and repeating backing away from the lie when pressed by people who actually looked at what they said, does not make the lie true or the backing away false.

Read for yourself, think for yourself. There's no point in rehashing the same arguments. This interview gives very little new - what is new does not substantively affect their claims or the position of FreeSync as nothing more than words on a press release.

But, of course, rather than accepting that AMD might just really be horrendously misrepresenting the true nature of FreeSync, it's far simpler to just attack me because you don't like what I'm saying. I mean, how could I possibly have useful things to say, if I make such a long post that people can't be bothered to read it?

 

[Mand]

I'm looking forward to this big time because it will give actual selection in screens that support it with it using part of the DP 1.2a standard which will come in all new monitors, not just one crappy one like gsync currently supports.

Open standards always trump proprietary and are the best solution for end-users.

Wrong and wrong.

It's not required. DP 1.2a is out right now - and they added something to the spec. They didn't just invalidate all the DP 1.2a devices sitting on peoples' desks by a spec update. Adaptive Sync is optional. It will not be in all new monitors.

And no, open standards do not always trump proprietary. I bet you have a Blu-Ray player. Guess what? That's a proprietary standard, and it beat out HD-DVD, which was an open standard.

 

[Madpacket]

So AMD has several monitor manufacturer's preparing gaming monitors to be ready within a relatively short period of time. I'm wondering if it's possible that some of the manufacturer's will offer both A and G sync in the same panel until A-Sync takes over the market and Nvidia is forced to drop G-Sync in favor of A-Sync.

That's how I see this playing out

 

[Atreidin]

Oh boy some information about Freesync. I am sure AMD will say more about it in the future. If not, then I guess Gsync will be the defacto winner.

Flying off the handle and shouting "LIAR!" at every word of every statement, even ones that can't possibly be proven one way or another yet, makes one look like a crazy person. The tone and spin put on the analysis of every sentence is extremely negative. If one applied the same analysis to every announcement from every company it would then appear that no company is ever going to release a new, worthwhile product or technology ever again.

 

[SolMiester]

WOW, I find it quite amazing that NV had a product out before AMD even mentioned their FreeSync theory and bogus demo, now they claim to have come up with the idea first....

 

[monstercameron]

WOW, I find it quite amazing that NV had a product out before AMD even mentioned their FreeSync theory and bogus demo, now they claim to have come up with the idea first....

I didn't read the interview but It's well within the scope of possibility that they did come up with the idea or concepts behind the idea before nvidia even started designing g-sync.

This is the company that was testing and designing gpus with stacked memory years ago [ http://semiaccurate.com/2011/10/27/amd-far-future-prototype-gpu-pictured/ ]. Just because it wasn't shown to the public doesn't mean a thing and gives no way to dispute it [atleast as an industry outsider].

 

[sontin]

So, it's like Apple? Take existing technology, patent it and then claim you invented it? :lol:

AMD hasn't been able to show a real product with Freesync. How can they even think to claim that they have worked longer on "variable refresh" than nVidia?!

 

[MisterMac]

So, it's like Apple? Take existing technology, patent it and then claim you invented it? :lol:

AMD hasn't been able to show a real product with Freesync. How can they even think to claim that they have worked longer on "variable refresh" than nVidia?!

...because it's AMD.

They might have - and might have just sucked so much that NVidia could figure out a solution to a dynamic refresh problem faster than them.


Either way you look at this statement from my angle - it looks like the same repeated mistakes AMD does all the fucking time.

They're quick to burst out claims without actually thinking what it means to their product.


A. It's a lie.

B. They're so incompetent NVidia owned them both on a working standard + time to market.

Which one is worse?

 

[dacostafilipe]

WOW, I find it quite amazing that NV had a product out before AMD even mentioned their FreeSync theory and bogus demo, now they claim to have come up with the idea first....

It's just Marketing. Do you think they would tell us "yeah, we just did what nVidia did"? No chance.

They did not come up with FreeSync before nVidia, but they did have the idea/patent of doing DRR long before. This patent dates 2009 if I remember well.

But bad move by AMD on this one!

This kind of articles will start to appear more and more. Let's hope more technical information will be available soon.