[TechPowerUp article] FreeSync explained in more detail

[3DVagabond]

Well I certainly don't disagree that free-sync is closer to vaporware and proof of concept. In fact I agree with you completely, sushi, that free-sync is vaporware at this point. I just wonder why they chose to demonstrate it when they did. Nvidia didn't show g-sync until it was done. I'd assume that AMD wouldn't show free-sync until it's done, but it's increasingly looking like it won't be around until 2015.

We'll see though. I do have to question the statement of nvidia using hardware. Because PCPer questioned AMD's vice president of visual computing, Kodira, and he stated and confirmed that AMD's free-sync will require the monitor to have a variable refresh control board just like g-sync does. Also, FPGA is only being used right now and it will transition to ASIC later in Q1. FPGA is expensive, ASIC isn't. At least, that's what nvidia is stating. So their assumption is that G-sync costs will be substantially lower by the end of Q1 2014 since monitor manufacturers will switch from FPGA to ASIC gsync modules. Does anyone have concrete data points on FPGA versus ASIC costs? If not i'll go dig that up.

Someone correct me if i'm wrong. Since free-sync requires monitors to have a variable refresh aware module, just like gsync (this is per AMD's Kodira, BTW), That would indicate that free-sync isn't free? Wouldn't that means that free-sync costs money in literally the same area that g-sync requires? I dunno. I don't expect free-sync to exist until 2015, so maybe costs will be substantially lower for the control board by that time. Free i'm not sure about however, unless someone can clarify that for me. I could be really missing something here, i'm not sure.

If you look at the G-sync how to video they are removing components and replacing them with others. Do you really think it costs that much extra for the G-sync board compared to the boards that were removed?

The 2 boards on the left are replaced by the single board on the right. It's not like they simply add $200 worth of components. They remove everything else first.

Looking at those two images, I can't believe that G-sync really costs any more. We'd need the BoP to know for certain, of course.

As far as the purpose of the Free-Sync demo, I believe it was simply AMD showing that hacking VBLANK to do real time variable refresh rate adjustment isn't hard. It's not the next great accomplishment. I hope they are right and it's simply added by the monitor companies to augment/replace vsync.

 

[3DVagabond]

sushiwarrior,

I'm not clear why the DP 1.3 standard that is the core idea (if there is one) behind FreeSync matters. If Nvidia can do it with DP 1.2, why do we need DP 1.3? FreeSync is still going to require new hardware in the displays themselves, it's not just a cable spec issue, so how does the DP 1.3 standard translate into actual use in a free way? If it requires the display manufacturers to include as-yet-undeveloped hardware to include in their displays (and it does require it) that doesn't exist yet, how is that cost not going to be translated to us, the consumer? How could it possibly be free?

I find it rather odd that people are demanding that Nvidia give them something they undoubtedly spent a good deal of money on internal R&D and hardware manufacturing on for free. They developed it, why shouldn't you pay for it if you want it? Isn't that how companies work - they make something you want, and so you then pay to get what you want? It's their effort, their IP, their product. Why do you deserve it for free? Why do their competitors deserve it for free?

If the hardware didn't exist they wouldn't have been able to take an off the shelf laptop and adapt it like they did. All they did was activate a feature already in their drivers.

 

[3DVagabond]

I'd be happy if Free-sync turns out to be as good as G-sync. Unfortunately, all the technical talk slips in hints that it will at least add additional latency.

For most, latency may not be as big of a deal, but to me, I get nausea due to latency.

The VBLANK interval is very short. If this info is sent over the VBLANK interval we won't notice any additional latency. It's only ~0.74ms@60Hz.

 

[bystander36]

The VBLANK interval is very short. If this info is sent over the VBLANK interval we won't notice any additional latency. It's only ~0.74ms@60Hz.

The issue is whether they have to set the next refresh rate before the refresh happens, or if they can simply tell it to refresh at a set point.

G-sync does allow them to call a refresh. I do not know if Freesync can say the same.

 

[3DVagabond]

Well sure, and quantum computing has been around for 34 years, yet I can't buy that anywhere.

Future products matter for those who either have patience or aren't capable of purchasing said $200 product. Obviously there is only one option now. For some people, that option is expensive and not what they want. We are discussing the alternative. I would suggest taking useless observations somewhere else, instead of derailing the thread.

It's not so much can't afford it, at least for me. It's being sick and tired of companies charging $100's of dollars for something that costs them pennies to implement. They can only do it if we let them and I'm not going to just accept the pricing without questioning it. Especially since AMD was able to recreate it from existing hardware so easily. It just wreaks.

 

[Lonbjerg]

It's not so much can't afford it, at least for me. It's being sick and tired of companies charging $100's of dollars for something that costs them pennies to implement. They can only do it if we let them and I'm not going to just accept the pricing without questioning it. Especially since AMD was able to recreate it from existing hardware so easily. It just wreaks.

AMD recreated what?
Once again empty PR is compared to reviewed products.

AMD should make a new slogan:
"Our empty PR will MATCH anything real!"

 

[blackened23]

If the hardware didn't exist they wouldn't have been able to take an off the shelf laptop and adapt it like they did. All they did was activate a feature already in their drivers.

We already covered this untrue statement. Didn't we? Didn't you bring this up twice or something? And we covered it twice? Or was that someone else? Just curious. If it wasn't you, apologies. If it was you, let's go through this again. I know we're here to grow and learn from mistakes . That's cool. Anyway, laptops use eDP. Desktop monitors do not use eDP.

If you think they do, PCPer asked AMD to provide a list of monitors that use eDP. The list provided to PCPer was as follows:

List of desktop monitors using eDP that are able to use free-sync now, as confirmed by AMD :
1.

This question was asked as soon as news of free-sync broke. PCPer's response to above on Thursday was #1. Then PCPer emailed for a followup, the response:

List of desktop monitors using eDP that are able to use free-sync now, as confirmed by AMD :
1.

PCPer then asked AMD's VP of visual computing what was up. This is THE man at AMD in charge of graphics/visual computing. And he said, basically, that free-sync required monitors to have a variable refresh control board just like g-sync. So any statement that free-sync is free is based on an untruth. It requires costs in the same area as g-sync. Or, we could pretend that nvidia's engineers are idiots and didn't explore this possibility. The truth is, nvidia's engineers did explore this per Tom Petersen. Not possible without a variable refresh control board installed on a monitor.

So here is AMD's list of monitors on the market NOW that can use free-sync. This also includes monitors available NOW that can be retrofitted for free-sync:

1.

By the way. Anandtech also inquired with AMD about which monitors use eDP and can use free-sync now. The answer? I'll let you take a guess.

 

[Despoiler]

If you look at the G-sync how to video they are removing components and replacing them with others. Do you really think it costs that much extra for the G-sync board compared to the boards that were removed?

The 2 boards on the left are replaced by the single board on the right. It's not like they simply add $200 worth of components. They remove everything else first.

Looking at those two images, I can't believe that G-sync really costs any more. We'd need the BoP to know for certain, of course.

As far as the purpose of the Free-Sync demo, I believe it was simply AMD showing that hacking VBLANK to do real time variable refresh rate adjustment isn't hard. It's not the next great accomplishment. I hope they are right and it's simply added by the monitor companies to augment/replace vsync.

FPGAs are not cheap.

We already covered this untrue statement. Didn't we? Didn't you bring this up twice or something? And we covered it twice? Or was that someone else? Just curious. If it wasn't you, apologies. If it was you, let's go through this again. I know we're here to grow and learn from mistakes . That's cool. Anyway, laptops use eDP. Desktop monitors do not use eDP.

If you think they do, PCPer asked AMD to provide a list of monitors that use eDP. The list provided to PCPer was as follows:

List of desktop monitors using eDP that are able to use free-sync now, as confirmed by AMD :
1.

This question was asked as soon as news of free-sync broke. PCPer's response to above on Thursday was #1. Then PCPer emailed for a followup, the response:

List of desktop monitors using eDP that are able to use free-sync now, as confirmed by AMD :
1.

PCPer then asked AMD's VP of visual computing what was up. This is THE man at AMD in charge of graphics/visual computing. And he said, basically, that free-sync required monitors to have a variable refresh control board just like g-sync. So any statement that free-sync is free is based on an untruth. It requires costs in the same area as g-sync. Or, we could pretend that nvidia's engineers are idiots and didn't explore this possibility. The truth is, nvidia's engineers did explore this per Tom Petersen. Not possible without a variable refresh control board installed on a monitor.

So here is AMD's list of monitors on the market NOW that can use free-sync. This also includes monitors available NOW that can be retrofitted for free-sync:

1.

By the way. Anandtech also inquired with AMD about which monitors use eDP and can use free-sync now. The answer? I'll let you take a guess.

Why would you expect AMD to have a list of monitors when they have no plans to productize Free-Sync? This was a proof of concept demo. Nothing more.

You realize that end to end DP 1.2 can bypasses the control board functions right?

http://en.wikipedia.org/wiki/DisplayPort

Can drive display panels directly, eliminating scaling and control circuits and allowing for cheaper and slimmer displays

The only thing you need is a board with DP 1.2 in and eDP out, which in my investigation was 100% of the boards.

http://www.winmate.com.tw/IndustrialDisplay/LcdSpecBoard.asp?Prod=01_0683&Typeid=B010113

http://www.st.com/web/catalog/mmc/FM128/CL1746/SC991/PF252131

If you want to investigate further
http://www.vesa.org/displayport-developer/certified-components/

 

[VulgarDisplay]

I still don't understand why everyone is turning "no public plans" to go to market into no plans to go to market.

AMD and nvidia have no public plans to launch 20nm GPUs but that doesn't mean they are not coming.

 

[Keysplayr]

I still don't understand why everyone is turning "no public plans" to go to market into no plans to go to market.

AMD and nvidia have no public plans to launch 20nm GPUs but that doesn't mean they are not coming.

I plan to launch my own brand of GPU come Q4 2014.

 

[bystander36]

I still don't understand why everyone is turning "no public plans" to go to market into no plans to go to market.

AMD and nvidia have no public plans to launch 20nm GPUs but that doesn't mean they are not coming.

It clearly is still in development based on a few things from them. I would expect some sort of launch eventually, but we really have no idea if and when they will.

It really is a proof of concept atm, but there definitely is an option for it to be released eventually. I do think it is likely they do, but we can't say for certain either.

 

[Despoiler]

I still don't understand why everyone is turning "no public plans" to go to market into no plans to go to market.

AMD and nvidia have no public plans to launch 20nm GPUs but that doesn't mean they are not coming.

It's because they have never been involved in corporate product development and as such are ignorant of what it entails. Engineers came up with a proof of concept known as Free-sync. The other side of that is product development. PD is what takes the PoC and fleshes out what is presented and sold to consumers. All AMD said is that the PD side of things hasn't been started. However, someone is going to write up a business case for Free-Sync and that business case will decide if Free-Sync is put into AMD's PD cycle.

 

[dacostafilipe]

The issue is whether they have to set the next refresh rate before the refresh happens, or if they can simply tell it to refresh at a set point.

G-sync does allow them to call a refresh. I do not know if Freesync can say the same.

It's the same on both systems. Both connect to the LCD on the same way (eDP). Both use VBLANK for "variable refresh".

 

[BallaTheFeared]

I plan to launch my own brand of GPU come Q4 2014.

After you sign the contract to use my EzBake Oven fab.

 

[sontin]

Because AMD said that nVidia doesn't support eDP:
There Kepler cards supports eDP 1.2 and so they are able to use VBLANK:
http://www.geforce.com/hardware/notebook-gpus/geforce-gtx-780m/specifications

So, why did AMD lie to Techreport?!

 

[Leadbox]

Because AMD said that nVidia doesn't support eDP:
There Kepler cards supports eDP 1.2 and so they are able to use VBLANK:
http://www.geforce.com/hardware/notebook-gpus/geforce-gtx-780m/specifications

So, why did AMD lie to Techreport?!

This is what was said:

The quantization problem can only be completely resolved via dynamic refresh rates. However, the exec initially expressed puzzlement over why Nvidia chose to implement them in expensive, external hardware.

The exec's puzzlement over Nvidia's use of external hardware was resolved when I spoke with him again later in the day. His new theory is that the display controller in Nvidia's current GPUs simply can't support variable refresh intervals, hence the need for an external G-Sync unit. That would explain things. I haven't yet had time to confirm this detail with Nvidia or to quiz them about whether G-Sync essentially does triple-buffering in the module. Nvidia has so far been deliberately vague about certain specifics of how G-Sync works, so we'll need to pry a little in order to better understand the situation.

One can always be overly dramatic and call it lying, if one chooses

 

[Shivansps]

If you look at the G-sync how to video they are removing components and replacing them with others. Do you really think it costs that much extra for the G-sync board compared to the boards that were removed?

The 2 boards on the left are replaced by the single board on the right. It's not like they simply add $200 worth of components. They remove everything else first.

Looking at those two images, I can't believe that G-sync really costs any more. We'd need the BoP to know for certain, of course.

As far as the purpose of the Free-Sync demo, I believe it was simply AMD showing that hacking VBLANK to do real time variable refresh rate adjustment isn't hard. It's not the next great accomplishment. I hope they are right and it's simply added by the monitor companies to augment/replace vsync.

OMG, you dont know how much FPGA costs?

 

[NTMBK]

After you sign the contract to use my EzBake Oven fab.

Make sure he doesn't tie you into a WSA!

 

[BrightCandle]

Freesync is dependent not on AMD but on the monitor makers. eDP has supported variable refresh rate for 5 years by my count and yet at no point has this technology been brought to the main market for the purpose of replacing the vsync 60hz cycle. AMD simply isn't in charge of the release cycle of freesync at all.

NVidia decided to take matters into its own hands, built a controller for a particular panel and demoed real hardware that would be available. In the process they convinced a number of monitor manufacturers to get onboard and support their technology. From my research into how they both work gsync is superior for the purpose of gaming, it has anywhere from 16-33ms less of lag at 60hz and supports low persistence and 3D as well as requiring no prediction of upcoming frame rate and allows the monitor to refresh itself when the image being held would be a problem.

I hate Nvidia's plan, they have no regard for the industry as a whole and have carved out an expensive but perfect solution to the problem at hand and have made it hard for their competitor to also use the monitors. But AMD is also much earlier into the process, hasn't got any monitor manufacturers convinced to change their implementation.

To me this always boils down to the same thing. In the end industry standards win, but very rarely is innovation done via standards. So a company makes a thing, sells it and makes back its research costs and makes some good money and then the industry comes up with a standard out of the competing components. But that standards process often takes years, in some cases a decade to occur. AMD's plan is for the coming years, not really something for now.

 

[Cookie Monster]

If you look at the G-sync how to video they are removing components and replacing them with others. Do you really think it costs that much extra for the G-sync board compared to the boards that were removed?

The 2 boards on the left are replaced by the single board on the right. It's not like they simply add $200 worth of components. They remove everything else first.

Looking at those two images, I can't believe that G-sync really costs any more. We'd need the BoP to know for certain, of course.

As far as the purpose of the Free-Sync demo, I believe it was simply AMD showing that hacking VBLANK to do real time variable refresh rate adjustment isn't hard. It's not the next great accomplishment. I hope they are right and it's simply added by the monitor companies to augment/replace vsync.

First and foremost, its implemented via FPGA... they are quite expensive and it clearly shows that the first G-SYNC modules are pretty much "prototypes" for a lack of a better term. Once they take the design and implement it in an ASIC, it will be cheap but that takes a lot of time, risk, engineering manpower and investment so I suspect they have decided to do this now so that it hits the market earlier.

Secondly with the BOM cost aside, nVIDIA probably has already invested in some engineering resources into tackling this issue which costs $$$. Nothing is ever free and in this sort of environment no one will do this if there are no returns/incentives. So many forum members are quite naive in this matter, as if these companies are charities.. It just simply does not work like this at all.

To me this always boils down to the same thing. In the end industry standards win, but very rarely is innovation done via standards. So a company makes a thing, sells it and makes back its research costs and makes some good money and then the industry comes up with a standard out of the competing components. But that standards process often takes years, in some cases a decade to occur. AMD's plan is for the coming years, not really something for now.

Pretty much sums it up.

 

[Mand]

To me this always boils down to the same thing. In the end industry standards win, but very rarely is innovation done via standards. So a company makes a thing, sells it and makes back its research costs and makes some good money and then the industry comes up with a standard out of the competing components. But that standards process often takes years, in some cases a decade to occur. AMD's plan is for the coming years, not really something for now.

Yep. It'd be nice if they didn't feel the need to deceptively throw dirt on their competitor's genuine innovation while they wait for a standard (that wouldn't have happened at all if not for the release of G-sync).

 

[bystander36]

It's the same on both systems. Both connect to the LCD on the same way (eDP). Both use VBLANK for "variable refresh".

While both use VBLANK for "variable refresh", Nvidia's system also releases an existing refresh to start a new one. AMD has not mentioned anything of the sort and they also said they'd need to use triple buffering to accomplish what G-sync does.

Now they haven't said specifically what their plan is, so we have to wait, but it doesn't sound like they can do the same thing. Time will tell.

 

[VulgarDisplay]

I plan to launch my own brand of GPU come Q4 2014.

Still don't get the difference. You just publicly announced something you can't possibly deliver on.

Completely different than AMD not having public plans to launch something they demonstrated on off the shelf crap laptop screens.

Product development takes time.

 

[blackened23]

Still don't get the difference. You just publicly announced something you can't possibly deliver on.

Is it different though? I'm trying to think of things that AMD didn't deliver on. Such as. DX9/CF and eyefinity/CF Frame pacing fixes on 79xx and earlier cards, that was supposed to be fixed in 2012. Oh but wait. We'll fix it in 2013. Then it was Jan 2014. AMD seemed to tell us originally that they would deliver a frame pacing fix for CF DX 9 and CF eyefinity on Tahiti and older cards in 2012. Lemme look at my calendar again. Seems to say 2014. I could be wrong though. Someone add the number of months up here that eyefinity CF was broken all Tahiti and earlier cards.

Let's see what else. Did AMD possibly deliver on HD3D? Well there's the fact that they abandoned it. After they pitched it as a free API for developers. (sound familiar? Free and open standards anyone? Did I hear that during the free-sync press tech thingy? Someone remind me). Oh yeah. Free and open standards with HD3D. Open to all developers. But now , you gotta pay tri-def for HD3D. AMD doesn't provide the gaming driver, and doesn't support it. Whoops.

I dunno bro. Some would argue that there's no possible way AMD could deliver on free-sync based on their prior history of....never delivering. Definitely not in 2014. There's always faith and hope though while playing the AMD waiting game. Like, I have faith and hope in keys' GPU being released in Q4.

 

[Despoiler]

Yep. It'd be nice if they didn't feel the need to deceptively throw dirt on their competitor's genuine innovation while they wait for a standard (that wouldn't have happened at all if not for the release of G-sync).

Umm no. You can do this with DP 1.2 as Nvidia and AMD have shown, but DP 1.3 fully supports a controller that understands variable refresh. It's going to be ratified with in the next 2-3 months. It's been 3 years since DP 1.2 was released so the 1.3 standard was being developed that long. The only thing G-Sync is doing is attempting to lock people into Nvidia before the standard hits. They definitely did not come up with the concept of variable refresh if it's going into a standard.