AMD's under reported dynamic refresh rates

Subject: General Tech | January 7, 2014 - 12:45 PM |
Tagged: g-sync, free sync, dynamic refresh rate, amd

The Tech Report learned yesterday of a feature that AMD has been offering over the past three generations of GPU called dynamic refresh rate, which they billed as a power saving feature.  Skipping an unnecessary vertical screen refresh would certainly save you a bit of power but really isn't that attractive a feature.  NVIDIA looked at this feature in the opposite way, not to save power but to refresh your screen as quickly as it is capable of to provide much smoother graphics and that marketing has had G-SYNC on everyone's lips.  Assuming AMD can get the word out, their variable refresh rate technology should be compatible with most new mobile/laptop products based on the EDP specification, no additional costs or equipment required. 

View Full Size

"During an impromptu meeting in a hotel ballroom this morning, we got an eye-opening demo of a dynamic refresh rate capability that's been a part of Radeon GPUs for several generations. AMD thinks this feature can be combined with triple buffering to deliver G-Sync-like animation smoothness without the cost associated with specialized display hardware."

Here is some more Tech News from around the web:

Tech Talk

January 7, 2014 | 01:11 PM - Posted by arbiter

So wait, skipping a refresh cycle? They want to use triple buffering which add's a ton of input lag in to the pipe line, yea its free but A lot of people I think would take screen tearing over that.

January 7, 2014 | 04:04 PM - Posted by Anonymous (not verified)

In most cases triple buffer actually remove input lag caused by vysnc

January 7, 2014 | 04:07 PM - Posted by Anonymous (not verified)


January 7, 2014 | 04:59 PM - Posted by arbiter

To re-quote a quote i have in my profile on another site,
"Why is it acceptable for people to be idiots but not acceptable for me to point it out?"

January 8, 2014 | 08:16 AM - Posted by Anonymous (not verified)

"because your the one wrong?"
example of the explainatin as to why anon is right

January 8, 2014 | 08:45 AM - Posted by Anonymous (not verified)

that article is a mixed, but first few explain most of what is causing the confusion, flip que and render ahead buffer is getting confused for triple buffer which does not cause input lag, however TRUE triple isn't used in alot of situations in games

January 7, 2014 | 01:28 PM - Posted by me (not verified)

Yes this is very under reported. I've been trying to piece together info on this but it's sparse.

requires variable vblank display that's implemented correctly (possibly through firmware flash) - from Anand's coverage
Triple buffer implementation - from Tech Report

Also manipulates vblank - from pcper & Anand's gsync review
requires gsync module in monitor (presumably to implement the variable vblank correctly in the display)

my questions:
1) How simple would it be to update firmware to the correct variable vblank implementation in existing monitors? (maybe on of the monitor manufacturers could comment?)
2) How does the visual experience compare? (whip Allyn into action, or more action :D)
3) Do they have a variable framerate example? From all the videos I'm only seeing the windmill which is a nice proof of concept but does it vary the framerate like a game?

Thanks guys.

January 7, 2014 | 02:09 PM - Posted by renz (not verified)

yup i want to see how does it work in real games as well. so when can we expect amd to show this with real game running?

January 7, 2014 | 03:04 PM - Posted by arbiter

Only fessible way i could tell you could do a firmware update is if monitor supported it via USB (if the monitor has the ports) other then that don't think there is a way to do it your self so for majority of people its new hardware anyway.

January 7, 2014 | 07:01 PM - Posted by Anonymous (not verified)

I think PowerStrip can alter or update a monitor´s firmware.

January 7, 2014 | 02:10 PM - Posted by Penterax (not verified)

AMD downplays the importance of getting monitor manufacturers to build their monitors with the capability of using "free sync". The fact is, hardware support IS required. Whether or not a "firmware update" will make it possible or not isn't the issue, the issue is getting monitor makers to adhere to the necessary standard.

Assuming what AMD says they (have been doing) is exactly what Nvidia is doing, I think Nvidia's solution makes a lot more sense, because it forces manufacturers to make a specific decision to include the capability required. AMD is saying "Well, this should work and manufacturers should support it", Nvidia is recognizing the politics of dealing with those manufacturers and designed a system to make sure they are on board with the technology.

I don't think the G-SYNC technology should effect the price the end buyer pays, or at least not by much. It essentially replaces the control hardware for the monitor and doesn't necessarily cost more to build.

What I'm tired of in this whole thing is AMD reacting to what they should have been doing all along. If they really cared about the quality of the visual aspects of the games we play, they would have been behind this long ago. Instead they only get behind it when Nvidia and hardware sites start talking about problems and solutions. Seems to me that if graphics quality was left up entirely to AMD, we would never make any improvements.

January 7, 2014 | 03:21 PM - Posted by arbiter

damn double

January 7, 2014 | 03:21 PM - Posted by arbiter

I noticed that as well, yea AMD goes cheap route to keep costs down but they just expect companies to add support to it. AMD also tends to use software solutions to problems (ex. the whole crossfire/frame pacing bit), where As nvidia though does tend to cost more but really is the best way to go about it, a hardware solution to fixing something.

" If they really cared about the quality of the visual aspects of the games we play, they would have been behind this long ago"

Yea if they really did care they would address the whole issue with people complaining about crossfire years ago instead of just ignoring the problem. They didn't even address it til FCAT which was made by Nvidia came out and revealed the scope of the issue.

January 7, 2014 | 04:53 PM - Posted by Anonymous (not verified)

"AMD downplays the importance of getting monitor manufacturers to build their monitors with the capability of using "free sync". The fact is, hardware support IS required."

It's in the d*mn standard! EVERY monitor should support it!

January 7, 2014 | 06:02 PM - Posted by Anonymous (not verified)

Is it a mandatory part of the standard? We all know that if it's something optional, chances are very high that it can't be solved by any kind of firmware upgrade.

Is it actually a ratified part of the standard or is it more of a draft?

How much room for interpretation is there in the standard regarding the feature(s) that AMD are using in unintended ways to create "free-sync"?

January 7, 2014 | 07:02 PM - Posted by Anonymous (not verified)

The number of displays compliant to the standard are more than those compatible with the g-sync add-on, if you can buy a laptop that has one.

January 7, 2014 | 09:21 PM - Posted by nobodyspecial (not verified)

Not if it was created with AMD help just to show this, or paid for by AMD to get the tech into the laptop for this announcement. I mean even NV had to pick someone to show their tech with first. They picked ASUS. You have to get it right with one, before you try applying it to the whole planet's products.

I'm skeptical because they displayed no games and it's AMD we're talking about here, who always seem to skimp these days. If this was so great, and in everything why haven't EITHER side used this standard to fix something ALL gamers have hated for years (stutter, lag, tearing etc)? Also I note someone said it MAY be possible for a firmware update to fix it, but maybe just means "I will believe it when I see it" to me and I no longer Pre-order anything...LOL.

Has anyone seen AMD's 45w Richland they LAUNCHED? Neither have I. :( Be careful what you believe ;)
Inbound July article date. More articles can be found for Aug/Sept. Still waiting I think unless they got one out somewhere and I missed it. It was supposed to go on sale in Aug at end of the month.,23966.html
I don't even think you can buy it in a system. Newegg doesn't sell them, amazon etc in box. How long before phase 2 or phase3 drivers?

If a monitor maker actually had the tech inside, I'll take bets against anyone thinking they'll add this via firmware updates, instead of telling you to BUY a new monitor from them that has freesync...LOL. I can see NV's working, and in games too. I see AMD saying it isn't ready and not even planning to market it yet? Sounds like they're counting on a lot of people just helping out because they like AMD and giving out free stuff that stifles their monitor sales. Not going to happen.

January 7, 2014 | 02:30 PM - Posted by Lord Binky (not verified)

If I'm remembering correctly I think my Vblank is set to 1* already to allow 2560X1440 @120Hz (over Dual DVI). I'd have to drop my refresh rate down to give it room to increase Vblank. An interesting challenge to find an optimal balance.

*Using Toasty X's CRU tool. I use it with any monitor since you can overclock most monitors at least an extra 10Hz and the jump between 60 & 80 Hz is the most impressive jump if you go by 20Hz increments.

January 7, 2014 | 02:38 PM - Posted by madhatter256 (not verified)

"Ah AMD, you really are your own worst enemy sometimes."

That statement pretty much summed up the whole company...

Great opening statement.

January 7, 2014 | 05:19 PM - Posted by David (not verified)

Oh really?

If the Freesync/G-Sync thing was so obvious, it would have been done a long time ago by either AMD or Nvidia. All important things seem obvious in the backrear mirror.

Plus, the author of this article takes for granted that Freesync would be on par with G-sync, something that is far from certain.

And without that certainty, his statement falls apart. It's very easy being 100% right on the internet after a fact (G-Sync). Would Jeremy or you come up with a similar idea? Hah!

January 7, 2014 | 03:37 PM - Posted by Nilbog

This sounds surprisingly similar to options already available in Radeon Pro. Which i, and many others, have been using to help fix the frametime issues with CrossfireX.

It actually seems to work better than the official "Frame Pacing" solution.

"Dynamic Vsync" and "Dynamic Framerate Control"

I would really love to hear your guy's thoughts on this.

January 7, 2014 | 04:54 PM - Posted by Anonymous (not verified)

So wait, you said that gsync should change the industry, should be in every monitor and gpu. So now it's know that it was feasible for 3+ years, is free and depends only on a standard feature that monitor makers should support ANYWAY...

... and you are not praising the hell of it???

January 7, 2014 | 05:23 PM - Posted by Nilbog

Probably because they have been sitting on it, doing nothing with it....

January 7, 2014 | 06:51 PM - Posted by Anonymous (not verified)

Everything take man-hours to implement. AMD has a lot in their hands, HSA, gaming evolved, mantle, drivers, frame-rating. AMD isn't the only one that have access to the standard. So where is nvidia? AMD supports it for 3 generations! If people are complaining because AMD was sitting on it, what about nvidia?

What is the excuse for nvidia making hardware and charging the end-user for it while it could be done by software?

AMD just made A LOT of people look like clowns.

January 7, 2014 | 05:16 PM - Posted by Anonymous (not verified)

Sounds like the Monitor OEMs are going to have to Implement some form of universal plug/interface/standard for variable monitor refresh synchronization hardware from any GPU device OEM! The legal implications of only one GPU makers' hardware/firmware solution being available to only users of that GPU's OEM hardware, and not available to others, could be problematic, and even it is not, the Monitor OEMs are going to have to provide a default interface that can be use for non G-Sync devices and the G-Sync interface for devices that do, two interfaces are going to cost more to design and implement for OEM monitor makers, so it is either an industry standard hardware firmware interface(least costly), or provide a hardware slot on all external monitors to accept a GPU device OEM specific control card.

January 7, 2014 | 07:33 PM - Posted by Anonymous (not verified)

Video Electronics Standards Association

CVT Standard

Coordinated Video Timings Standard
Version 1.1
September 10, 2003


The VESA Coordinated Video Timings (CVT) is a method for generating a consistent and coordinated set of standard formats, display refresh rates, and timing specifications for computer display products, both those employing CRTs and those using other display technologies. The intention of the CVT specification is to give both source and display manufacturers a common set of tools to enable new timings to be developed in a consistent manner that ensure greater compatibility.


The CVT Standard defines rules and methods by which new and existing formats can be defined. Based primarily on the VESA GTF Standard, CVT defines restrictions to the pixel clock modularity, refresh rate and aspect ratio. It also specifies a new equation-based method for developing Reduced Blanking timings, designed for use with non-CRT displays that can accept Reduced Horizontal Blanking times.

January 7, 2014 | 10:18 PM - Posted by Anonymous (not verified)

Links to non paywalled white papers, and any white papers dealing with, GPU frame rate and display refresh control over display port standard protocols, would be helpfull. Also the total industry adoption rate for these VISA standards among the total OEM display device makers! The orginal hardware in a monitor may be VESA compliant, but what about the Nvidia hardware/firmware that has to be added to the display, to make it work with the GPU hardware. This is all being done over display port, but is it being done using a standard control packet, that is taged for, and needs to be fruther post processed (in a non standard/open way) by Nivida specific hardware/firmware to be of any use for control of the refresh rate. There may be VESA standard ways of encapsulating control packets that need to be fruther post processed on a non standard Nvidia(or Any) Hardware/firmware, and not open to others, G-sync way, but there needs to be a set of standard control packets that every GPU maker must use as agreeded upon protocal so that all monitors can be made to work with any GPU OEMs product without the need for special hardware/firmware!

January 7, 2014 | 09:52 PM - Posted by praack

i hear a couple of things in this forum surrounding the standards - one that they are currently in place but not followed. so if they were followed we would have the ability to see what AMD's dynamic refresh rate is and compare it to Gsync
the other is that AMD is taking the cheap way out

usually i see companies looking at not what they can deliver free but where they can make profit. So if they have an item that can be added to the monitor that would be free with little return value compared to a hardware module that immediately pumps the price of the unit up and profits increase- which way would a company go?

It could be that AMD is just pointing out the realities of the market place - we could have had something similar for free- but instead we get something more complicated for more of a cost

January 7, 2014 | 10:33 PM - Posted by Anonymous (not verified)

Nvidia developed a module and instead of selling it to a panel producers its selling it to a monitor sellers.

Something is definitely wrong there.

AMD could just be pointing it out that the standards are there to follow but doesn't want to go into the panel business.

Nvidia might be more then happy to just make a profit from it until panel manufacturers adhere to the standards. If the standards are there the panel producers will include it in revised version of there products and take all the profit away from Nvidia.

January 8, 2014 | 11:44 AM - Posted by Lord Binky (not verified)

I find it odd the transition from asynchronous refresh to synchronous refresh is so troublesome.

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

By submitting this form, you accept the Mollom privacy policy.