My initial reaction to XFR was severe disappointment.
If you see Dr. Su's introduction to XFR announcing the R7 launch she does not appear to be hyping a mere 80Mhz over the boost clock, and so this is why I now think more is to come.
I maintain that XFR works on thermal headroom because this is what determines whether the voltage can be increased. As you know, the amount of voltage needed and the concomitant rise in temps is what is limiting manual overclocks. I believe this is also what is stopping XFR.
1) if you are running stock, there's performance boost option in bios (most people don't use it anyway) as they overclock on their own. XFR won't work if that is turned off. (basically its a turbo)
I think you are completely misunderstanding of what XFR is.
Basically the turbo that amd does on 1700x is from 3.4GHz to 3.8GHz (XFR will give you better range 3.9-4.1 etc for couple seconds if your cooling allows for it) Without offset it would likely to go higher, but may not be stable; you wouldn't want your system to crash. Thus they created a safety margin of +20'C.
Look at this amd graph, and how long did XFR was working; it lasted pretty much 2-3 seconds to gain that little bit of extra power; and thats how it works. It only overclock slightly when needed, and its only for short bursts.
in most cases you wouldn't even notice that your cpu XFR kicked in giving you 200MHz on short burst.
1 Like
Except it isn't.
Temperatures are largely fine. it just isn't stable. You can keep it cool enough with current solutions at 1.5V. It is doesn't shut off from temperatures. It just isn't stable and even when it is that much voltage degrades the CPU extremely quickly. Thermal limits are only coming into play at 1.6V+. With LN2 and nearly 1.9V they were able to keep it stable for a few runs on a world record. Beyond that nope.
And if the required voltage is what is the problem then that is an architectural or process limitation. If that is the case then it doesn't matter what XFR does as it would never be able to be used regardless.
Sorry.
as @anon5205053 said, it is just a higher boost state IF cooling allows. The number is fixed. More cooling doesn't mean more performance beyond a certain set point. It is only on one core anyway so largely worthless.
XFR has been bottlenecked by the same factors that are currently limiting manual overclocks.
The reason XFR is appealing is that only a small minority overclock.
wat.
So much wat in this post.
wat
No. So much no.
Yeah you can change BCLK to achieve higher memory clocks or do overclocking but that isn't the primary method. XFR and most manual OCing on Ryzen works by changing the multiplier which can be done with high granularity.
All Ryzen CPUs are unlocked and X370/B350 chipsets can do it fine
1 Like
OK, I think we are saying the same thing but using different words here.
The idea that because the XFR is such a small boost, and that it is limited to only a single core is precisely why I believe this is a work in progress.
Get it straight:
Not "wat", but watts!
thats what it was designed for... something extra on top of your typical turbo boost...
1 Like
We aren't.
Well no shit lol. Of course they will improve the technology as generations of CPUs improve but XFR is not at all what you think. It is nothing more than a boost clock. It's limited because putting it to any more cores or higher would exceed the TDP and power delivery of the chip.
It is NOT "automatic overclocking." It is just another boost state within the confines of TDP.
Such pessimism, where are all the optimistic enthusiasts?
It's really unfortunate that some people think this is just a boost clock.
I guess I can't blame them as what we have currently is so disappointing, however I believe this will be improved in this generation into what it was intended to be.
I'm optomistic about Ryzen as a whole. Some little bolt on that turns off as soon as I manually OC? Who cares.
Because it is....... That is how it functions.
It is disappointing because you misunderstood what it was and what it does. AMD's documentation makes it clear that it is just an advanced boost state.
It won't though. Ryzen tops out at 4.1-4.2 Ghz. No matter how much reasonable voltage you dump into it.
I'm done responding to this. You clearly have no idea what you're talking about and are seriously misinformed about what XFR is or how it functions. I'm sorry it is disappointing to you but that is what it is. It won't change on this generation of chips. Just accept it. Move on. Enjoy Ryzen for the great value it is
Have a nice day
I know because i talked to the engineers that it is currently a very well optimised boost state.
That will automatically go to if the CPU isnt overheating and etc.
There may be different implementations of it in the future but that is how it stands right now. And it is most definitely not auto overclocking, or a work in progress.
The tuning they are doing now is to optimise the boost states to work well in Win10, so there are no sleep mode issues, etc. You saw the same kind of thing with Win7 and the first few gens of i7s that had turboboost.
2 Likes
If what you are saying is true then XFR is truly nothing more than a gimmick, and I just don't believe this is the case.
The engineers you contacted are likely limited to what can be divulged at this time.
I know the info because I have direct contact with AMD, as I am a part of red team plus.
It isn't a gimmick either. It does boost the speed, and intelligently. But on this gen, not very much.
So its sorta comparable to Turboboost.
I'm not doubting your word.
I just don't think that AMD or any company is going to disclose everything they are working on, especially a feature that is as incomplete as XFR.
It
isnt.
incomplete.
It is a fully functioning tool that integrates well into the CPU and other boost states.
Calm down Caveman, don't let ill-informed forum members worry you. Chances are they don't want to learn, just complain
1 Like
yeah, lol. I'm not angry just making a point, I was going to give up anyway if he kept arguing lol
For Ryzen 1.0, it is complete. Maybe in future generations it will be iterated on, but for Ryzen 1.0 this is as far as it is going. Period. The silicon does not have the overhead to support boosting higher than at best 200Mhz on one core, on only select skews. It will not be implemented on multi-core on Ryzen 1.0 either. The TDP constraints and constraints of the silicon itself prohibit it from possibility. Voltages would have to be hiked to support this and that work would probably also need to be done at the factory to determine those voltages.
The other caveman would know this best and he has articulated that this is as far as the feature will go for Ryzen 1.0. He has direct access to AMD. He's had private and public meetings with AMD officials due to his involvement in AMD's official Red Team Plus program. If an engineer from AMD said to him that this is all it is, than that's that.
Red Team Plus ? The PR advertising stunt?
Oh, I thought it was something with serious inside access?
Those engineers will keep it light and fluffy with the public, they aren't allowed to say much else.