AMD "RYZEN"

[Charlatan 68]

If those single core figures are correct, and the CPU is supposed to clock well things could be about to get very interesting

[BaldyB 20]

If those single core figures are correct, and the CPU is supposed to clock well things could be about to get very interesting

 

Many mobo makers have the BIOS authors write custom BIOS's to facilitate O Clocking. I think Ryzen at least for me is the only CPU that's supposed to do it at the chip level. However, it depends on the cooling capacity and temp. If you cooling isn't good enough it is supposed to throttle back to prevent burnout. If you have a higher end system like the Noctoa (Did I spell it right?) or a water cooled system it should overclock by itself. The heat sink builders are coming out with high end Ryzen coolers for a reason!  

I think their will be some surprises. Just because AMD is a much smaller company than intel doesn't mean they are stupid. They did reach 1 GH first. They did come up with a 64 bit system first. The only mistake they made IMHO was going the multi core route too early. The software people never followed in any meaning full way.

Best

BaldyB

[swiesma 625]

The problem for AMD was that the Core Architecture by Intel was pretty good and they did not have something against this.

I owned a lot of AMD CPUs some yaers ago, starting with a K6. @BaldyB is right, AMD is not a bunch of stupid people... And I really hope that they get something going.

 

More competition is good for us!

 

I see the CPU market like the car market.

Nothing really new until Tesla came and brought new winds...

 

I hope AMD can do the same and get Intel to work again and not selling 5%-Performance-Upgrades each year.

[martin-w 3,234]

Many mobo makers have the BIOS authors write custom BIOS's to facilitate O Clocking. I think Ryzen at least for me is the only CPU that's supposed to do it at the chip level.

BaldyB

 

 

Not sure what you mean there. Motherboards do indeed have BIOS's that "facilitate" [make easier, more convenient] overclocking. But how does the AMD CPU itself make overclocking more convenient? Intel CPU's have "turbo Mode" that automatically overclocks by a few hundred megahertz and AMD CPU's have the equivalent. Is that what you mean? If so it's no different to many generations of Intel CPU's.

 

Or are you referring to something else? Can you elaborate.

 

 

If you have a higher end system like the Noctoa (Did I spell it right?) or a water cooled system it should overclock by itself. 

 

 

 

It's "Noctua" Baldy.  :smile:  Again, from the above you seem to be referring to AMD's equivalent of Intel's "Turbo Mode". AMD call it "AMD Turbo Core". In which case that's no different to Intel CPU's.

[BaldyB 20]

Not sure what you mean there. Motherboards do indeed have BIOS's that "facilitate" [make easier, more convenient] overclocking. But how does the AMD CPU itself make overclocking more convenient? Intel CPU's have "turbo Mode" that automatically overclocks by a few hundred megahertz and AMD CPU's have the equivalent. Is that what you mean? If so it's no different to many generations of Intel CPU's.

 

Or are you referring to something else? Can you elaborate.

 

In AMD's published documentation:

Pure Power – more than 100 embedded sensors with accuracy to the millivolt, milliwatt, and single degree level of temperature enable optimal voltage, clock frequency, and operating mode with minimal energy consumption;

• Precision Boost – smart logic that monitors integrated sensors and optimizes clock speeds, in increments as small as 25MHz, at up to a thousand times a second;
• Extended Frequency Range (XFR) – when the system senses added cooling capability, XFR raises the Precision Boost frequency to enhance performance;

My read.

 

​1: Precision Boost varies clock speeds on the fly!

2: XRF Determines, "when the system senses added cooling capability, XFR raises the Precision Boost frequency to enhance performance;"

 

Now you know!

 

Best

BaldyB

[Nuno Pinto 1,394]

Well said Nuno.

At the end of the day it doesn't matter which company supplies the CPUs...

All that matters is bang for the bucks :-)

 

Exactly. Let's hope indeed it makes a difference this time around.

[TechguyMaxC 516]

I'm very excited for this processor release.  I don't necessarily expect it to outperform Intel for MSFS/P3D because these sims are still largely constrained by single thread CPU performance and that is at this point in time mostly limited by clock speed.  X-Plane might be a different story though, I've never tested it extensively to see how it responds to higher thread count.  All that being said, the fact that Ryzen's prices are so much better than Intel's makes me highly inclined towards purchasing an 8 core model at launch.  

[OmniAtlas 93]

Here is the boxart with all the known specs and prices. Exciting!

 

http://www.digitaltrends.com/computing/amd-ryzen-cpu-box-art-prices-speeds/

[martin-w 3,234]

In AMD's published documentation:

Pure Power – more than 100 embedded sensors with accuracy to the millivolt, milliwatt, and single degree level of temperature enable optimal voltage, clock frequency, and operating mode with minimal energy consumption;

• Precision Boost – smart logic that monitors integrated sensors and optimizes clock speeds, in increments as small as 25MHz, at up to a thousand times a second;
• Extended Frequency Range (XFR) – when the system senses added cooling capability, XFR raises the Precision Boost frequency to enhance performance;

My read.

 

​1: Precision Boost varies clock speeds on the fly!

2: XRF Determines, "when the system senses added cooling capability, XFR raises the Precision Boost frequency to enhance performance;"

 

Now you know!

 

Best

BaldyB

 

 

 

 

So basically it's the same as Intel's Turbo Mode and AMD Turbo Core, just a bit more sophisticated. Fit a high end cooler and it Turbo's a bit higher.

 

I doubt it's a case of, fit a high end cooler and suddenly you'll see a 15-20% overclock, It will just tweak higher. After all, it can't overclock appreciably as it couldn't determine if the overclock was stable as no stress tests are run. 

 

It would have no idea if it were a golden chip or a not so impressive chip. Therefore it would have to make sure the temp was well under TJ Max and apply a generic voltage that applied to "most CPU's". I suspect a percentage of chips would be unstable with this method requiring more voltage and indeed some needing less voltage. 

 

You will still need to overclock manually and run stress tests, or with a Windows Utility, for what we enthusiasts would regard as a significant overclock at the lowest stable voltage.

 

From what I've read this morning, AMD are quoting the turbo frequencies, but they are keeping very quiet in regard to the max frequencies that XFR will go to. All they are saying is that XFR will go "beyond" a certain frequency with adequate cooling. Don't be surprised if it's not much.

 

Anyway, we will see how it functions in practice when it's released. I will be intrigued to see how this functions.

[BaldyB 20]

 

 

So basically it's the same as Intel's Turbo Mode and AMD Turbo Core, just a bit more sophisticated. Fit a high end cooler and it Turbo's a bit higher.

 

Not quite.

They both have Turbo Core/Mode. AMD's added feature is the ability to sense additional cooling capacity and adjust the OC limits, again at the chip level, not a BIOS function!

With some BIOS's you can enable or lock out Turbo Mode/ Core. Only AMD can automate this function. In essence the upper Turbo OC limits are dependent on your cooling capacity with sensors on the chip, not a BIOS function!

 

Best

BaldyB

[martin-w 3,234]

They both have Turbo Core/Mode. AMD's added feature is the ability to sense additional cooling capacity and adjust the OC limits, again at the chip level, not a BIOS function!

 

 

 

 

Well yes, I'm aware of that Baldy. Incidentally, Intel Turbo Mode is via the chip as opposed to the BIOS too. XFR may be an additional feature to AMD Turbo Core Mode, but it's the same principle. Namely the CPU overclocking to a degree. As I said, it's the same as Turbo, an auto overclock, but more sophisticated as it monitors temps etc. 

 

 

 

With some BIOS's you can enable or lock out Turbo Mode/ Core.

 

 

 

 

And It's expected that the same function will be included in the BIOS for Ryzen's XFR too, from what I was reading this morning.  

 

 

 

In essence the upper Turbo OC limits are dependent on your cooling capacity with sensors on the chip, not a BIOS function!

 

 

 

Again, I was aware of that. My point is that to enable such a feature, it must assign a voltage that is higher than most chips require. The silicone lottery dictates that some chips will require higher voltage for a given overclock. Thus AMD have no choice but to configure this technology to over volt to a degree, in order to accommodate those chips. Thus a "more efficient" overclock is going to be achieved by manually overclocking and running stress tests, or indeed via a windows based auto overclock technology that also runs stress tests.

 

I don't wish to rain on AMD's parade, but I don't expect this technology to do what I "suspect" you think it will. It's a more sophisticated version of Turbo, that will boost the CPU's frequency higher if you have adequate cooling but there's no information anywhere to confirm this hypothesis that it will boost the OC "significantly". Higher than Turbo yes, but I don't expect it to be  an enthusiast level overclock.

 

I feel it's significant that AMD are refusing to release any other details except "it will overclock higher than the Turbo Core frequency". Could be 100 MHz more, 200, 300, we don't know.

 

What I am pretty sure about, is that the 4 Core version of Ryzen won't be overclocking automatically to Kaby Lake's 5 GHz average overclock, regardless of how awesome your cooling is.

 

I could of course be completely wrong. We shall see.  :smile:

[TechguyMaxC 516]

Forgive me for getting technical here but allow me to explain why, in fact, AMD can auto-OC the new Ryzen chips:

 

In electrical engineering there's this thing called a Shmoo plot.  A Shmoo plot allows you to extrapolate information from a range of data points in order to determine the voltage necessary to run at a target clock speed (for example).  If I recall correctly you need 2 pairs of voltage + clock speed in order to perform this extrapolation. Since AMD has a multitude of sensors built into Ryzen, they can use this principle to determine the appropriate voltage for a given clock speed and the particular chip in question, then apply these settings in real time.  

 

Now, there is variation in silicon so settings that work for one chip won't necessarily work for another.  Example: if you go buy a Ryzen 1800 X on launch day from Newegg and I get one from my local Micro Center, we could very well end up with chips with very different capabilities.  Your chip may only overclock to say 4.4GHz whereas mine might OC to perhaps 4.6GHz, both at the same (or similar) voltage even.  To account for this, each chip is "tagged" with the necessary voltage(s) for default operation in a value called "VID" (at least that's what Intel calls it, not sure about AMD).  Each chip knows this value and applies it as necessary for a desired clock speed.  Using the VID of each particular chip and the base and boost clocks they can essentially calculate a Shmoo plot (either by means of testing at the factory and then programming the values into a table on the chip, or in real time) to determine precisely how much voltage is necessary for a target clock speed.

 

I hope that all makes sense.  If not, take my word for it?  

[martin-w 3,234]

 Since AMD has a multitude of sensors built into Ryzen, they can use this principle to determine the appropriate voltage for a given clock speed and the particular chip in question, then apply these settings in real time.

 

 

 

 

Right, so in essence my belief that the Ryzen chip can't determine it's own position in the overclockability [did I invent a new word?] hierarchy, whether it's a golden chip or a crap chip, is in essence wrong?

 

Ryzen can determine, all by itself, that it is less capable or more capable and set voltage accordingly?

 

If yes, I am enlightened and grateful for your contribution.  :smile: 

[TechguyMaxC 516]

 

 

 

Right, so in essence my belief that the Ryzen chip can't determine it's own position in the overclockability [did I invent a new word?] hierarchy, whether it's a golden chip or a crap chip, is in essence wrong?

 

Ryzen can determine, all by itself, that it is less capable or more capable and set voltage accordingly?

 

If yes, I am enlightened and grateful for your contribution.  :smile: 

 

Essentially, yes.  OC capabilities are somewhat determined at the factory when the VID for each chip is established and programmed in.  There are limits, of course.  The temperature of the chip has a sizable impact on the voltage requirements (which in turn effects temperature so it's a feedback loop).  This is why AMD is stressing the need for good cooling in order to take advantage of this feature.  

 

I will say this.  I do not personally believe AMD is going to auto-OC Ryzen by more than the range of frequency you have already stated.  If they're too aggressive they could end up with a lot of dead chips on their hands and at that point AMD will be fighting to stay in business.  

 

Edit: I want to clarify something from my previous post.  When I gave example clockspeeds for overclocked chip potential, I was in fact referring to manual overclocking which an enthusiast might perform, rather than Ryzen's use of XFR.  I would honestly be quite surprised if XFR ever approached these clock speeds (with the SKUs that have been announced so far, anyway).  That being said, XFR's clock speed scaling will also likely vary from chip to chip.  As you indicate, golden sample chips will hit higher clocks than standard or sub-standard Ryzen samples.  

[BaldyB 20]

 

 

Each chip knows this value and applies it as necessary for a desired clock speed. Using the VID of each particular chip and the base and boost clocks they can essentially calculate a Shmoo plot (either by means of testing at the factory and then programming the values into a table on the chip, or in real time) to determine precisely how much voltage is necessary for a target clock speed.

Ditto!

 

Best

BaldyB