FEATURED AMD ZEN Discussion (Previous Rumor Thread)

Let me put it this way, unless you work at Intel/AMD in this area, you will never understand how boost works. There are so many different factors inside a CPU that could cause a CPU to change its multiplier. To guess this would be based off of complete ignorance of the design process. Operate under the assumption that boost has its own mind that you cannot control. Utilize based of series of tests and data points that determine if the tool should be on or off for a particular test.

This isn't a car's turbo where you have only have a handful of variables. We are talking about a system of variables that requires years of direct experience to understand.

I think he's referring to the 128-bit AVX vs. 256-bit AVX.

http://forum.hwbot.org/showthread.php?t=167111

I'm still wondering where the R5 release info was... I don't recall reading that...

Helgaiden said:

Speculation on Temps has been rather quiet. Anyone want to comment on that, guys? I have a few friends playing devil's advocate saying these may run hot since temp conversations have been rather muted that its like AMD is hiding something. Other guy speculates chip separation (whatever that means...) due to heat (didnt even happen with prescott) and is advising to wait 6 months to a year to see how everything plays out with these new chips....

im just left scratching my head. These things simply cannot run any hotter than their previous gen counterparts, right? And imo, inclusion of XFR points to AMD being pretty confident of operating within a good enough range to extend past specified boost ratings. Thats a positive sign.

Click to expand...

XFR goes 100mhz over... that's it. My concern is 4 ghz boost, 4.1ghz XFR, and 5.2 ghz on ln2... that is a steep slope of cooling for 1ghz overclock....

BTW

http://www.intel.com/content/www/us...ology/turbo-boost/turbo-boost-technology.html

Due to varying power characteristics, some parts with Intel® Turbo Boost Technology 2.0 may not achieve maximum turbo frequencies when running heavy workloads and using multiple cores concurrently.

Click to expand...

And the only white paper Intel has on their Turbo tech, released in 2010.

Johan45 said:

Yeah what he said^^^. The cpu either runs the instructions or it doesn't, so unless it downclocks automatically when it reads them that statement doesn't seem to make any sense to me.

Click to expand...

Intel does 256-bit operations in one go. Ryzen will need to take two 128-bit bites. Power per throughput should be comparable.

If Ryzen does keep up with Intel, I'd be really happy to be wrong on this one! This will possibly be the first bench I do after setting up the system.

I went ahead and took the pre-order plunge
I ordered the following:
R7 1800x - Will be released on the 2nd + 2-3 day shipping.
Asus CHVI-H - Will be released on the 2nd + 2 day shipping.

I thought I read on here that mobo's would be released on the 25th??????

EarthDog said:

XFR goes 100mhz over... that's it. My concern is 4 ghz boost, 4.1ghz XFR, and 5.2 ghz on ln2... that is a steep slope of cooling for 1ghz overclock....

Click to expand...

All this XFR hype for only 100mhz? Blah.

- - - Updated - - -

Bluefalcon13 said:

I'm still wondering where the R5 release info was... I don't recall reading that...

Click to expand...

I know the pic is from WCCFtech but its a slide from the event so it should be legit

Intel max turbo clock is a myth on current OS. OS like new Windows will balance load and CPU will never go up to max turbo frequency. Look at the laptops. To see max clock in CPU-Z you have to disable all other cores or force it other way. The same is on Xeons. Even 2nd step isn't easy. At the end mobile i5 are for most of the time as good as i3.
When you overclock something manually then you force max ratio but at auto, turbo boost will never reach it in typical environment.

I really don't care how high is turbo frequency when I set everything manually anyway. My only concern is possible difference between all higher chips like maximum frequency or any changes in architecture which are not described in general specification. I wouldn't be happy to buy a CPU and 1-2 weeks later find out it won't overclock as high as higher version which I could buy but I wanted to save $50-100. Like Janus said, all are pre-ordering these chips but really we don't know anything except what marketing want's us to know. There were no real tests of Ryzen but somehow AMD is pushing pre-orders as official big thing.

My point of view is a bit different ( and some here share that ). I will buy it anyway regardless how it will perform also because I'm curious how is works and I know that in most reviews I won't read all I want because most reviewers are only reviewers and have no additional knowledge about hardware. There are single websites which are providing good articles and most others are only copy/paste marketing stuff.

Funny is to see how all believe in most leaks but most won't even stop for a moment and think why all "leaks" are provided by 2-3 websites. I'm don't want to create next conspiracy theory but most of that is controlled by AMD and those who work with marketing know that from most previous premieres. They made a great job considering how many pre-orders were already made.

Dolk said:

Science time!

With the introduction of 14nm FET sizes, we kinda entered a new realm of material breakdown. We are very close to the point where electron drift (the instantaneous moment when electronics can freely float between atoms) will cause erratic behavior inside the materials. Now some materials account for this behavior and Semiconductors like to exploit it. However when the behavior becomes so intense, it becomes very detrimental affect to the CPU. This can cause lower resistance in gates, creating current leakage or it might cause the very well defined interconnects that must have a specific timing requirement to move faster (memory speed mismatch). There are a lot of things that can start to happen.

For more information on these challenges you should read this article: http://semiengineering.com/interconnect-challenges-grow-3/

But the long and short of it is this: Expect clocks to start to stabilize around a bell curve, influence of temperatures will not matter. The only way to circumvent this is by moving to newer materials.

And for those that really care: Maths https://web.stanford.edu/class/ee311/NOTES/InterconnectScalingSlides.pdf

Click to expand...

So after looking at the links you posted, for overclocking was it the increase of electrical resistance do to heat or feature size change and distortion do to heat in past that had a determinate factor how far you could overclock, also what does temperature change effect in the Die to allow a stable overclock at the same Vcore with a lower temperature compared to the same overclock unstable at higher temperature?

ATMINSIDE said:

If that's accurate... scaling with clocks is fantastic.

Click to expand...

Please explain/elaborate on how you get your answer....

MaddMutt said:

Please explain/elaborate on how you get your answer....

Click to expand...

Because it is exceeding the score at a lower clock???

wingman99 said:

So after looking at the links you posted, for overclocking was it the increase of electrical resistance do to heat or feature size change and distortion do to heat in past that had a determinate factor how far you could overclock,

Click to expand...

Yes to both. New feature size technology, materials, increase in layer stacking, and a whole host of other introductions that doesn't even have to do with the materials will cause the CPU to be different than its last generation and can contribute toward overclocking potential. Now the largest factor has been feature size shrinkage.

also what does temperature change effect in the Die to allow a stable overclock at the same Vcore with a lower temperature compared to the same overclock unstable at higher temperature?

Click to expand...

Right here you are getting into theory of operation. Technically the material will give a specific property under a specific load, thus you can conclude that if the delta temperature is the same dependent on a change in clock and power, the scaling factor should be linear. Unfortunately this is not how the real world works, perfect physics doesn't exist. Instead we have to either go with non-linear step loading or bell curves because the behavior of a system will depend on ranges from various variables. The closer we get to maximizing Moor's Law, the more we will become "stable" in specific areas because of the vast complexity of the system itself.

^^ uhhh......HUM....what ever he just said.

Ryzen 1700 overclocked.

http://wccftech.com/amd-ryzen-7-1700-overclocked-4ghz/

again the same source ... also 4GHz ? it's like stock and doesn't look good:
"We just tested a 1700, it hit 4.0GHz stable in everything, but ONLY in the Crosshair mainboard, the lower-end boards it was hovering around 3.80GHz as the VRM’s were cooking with extra voltage. It however was maxing around 4050MHz, so I’d say 1700 can do 3.9-4.1GHz, of course the 1800X will probably do 4.1-4.3 as no doubt better binned, but if your clocking the motherboard has a big impact on the overclock and so far Asus Crosshair and Asrock Taichi seem the best two."

So it's what I'm talking about regarding motherboards. 65W TDP and problems with temps ? Is this a joke ? There are boards like X99 ITX from ASRock which support 18 core Xeons so what is causing 65-95W TDP CPU to cause so much issues on X370 boards ?

Will wait for B350 board reviews. Will drop Ryzen 1500 6C/12T into it and aim for 3.8-4Ghz. Will have best value gaming machine in town

Crosshair 4 is listed as $349 AUD here. Asus Prime B350-Plus is listed as $149 AUD

Don't need any of the extra features of a more expensive board honestly.

These X370 boards aren't so special looking at all the features ... manufacturers didn't even care to add RGB LEDs like on Intel even though it's so popular in higher series. The best boards have RGB single spot lighting like chipset heatsink only. I was counting on some good X370 mATX or ITX board after premiere but nothing like that is listed for release in first weeks.

Woomack said:

These X370 boards aren't so specials looking at all the features ... manufacturers didn't even care to add RGB LEDs like on Intel even though it's so popular in higher series.

Click to expand...

That's the best feature, IMO; Hopefully the trend will continue. Maybe I'm getting old, but I could care less about having my computer case look like a Christmas tree

Bishoff said:

That's the best feature, IMO; Hopefully the trend will continue. Maybe I'm getting old, but I could care less about having my computer case look like a Christmas tree

Click to expand...

Agreed, keeping prices down is a good thing, and I'm sure bling boards will become available.

Am actually considering the Asus Prime X370 Pro for $239 AUD.

Pro Carbon/Aorus etc are around $300+

https://www.pccasegear.com/category/138_1874/motherboards/amd-socket-am4

Those processor prices though..........

https://www.pccasegear.com/category/187_1873/cpus/amd-socket-am4