Navi/Ryzen 3000 launch Megathread

[Blademaster91]

11 minutes ago, imreloadin said:

Because you've just gotta have an excuse for these not hitting 5 Ghz like all your previous claims said they would right

Also interesting reviewers aren't showing off overclocks in their initial reviews, if they were getting 5Ghz as the "leakers" were claiming, then reviewers would be showing it off and not saving it for another video.

9 minutes ago, S w a t s o n said:

>The socket literally will not deliver more power
>Excuse

Yes, yes that's how that works. It's definitely my bad that AMD set the socket power limit to 140W on AM4....

Except the board can deliver way more than 140w, OEM's wouldn't be wasting designs if the socket couldn't handle more than 140w. Hopefully a BIOS patch could fix the limitation?

[INABLITY]

I saw topic with list motherbords compatibility for 3gen and my x370-f in this list. But i still don't have any updates from ASUS about this theme..

[S w a t s o n]

Just now, imreloadin said:

Because that's how it worked with previous TR versions right

Previous TR versions werent power constrained. So it wasnt an issue.

It's like the old RAM meme. Computer is slow, buy more RAM. Except that only works when you NEED more RAM.

In this case, the CPU's NEED more power, at least to OC past fucking 4.3GHz all core. We literally know the silicon goes up to 4.7GHz CONFIRMED minimum (3950x), more power allows the cores to scale higher.

[LukeSavenije]

Just now, ybriK said:

As expected, still can't beat my 5.1GHz 9900K.

well.... numbers show otherwise...

[leadeater]

12 minutes ago, S w a t s o n said:

>The socket literally will not deliver more power
>Excuse

Yes, yes that's how that works. It's definitely my bad that AMD set the socket power limit to 140W on AM4....

Very much doubt it's a socket limit, power levels are far too low and this is a known observed characteristic of the Zen architecture. Overall it's a much more complicated situation but there is simply no way the socket, VRM or bios is limiting the power to the CPU preventing higher clocks.

 

Given all the characteristics of a silicon chip there is a required temperature to achieve a given clock, this is more typically looked at in XOC than ambient/general usage but suffice to say LN2 isn't allowing the socket to deliver more power or stopping it from melting, the silicon die requires that lower temperature to function at all.

 

Ryzen is no different, Intel isn't any different, Nvidia nor Radeon because they are all silicon architectures but the differences in regards to what clocks they can achieve are architecture differences not fab or silicon. All parts make a whole so don't take it as the fab node doesn't matter either but everything is a balance so a node alone isn't going to allow higher clock and neither is an architecture tweak.

 

Basically yes 5Ghz is possible on Zen2, at what temperature this is possible at GN will find in their upcoming LN2 overclocking.

[S w a t s o n]

3 hours ago, Blademaster91 said:

Except the board can deliver way more than 140w, OEM's wouldn't be wasting designs if the socket couldn't handle more than 140w. Hopefully a BIOS patch could fix the limitation?

That's my question. Why do $700 boards exist when you run into a 140W socket power wall

 

3 hours ago, leadeater said:

-snip-

Anandtech themselves confirmed that socket power was limiting



Note that only 105W CPUs get to use up to 140W. The limit goes down as the TDP goes down.

[Blademaster91]

1 minute ago, ybriK said:

As expected, still can't beat my 5.1GHz 9900K.

Tbh the ryzen 3000 chips don't need to beat Intel, being close enough in gaming is more than sufficient for AMD.

[RobFRaschke]

3 minutes ago, ybriK said:

As expected, still can't beat my 5.1GHz 9900K.

In what workload?

[dfsdfgfkjsefoiqzemnd]

4 minutes ago, ybriK said:

As expected, still can't beat my 5.1GHz 9900K.

It's still early.  We'll see what happens once people start overclocking Zen2 and BIOSes and drivers get better. 

[leadeater]

3 minutes ago, S w a t s o n said:

Anandtech themselves confirmed that socket power was limiting

And yet GN has EPS12V power draw figures at ~175W. Edit: Removing a power limit doesn't allow you to achieve a clock not possible at the temperature you are trying it at. If it's not possible at the temperature then it's not possible, no amount of extra power pushed in to the chip will make it happen.

[S w a t s o n]

Just now, leadeater said:

And yet GN has EPS12V power draw figures at ~175W.

Can I see?

 

[Ravendarat]

Its interesting that Bitwits benchmarks tell a different story than Jayz and LTT. I wonder why the results are so different

[RobFRaschke]

Just now, Blademaster91 said:

Tbh the ryzen 3000 chips don't need to beat Intel, being close enough in gaming is more than sufficient for AMD.

Especially considering the value proposition. the 3600 is 20% cheaper than the 9600K and while the 9600K does win in gaming, it's by 10% or less in gaming, and the lack of hyperthreading/SMT on the 9600K means the 3600 destroys the 9600K in productivity.

[LukeSavenije]

2 minutes ago, leadeater said:

And yet GN has EPS12V power draw figures at ~175W.

i'm confused

 

tho i have a short theory

 

stock lock

[S w a t s o n]

3 hours ago, imreloadin said:

-snip-

Yea it's not like I dont have an entire page of Anandtech's in depth analysis of how the various power/current limits work is evidence right?

https://www.anandtech.com/show/14605/the-and-ryzen-3700x-3900x-review-raising-the-bar/18

[leadeater]

1 minute ago, S w a t s o n said:

Can I see?

 

 

Sorry it was 170.4 not, ~175W. Rounded up but yea my mistake on where that 4 was.

[S w a t s o n]

Just now, leadeater said:

 

Sorry it was 170.4 not, ~175W. Rounded up but yea my mistake on where that 4 was.

I did find it, on the blender run. I would like to know their accuracy and methods/variables. I know they are using a current clamp but afaik this is an outlier. Even if it's actually reaching 170W, that's not that much more than 140W. For example 9900k can go over 200W easily at stock. Anything approaching 5GHz on all cores was never going to sip power
https://www.anandtech.com/show/14605/the-and-ryzen-3700x-3900x-review-raising-the-bar/18
 

[cj09beira]

10 minutes ago, S w a t s o n said:

Previous TR versions werent power constrained. So it wasnt an issue.

It's like the old RAM meme. Computer is slow, buy more RAM. Except that only works when you NEED more RAM.

In this case, the CPU's NEED more power, at least to OC past fucking 4.3GHz all core. We literally know the silicon goes up to 4.7GHz CONFIRMED minimum (3950x), more power allows the cores to scale higher.

considering the voltage they are needing to even achieve 4.4ghz (1.4-1.5v) (4.3 isn't much better either) i am quite confused as to how they are able to boost to 4.6ghz, the difference is so large 

[RobFRaschke]

1 minute ago, S w a t s o n said:

I did find it, on the blender run. I would like to know their accuracy and methods/variables. I know they are using a current clamp but afaik this is an outlier. Even if it's actually reaching 170W, that's not that much more than 140W. For example 9900k can go over 200W easily at stock. Anything approaching 5GHz on all cores was never going to sip power
https://www.anandtech.com/show/14605/the-and-ryzen-3700x-3900x-review-raising-the-bar/18
 

I'm really not trying to get into this argument, but 170W is over 20% more than 140W. That's quite a bit much more.

[Delicieuxz]

How is this chart to be read - what is the correlation between the build price and the % of AMD's value against Intel?

 

 

When 1%, 0.1% lows / 99th percentile metrics started being reported in benchmark videos I had a similar confusion. I sort of wish ambiguous metrics would have explanations to go along with them somewhere on the screen, like maybe in some small print or a link to an explanation.

[S w a t s o n]

1 minute ago, RobFRaschke said:

I'm really not trying to get into this argument, but 170W is over 20% more than 140W. That's quite a bit much more.

Not for overclocking, power scaling goes out the window.

Quote from Anandtech
 

Quote

In POV-Ray, running the 3900X at a flat 4.3GHz gives it a 8.2% performance boost over stock. Enabling PBO doesn’t make much difference in multi-threaded workloads for the 3900X as it’s still being limited by the 142W PPT limit.

Unfortunately we weren’t able to further investigate raising the PPT limit for this article due to time contraints as well as currently non-final firmware version for X570 motherboards from the vendors.

 

[RobFRaschke]

Just now, cj09beira said:

considering the voltage they are needing to even achieve 4.4ghz (1.4-1.5v) (4.3 isn't much better either) i am quite confused as to how they are able to boost to 4.6ghz, the difference is so large 

4.6 on a single core requires MUCH less power than overclocking 8 or 12 cores to 4.3.

[RobFRaschke]

3 hours ago, Delicieuxz said:

How is this chart to be read - what is the correlation between the build price and the % of AMD's value against Intel?

 

When 1%, 0.1% lows / 99th percentile metrics started being reported in benchmark videos I had a similar confusion. I sort of wish ambiguous metrics would have explanations to go along with them somewhere on the screen, like maybe in some small print or a link to an explanation.

Looking only at the CPU cost itself is the Upgrade metric. The rest of the numbers are based on a full system build at that cost, with the CPU being only a decreasing portion of the cost of the build.

[LukeSavenije]

@S w a t s o n @leadeater from what i can see is the 140w lock just for stock clocks, so you can easily go past it if you want to oc it

[leadeater]

1 minute ago, S w a t s o n said:

I did find it, on the blender run. I would like to know their accuracy and methods/variables. I know they are using a current clamp but afaik this is an outlier. Even if it's actually reaching 170W, that's not that much more than 140W. For example 9900k can go over 200W easily at stock. Anything approaching 5GHz on all cores was never going to sip power
https://www.anandtech.com/show/14605/the-and-ryzen-3700x-3900x-review-raising-the-bar/18

Well we can apply some simple logic to this. You can, like almost always disable or increase power limits, this is a stock package limit not socket or platform because Zen+ as shown draws more and the 3950X will draw more than current released Zen2 products.

 

But the real kicker is if it were possible on a 240/280 CLC why is there an absence of any reviews showing it? If it were possible it would have been done, GN went through the effort of overclocking the 3900X and specifically stated that is the maximum possible and products are coming out stock closer to the limit than before. If all signs point to what has been shown as what you can achieve on ambient then there is no reason to believe it is not the case. I know GN review was on the 3600 but there was enough information shown and enough explanation in that video to conclude even the best (pre tested by AMD) chips maxing out at 4.3GHz is what you will get.