Ryzen: Strictly technical

[wahdangun]

It appears that you can use bclk overclocking to manage to overclock the processor while still maintaining the processor's own automatic overclocking abilities. Depending on the silicon lottery, and the power delivery of your motherboard, single core boost can take an excursion over 4.5 GHz.

There are problems with it though. Past about 103 MHz, the PCI-E bus drops back to 2.x speeds and some NVMe drives begin to have controller issues. The processor PB/PB2 tends to stop a few MHz past that from what I've read as well.

I'm actually quite interested in seeing how a 2400g behaves with Bclk overclocking. I want to think that the overall SOC overclock will prove to be a nice improvement to system performance. With little need for add in cards, PCIE behavior may be different and not impactful. Bclk OC also seems to be helpful in memory overclocking as well.

But how about motherboard with external clock generator ?

Some of redditor can use bclk up to 105, and the hard limit was like 104,9.

But I'm never see reviewer achieve that and some benchmark number, my guess it can be pretty neck to neck with intel 8700k.

 

[burninatortech4]

Has the Asmedia chipset received an optical shrink from 55nm (x370/b350) to something smaller for x470? There were rumors that was what AMD meant by "Lower Power" on their marketing slides. If there were no changes, do we expect any new features on the 500 series chipsets? Any thoughts/predictions?

 

[The Stilt]

Has the Asmedia chipset received an optical shrink from 55nm (x370/b350) to something smaller for x470? There were rumors that was what AMD meant by "Lower Power" on their marketing slides. If there were no changes, do we expect any new features on the 500 series chipsets? Any thoughts/predictions?

Could be wrong, but AFAIK the lower power consumption is achieved through more aggressive power management in the firmware.

 

[bitxan]

Hello, I read all the thread and I see that the maximum admissible voltage for ddr4 would be 1.50 volts to avoid damaging the IMC (as he says Robert Hallock) , but ..... what is the maximum frequency at which Infinty Fabric can work without risk of damage?
What is the maximum frequency limit at which infinity Fabric AMD 2700x can work?

 

[Markfw]

Can somebody tell me what the symbols at the top of Ryzen master mean ?
PPT 83%
TDC 79%
EDC 97%
PTC 95C

I have been having a problem with running hot, I now have a Noctura NH-D15 on it and checked the thermal paste. Its STOCK and running 92c @ 1.35 vcore (about)
And yes, the 2 fans are both spinning. This is troubling. Could my CPU be defective ? When running Rosetta it was 68c, under Asteroids@home its now 92c..

I did update my Taichi to 1.35a beta bios, hoping it was a reporting problem. And there is hot air coming from the fans on the Noctura. OPEN CASE.

The only difference is that now the 2 video cards are under 100% load (1050TI and 1060) but I put the case with the open side up, so the heat from those should not be an issue.

I just noticed, I did not include tech details:

2700X at stock, but it says its running around 1.35 vcore.
X470 Taichi, bios 1.35a

And last, it was running 4 ghz all core@stock running Rosetta, but on Asteroids, its only running 3.75-3.8. I think that tasks say something about AVX, would that make it run hotter and slower ?

 

[BeepBeep2]

Can somebody tell me what the symbols at the top of Ryzen master mean ?
PPTn 83%
TDC 79%
EDC 97%
PTC 95C

I have been having a problem with running hot, I now have a Noctura NH-D15 on it and checked the thermal paste. Its STOCK and running 92c @ 1.35 vcore (about)
And yes, the 2 fans are both spinning. This is troubling. Could my CPU be defective ? When running Rosetta it was 68c, under Asteroids@home its now 92c..

I did update my Taichi to 1.35a beta bios, hoping it was a reporting problem. And there is hot air coming from the fans on the Noctura. OPEN CASE.

The only difference is that now the 2 video cards are under 100% load (1050TI and 1060) but I put the case with the open side up, so the heat from those should not be an issue.

I may be wrong:
PPT - Package Power Tracking, how close to VRM limit the CPU package power draw is, 100% (maximum) should throttle.

TDC - Thermal Design Current, maximum current available in a load condition based on the thermal situation (sustained / avg VRM capacity when hot, 100% is maximum)

EDC - Electrical Design Current, maximum peak current the VRM can supply for short periods / transient conditions (100% is maximum)

PTC - If I'm not mistaken it will be package temperature, tCTL(?)

Hello, I read all the thread and I see that the maximum admissible voltage for ddr4 would be 1.50 volts to avoid damaging the IMC (as he says Robert Hallock) , but ..... what is the maximum frequency at which Infinty Fabric can work without risk of damage?
What is the maximum frequency limit at which infinity Fabric AMD 2700x can work?

IF frequency is tied to DRAM, running a higher frequency will not cause damage. Keep SoC voltage under ~1.2v, CPU may exhibit negative DRAM frequency scaling well below that. (1.05-1.15v).

You only need to worry about stability at high memory clock, and not to use too much voltage.

 

[CrazyElf]

The interesting question becomes what the IPC of Zen 2 will be.

AMD claimed that they had major changes that they could utilize to improve Zen 2, perhaps in time to oppose an Intel 10nm Skylake shrink.

I'd be very interested to see everyone's thoughts on the type of IPC that we might see in Zen 2. I think that we may see better than Skylake IPC, at least in some benchmarks. On average though, it is hard to say, but certainly the gap could close.

The design was completed months ago, which means they probably have the floorplan all ready to go, except for some minor tweaks.

Makes me wonder about those 5+ GHz claims on 7nm by GloFo if it ever shows up in H2 2018/Q12019.

Historically, IBM has been very good with clockspeeds and we know that there has been a lot of collaboration between IBM, AMD, and GF on this front.

It really isn't possible to say what the next generation is. Past performance is no assurance of future success. Even the people who themselves are working on it right now might not know. There might be unforeseen difficulties with 7nm or it might go smoothly. Certainly Intel (and their 10nm process is comparable to the TSMC/GF 7nm nodes) has experienced a ton of challenges ramping up their 10nm, leading to an unprecedented 3 year delay.

Another possibility is that AMD might elect to go with TSMC instead for Zen 2. That would be a major departure, but quite possible. Alternatively, AMD may want a second supplier, much like how Nvidia has done so with Samsung on their GPUs.

One thing I am noticing from the VRM on some of the high-end X470 boards is they are extremely overkill.

http://www.overclock.net/forum/27199649-post2954.html
https://www.hardwareluxx.de/community/f12/pga-am4-mainboard-vrm-liste-1155146.html

Excepting the ASUS TUF, anything but the flagship Gigabyte board, and any MSI board, of course...

Could it be that Ryzen 2 next year will have 12 core CPUs for the desktop platform?

The way the CCX is made, this would require a more complex design for the L3 caches. In a fully connected node, the number of connections = n(n - 1) / 2.

For a 4 core CCX, that is 4 x (4 - 1)/2, or 6 connections. In a 6 core CCX, that would require 6 x (6 - 1)/2, or 15 connections. That is 2.5x as many connections, but only 1.5x as many cores. You'll need a lot more L3 cache too. If they do go the 6 core route, they may be forced to abandon the fully connected topology in favor of something else, perhaps a ring or a partial connection that minimizes the number of hops.

Perhaps a 3 CCX design of 4 cores might be another option. I think that AMD should add an L4 cache to facilitate inter-CCX connectivity. It doesn't have to be a big cache, but it might reduce latency.

In many ways, this resembles network topology design, with very similar trade-offs.

 

[stAbb]

@Markfw,

Under Windows, I always start my troubleshooting with System Information Viewer.
With my Ryzen 1600x it shows the temperatures, both on package and on die, and voltages of the CPU as well as of each individual core.
I am sure other programs work just as well, but this program works on everything from 9x to windows 10.

Having one tool is always preferable for me, as I can then spend the time to get to know it properly.

As for Ryzen specific quirks, I tested overclocking on my 1600x by fiddling with C0, C1 and C2 states, setting up custom speeds and voltages.
It only uses the C0 voltage while under load though, even when I down clock it to the C1 state by using a custom windows power profiles.
When I don't touch the C0, C1 and C2 states the voltages go down with clock-speed as expected.
Perhaps the same thing is happening to you where the CPU only uses the C0 voltage?

For me, when this happens, System Information Viewer will show the C0 Voltage (I presume around 1.35 Volts, judging from the Ryzen Master readout) on the individual cores, regardless of load.
In my case the CPU voltage does go down when idling.

Hope this helps and with kind regards,
stAbb

System Information Viewer: http://www.rh-software.com/
PS: I've seen the FSB readout go down after setting the C0 state to a higher clock-speed from the OS. This is a bug in the readout as my performance clearly doesn't go down.
PPS: If you manage to fix the issues and feel like overclocking your system later, I'd start with upping the FSB (up to 103) and see if that doesn't mess up the C state voltages.

 

[Markfw]

OK, help me out here... Threadripper at stock, 3.4 ghz running 3200 memory cas 14 b-die

vs

2700x @ 4.1 ghz with 3600 memory@3466, cas 15. b-die.

In cinebench SINGLE core, the 2700x should easily win. But NO. 173 for threadripper, 170 for the 2700x.

How is this possible ?

 

[BeepBeep2]

OK, help me out here... Threadripper at stock, 3.4 ghz running 3200 memory cas 14 b-die

vs

2700x @ 4.1 ghz with 3600 memory@3466, cas 15. b-die.

In cinebench SINGLE core, the 2700x should easily win. But NO. 173 for threadripper, 170 for the 2700x.

How is this possible ?

Single core turbo (1950X) is 4.2 GHz.

 

[Markfw]

I forgot to check for that, thanks !! The 2700x is set for 4100 all-core.

Edit: I reran while running Ryzen master, and yes, it hit 4.2

 

[gupsterg]

The "Precision Boost Override" feature available on 400-series motherboards allows increasing the physical limiters mentioned earlier. On SKUs belonging to the 105W TDP infrastructure group, the default limiters are following: PPT 141.75W, TDC 95A, EDC 140A and tJMax of 85°C (absolute, excl. offset).

When "Precision Boost Override" mode is enabled (AGESA default), PPT becomes essentially unrestricted (1000W), TDC is set to 114A and EDC to 168A. These limits can be customized by the ODM so that the new limits will comply with the electrical characteristics of the motherboard design in question.

Thanks for Pinnacle Ridge Chapter .

When PBO is enabled what happens in context of tJMax, is it still 85°C or extended?

Is the 85°C a throttling temp limit or shutdown? if not shutdown what is shutdown temp limit?

 

[The Stilt]

Thanks for Pinnacle Ridge Chapter .

When PBO is enabled what happens in context of tJMax, is it still 85°C or extended?

Is the 85°C a throttling temp limit or shutdown? if not shutdown what is shutdown temp limit?

With stock configuration:

- 85°C target temperature.
- 115°C shutdown temperature.

Both relative to tDie, i.e. 95°C & 125°C tCTL e.g. on 2700X (10°C offset).

In normal operating mode these CPUs don't hard throttle, they pull back.
Meaning they will reduce the frequency until the temperature stays at or below the threshold.
For example if the CPU maintains 3950MHz average frequency on all cores at 80°C tDie, at 85°C tDie the average frequency will be slightly lower.

If the CPU is running in OC-Mode then the temperature target becomes obsolete.
In OC-Mode the shutdown temperature of 115°C is the only temperature related limit, unless the ODM chooses to manually include a separate "hard throttling" limit.
When the "hard throttling" threshold is reached, the CPU frequency drops to the minimum frequency state, which typically is 550MHz.
By default such limit is not used.

The shutdown temperature cannot be changed, however the temperature target can be changed through the bios (typically CBS menu).

 

[Bubbleawsome]

Wow zen can hit 115C? That’s impressive

 

[Insert_Nickname]

Wow zen can hit 115C? That’s impressive

The old Socket 462 Athlons could hit 370C. Though they didn't function too well afterwards...

Seriously though, I don't think 115C is good for the CPU long term. But I think that goes without saying.

 

[Markfw]

Wow zen can hit 115C? That’s impressive

Well, my 2700x says 95c, and when it hits that it WILL downclock to keep the temp at 95c or under.

 

[virpz]

I am curious on how different these 12 nm can be between them... Anyone was able to stabilize 4.2GHz with vcore under 1.28V on a 2700X ?

 

[Markfw]

I am curious on how different these 12 nm can be between them... Anyone was able to stabilize 4.2GHz with vcore under 1.28V on a 2700X ?

Well, I have 4.1@1.23 vcore. I can't imagine 4.2 would be much harder

 

[The Stilt]

Well, I have 4.1@1.23 vcore. I can't imagine 4.2 would be much harder

In most cases it is, if not impossible altogether.
The frequency / voltage curve is nearly vertical after ~3.9GHz on 14nm LPP and not much different on 12nm LP beyond 4.1GHz.

 

[CatMerc]

In most cases it is, if not impossible altogether.
The frequency / voltage curve is nearly vertical after ~3.9GHz on 14nm LPP and not much different on 12nm LP beyond 4.1GHz.

I suspect any chip that can do 4.2GHz at <1.3V is being binned for Threadripper 2. Same happened with Threadripper 1.

 

[virpz]

In most cases it is, if not impossible altogether.
The frequency / voltage curve is nearly vertical after ~3.9GHz on 14nm LPP and not much different on 12nm LP beyond 4.1GHz.

I am asking because for 1700 and 1800X my minimum vcore vs frequency curve was pretty much in agreement to your findings. For the 2700X I got 4.2GHz@1.25V for a month now and it has been nothing but rock solid.
I will try the "FIT" vcore consultation you described to see what comes out.

I suspect any chip that can do 4.2GHz at <1.3V is being binned for Threadripper 2. Same happened with Threadripper 1.

Seems like one of those dies slipped to me.

 

[CatMerc]

I am asking because for 1700 and 1800X my minimum vcore vs frequency curve was pretty much in agreement to your findings. For the 2700X I got 4.2GHz@1.25V for a month now and it has been nothing but rock solid.
I will try the "FIT" vcore consultation you described to see what comes out.


Seems like one of those dies slipped to me.

Try to run Realbench overnight and see if it doesn't have any errors.

 

[virpz]

Try to run Realbench overnight and see if it doesn't have any errors.

I had a 8 hours run at prime95-Small FFTs, will give realbench a try.

have a look at other not-so-stressed-overclocks.

Spoiler

https://youtu.be/hypxsMUkzSI

 

[Elfear]

I have a CH6 motherboard and am using the Stilt's excellent memory timing presets. With the 3466 preset, HWiNFO shows my dram voltage at 1.417V. Is that going to be an issue long term?

 

[The Stilt]

With the 3466 preset, HWiNFO shows my dram voltage at 1.417V. Is that going to be an issue long term?

Not at all.
All DDR4 ICs must be able to sustain up to 1.5V according to JEDEC specs.