Speculation on Ryzen Overclocking

[IEC]

Plot twist: Even the 1700 supposedly supports XFR (just tries to stay within its 65W TDP unless you manually OC it)

Source is Gibbo and 8pack on OCUK:
https://forums.overclockers.co.uk/posts/30535235

 

[unseenmorbidity]

Looks like I missed the little detail about the Crosshair VI Hero being AM3 mount compatible. I suppose that means if I get lazy or my custom loop water cooling parts get delayed in the postal system I can always slap my Swiftech H220 on there for preliminary evaluation.

Looks like a questionable design to me...

 

[tnt118]

Is there any way to gauge what the potential of Gigabyte's AM4 motherboards might provide in terms of VRM? I'm not finding any power specifications on their website.

 

[psolord]

Looks like a questionable design to me...

Indeed, two holes so close together, could be weakening the pcb.

Everyone knows that anything that packs two holes so close together, is not to be trusted anyway.

 

[JimmiG]

You're starting at 95W TDP. Once you add frequency (~linear increase) and voltage (exponential increase), you could easily go past 200W with a hefty all-core OC. Meaning VRMs need to be beefy (as they appear to be on the Crosshair VI Hero).

It's mostly marketing, IMO. The fancy heatsinks, the LEDs and the neon glow appeal to the "gamer" crowd, and motherboard vendors are happy to sell the idea that you *need* a $300+ mobo that looks like a space ship in order to reach a decent OC. Usually, the CPU itself or the CPU cooling will become a limitation first, and the actual efficiency of those heatsinks is debatable (they look fancy but don't provide a lot of surface area due to their flattened shapes, and the TIM underneath might not be the best).

You'll be fine with a mid-range board unless you're trying to do something extreme like OC on LN2...

Not saying you should skip VRM heatsinks altogether, but if someone made a proper high-end board that didn't look like the computer equivalent of a ricer, I'd buy it.

 

[IEC]

From 8 Pack on OCUK:

3K+ dimms working fine on Asus Cross hero.

Source: https://forums.overclockers.co.uk/posts/30535497

 

[The Stilt]

Indeed, two holes so close together, could be weakening the pcb.

Everyone knows that anything that packs two holes so close together, is not to be trusted anyway.

Weakening the PCB?
6-layer PCBs are touch as hell and besides there are no forces subjected to the mounting holes directly. The back plates exist for a reason.

 

[IEC]

Looks like a questionable design to me...

Properly designed back plates are the load-bearing portion of the mount...

Would not make any difference for AIO or water block mounting even if it were to bear some of the load.

 

[Rifter]

Looks like there will be very limited overclocking on these, maybe 4.2Ghz on air max so far from the leakes we have seen. On the flip side they seem to perform great out of the box.

 

[PG]

Looks like there will be very limited overclocking on these, maybe 4.2Ghz on air max so far from the leakes we have seen. On the flip side they seem to perform great out of the box.

I think it depends on how you look at it. One guy said that they got a non-x 1700 to around 4 or 4.1. Base clock on that is 3.0, so that's pretty good if you look at it that way. It's a 1 Ghz OC.



Another general thought:
Somebody mentioned layers when talking about the mounting holes, and it got me thinking about how some cheap boards I've seen in the past seemed thin. They must have had less layers or something. Are all these AM4 boards going to have the same number of layers? Will they all be 6 like The Stilt mentioned or would some cheap ones have just 4 or something? I would like to avoid the cheap and thin boards if possible.

 

[The Stilt]

ASRock PCBs are generally so thin that they almost flap in the breeze.

Generally only high-end boards use >= 6-layer PCBs.

 

[imported_jjj]

ASRock PCBs are generally so thin that they almost flap in the breeze.

Generally only high-end boards use >= 6-layer PCBs.

You might be underestimating them, not 100% sure but i think they use 8-layer 2oz in high end.
They don't list layers but all their X370 boards listed so far use 2oz copper. http://www.asrock.com/microsite/AMDAM4/

 

[scannall]

Is there any way to gauge what the potential of Gigabyte's AM4 motherboards might provide in terms of VRM? I'm not finding any power specifications on their website.

This says a 6+4 VRM. http://www.gigabyte.us/Motherboard/GA-AX370-Gaming-K7-rev-10#kf I've always had great luck with Gigabyte boards in the past.

 

[Absolute0]

I think it depends on how you look at it. One guy said that they got a non-x 1700 to around 4 or 4.1. Base clock on that is 3.0, so that's pretty good if you look at it that way. It's a 1 Ghz OC.
.

Yeah there's some good value buys for overclockers to exploit. However, it seems there's little headroom beyond what the tops models are clocked at. That ~4.1 result was on high end mobo and watercooling. The hype-driven segment of the community (probably just a few loud people but we've all seen these posts) which has been speculating ~4.5 Ghz on aircooling is going to be disappointed.

Maybe why AMD has been packaging 500g beefcake coolers here, these chips have been pushed to beat everything in Intel's lineup.

To stay in perspective however, Ryzen doesn't need to hit 4.5 Ghz to impress. 8 cores at 4 Ghz is pretty fantastic to be fair .

**Possible that some later steppings improve OC-ability . Being AMD64/Opteron OCer from back in the day I have observed certain steppings/batches to OC 500 Mhz better than others...

 

[Rifter]

Yeah there's some good value buys for overclockers to exploit. However, it seems there's little headroom beyond what the tops models are clocked at. That ~4.1 result was on high end mobo and watercooling. The hype-driven segment of the community (probably just a few loud people but we've all seen these posts) which has been speculating ~4.5 Ghz on aircooling is going to be disappointed.

Maybe why AMD has been packaging 500g beefcake coolers here, these chips have been pushed to beat everything in Intel's lineup.

To stay in perspective however, Ryzen doesn't need to hit 4.5 Ghz to impress. 8 cores at 4 Ghz is pretty fantastic to be fair .

**Possible that some later steppings improve OC-ability . Being AMD64/Opteron OCer from back in the day I have observed certain steppings/batches to OC 500 Mhz better than others...

Sure 4Ghz is great im just not sure i can pull the pin on something that is not clocking faster than my 7 yerr old nehalem, even if its got twice the cores and is faster to start off with IPC wise.

Im hoping the 6 cores can push farther, or that the next stepping adds some headroom. Time will tell im not in a huge hurry but want to upgrade by summer.

 

[unseenmorbidity]

Sure 4Ghz is great im just not sure i can pull the pin on something that is not clocking faster than my 7 yerr old nehalem, even if its got twice the cores and is faster to start off with IPC wise.

Im hoping the 6 cores can push farther, or that the next stepping adds some headroom. Time will tell im not in a huge hurry but want to upgrade by summer.

By that logic, you should use a single core CPU. I think Ryzen will hit 6 Ghz on a single core!!

 

[CHADBOGA]

Everyone knows that anything that packs two holes so close together, is not to be trusted anyway.

 

[Absolute0]

Sure 4Ghz is great im just not sure i can pull the pin on something that is not clocking faster than my 7 yerr old nehalem, even if its got twice the cores and is faster to start off with IPC wise.

Im hoping the 6 cores can push farther, or that the next stepping adds some headroom. Time will tell im not in a huge hurry but want to upgrade by summer.

That sounds like a psychological requirement if you want > 4.2 Ghz. Looks like you've extracted good value from the i7 930 as it is but the IPC of new chips whip Nehalem out of the park. Anyway, fewer cores are only likely to clock higher because the thermals will be easier to manage.

 

[scannall]

Being able to talk about an 8 core, 16 thread CPU going over 4 Ghz at all is a great thing. 4.5 seems pretty unlikely, but my speculations have been wrong before. 4.2? Yeah, no problem.

 

[FIVR]

The 1700 is a 65W SKU binned for low power and 3Ghz clocks and still hits 4.1Ghz.That means the 1800X at 95W will probably hit 4.5Ghz easily. Keep concern trolling about worse than nehalem Absolute0, you're really making a good reputation for yourself here with that attitude.

 

[IEC]

The 1700 is a 65W SKU binned for low power and 3Ghz clocks and still hits 4.1Ghz.That means the 1800X at 95W will probably hit 4.5Ghz easily. Keep concern trolling about worse than nehalem Absolute0, you're really making a good reputation for yourself here with that attitude.

I think 4.5GHz all-core turbo is quite optimistic under air. But we will see soon enough.

 

[lolfail9001]

The 1700 is a 65W SKU binned for low power and 3Ghz clocks and still hits 4.1Ghz.

There is no evidence 1700, 1700X or 1800X are actually binned. 1700X/1800X just lift the base clock and TDP up, that's it.

That means the 1800X at 95W will probably hit 4.5Ghz easily.

Just make sure to cool your VRMs with a chiller.

Keep concern trolling about worse than nehalem Absolute0, you're really making a good reputation for yourself here with that attitude.

Oh, the irony.

 

[FIVR]

I think 4.5GHz all-core turbo is quite optimistic under air. But we will see soon enough.

Optimistic because?



1700 has 3.0Ghz Base clock and we know for a fact it cocks to 4.1Ghz
1800X has 3.6Ghz Base clock and somehow 4.5Ghz is supposed to be impossible?


How about 4.5Ghz likely, and golden samples will likely reach 4.8Ghz all core on air.

 

[WhoBeDaPlaya]

^
https://www.reddit.com/r/Amd/commen...y_in_a_top_tier_mobo/?st=izj793mg&sh=cef8c641

 

[FIVR]

lolfail9001 how much experience do you have on this topic... your post is non-sensical. Of course they are binned. They have 600mhz difference in base clock and 30W difference in TDP. You must not have much knowledge about this topic?