AMD Ryzen 7 1800X hits 5.2 GHz, breaks benchmarking record

[Misanthrope]

3 minutes ago, Bouzoo said:

That is admirable on its own, LN2 doesn't play a role here. I'm asking specifically what is impressive about this OC?

Well honestly I don't find LN2 OC impressive I think I should probably pass on further comments.

[DocSwag]

It's a little low but not horrible considering it's an 8 core. If I recall when Intel did an ln2 demo for the 4790k they only got to 5.5 ghz yet the 4790k still could reach 4.8 ghz or so on water.

[MageTank]

6 hours ago, TrigrH said:

how is the bios going to "huge"ly change how far a CPU can OC?

For as much as you keep bringing up the 7700k, one would think you'd already have the answer. Both MSI and Gigabyte had bad Z270 BIOS's that prevented 5ghz on launch. After the bios update, people were getting 5.1/5.2 on the exact same boards. 

 

We don't know what was limiting this chip. Not enough samples to know if the bin of the chip stopped them, or if even the power delivery of the board was robust enough to handle it. That being said, LN2 ceilings tells us nothing about real world voltage scaling. This doesn't mean 4.5ghz on water is out of the question, as we don't know how the voltage scaling works. Others are correct in saying it's not linear. We will know in a couple weeks once the vast majority of overclockers get their hands on it. After that, we will know in a few months whether or not their overclocks are actually safe/stable for longterm.

[Bananasplit_00]

1 hour ago, That Norwegian Guy said:

These days, there's almost never a case where water cooling will reach higher clocks than a top of the line air cooler. Because the dielectric barrier is hit way earlier than the thermal barrier. Since Ryzen promises to be very efficient just like recent Intel chips, there's no reason for this to not continue to be the case here.

maybe, i dont know much about high end overclocking exept that its often quite limited, the water estimate may be a bit high but i would expect you to get another 200MHz or so from going to water from air

[That Norwegian Guy]

30 minutes ago, Bananasplit_00 said:

maybe, i dont know much about high end overclocking exept that its often quite limited, the water estimate may be a bit high but i would expect you to get another 200MHz or so from going to water from air

No the difference will almost always be 0 mhz, but you'll get around 4°C-7°C temperature difference. And that temp difference is not enough to stabilize beyond unsafe voltages to get more clock speed. You don't even need the 'best' air coolers to run the maximum safe voltage limit of any given chip.

 

For example a standard 6-pipe tower cooler with a 120mm fan is enough to keep temperature below max on a 3570K running at its max usable voltage (1.35v)

If you're using overkill cooling, you're not really improving anything. There's no extra benefit from running 7 degrees cooler if you were already below the max threshold. It doesn't extend the life time of your chip or motherboard or anything of that nature.

 

Hell, I've been running a Core2Duo E8400 for almost 10 years overclocked to 4Ghz (1Ghz OC) on a stock cooler, just 4 degrees (load) to 16 degrees (idle) below the thermal limit of the chip and that system is still pristine to this day despite hundreds of thousands of power-on hours.

[Curufinwe_wins]

48 minutes ago, That Norwegian Guy said:

No the difference will almost always be 0 mhz, but you'll get around 4°C-7°C temperature difference. And that temp difference is not enough to stabilize beyond unsafe voltages to get more clock speed. You don't even need the 'best' air coolers to run the maximum safe voltage limit of any given chip.

 

For example a standard 6-pipe tower cooler with a 120mm fan is enough to keep temperature below max on a 3570K running at its max usable voltage (1.35v)

If you're using overkill cooling, you're not really improving anything. There's no extra benefit from running 7 degrees cooler if you were already below the max threshold. It doesn't extend the life time of your chip or motherboard or anything of that nature.

 

Hell, I've been running a Core2Duo E8400 for almost 10 years overclocked to 4Ghz (1Ghz OC) on a stock cooler, just 4 degrees (load) to 16 degrees (idle) below the thermal limit of the chip and that system is still pristine to this day despite hundreds of thousands of power-on hours.

Sorry to burst your bubble here, but that actually isn't true.

 

Electron leakage scales exponentially with temperature (via the Maxwellian). 

 

You literally get more stability by being colder with every single modern chip. The smaller transistors become (the more stability is QM leakage dominated), the more decreasing temperatures helps keep the system stable.

 

Your c2d is running what a 45 process? Electron leakage from that would be about 4 orders of magnitude lower than it is at 14nm (and yes... orders of magnitude. When you want to do the square well potential problem in QM come back to me and we can talk about it.)

 

 

Which is how you get something like this: http://hwbot.org/submission/3324224_a39a_3dmark___fire_strike_geforce_gtx_1080_25080_marks

 

[Doobeedoo]

Well info about OC is still pretty vague anyway. I'm mainly interested to see how AiO will handle OCs after release. And how well will like lowest core Ryzen CPU OC too.

Also Zen is on LPE and not LPP 14nm no? 

[That Norwegian Guy]

9 minutes ago, Curufinwe_wins said:

Sorry to burst your bubble here

 

You're not bursting any bubbles, electron leakage is insignificant to the life time of CPUs because cache and other package components will die off long before the silicon itself. Of more importance is dielectric friction, which is constant with voltage and is not affected by temperature whatsoever.

[cangath]

7 hours ago, Pohernori said:

 

Woah woah woah you want 5ghz on water 24/7 stable on all 8 cores? 

 

We're quite far from that still. You don't even see 5ghz 6900ks very often. 

Well saying quite far is a little off. Clock speeds haven't changed that much in the past years because manufacturers are more concerned with tdp. 

[CTR640]

I need an army of slaves who can keep cooling down the CPU to run it 24/7 on 6+GHz

[Deletist_Jerk]

Why are people bringing up the 7700K? It doesn't matter how much that CPU can overclock. It's way behind 1800X in IPC. 

[Curufinwe_wins]

29 minutes ago, That Norwegian Guy said:

You're not bursting any bubbles, electron leakage is insignificant to the life time of CPUs because cache and other package components will die off long before the silicon itself. Of more importance is dielectric friction, which is constant with voltage and is not affected by temperature whatsoever.

Wow, you really don't understand electron leakage...

 

It isn't something that reduces lifespan, it's something that reduces frequency and stability  (electron leaks out of the transistor and thus fails to properly run on that one cycle). It is the same idea as solar radiation causing bit rot in ram. It doesn't prevent ram from working indefinately, it merely causes one set of data to be corrupted.

 

Electron leakage is the single DOMINATE factor in transistor scaling right now. In fact, it was the reason finfet was invented. 

 

 

Edit: with a sufficiently high electron leakage current, you can indeed permanently damage a transistor, but instability occurs far far before that.

[vanished]

5 hours ago, cj09beira said:

When you lower the temperature of the cpu that much you need to raise voltage just to conpensate for the higher resistance. Go check graphs of silicon resistence in relation to temperature.

Actually I'm pretty sure lowering temperature lowers the resistance, and a quick google search seems to confirm this...

[cj09beira]

9 minutes ago, Ryan_Vickers said:

Actually I'm pretty sure lowering temperature lowers the resistance, and a quick google search seems to confirm this...

its the opposite for semiconductors

[ZackBarletto]

5 hours ago, Ramaddil said:

 

I really agree with this post.  I never bought a CPU simply for gaming.  I think everyone has come to a place where the only reason they buy a CPU is for a single function when in fact they can do more than one thing.  Yes Ryzen probably will not have the best single core performance and might not get the best FPS in every game, but seriously there comes a point of diminishing returns when it comes to gaming and FPS anyway.  I have always bought AMD, not because I am a Fanboy, but because they were reasonably priced and could run the games I played, and I was able to use the CPU for other things.  Granted AMD has been far behind the curve in the CPU market, I am excited about Ryzen because it gives me a opportunity to upgrade my home server PC which does more than just streaming and need more cores and threads.  

 

I think everyone needs to look at the whole picture instead of just focusing on one thing (gaming).  I also agree that as time goes on and given that the limits have almost been reached regarding single core performance gains, the future is multi core and we should embrace it.

Thank You! a part of me has felt this was for a long time. but i just called myself a fanboy and accepted 2nd place every time. lol

 

But i have always tooted the value of AMD. Its always been riiiight there, but always much cheaper.

 

And yes, this Rysen launch has me excited to upgrade my pc and continue to use AMD for years to come.

[vanished]

6 minutes ago, cj09beira said:

its the opposite for semiconductors

can you link the charts you're looking at then?  

[cj09beira]

55 minutes ago, Ryan_Vickers said:

can you link the charts you're looking at then?  

https://images.duckduckgo.com/iu/?u=http%3A%2F%2Ffile.scirp.org%2FHtml%2F4-7500658%2F5439728b-f8a6-4be1-9f94-19b823c0a243.jpg&f=1

this is conductivity so resistance  is the opposite 

[Humbug]

14 hours ago, Deletist_Jerk said:

Why are people bringing up the 7700K? It doesn't matter how much that CPU can overclock. It's way behind 1800X in IPC. 

no it's not. It beats ryzen in both ipc and clock speed.

it will lose to ryzen in parallel workloads as Ryzen has more cores and superior SMT implementation.

[vanished]

10 minutes ago, cj09beira said:

https://images.duckduckgo.com/iu/?u=http%3A%2F%2Ffile.scirp.org%2FHtml%2F4-7500658%2F5439728b-f8a6-4be1-9f94-19b823c0a243.jpg&f=1

this is conductivity so resistance  is the opposite 

oh, that's weird...

[done12many2]

2 hours ago, Deletist_Jerk said:

Why are people bringing up the 7700K? It doesn't matter how much that CPU can overclock. It's way behind 1800X in IPC. 

 

It falls somewhere between Haswell and Broadwell, but it definitely isn't betting Kaby Lake in IPC.  

[Curufinwe_wins]

1 hour ago, Ryan_Vickers said:

oh, that's weird...

Not that I'd expect random people to know this, but it's actually fairly intuitive...

 

Until you reach the band gap, semi-conductors function as insulators...

 

If you can think about it in this way (it's more than a bit of an oversimplification, but still), increasing the temperature of a semi-conductor increases the electron mobility and thus makes the band gap easier to reach, thus lowering resistance.

 

But THE important issue with regards to lowering temperature is lowering leakage/tunneling through transistors. It singlehandedly is the main reason why LN2 out preforms DICE even though both of them can in theory handle the exact same amount of heat input.

[vanished]

2 minutes ago, Curufinwe_wins said:

Not that I'd expect random people to know this, but it's actually fairly intuitive...

 

Until you reach the band gap, semi-conductors function as insulators...

 

If you can think about it in this way (it's more than a bit of an oversimplification, but still), increasing the temperature of a semi-conductor increases the electron mobility and thus makes the band gap easier to reach, thus lowering resistance.

 

But THE important issue with regards to lowering temperature is lowering leakage/tunneling through transistors. It singlehandedly is the main reason why LN2 out preforms DICE even though both of them can in theory handle the exact same amount of heat input.

Interesting... It does make more sense when you put it like that

[AnonymousGuy]

So basically it's a dog.  6950X, 25% more cores, 5.325Ghz in Cinebench on LN2.

 

Paired with their shitty 100mhz turbo boost whatever they're calling it, Ryzen ain't going to be overclocking for shit.

[vanished]

9 minutes ago, AnonymousGuy said:

So basically it's a dog.  6950X, 25% more cores, 5.325Ghz in Cinebench on LN2.

 

Paired with their shitty 100mhz turbo boost whatever they're calling it, Ryzen ain't going to be overclocking for shit.

The people willing to spend whatever it costs on a 6950X and the necessary cooling to get it to that speed are a completely different segment (and a very very small one I might add) to those interested in RyZen

[done12many2]

12 minutes ago, AnonymousGuy said:

So basically it's a dog.  6950X, 25% more cores, 5.325Ghz in Cinebench on LN2.

 

Paired with their shitty 100mhz turbo boost whatever they're calling it, Ryzen ain't going to be overclocking for shit.

 

 

I've got a Ryzen 7 1800x reserved for pickup on the 2nd at my local Micro Center.  I plan to see this first hand, but suspect that single-thread and overclocking performance are going to be underwhelming.