Navi/Ryzen 3000 launch Megathread

[S w a t s o n]

5 minutes ago, Drak3 said:

That’s an entire generation where you’re wrong.

 

And Ryzen 1 was still not an “almost always.”

Irrelevant to Swatson’s claim. For older Ryzen chips, it’s outright wrong.

You mean an entire refresh where it was possible but not always. And I said almost always. And when you couldnt it was like 100MHz 1 core vs all core. Wow you got me???
 

Ryzen 1 was like 50/50 and was definitely capped much harder than Zen2 here. Zen2 has the advantage. To look at Zen1/+ and say "oh but look your all core is 100mhz lower than single core XFR" on a chip that only does 4GHZ in the first place isnt really shutting down my argument

[williamcll]

Any news on the 3800X? It might just have enough clock speed to beat the 9900k

[cj09beira]

Just now, williamcll said:

Any news on the 3800X? It might just have enough clock speed to beat the 9900k

you will have to wait a little for reviews on that one, as amd didn't send those to reviewers, so they will have to buy their own, maybe in a day we will know how well that one does

[leadeater]

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

Uh what? You can almost always hit boost on every core when OCing. As you said Zen could, intel can.. who else are we comparing to? Also you mentioned that zen was extremely frequency capped but these only hitting 4.2-4.3GHZ is the same effective cap as Zen1 even though we know it's not actually a frequeuncy wall. In my eyes that supports my opinion that the all-core XFR clock is reachable once constraints are removed.

XFR was barely above standard boost, Zen2 boost and all core boost gap is much wider. What is possible on one may not be on another, because again Zen/Zen+ is actually useless to look at in this regard but it was entirely frequency capped. There wasn't any difference between stock (X variants), all core boosts and all core OCs that are at all worth talking about which is why you'd never by an X variant in the first place and just use all core OC to hit the hard limit at ambient. You can OC Zen/Zen+ much higher but only by lowering temperatures, because that is a thing as I explained.

 

As for Intel this is true for them as well, not all 9900K's can hit 5.0GHz all core even though single core stock spec is 5.0GHz. Can do, will likely do are all probabilistic terminology but what makes you think Zen2 doesn't hit a frequency wall just as hard as Zen/Zen+ did? Also Silicon Lottery has only 35% of 9900K's being able to 5.0GHz all core, though I think that was higher before the new SKU introduction.

 

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

So we can agree that yes either the socket power limit is the constraint, or maybe heat desnity of the small AM$ socket? So you agree with me now?

Not for the socket or VRM, none of these are electrically or thermally the limit or a reason for 140W package limit. The chips being largely at the upper limit stock however is a likely reason for it.

 

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

Current clamps are not 100% accurate, and have variance just in literally how you position it over the wire. 170W into the VRM at say 90% efficiency means the package is at 153W or basically the 140W limit I speak of is basically true.

Current clamps are highly accurate, if you don't have a garbage one and a terrible scope. We'd never be able to make functional products if they weren't.

[Drak3]

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

To look at Zen1/+ and say "oh but look your all core is 100mhz lower than all core on a chip that only does 4GHZ in the first place isnt really shutting down my argument

It does when your argument is that one almost always meets or exceeds single core. And 100MHz better single core in the 4GHz and below range is worth considering over multicore if a workload benefits more from single core.

 

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

You mean an entire refresh where it was possible but not always

False. Zen+ is second generation Ryzen, and while it is technically possible on a microscopic selection and at voltages that kill chips exponentially faster, it’s not pragmatically possible.

 

 

You’ve made a false claim. I pray to god no one ever takes any advice you give on processors.

[S w a t s o n]

1 minute ago, leadeater said:

XFR was barely above standard boost, Zen2 boost and all core boost gap is much wider. What is possible on one may not be on another, because again Zen/Zen+ is actually useless to look at in this regard but it was entirely frequency capped. There wasn't any difference between stock (X variants), all core boosts and all core OCs that are at all worth talking about which is why you'd never by an X variant in the first place and just use all core OC to hit the hard limit at ambient. You can OC Zen/Zen+ much higher but only by lowering temperatures, because that is a thing as I explained.

I'm just using XFR to denote turbo/boost, my point was that ALMOST (read that again ALMOST) always you can reach turbo/xfr/whatever you want to call it, on every core. This is widely true and was like not ALWAYS true for zen due to the process node and them being clocked to the wall. We can agree on this.

Quote

s for Intel this is true for them as well, not all 9900K's can hit 5.0GHz all core even though single core stock spec is 5.0GHz. Can do, will do, will likely do are all probabilistic terminology but what makes you think Zen2 doesn't hit a frequency all just as hard as Zen/Zen+ did? Also Silicon Lottery has only 35% of 9900K's being able to 5.0GHz all core, though I think that was higher before the new SKU introduction.

 

While it's true that not every 9900k can do it, I am not claiming the silicon lottery does not exist. Just that you can reasonably assume an all-core matching single core is possible. Clearly we arent seeing a silicon lottery problem here.
 

Quote

Not for the socket or VRM, none of these are electrically or thermally the limit or a reason for 140W package limit. The chips being largely at the upper limit stock however is a likely reason for it.

 

Current clamps are highly accurate, if you don't have a garbage one and a terrible scope. We'd never be able to make functional products if they weren't.

 

Uh you yourself just agreed the socket power limit is the problem. I'm not claiming the fucking 14 phase VRM's cant do it. We both agree that the socket power limit set by AMD is the problem. As for the current clamp, like I said it's human error that's possible. I'm not disputing the 170W because it seems highly possible given that efficiency losses bring it to the 140W limit we AGREE exists.

 

20 minutes ago, leadeater said:

Either you can't remove the package power limit, meaning 5GHz is impossible so that ends that debate, or no one removed the package power limit during OC for the reviews, or they were not allowed to for some reason for the reviews, or the current bios's do not allow it to be removed yet. If find it odd that if the option exists now not a single person did so, even if somehow not allowed under reviewer guides (why would AMD impose this?).

[Drak3]

Just now, S w a t s o n said:

my point was that ALMOST (read that again ALMOST) always you can reach turbo/xfr/whatever you want to call it, on every core

And that is wrong when dealing with Ryzen.

[S w a t s o n]

12 minutes ago, Drak3 said:

It does when your argument is that one almost always meets or exceeds single core. And 100MHz better single core in the 4GHz and below range is worth considering over multicore if a workload benefits more from single core.

 

False. Zen+ is second generation Ryzen, and while it is technically possible on a microscopic selection and at voltages that kill chips exponentially faster, it’s not pragmatically possible.

 

 

You’ve made a false claim. I pray to god no one ever takes any advice you give on processors.

Yikes you are really reaching here.

Zen+ is a refresh, not a fully fledged second gen architecture really, not even AMD tried that shit. And no the chips arent dying... what are you going on about. It's very pragamatically possible. Running any CPU with an overvolt carries risk but to reach all core turbo speeds on Zen+, I would not say the risk is any higher than a 9900k at 5GHz.

I pray to god you stop posting on this forum

[Lathlaer]

1 hour ago, Ravendarat said:

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

Only two games from Bitwit and LTT overlap (BV and SotTR) and those actually kinda align (in the sense that Intel is winning there).

 

I don't have the time to compare the rest though, so maybe.

[leadeater]

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

While it's true that not every 9900k can do it, I am not claiming the silicon lottery does not exist. Just that you can reasonably assume an all-core matching single core is possible. Clearly we arent seeing a silicon lottery problem here.

Well I don't consider 35% chance reasonable to assume though. Compound that with Intel known to have an architecture that overclocks very well and the best fab technology and optimization in the industry (10nm issues doesn't make this not true). A 50/50 chance is one thing, still not all that safe to assume, but I'm not expecting it to be that high given what we have seen with Zen in the past.

[S w a t s o n]

Just now, leadeater said:

Well I don't consider 35% chance reasonable to assume though. Compound that with Intel known to have an architecture that overclocks very well and the best fab technology and optimization in the industry (10nm issues doesn't make this not true). A 50/50 chance is one thing, still not all that safe to assume, but I'm not expect it to be that high given what we have seen with Zen in the past.

I am talking about the Zen2 lineup as a whole. Just like most people would say, yea sure 5ghz all core is possible on 9900k. We might see silicon lottery play into it once power limit is not longer the constraint. Just saying it should definitely be possible within reason.

[Drak3]

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

Zen+ is a refresh, not a fully fledged second gen architecture really

By that argument, We’re still on 6th gen Core i.

 

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

not even AMD tried that shit.

AMD called Ryzen 2xxx second generation. 3xxx is third generation.

 

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

And no the chips arent dying... what are you going on about.

The voltage required is only relatively safe to run one or two cores at most. Running all cores at the voltages XFR runs at takes years upon years off of the life of the chips subjected to that abuse.

 

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

It's very pragamatically possible. 

No, it’s not. Ryzen chips getting to XFR was, at best, iffy. Ryzen 2, you’re pumping unsafe voltages and even then, you’re not going to achieve it often.

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

Just like most people would say, yea sure 5ghz all core is possible on 9900k.

Probably because on the 9900K, a significant number if chips actually reach that.

[cj09beira]

3 minutes ago, Jack_of_all_Trades said:

Quite whelming, is there any review that focuses on power consumption while comparing amd and intel ?

the 3700x uses around the same as a 2600 and the 3900x the same as a 2700x

[cj09beira]

Just now, Jack_of_all_Trades said:

 

Under full load ? If so, quite impressive.

yes full load

[S w a t s o n]

5 minutes ago, Drak3 said:

By that argument, We’re still on 6th gen Core i.

 

AMD called Ryzen 2xxx second generation. 3xxx is third generation.

 

The voltage required is only relatively safe to run one or two cores at most. Running all cores at the voltages XFR runs at takes years upon years off of the life of the chips subjected to that abuse.

 

No, it’s not. Ryzen chips getting to XFR was, at best, iffy. Ryzen 2, you’re pumping unsafe voltages and even then, you’re not going to achieve it often.

Probably because on the 9900K, a significant number if chips actually reach that.

Literally everyone says kabylake - coffee lake are skylake baseline cores......

Thanks for pro fucking tip. Ryzen 2000 existing doesnt mean it's truly the second generation architecture. Literally check any materials or quotes from AMD.

This is how overclocking works, more voltage = faster degredation. This applies to every CPU to ever exist.

Yes it is, possible =/= every chip can do it.

9900k hits 5GHz all core at about 35%, pretty good. I'm expecting that Zen2 can do similar (all core of boost clock) but of course we dont know yet. This is a new COMPLETELY different node, and a new arch.

[Drak3]

Just now, S w a t s o n said:

Literally everyone says kabylake - coffee lake are skylake baseline cores......

Doesn’t change the fact that the 7700K is a 7th gen product, 8700K is 8th gen, 9900K 9th gen.

 

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

This is how overclocking works, more voltage = faster degredation. This applies to every CPU to ever exist.

There is a threshold where that degradation accelerates faster and faster. XFR operates past that threshold, and it can only do so safely because it only boosts one to two cores when necessary.

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

Yes it is, possible =/= every chip can do it.

For it to be pragmatically possible, it can’t rely on unsafe voltages and be a microscopic sliver of silicon, or rely on extreme cooling.

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

I'm expecting that Zen2 can do similar

That’s extremely stupid. Everything we’ve seen so far is that Zen 2 can’t do that.

[S w a t s o n]

9 minutes ago, Drak3 said:

Doesn’t change the fact that the 7700K is a 7th gen product, 8700K is 8th gen, 9900K 9th gen.

 

There is a threshold where that degradation accelerates faster and faster. XFR operates past that threshold, and it can only do so safely because it only boosts one to two cores when necessary.

For it to be pragmatically possible, it can’t rely on unsafe voltages and be a microscopic sliver of silicon, or rely on extreme cooling.

That’s extremely stupid. Everything we’ve seen so far is that Zen 2 can’t do that.

Yikes, so because intel refreshes skylake for 3 years and increases the number on the box that somehow means that it's actually different? I implore you to check with any sources that deal with semiconductors or actual architecture and not just marketing materials.

The voltages are safe and the cooling is not extreme so, seems pragmatic to me.

Everything you've seen of Zen2 is:

1. Pre-release bios
2. Locked to 140W socket power limit that all the 105W cpu's are stuck under and are hitting under all core overclocks or just PBO.
3. On a small socket (16 cores in the area of AM4 is very heat dense, 3 dies instead of 1 with unusual placement compared to any other desktop CPU for which coolers are designed.

So basically you haven't seen Zen2 even stretch it's legs. Limit a 9900k to 140W and try to reach 5GHz all core.

[leadeater]

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

Uh you yourself just agreed the socket power limit is the problem. I'm not claiming the fucking 14 phase VRM's cant do it. We both agree that the socket power limit set by AMD is the problem. As for the current clamp, like I said it's human error that's possible. I'm not disputing the 170W because it seems highly possible given that efficiency losses bring it to the 140W limit we AGREE exists.

Well if you mean package power limit that is set in the microcode yes, not socket power limit. Socket power limit implies no matter the SKU that is the power limit of the socket, there's a differentiation between socket and package. We know the socket and AM4+ platform families can do more than 140W. My expectation around that 140W value is that it was chosen as that was where the product was most stable and yielded the most viable products. If Zen2 was broadly capable of more than that then I would expect it to be higher, say 150W. But even at 150W that may not be high enough to give a meaningful clock increase for all cores if every Zen2 die could achieve it in the first place.

 

There is a lot of design thought and reasoning that go in to product specs and parameters, AMD isn't about to unnecessarily tie an anchor to itself just to look slightly better in the power draw graphs when everyone knows performance is king. People will, and have, closed their eyes to power draw when the performance is there.

 

Zen2 products are absolutely excellent as they are, we don't need to expect much more from them and to do so is only setting yourself up for disappointment. Hitting high clocks is just some arbitrary goalpost with little meaning behind it, an all core OC of 4.7GHz should yield some good performance gains and while even I consider that to be a though ask for most Zen2 products something much more achievable than 5GHz. Such a huge jump in clocks from one generation is unprecedented in the last 1 to nearly 2 decades and smaller nodes actually start to hurt clocks not improve them due to that resistance increase, a situation only encountered on very small nodes. I know that is a weird thing to think about since in the past node shrinks have meant clock increases but resistance wasn't an issue for those, we have officially tipped the scales but have found ways to balance them.

 

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

I'm not disputing the 170W because it seems highly possible given that efficiency losses bring it to the 140W limit we AGREE exists.

A typical motherboard VRM will be operating in the 90% efficiency range, below that is considered bad and that would happen when you are drawing too little current or way significantly higher currents than optimal. 82% is in the VRM near death from over current draw territory.

 

The package power limit likely isn't as hard limiting at it implies so the CPU actually drawing ~150W isn't unreasonable even if it's only short term or some kind of peak type thing.

[suits]

Did anyone have an OC'd 9900k in their benchmarks to at least 4.8 GHz? All of them I've seen have stock 9900k. I dont think anyone bought a 9900k and didn't OC it...

[Drak3]

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

Everything you've seen of Zen2 is:

And? The odds of the magical 5GHz you’ve been preaching as fact until this thread, is pathetically small.

 

We don’t see that mythical uplift in other products, Older Ryzen included. No reason to think that will change. 

 

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

I implore you to check with any sources that deal with semiconductors or actual architecture and just marketing materials.

Why? I’m well aware of the fact that there are basically no meaningful architectural changes. If you weren’t so egotistical, you’d know that I’ve said that The Lake lineup is little more than tweaked. Broadwell, which is die shrunk Haswell.

 

That is irrelevant to product generations.

[MMKing]

Edit:

Edited out erroneous information

 

@Drak3 I heard the 290x and the 390x was not the same product. You're arguing with @S w a t s o n over terminology. The 7700k and the 8700k is not identical, but they are made on the same process. In one way, they are different albeit similar, in another way they are identical.

[S w a t s o n]

8 minutes ago, leadeater said:

Well if you mean package power limit that is set in the microcode yes, not socket power limit. Socket power limit implies no matter the SKU that is the power limit of the socket, there's a differentiation between socket and package. We know the socket and AM4+ platform families can do more than 140W. My expectation around that 140W value is that it was chosen as that was where the product was most stable and yielded the most viable products. If Zen2 was broadly capable of more than that then I would expect it to be higher, say 150W. But even at 150W that may not be high enough to give a meaningful clock increase for all cores if every Zen2 die could achieve it in the first place.

 

There is a lot of design thought and reasoning that go in to product specs and parameters, AMD isn't about to unnecessarily tie an anchor to itself just to look slightly better in the power draw graphs when everyone knows performance is king. People will, and have, closed their eyes to power draw when the performance is there.

 

Zen2 products are absolutely excellent as they are, we don't need to expect much more from them and to do so is only setting yourself up for disappointment. Hitting high clocks is just some arbitrary goalpost with little meaning behind it, an all core OC of 4.7GHz should yield some good performance gains and while even I consider that to be a though ask for most Zen2 products something much more achievable than 5GHz. Such a huge jump in clocks from one generation is unprecedented in the last 1 to nearly 2 decades and smaller nodes actually start to hurt clocks not improve them due to that resistance increase, a situation only encountered on very small nodes. I know that is a weird thing to think about since in the past node shrinks have meant clock increases but resistance wasn't an issue for those, we have officially tipped the scales but have found ways to balance them.

 

A typical motherboard VRM will be operating in the 90% efficiency range, below that is considered bad and that would happen when you are drawing too little current or way significantly higher currents than optimal. 82% is in the VRM near death from over current draw territory.

 

The package power limit likely isn't as hard limiting at it implies so the CPU actually drawing ~150W isn't unreasonable even if it's only short term or some kind of peak type thing.

Ok cmon now you're getting into semantics. The limit that AMD artificially placed on the power the package can draw is what I was always referring to. I have never claimed it was electrically impossible.

As for 90% efficiency, I know...
 

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

 170W into the VRM at say 90% efficiency means the package is at 153W or basically the 140W limit I speak of is basically true.

 

[LukeSavenije]

I'm stopping to follow the thread

 

mention me when you got additional sources

[S w a t s o n]

1 minute ago, Drak3 said:

And? The odds of the magical 5GHz you’ve been preaching as fact until this thread, is pathetically small.

 

We don’t see that mythical uplift in other products, Older Ryzen included. No reason to think that will change. 

 

Why? I’m well aware of the fact that there are basically no meaningful architectural changes. If you weren’t so egotistical, you’d know that I’ve said that The Lake lineup is little more than tweaked. Broadwell, which is die shrunk Haswell.

 

That is irrelevant to product generations.

If I wasnt egotistical (nice, thanks), I would be psychic? Cool story bro. And yet you tried to claim that we are really past skylake when we arent. Wow the box says 9th gen. WHO CARES?

[JoeCoke]

I wonder what good ol' Zmeul is up to right about now...