• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD Zen 2 (Ryzen 3000) - *** NO COMPETITOR HINTING ***

Those leaks were from Adored and he doubled down on them in the past couple of weeks, stating that the 16 core was going to be, I forget what, something like 4.3Ghz boost and he said straight up the 12 core would be 5Ghz. He's now since getting basically everything to do with Navi and this launch completely wrong, taken to twitter to defend himself and is now saying it was 5Ghz overclocks the 12 core will do. That Navi isn't the architecture he thought, the massive price differences, the lack of a 16 core... opps.

I don't think he was that far off and think there are still some surprises to be had. At the same time I think you would agree that all they have to do is win. They don't need to show their very best when there isn't any competition for it. I still think that we will eventually see 5ghz Zen2 chips, don't you?
 
From your link

MSI Z390 Godlike (ET Auto) Turbo Duration vs. ASUS Maximus XI Hero (MCE Off)

YnOqVKa.png


So the Asus board has MCE disabled and TDP limit enabled, yes there is a brief 10 second boost before it climbs down to a 95 watt limit, at about 2.4Ghz under AVX load. the MSI board is set to let it boost over spec and its running at 4.7Ghz near 200 Watts, the Asus at 4.2Ghz has the TDP limit taken off and its running 4.2Ghz at 160 Watts.
The graph you need to look at is the one immediately preceding that (IIRC). That'll be the one that demonstrates what Intel spec behaviour is supposed to look like. That it is only possible with the Asus Maximus Hero XI is a whole different matter.

Your assertation that base clocks operate at TDP limit defies the laws of physics; there can be no thermal headroom for boost to operate if it operates at its thermal limit whilst at base clocks. You don't have to believe me or Gamers Nexus, just go look at how Intel details how it's turbo boost works.
In the real world though, 9900K owners don't pair it with a 95w rated cooling solution. Mobo makers know this, hence why default to ignoring Intel spec; they know that the 9900K can perform significantly better if allowed to default out of spec. None of that detracts from the fact that at spec the 9900K does meet the 95w TDP limit.
If you want to argue that Intel's spec is not truly representative, then you should lobby Intel to produce a spec that is more representative of actual performance.
 
Those leaks are about 7 months old, which makes the information about 8 months old. A lot can change in that time, and a lot did change in that time.

Whether or not real-world silicon actually lived up to projections remains to be seen, but certainly I don't think AMD expected the relentless barrage of vulnerabilities and exploits hitting Intel CPUs. I don't think AMD expected Intel 10nm to still be a pipe dream. And I certainly don't think AMD expected Intel's responses to Ryzen and Threadripper to be a gluing MOAR COREZ onto Coffee Lake, a cherry-picked 5GHz glued Coffee Lake, and a joke Xeon with the clock multiplier removed.

And because of all this, AMD have clearly stepped back and gone "Intel have nothing right now, we actually don't need to release all this right now". The 3800X is likely going to compete with the 9900KS in every metric that matters to non-children and fanbois, the 3900X is going to salt the earth afterwards and a low-clocked 16 core desktop Ryzen is beating an Intel HEDT.

16 cores exist. 5GHz boost exists. AMD don't need to release them now. It pains me to say that because it shows Intel are so much on the back foot now AMD can already start drip-feeding and controlling technology release. I didn't think they'd be able to pull that business trick and start milking until Zen 3.

Agreed, IMHO AMD are / were under a lot of pressure. Big changes to the design and on a new node. A lot of pressure to get that out of the door. I think their best case scenario was 200 / 300 Mhz. higher but given the situation (and the performance increases) Lisa said ok, let's ship it. There could be quite a bit more to come from these chips. Anyone for the 3650x, 3850x etc in about 6 months?
 
I don't think he was that far off and think there are still some surprises to be had. At the same time I think you would agree that all they have to do is win. They don't need to show their very best when there isn't any competition for it. I still think that we will eventually see 5ghz Zen2 chips, don't you?

I think we will see close to 5ghz with the Zen2 refresh if it happens, if not with Zen3.

As for AdoredTV, i actually liked his videos, but after reading how he responded to criticism on twitter i have zero sympathy for him. He become extremely hostile and childish, swearing at people and sneering at other tech accounts follower amounts etc, extremely childish behaviour. If he had held his hands up and said it is what it is, i was wrong for whatever many reasons, i think he would have garnered a ton of respect from the community. As it stands now people who were adamant he was correct and now laughing at him, people that were likely to forgive him for this are now calling him out, because of the way he reacted. Its a shame, but if your going to put your neck on the block, you need to have the balls to suck it up and apologise if wrong, not go on the hostile attack to anyone who criticises you.
 
It was my thinking too.
All-core overclock is more difficult. I would rather increase BCLK (and offset voltage) and benefit from XFR, just bumped higher.


What does it make 3900X?
With default clocks being higher than 3600X and default power draw not doubling with number of cores, surely it must get best binned 6 core dies?

Thes speculations are nice, but I am waiting for proper reviews. Very interested in web browser benchmarks from anandtech, they are the best single-core/gaming/latency indicator.

Also waiting for mATX motherboard. Any news on B550? Else what is the best B450 mATX? MSI mortar?

One of the difficulties is the 14nm I/O chip, with 16 cores pushing hard you're really pushing the I/O chip. Could be simply that the I/O chip was using a hell of a lot of power and could be the limitation in stable clock speeds, it could also be that it wasn't really overclocked, and it was just a 16 core on water because they wanted to show what it could do stock, quiet, etc.


I don't think he was that far off and think there are still some surprises to be had. At the same time I think you would agree that all they have to do is win. They don't need to show their very best when there isn't any competition for it. I still think that we will eventually see 5ghz Zen2 chips, don't you?

He wasn't far off except naming, clock speeds, what cores would be available, time they would be launched, etc. His initial leaks all said Navi and Zen being talked about at CES, then he ran a story about how Navi got delayed to defend him saying it would launch though he still got loads of information wrong about it being straight GCN. Then he doubled down on 12 core 5Ghz and 16 core launching at Computex and was wrong again.

He changed his mind shortly before CES and stated that all three sources told him that Ryzen wouldn't be chiplets even though if you read the actual messages from the sources none of the three of them actually said that. He has videos contradicting himself and every time it's wrong he blames the sources even though regardless of if they are true you can see from his own reading of what they told him he simply drew conclusions entirely unsupported by what he was told (if the sources were real or not).

The only thing he was 'right' about was Rome being chiplets, when AMD has been releasing papers on chiplet designs for the past 7-8 years and with EPYC 1 being an half way step and with 7nm known to be a breaking point when it comes to I/O scaling. People have been talking about the industry being going to chiplets, interposers and on package memory over the last 7-8 years a lot, it started with Fury X and AMD had samples of an APU with HBM as far back as 2011. In other words a full chiplet style chip was always coming, when wasn't as clear but AMD already made a huge step with EPYC being made of 4 dies.

Before the Rome call he got Polaris completely wrong, etc.
 
Those leaks were from Adored and he doubled down on them in the past couple of weeks, stating that the 16 core was going to be, I forget what, something like 4.3Ghz boost and he said straight up the 12 core would be 5Ghz. He's now since getting basically everything to do with Navi and this launch completely wrong, taken to twitter to defend himself and is now saying it was 5Ghz overclocks the 12 core will do. That Navi isn't the architecture he thought, the massive price differences, the lack of a 16 core... opps.
All of the SKUs announced were in the original leak. The naming matched the pricing, though it didn't match the listed SKUs. That indicates they moved the stack up a price point, likely because of the increased awareness they were getting towards their products.
I wouldn't write off Adoredtv just yet. The only thing missing really is the 5GHz clock SKUs.
 
If we're going to have a TDP discussion then the right comparisons to make are: a) under a sustained load, what is the 24/7 frequency that the CPU can maintain within it's TDP limit? b) at that 24/7 frequency, what is it's performance level?
For me, the 9900K marginally beats the 2700X here, and the 3800X will comfortably beat both.
Whether or not the 3800X can beat the 9900K in absolute performance, with neither CPU being held back, remains to be seen. I think it will, but not necessarily by a comfortable amount.

Now, the 9900KS. Intel is talking that up as an all core 5GHz all of the time CPU. With that in mind, it'd be wrong for them to claim a TDP that didn't reflect it's actual power profile; 5GHz on all 8 cores at all times under a sustained load.
 
All of the SKUs announced were in the original leak. The naming matched the pricing, though it didn't match the listed SKUs. That indicates they moved the stack up a price point, likely because of the increased awareness they were getting towards their products.
I wouldn't write off Adoredtv just yet. The only thing missing really is the 5GHz clock SKUs.

The names were there...

you mean with a 1800x, and a 2700x, there was a 3700x... omg, how did he ever come up with those names, it seems almost impossible for someone to add a 1 to the start of the previous numbers.

I mean, lets take this, what do you think the 10 core Intel chip will be called, a ZX64, a P9980, or you know, an i11 1190k?

The only thing missing were the actual models, the actual clock speeds, the date they'd announce, the date they'd launch... everything that a three year old couldn't guess the day the 1800x name came out?

RX5700 is less predictable because they've moved to a new architecture and a new naming scheme. Zen 3 following Zen 2/1 naming was a given.
 
The names were there...

you mean with a 1800x, and a 2700x, there was a 3700x... omg, how did he ever come up with those names, it seems almost impossible for someone to add a 1 to the start of the previous numbers.

I mean, lets take this, what do you think the 10 core Intel chip will be called, a ZX64, a P9980, or you know, an i11 1190k?

The only thing missing were the actual models, the actual clock speeds, the date they'd announce, the date they'd launch... everything that a three year old couldn't guess the day the 1800x name came out?

RX5700 is less predictable because they've moved to a new architecture and a new naming scheme. Zen 3 following Zen 2/1 naming was a given.
It makes sense though the rx 5xx series to the rx 5xxx series
 
I don't see the point in buying, as well, in order to get more for your hard-earned money, honestly think that if AMD can afford to do this, they are no longer on the road to recovery but safe and secure, and honestly hope that intel’s 3rd Core X HEDT Lineup in the autumn will return them back to the Earth from Mars.

if i was you id settle on a ryzen 5 and get a better gpu sorted out and stop looking at products that will be expensive because its obvious you have no intent to buy any of them and just seem hell bent on moaning that things arnt super cheap. move on lifes unfair and top of the line stuff always costs more than everyone wants to spend.
 
Did she?
Doesn't look as leathery as Huang's.

Clearly needs up upgrade :D

NTCufqB.png

AMD-Dr-Lisa-Su-CES-2019-Keynote-AMD-Ryzen-Under-the-Hood-IO-Die-for-PCIe-Gen4.jpg
 
if i was you id settle on a ryzen 5 and get a better gpu sorted out and stop looking at products that will be expensive because its obvious you have no intent to buy any of them and just seem hell bent on moaning that things arnt super cheap. move on lifes unfair and top of the line stuff always costs more than everyone wants to spend.

You can't explain to me what I need. I can afford a Ryzen 7 1700 for €145 and a Ryzen 7 2700 for €208. My GPU is the R9 380 4GB and it's much more than enough for me.
What I need is a 12-core CPU to last in the next years.

Will wait half a year if necessary but I'll get what I planned. AMD can't screw everything.
 
Back
Top Bottom