what happened to the 3850X

So we got the full expected lineup from the leak minus some of the graphics parts. He had big time caveats on pricing and clocks because there was no way to have final details that far ahead or know what market conditions would be. Honestly, given just how far ahead those leaks were I don't know how they could have been much more accurate.

I'll be picking up a 3700X. If Acer releases a bios for the helios 500 it'll replace the 2700 in that, if not it'll go in my sff to replace the 1700 that's in it. Then I wait for 16c threadripper. Should be 4.9 boost with quad channel ram and no latency penalty - plus a lot more power to play with fit oc'ing. Can't wait!
 
So we got the full expected lineup from the leak minus some of the graphics parts. He had big time caveats on pricing and clocks because there was no way to have final details that far ahead or know what market conditions would be. Honestly, given just how far ahead those leaks were I don't know how they could have been much more accurate.
Yes, scary accurate if you remove the marketing names and prices (which are 2 things that changes more often than a baby's diaper.)
Considering the leak was from December 2018... pwoah!
adoredvsamd.png
 
Expectation of 5ghz as the standard of performance tells me youre (general statement, not any one specific person) either a dumb puppet who was brainwashed by a number or 2 you have no idea how processors work and are still brainwashed.

5ghz is the number that Intel has settled on as the defacto cap per se of thier configurations performance peak.

AMD clearly doesn't need 5ghz to reach the same performance thus making 5ghz nothing but an arbitrary number based on another companies design results.

I would give an analogy but its already retarded seeing this expectation for 5ghz and disregard for the massive strides AMD has made at less power and clocks than Intel.

Anyone who says AMD sucks because zen 2 is not at 5ghz is a fool especially given we see how fast and power saving the new architecture is at lower clocks.
 
Last edited:
Yes, scary accurate if you remove the marketing names and prices (which are 2 things that changes more often than a baby's diaper.)
Considering the leak was from December 2018... pwoah!
View attachment 167331

The fake "leaks" said the the 12 core part would turbo 5.0 GHz and 16 core part would turbo to 5.1 GHz.

Then there are the Zen 2 APUs which aren't even't supposed to show up until next year, and the Zen+ APUs that are missing.

The only reason the fake "leaks" seems close in your chart is that you only cherry-picked things that are close and disregarded everything else.

_______________________________________________________________________________

Imagine I got 20% on a exam and only cherry-picked the problems that I got right to say that I did great!

That's exactly what you are doing.
 
The fake "leaks" said the the 12 core part would turbo 5.0 GHz and 16 core part would turbo to 5.1 GHz.

Then there are the Zen 2 APUs which aren't even't supposed to show up until next year, and the Zen+ APUs that are missing.

The only reason the fake "leaks" seems close in your chart is that you only cherry-picked things that are close and disregarded everything else.

_______________________________________________________________________________

Imagine I got 20% on a exam and only cherry-picked the problems that I got right to say that I did great!

That's exactly what you are doing.

Exactly. The leaks said 5.1GHz turbo, 16 cores, 15% IPC improvement, at 105w TDP. All of that combined was too good to be true.
 
Exactly. The leaks said 5.1GHz turbo, 16 cores, 15% IPC improvement, at 105w TDP. All of that combined was too good to be true.

Actually, I believe the 16-core part Jim talked about was stated as having a TDP of 135W (and 125W for the 12-core)...?
 
Cool part, I honestly do not see use for it in any of my setups. Unless it supports ecc and can run on the ASRock X470D4U.

As for taking away from TR, I do not really see that either. They both may be 16 cores parts, but they are designed for completely different uses. Well... I see it may take some sales from TR, and that would be the overkill 16 core gamer who believes he needs all of that power.
 
Cool part, I honestly do not see use for it in any of my setups. Unless it supports ecc and can run on the ASRock X470D4U.

As for taking away from TR, I do not really see that either. They both may be 16 cores parts, but they are designed for completely different uses. Well... I see it may take some sales from TR, and that would be the overkill 16 core gamer who believes he needs all of that power.

Threadripper still has way more memory bandwidth and more PCIe lanes. That's of value to people who need it for workstation applications.
 
  • Like
Reactions: Boil
like this
Threadripper still has way more memory bandwidth and more PCIe lanes. That's of value to people who need it for workstation applications.

And WAY more power delivery when the inevitable zen2 refresh comes. Even the matx x399m can push a 2990wx to 4.1 with active vrm cooling - meaning it can push 400-500w. I'm just hoping they release a 16c version, tr has traditionally had higher max boost than am4 due to better silicon, and with no more latency/numa issues it'll be a total ass kicker. Even if they don't, we should get a quad chiplet 24c that's around 3.9 base, 4.8 or 4.9 boost.
 
That's not true. It's an upgrade if the IPC improvement claims are true. A 15% upgrade and boost clocks up to 4.6GHz would put Zen 2 in a pretty good spot. Especially with 8c/16t and 12c/12t parts on the high end. Node shrinks aren't magic. As I said, when your improving processors, you can generally go for higher clocks, or higher IPC, but generally not both. AMD did manage both from what we've seen thus far. AMD managed to not only improve IPC by 15% (which is huge) but increase boost clocks fairly significantly. They did this while keeping a 105w TDP and increasing core counts. All that's damn near too good to be true as it is. Wanting 5.0GHz and all that just isn't realistic.

I wouldn't hold out hopes of achieving 5.0GHz through overclocking either. Like I said, you'll probably see 100-200MHz and load all cores if your lucky. Your TDP will go through the roof, but that's probably the best you can hope for.



The embargo date is 7/7/19. Its unlikely you will see anything before then. AMD tends to sample chips only a week or two out from the embargo date. Reviewers don't have their hands on them yet. I'm working with motherboard vendors to get X570 samples which aren't available yet and won't be for at least a couple weeks or more. Reviewers aren't often left with much time ahead of launches to work with the hardware.



German overclocker Der8bauer has X570 motherboards and 3000 chips in hand he gave a glance on his site yesterday in a video.
 
German overclocker Der8bauer has X570 motherboards and 3000 chips in hand he gave a glance on his site yesterday in a video.

I'm actually surprised anyone would have them this early, but that's a very good sign.
 
It is just not you that does not know what the word speculation means ;)

We have this kind of conversation with every knew major product release. Speculation is great provided everyone is aware that it's speculation. We like to talk about this stuff, otherwise we wouldn't be here. Great! It's only a problem when people get emotionally attached to a certain result, and come to expect a certain outcome as opposed to discussing the possibility of a certain outcome.

I dislike both the notion that speculation is bad because it sets expectations (it doesn't, unless someone is being dumb), but also dislike the notion that it's anything more than what it is. Rumors/leaks... these are speculation ONLY. Never take them too seriously!
 
These companies spend a lot of time going through every possible iteration of product lineups, so I'm sure a 3850X was a possibility that was simply passed over for xyz reasons.
 
We have this kind of conversation with every knew major product release. Speculation is great provided everyone is aware that it's speculation. We like to talk about this stuff, otherwise we wouldn't be here. Great! It's only a problem when people get emotionally attached to a certain result, and come to expect a certain outcome as opposed to discussing the possibility of a certain outcome.

I dislike both the notion that speculation is bad because it sets expectations (it doesn't, unless someone is being dumb), but also dislike the notion that it's anything more than what it is. Rumors/leaks... these are speculation ONLY. Never take them too seriously!

I talked about it earlier on the Navi discussion , it can hurt as well. But take the Zen 2 case where loads (too many for my liking) of people just type where is my 5ghz cpu ?

People keep referring to it as 5ghz and it is being interpreted as being all core boost not as a single core one.
It seems that this reality we are in people see what they want to see and hear what they want to hear. And everything dipped in absolutes (great binary thinkers :) ) ....
 
These companies spend a lot of time going through every possible iteration of product lineups, so I'm sure a 3850X was a possibility that was simply passed over for xyz reasons.

Way to read the thread title & then blindly comment...

The hypothetical 3850X is now known as the Ryzen 9 3950X...

You know, exactly what we have been discussing in the last two pages...
 
Good possibility that there will still be a Matisse refresh:

https://www.techpowerup.com/267523/...en-7-3850x-and-3750x-processors?cp=2#comments

Although Zen 3 is not too far behind, a higher clocked Zen 2 or 'Zen 2+' could keep potential buyers from drifting their eyes toward Comet Lake.

I saw that. Looks like these are going to be called the 3600XT, 3800XT, and 3900XT. I guess it comes down to how much more headroom they have and the price. Hopefully not vaporware like the 9900KS.
 
I saw that. Looks like these are going to be called the 3600XT, 3800XT, and 3900XT. I guess it comes down to how much more headroom they have and the price. Hopefully not vaporware like the 9900KS.

But the best we're going to see is 5% clock sped improvement, just like Zen+

Everyone here knows that the real meat is in Zen 3, and this is just to satisfy demand NOW.

Since Zen 3 is coming from the same process node, it would be mind-boggling if the new core didn't at eat match th e frequency pf the refresh!
 
I saw that. Looks like these are going to be called the 3600XT, 3800XT, and 3900XT. I guess it comes down to how much more headroom they have and the price. Hopefully not vaporware like the 9900KS.

The 3600XT is a good move as the new i5s seems to be strong competition. I would hope that the new Renoir APUs get the same advancements.
 
Unfounded yes, but that's what they needed to actually be an upgrade from zen+. The best description of zen 2 is, it's not slower. The twelve core is pretty cool too for people who can load 24 threads, but with the node shrink I was really hoping.
5ghz seems to be the silicon stone wall. I bet zen2 will oc up there with proper cooling.
Did you even look at benchmarks? They were a pretty jump in performance from zen+ to zen2, especially in things that were sensitive to latency (like games).
Geometric mean of 1540 benchmarks:
3700x is 34.6% faster than 1700x, and 21.8% faster than 2700x. I feel a 20% increase in performance on average is pretty good considering what we've been seeing. It's not slower is what I called the the 6700k to 7700k release.
That said, the 3850x (probably going to be released as the 3800xt) will probably be a slightly higher clocked 3800x for the same $ or slightly cheaper.
 
Did you even look at benchmarks? They were a pretty jump in performance from zen+ to zen2, especially in things that were sensitive to latency (like games).
Geometric mean of 1540 benchmarks:
3700x is 34.6% faster than 1700x, and 21.8% faster than 2700x. I feel a 20% increase in performance on average is pretty good considering what we've been seeing. It's not slower is what I called the the 6700k to 7700k release.
That said, the 3850x (probably going to be released as the 3800xt) will probably be a slightly higher clocked 3800x for the same $ or slightly cheaper.

cough that post is from june 2019 and zen 2 launched a month later cough .. :p
 
  • Like
Reactions: travm
like this
cough that post is from june 2019 and zen 2 launched a month later cough .. :p

I'm sure there was some sort of placeholder and then they edited it after Zen2 launched. I mean the Data shouldn't be that much different anyway.
 
I’m down to price shift rather than push the boundaries.
I’d like to see a $105 2700AF I can pair with a top quality $125 B550 board.
 
Did you even look at benchmarks? They were a pretty jump in performance from zen+ to zen2, especially in things that were sensitive to latency (like games).
Geometric mean of 1540 benchmarks:
3700x is 34.6% faster than 1700x, and 21.8% faster than 2700x. I feel a 20% increase in performance on average is pretty good considering what we've been seeing. It's not slower is what I called the the 6700k to 7700k release.
That said, the 3850x (probably going to be released as the 3800xt) will probably be a slightly higher clocked 3800x for the same $ or slightly cheaper.
1590321904478.png

Actually, 10% seems to be the magic number. +/- some. You could average those I guess and have a number just slightly over 10.

Have to admit, I was pleasantly suprised with the increase. I did not expect the IPC gains they made, and the preliminary benchmarks released by AMD suggested they were cherry picking benchmarks and running them balls to the wall on clock, then comparing it to a slower clocked Zen 2xxx. That was the basis of that post, from June 2019.
 
I’m down to price shift rather than push the boundaries.
I’d like to see a $105 2700AF I can pair with a top quality $125 B550 board.

i doubt we'll see those, tsmc has no reason to produce them. there's only 2 reasons i could see for why the 1600AF even exists. 1. an oem ordered them and never ended up using them or 2. it was an agreement between AMD and glofo so that they could keep producing products on 12nm along side the chipset and IO dies for zen 2.
 
But the best we're going to see is 5% clock sped improvement, just like Zen+

Everyone here knows that the real meat is in Zen 3, and this is just to satisfy demand NOW.

Since Zen 3 is coming from the same process node, it would be mind-boggling if the new core didn't at eat match th e frequency pf the refresh!

The flagship Zen 3 is rumored to be October, but the more mainstream CPUs could be closer to the New Year according to the latest rumors. That is just too far away for some.

New leaks of the XT lineup show some respectable clocks:
https://www.techpowerup.com/267615/...n-matisse-refresh-xt-sku-clock-speeds-surface

Releasing the XT lineup at the current SKU prices would really put the hurt on Comet Lake 'K' SKUs, especially with B-550.

Dropping the price of the current AMD lineup would make short work of the remaining CML cpus.
 
Also, it wouldn't be 'mind boggling' if Zen 3 Cpus didn't match Zen 2+ frequencies. Architecture plays as much of a role in clock speeds as does lithography node.

Same goes for Rocket Lake. I would be VERY surprised if any Willow Cove CPUs hit 5.3 ghz turbos.
 
Also, it wouldn't be 'mind boggling' if Zen 3 Cpus didn't match Zen 2+ frequencies. Architecture plays as much of a role in clock speeds as does lithography node.

Same goes for Rocket Lake. I would be VERY surprised if any Willow Cove CPUs hit 5.3 ghz turbos.


Sorry dude, this is going to be the simplest IPC increase ever: they're just combining the two 4-core CCXs into one 8-core chip! And the most likely 5% addition is double the side of L3 cache again!

If you think that SPEEDING UP that slow inter-core interface is going to slow down your maximum clock speeds, I have a bridge to sell you! AMD iskeepping thing s simple for Zen 3, They're saving the massive chnages for Zen 4 on DDR45.
 
Last edited:
Sorry dude, this is going to be the simplest IPC increase ever: they're just combining the two 4-core CCXs into one 8-core chip! And the most likely 5% addition is double the side of L3 cache again!

If you think that SPEEDING UP that slow inter-core interface is going to slow down your maximum clock speeds, I have a bridge to sell you! AMD iskeepping thing s simple for Zen 3, They're saving the massive chnages for Zen 4 on DDR45.

Then I will buy your proverbial bridge:
https://www.igorslab.de/en/new-deta...us-have-appeared-the-image-is-getting-denser/

It's looking like RKL will have a similiar slowdown.
https://www.techpowerup.com/267526/intel-rocket-lake-cpu-appears-with-6-cores-and-12-threads

So while both Willow lake and Vermeer may offer 15-20% IPC improvements, real-world performance increases may be much lower.

In both cases, 6 months is a long time to wait for a potentially 5-10% performance boost.
 
Then I will buy your proverbial bridge:
https://www.igorslab.de/en/new-deta...us-have-appeared-the-image-is-getting-denser/

It's looking like RKL will have a similiar slowdown.
https://www.techpowerup.com/267526/intel-rocket-lake-cpu-appears-with-6-cores-and-12-threads

So while both Willow lake and Vermeer may offer 15-20% IPC improvements, real-world performance increases may be much lower.

In both cases, 6 months is a long time to wait for a potentially 5-10% performance boost.


From that link Vermeer is seeing 4.6 ghz boost on A0 silicon. It should be faster by the time it sees A1 release. That will make it faster-clocked than the 3950X

I don't care about the shit that is Rocket Lake, as it's obvious rushed backport of Tiger Lake. one has a stable process node, while the other one obviously does not!

PRETENDING THAT ONE OF THESE BEING AN OBVIOUS FAILURE AUTOMATICALLY MEAN PROBLEMS FOR THE AMD CORE IS CALLED FALSE EQUIVALENCY. Look it up, genius!

Because AMD is obviously going to suddenly start having issues on a process node + small arcitecctursal tweak they've already been shipping in massive quantities for twelve months?
 
Last edited:
From that link Vermeer is seeing 4.6 ghz boost on A0 silicon. It should be faster by the time it sees A1 release. That will make it faster-clocked than the 3950X

I don't care about the shit that is Rocket Lake, as it's obvious rushed backport of Tiger Lake. one has a stable process node, while the other one obviously does not!

PRETENDING THAT ONE OF THESE BEING AN OBVIOUS FAILURE AUTOMATICALLY MEAN PROBLEMS FOR THE AMD CORE IS CALLED FALSE EQUIVALENCY. Look it up, genius!

Because AMD is obviously going to suddenly start having issues on a process node + small arcitecctursal tweak they've already been shipping in massive quantities for twelve months?

Which is slower than the 4.8 ghz boost of the 3900xt on the link I gave. Also, calm the F down.

You have no links or sources to any of your claims.
"Zen 3 is just Zen 2 with combining 4 core CCXs" Where did you get this info?

"Rocket is just a rushed backport of TGL." Sorry, Intel knew LONG ago that 10nm wasn't going to cut it for desktop use. After all, the decision to pull Icelake from desktop was made long ago.

Under the same process node, Broadwell could never hit the same clocks that Skylake 1.0 could.

Not sure of the false equivalency claim. It just sounds like the Intel failures are causing you to lash out.
 
Back
Top