Ryzen 3000 Leaks

[Shreyas1]

I just got my 2600 darn it

[Fnige]

3 minutes ago, CarlBar said:

Why would AMD do that when they can go for the top end

 

[Cereal5]

I’m still not a fan of a multi-die chip or infinity fabric... I guess that was kind of the way to go though due to shrinks becoming increasingly hard to do.

[BachChain]

Good god, even Vega's rumors weren't this absurd

[mr moose]

Just now, CarlBar said:

 

No i quoted the right person. The point is AMD would have to severely undercut what the silicon is capable of delivering to not get vastly better superior performance. hast just how big a deal the new node + chiplets are in terms of being able to produce practical desktop chips with this kind of bump.#

 

 

Why would AMD do that when they can go for the top end.

 

Righto then, wait for the results of benchmarks.    Personally I won't be surprised if it falls into the usual delta of performance improvements for a node upgrade and refinement.  If it is more than double the historical improvements from suchreefinements I'll be surprised.

[marldorthegreat]

2 minutes ago, BachChain said:

Good god, even Vega's rumors weren't this absurd

But this is something I would believe. 

[TetraSky]

If this is legit, I'm glad I waited instead of pulling the trigger on a new build on black friday even though it was so tempting. (Worst case, there's still Boxing day if we get info that it will sucks before then)
Now I'm interested in either the Ryzen 5 3600X or the Ryzen 7 3700.

 

But, I fully expect this to be false.

Guess we'll know in 1 month at CES.

[PlayStation 2]

(((leaks)))

[ravenshrike]

Redgamingtech posted a video regarding the increased core count yesterday and had possibly a separate source from where the guys at reddit got it since it was given to him before the reddit leaks dropped.

 

 

 

[BuckGup]

33 minutes ago, TigerHawk said:

I dunno man...as much as I wish this were true, most of this seems like it has to be BS. 5.1ghz on 16c/32t model at only 135W? How tf?

 

Some of them seem pretty reasonable though. To me the realistic boost speeds I expect for Ryzen 2(3000 series) is 4.5 to 4.7 ish tops. 5+ is like......I dunno man. Is the shrink to 7nm really going to make THAT much of a difference? And at such a good bin rate that you can get a 16 core 5.1ghz(probably dual core boost) part for under $500 US? Holy hell?

 

Edit: To be clear, not really doubting the core counts, more the clock speeds.

 

Edit2: actually yeah I am doubting the core counts now in the Ryzen 3 range. This means that any silicon that doesn't make it through with at least 6 cores with SMT enabled is just tossed in the trash? Unless they are going to have something below Ryzen 3 to use them for? Doesn't make sense.

I wouldn't doubt anything except the tdp and maybe price. They now can shove double the number of cores on the same CCX as shown with Epyc. The highest end Ryzen CPU had 8 cores and Epyc was at 32. So with how AMD manufactures CPUs unlike Intel they are all the same so it wouldn't make any sense to make another lineup just for Ryzen. As for clock speeds 14nm to 7nm is a big jump and Samsung vs TSMC. Threadripper refresh could achieve 4.3ghz so another 800mhz is a lot but not too far fetched as it will have new memory controller and so on

[CarlBar]

3 minutes ago, mr moose said:

 

Righto then, wait for the results of benchmarks.    Personally I won't be surprised if it falls into the usual delta of performance improvements for a node upgrade and refinement.  If it is more than double the historical improvements from suchreefinements I'll be surprised.

 

It's going to depend a lot on the workload. Gaming they probably won't do better than match intel because they're not really pushing the clocks too amazingly. Anything that wants lots of cores though. That would benefit a lot.

 

Also like i said i'm sure those thermal numbers are a tad off.

 

Based on what they revealed so far we know they could build a 7nm copy of the 2700x in terms of core count and frequencies at around the 60w TDP with the new 7nm Zen2 architecture and still get a noticeable performance uplift from IPC, memory speed increases, and lowered intercore latency, all of which would likely leave the 9900k as the only better CPU intel has. And 60w TDP is less than what the R5 uses atm. So they actually had room for some frequency uptick without going over their current TDP's for the R5.needless to say there's plenty of room in the 105w TDP of the R7 series to go to more cores and more frequency.

[poochyena]

There is no way
That Ryzen 7 3700X will be an absolutely incredible deal if true. 12 cores????? 5 ghz???

My 6600k will look like dirt compared to that thing.

[Okjoek]

6 and 8 core navi APUs YESSSSSSSSS!

[Curufinwe_wins]

I'm very skeptical about the clock speeds, but obviously if they can do it that is awesome news for everyone involved.

[Fnige]

Im waiting for techlinked to show this and maybe have it as their title

[leadeater]

AMD rumors = Rumor / 50% = reality.

[Guest]

1 hour ago, Techstorm970 said:

Not to mention that you could OC a cheaper FX-8350 and get the same clocks & performance with less heat and power draw.

It was the same chip.

[cj09beira]

1 hour ago, TigerHawk said:

I dunno man...as much as I wish this were true, most of this seems like it has to be BS. 5.1ghz on 16c/32t model at only 135W? How tf?

 

Some of them seem pretty reasonable though. To me the realistic boost speeds I expect for Ryzen 2(3000 series) is 4.5 to 4.7 ish tops. 5+ is like......I dunno man. Is the shrink to 7nm really going to make THAT much of a difference? And at such a good bin rate that you can get a 16 core 5.1ghz(probably dual core boost) part for under $500 US? Holy hell?

 

Edit: To be clear, not really doubting the core counts, more the clock speeds.

 

Edit2: actually yeah I am doubting the core counts now in the Ryzen 3 range. This means that any silicon that doesn't make it through with at least 6 cores with SMT enabled is just tossed in the trash? Unless they are going to have something below Ryzen 3 to use them for? Doesn't make sense.

that tdp isnt affected by single core boost.

about the clocks being high, well they are but it could be feasible, we have various things changing at the same time, there is a move from 20 finfet to 10nm finfet (in intel sizes which means they skipped a full node) while at the same time moving from a pretty low clocking node to a more high performance node, 

plus we already know how good the node is as vega 7nm has a spec of 1800mhz which is a huge increase from vega 10's 1500mhz, thats a super solid real world 20% increase in clocks, plus vega could already reach 1700s when oced which if the same oc head room exists, vega 20 might be able to pass 2Ghz on air ( or water as that would be a huge thermal load), apply that same change to the 1800x (as vega is 14nm not 12nm) and you get a 4.32ghz base 4.8ghz boost (that would be at stock), add the fact that much better binning can happen due to the small nature of the chiplets and you can go even further.

 

about the 6 core thing, due to the size of the chip having more than 2 cores not working will be very rare, and yields should be very very good.

[pas008]

49 minutes ago, leadeater said:

AMD rumors = Rumor / 50% = reality.

i think core count be true

clocks unsure of

and ipc there was a thread in news on this https://bit-tech.net/news/tech/cpus/amd-downplays-29-percent-zen-2-ipc-boost-reports/1/ on amd downplaying the amount of the increase

 

 

but all these together should put amd on top almost all around for awhile

 

[Fasterthannothing]

If this is true I can't wait 5ghz on 12 cores bye Intel

[dgsddfgdfhgs]

only wanted to see cheap apu to compete with 1050ti on 6c12t

[Crunchy Dragon]

1 hour ago, SlimyPython said:

FX_9590 was a very hot CPU

It had 8 cores and ran at 4.7ghz and is only $200aud though its TDP (i think that is what it is) was 220w

Correct, 220W.

[COTG]

While I am extraordinarily bullish for ryzen 3000, and the move to 7nm overall, as an amd investor for now two years I have to say that all this hype is great when its still in moderation but when it gets out of hand like this it tends to hurt amd more than help it. For whatever reason people take those rumours to heart, no matter how ludicrous they are, and if amd misses on any of the insane points listed they will be disappointed and likely stray away from amd. Not saying that ryzen 3000 will not be a beast, because it definitely will be, but probably not of such magnitude. 
Im not saying that there isnt a chance for these rumours to be true but looking at what they are proposing, and especially for what price, I would definitely remain skeptical...

[leadeater]

1 hour ago, pas008 said:

i think core count be true

clocks unsure of

and ipc there was a thread in news on this https://bit-tech.net/news/tech/cpus/amd-downplays-29-percent-zen-2-ipc-boost-reports/1/ on amd downplaying the amount of the increase

 

 

but all these together should put amd on top almost all around for awhile

 

Was more of a joke, that's basically how AMD rumors typically turn out but that's more in the GPU space.

[ApolloFury]

Remember the original Zen leaks back in 2016 (8 cores for $320) and people thought it was too good to be true?