Speculation: Ryzen 3000 series

Page 114 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

What will Ryzen 3000 for AM4 look like?


  • Total voters
    230

dnavas

Senior member
Feb 25, 2017
355
190
116
Yes but also completely unrealistic imo. A 1700 using 85w is max 3.7 all core. I expect the highest frequency 8c to come in at 95w tdp non oc and slot into all existing boards sans 320.

Yeah. Another reason why TR3 might be deferred. A high frequency 48 core starts to sound like you could use it to iron shirts. There's a lot leaking, but learning the freq/heat curves is going to be worth the wait. I assume that happens well after Computex....
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,540
14,495
136
Not being a fan boy so relax.
The chart is amazing for intel too, they are using old tech with old manufacturing processes and still end up with a similar result.
a $1400 CPU must be overclock a LOT to equal the stock AMD 16 core ?

Well we will see soon, but good luck staying with the old tech and overclocking and paying $1400 for it. I just turn down an offer of $1400 for a 32 core 2990wx, waiting for a better deal.
 

amrnuke

Golden Member
Apr 24, 2019
1,181
1,772
136
Not being a fan boy so relax.
The chart is amazing for intel too, they are using old tech with old manufacturing processes and still end up with a similar result.

Isn't it a bit fanboi-ish to toot Intel's horn after it's been uncovered that they get their "similar results" by leaving gaping security holes in their architecture?
 

amrnuke

Golden Member
Apr 24, 2019
1,181
1,772
136
Look at my post history...

There's a lot non-technology stuff there, and while I don't disagree with it, I don't really care to sift through pages of it.

I didn't mean my response as an affront to you specifically. What I meant was that anyone looking at AMD's "new tech" and Intel's "old tech" and saying that the fact that they're even close is somehow admirable on Intel's part, without considering how Intel has maintained its position, is definitely fanboi-ish.
 
Feb 4, 2009
34,553
15,766
136
There's a lot non-technology stuff there, and while I don't disagree with it, I don't really care to sift through pages of it.

I didn't mean my response as an affront to you specifically. What I meant was that anyone looking at AMD's "new tech" and Intel's "old tech" and saying that the fact that they're even close is somehow admirable on Intel's part, without considering how Intel has maintained its position, is definitely fanboi-ish.

Hence the warning at the beginning but I’ll let it rest
 

Abwx

Lifer
Apr 2, 2011
10,936
3,439
136
........

1-1080.e6b8f6cd.jpg

https://www.computerbase.de/2019-05/amd-ryzen-3000-details-pcie-4.0-matisse-x570-chipsatz/#bilder
 
Oct 19, 2007
82
28
101
a $1400 CPU must be overclock a LOT to equal the stock AMD 16 core ?

Just to be clear, the new Zen 2 is not running stock in that chart - quoting the article:

wccftech said:
Based on what’s being purported, this chip actually has a base clock speed of 3.2 GHz and only boosts to 4.3 GHz. This however, did not stop someone somewhere from overclocking it to 4.2 GHz on all cores and running a Cinebench R15 benchmark on it with quite an astonishing result.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,540
14,495
136
Just to be clear, the new Zen 2 is not running stock in that chart - quoting the article:
Sorry, I missed that. So a $1400 cpu overclocked to the max just equals what is most likely a $500 cpu. I know the tech is old, but they are still charging for it like its todays tech. Not to mention, no mention of how much juice it was taking to get to 4.8 vs the AMD running at 4.2 equaling it.
 

Saylick

Diamond Member
Sep 10, 2012
3,123
6,289
136
View attachment 6529

I'm simply floored. Can't wait until the announcement next week!
So I hear that Cinebench mainly runs out of the caches and favors Zen. Does anyone know off the top of their head how well Cinebench translates to other benchmarks, particularly those that are correlated better for real world applications?
 

krumme

Diamond Member
Oct 9, 2009
5,952
1,585
136
Ok. So 570 is 24 x pci 4.0 and is presented next weeks. That moves am4 a bit closer to hedt market imo.
I mean if we get a 4200 CB 15 score and 24 x pci 4.0 the only major thing left of current hedt market is memory bandwidth.
It is a major leap and I have no doubt this is not a 500 usd cpu. More like 700. And the 570 boards will have Intel prices to boot.
Save up guys :)
 

Hans Gruber

Platinum Member
Dec 23, 2006
2,131
1,088
136
So I hear that Cinebench mainly runs out of the caches and favors Zen. Does anyone know off the top of their head how well Cinebench translates to other benchmarks, particularly those that are correlated better for real world applications?
Cinebench favors more cores. See Threadripper scores. An 8 core Zen 2 will not do nearly as well as a 16 core Zen 2 CPU.
 
  • Like
Reactions: realibrad

dsplover

Member
Nov 1, 2014
38
4
81
Anyone know if Ryzen Matisse G CPUs can be overclocked.
I’d really like to overclock the 65 watt 3300G.

Thanks
 

.vodka

Golden Member
Dec 5, 2014
1,203
1,537
136
if Zen2 doesn't have a negative 256bit SIMD offset then you would expect 128bit SIMD to run near max all core boost most of the time.

https://www.anandtech.com/show/1357...-4-an-interview-with-amd-cto-mark-papermaster

IC: With the FP units now capable of doing 256-bit on their own, is there a frequency drop when 256-bit code is run, similar to when Intel runs AVX2?

MP: No, we don’t anticipate any frequency decrease. We leveraged 7nm. One of the things that 7nm enables us is scale in terms of cores and FP execution. It is a true doubling because we didn’t only double the pipeline with, but we also doubled the load-store and the data pipe into it.

There should be no negative offsets of any kind.
 

krumme

Diamond Member
Oct 9, 2009
5,952
1,585
136
So I hear that Cinebench mainly runs out of the caches and favors Zen. Does anyone know off the top of their head how well Cinebench translates to other benchmarks, particularly those that are correlated better for real world applications?
As I understand it CB mirrors perfectly the performance in Maxxons 4d software suite. So it's more or less 100% correlated to a real application. So it's actually not synthetic in that way like eg specint/fps.
Software is different. I mean it doesn't correlate to old game engines that more or less rely on a dx11 path that is more memory latency limited than more fpu throughput limited.
That said the performance seems pretty brutal if the numbers is true that is.
What software are you interested in?
 
Last edited: