Speculation: Ryzen 3000 series

Page 142 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

TheGiant

Senior member
Jun 12, 2017
748
353
106
Will these be water cooled?

3700x looks like it's run semi-close-enough to peak perf/watt while 3800x is being pushed to the minimum perf/watt. With the dual chiplet 3900x at least the wattage is split up, so maybe 2x ~42 watts, which is close to what the 3700 chiplet is seeing. The 3800x seems like a significantly lesser binned chiplet that's being overvolted and with high amps gushing through.
we will see with the reviews
  1. if its factory overvolted or the 3700x
  2. all core underclocked undervolted 3800x parts, that cant make 100% the clocks of 3800x
  3. just set to lower tdp and thus lower all core turbo

wanna see really that power and VRMs temperature
 

B-Riz

Golden Member
Feb 15, 2011
1,595
761
136
Nice. I think the 3700x looks better than the 3800x (and my guess is it might overclock better too, but will have to wait a month or two for reviews and concensus on how the 3700x overclocks).

Well, it is the succesor of R7 1700. Of which, later mfg dates, late 2018 and on, OC way better at lower voltages than the launch ones.
 

fleshconsumed

Diamond Member
Feb 21, 2002
6,485
2,363
136
I'm thinking those chipset fans won't spool up except under extreme load (e.g. extreme nVME RAID). As long as it's silent or close to it under normal operation I don't think a fan is a negative. I have a X399 board that has a small fan near the VRMs, for example. Almost never hear it spool up except at boot.
People have been repeating no spinning unless under extreme load a lot, but AFAIK it's based on a single uncorroborated statement by a manufacturer (MSI?).

Why not wait until there are reviews out. Yes those fans are small, but they may not have to spin very fast to provide enough cooling. I don't like the idea of fans on a mobo either, but hopefully the execution is solid and they aren't noisy.
I'll wait, I'm just not optimistic about it. I have very low noise annoyance threshold. All my case fans spin at <600RPM under idle conditions. Up until a year ago 4TB 3.5" spinning hard drive was the loudest component in my primary desktop, I finally replaced it with SSD and now my PC is nearly silent and I do not want to do anything to jeopardize that silence.
 

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
If this was 2005, it would make sense to go "disruptive" on desktop lineup, and Intel would have been panic mode already and would find a way to squeeze 16 cores on mainstream lineup by any means possible. But this is 2019, Desktop sales have fallen in half compared its peak years decade ago, and will keep declining. It has gotten to the point where Intel seems like they are about to skip 10nm for desktop and it doesn't look as devastating as it would have been 10 years ago. Basically, going down on "distuptive" war path on desktop market would be pointless and would only cut into their potential profits. AMD is probably smart to use their best 8-core dies on EPYC since that is where the growth lies.
The problem with 16 cores on desktop for Intel is they'll have to abandon the ring for the mesh, which is not as game-friendly as the former. They may be able to do 12 cores, but as of yet the highest cores on a ring system is the Broadwell-E Core i7 6950x with 10 Cores. 3 years later, Intel is now talking about another 10 Core Comet Lake, no doubt also based on the ring architecture.
 

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
Nice. I think the 3700x looks better than the 3800x (and my guess is it might overclock better too, but will have to wait a month or two for reviews and concensus on how the 3700x overclocks).
This is how I see it:

Zen Zen2
1700 3700
1700x 3700x
1800x 3800x

The 1700x struggled to reach 1800x clocks, while the plain 1700 could reach 1700x clocks. So, based on this binning pattern (if we can call it that) I'd go for either the Plain Jane 3700 or 3800x. AMD bins these chips aggressively, it's almost impossible to see lower tier silicon overclocking better than an upper tier silicon. I think this will remain true for this release as well.
 

Panino Manino

Senior member
Jan 28, 2017
868
1,114
136
Maybe for Ryzen 4000 series. To be more serious, I could see 3300g being 3c/6t (maybe with half L3) now that there is a native dual core APU.

Don't know if anyone already talked about this and have any guesses.
How will the new "APU" be? Will it use two chiplets, one for CPU and other for GPU, or just one "APU chiplet"?
In this case I can image 4 cores + xCU, what leaves open the possibility to have salvaged 3 core APU.

But I don't want 3, I want the luck 7!
 

jpiniero

Lifer
Oct 1, 2010
15,154
5,686
136
How will the new "APU" be? Will it use two chiplets, one for CPU and other for GPU, or just one "APU chiplet"?

Based upon what AMD said, it's likely to be monolithic. That Intel is doing a 6 core Comet Lake U option might tempt AMD to do more than 4 but remains to be seen.
 

Shivansps

Diamond Member
Sep 11, 2013
3,875
1,530
136
I think the answer is simple, desktop APU are mobile SoCs. And at this point im still expecting to be monolithic.

The bigger problem right now for a APU is DDR4, moving to 8 cores will most likely result in lower performance. It is just too much having a RX460 sized IGP plus 8 Zen cores on dual channel DDR4.
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,705
1,231
136
Both Renoir(Ryzen)(Zen2/Vega20) and Dali(Athlon)(Zen3/Navi) should be monolithic.

Pic/Ren
4/8
1/4

Rav2/Dal
Homo/Hetero
2/3 : 2/8
 
Last edited:

IRobot23

Senior member
Jul 3, 2017
601
183
76
Whoaahh

This is just insane from AMD. They are pushing so hard.... cmon you just showed and will release 3000 soon... yet zen 3 is already making his way out for 2020.
I love they of X570 pretty much what everyone excpect
40 PCIE 4.0 lanes
All extra features
no redacted with redacted VRMs.
Or buy X470/b450 and use less lanes.

Only thing that I am worried is DRAM latency and DRAM OC. Most of people are talking about 4.4-4.6GHz (+15% IPC) all core turbo which is not the best, but still better than +5% IPC and 4,8-5GHz.

So all I need is to hit high IFabric clocks and DRAM OC above 4GHz

If we look at ryzen 1000 which OC to around 3.8-4GHz and now they get 4.4-4.6GHz with Ryzen 3000... comparing 3.8GHz vs 4.4GHz (*1.03*1.15 + IPC) its like 5.2GHz Ryzen 1000. So jumping from 3.8GHz to 5.2GHz is something to consider (+better latencies, more bandwidth). But 4.6GHz sound even better ~ 5.45GHz or ~ 5.15Ghz skylake.

Profanity is not allowed in
the tech forums.

AT Mod Usandthem
 
Last edited by a moderator:

Hans Gruber

Platinum Member
Dec 23, 2006
2,302
1,216
136
You have to consider Zen 2 is the end of life of the AM4 platform. Zen 2 will be DDR5 with a new socket I assume. Is PCI4 important with Zen 3 around the corner? I am thinking B450/X470.
 

Slaughterem

Member
Mar 21, 2016
77
23
51
https://translate.google.com/translate?hl=en&sl=auto&tl=en&u=https://www.chiphell.com/thread-2001057-1-1.html

View attachment 6723

some important information I can decode from this author:

1. 5Ghz single core is realistic, and sounds like all core 4.8Ghz is pretty realistic
2. Cinebench R15, 8 core Zen2 @ 4.5Ghz easily beats 9900K @ 5Ghz
3. OC mechanism is pretty like Zen+, for example OC R5-2600 4.1Ghz sometimes stronger than 4.3Ghz due to heat or other factor......Zen2 has similar characteristic
What I get from this in regards to OC they are talking more about memory as compared to cpu.
1. 5Ghz for memory is realistic but at one half of IF or 2500 which may not be possible.
2. IF maxed out I think and not sure at 1700 In last version of zen.
3. This is a new version of IF. We don't know what is the max Frequency.
4. I would speculate that 2200 is max IF and memory max is 4400.
5. If you go above 4400 IF will remain at 2200 and will not result in improved performance
 

JDG1980

Golden Member
Jul 18, 2013
1,663
570
136
If I were AMD, I would use the standard cpu chiplet, and make the GPU+IO as monolithic. easier chip to make, and adds more diversity.

The problem is that IO doesn't scale that well with die shrinks. Therefore, AMD wants to keep it on 14nm for cost reasons (it's probably a bit cheaper per mm^2, and they have to pay for those WSA wafers anyway). On the other hand, AMD will definitely want the iGPU logic to be on 7nm for efficiency reasons (and because otherwise they'd have to backport Navi to 14nm for that one application).

If AMD does go with chiplets for the APU, I expect it to be a 14nm IO chiplet (possibly a new design), the same 7nm Zen2 die as on the existing products, and a small Navi GPU die shared with low-end discrete cards. But that probably won't happen. It's more likely that it will be a monolithic die - probably with at least 6 cores, if not 8.
 

H T C

Senior member
Nov 7, 2018
586
424
136
Didn't they say the could enable PCIe 4.0 on some existing earlier gen boards? If the fan were required only because of PCIe 4.0, then those older boards with it enabled would also eventually require some extra cooling, no?

I suspect there is something coming with this new chipset that we're not aware of yet and that's why it requires a fan.

Kind of like PBO with Zen + VS Zen: you could use Zen + CPUs in 1st gen AM4 boards and have XFR2 but you'd lose out on PBO since that feature was exclusive to 2nd gen AM4 boards.



One can use Zen 2 CPUs with older gen AM4 boards but will miss out on whatever is coming with gen 3 AM4 boards that's specially tailored for Zen 2 CPUs if we do so.
 

moinmoin

Diamond Member
Jun 1, 2017
5,064
8,032
136
I'm thinking those chipset fans won't spool up except under extreme load (e.g. extreme nVME RAID). As long as it's silent or close to it under normal operation I don't think a fan is a negative. I have a X399 board that has a small fan near the VRMs, for example. Almost never hear it spool up except at boot.
Computerbase.de managed to get words from ASRock on the matter, it doesn't sound good: https://www.computerbase.de/2019-05/amd-x570-luefter/
"According to ASRock the PCH X570 gets hot in all operating states, even while idling. Due to that the fan will run continuously on all boards by ASRock. While the fan does trottle based on temperature, the manufacturer wasn't willing to give info on fan speed and noise level."
Laut ASRock wird der Platform Controller Hub (PCH) X570 in allen Betriebszuständen sehr warm, auch im Leerlauf. Darum werden die Lüfter bei allen ASrock-Platinen auch durchweg rotieren. Die Lüfter sind zwar temperaturgesteuert, was das für Drehzahl und Lautstärke bedeutet, konnte der Hersteller aber noch nicht verraten.
You have to consider Zen 2 is the end of life of the AM4 platform. Zen 2 will be DDR5 with a new socket I assume. Is PCI4 important with Zen 3 around the corner? I am thinking B450/X470.
According to the spec DDR5 will have the same number of pins as DDR4, just different usage. If all pins are connected I'd think the socket needn't change for that. While AMD only promised AM4 forward compatibility up to Zen 2 I can imagine they'll keep using it as a rolling window, ie. Zen 3 still work on older AM4 boards but only starting with e.g. the 4xx chipset gen, dropping 3xx.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
26,125
15,270
136
Didn't they say the could enable PCIe 4.0 on some existing earlier gen boards? If the fan were required only because of PCIe 4.0, then those older boards with it enabled would also eventually require some extra cooling, no?

I suspect there is something coming with this new chipset that we're not aware of yet and that's why it requires a fan.

Kind of like PBO with Zen + VS Zen: you could use Zen + CPUs in 1st gen AM4 boards and have XFR2 but you'd lose out on PBO since that feature was exclusive to 2nd gen AM4 boards.

View attachment 6742

One can use Zen 2 CPUs with older gen AM4 boards but will miss out on whatever is coming with gen 3 AM4 boards that's specially tailored for Zen 2 CPUs if we do so.
Somewhere in this thread it was mentioned that for older motherboards, only PART of the PCIE4 would work. Thats why its not a problem. I don't remember if its the video or the M.2 that is enabled, but only one of the two vs x570 has everything enabled.
 
Reactions: Drazick

Karnak

Senior member
Jan 5, 2017
399
767
136
Yeah the difference is the chipset, there you'll get PCIe 4.0 only with X570 boards. But due to Zen2 itself having PCIe 4.0 support with a total of 24 (16+4+4) lanes it's possible to implement PCIe 4.0 with a BIOS-Update for older MB.

You'll always get the 16 lanes for the GPU but if you'll get the 4 lanes for the first M.2 slot depends on your MB layout. There are another 4 lanes but they don't matter because they're directly for the chipset and that's only relevant for X570.
 

ozzy702

Golden Member
Nov 1, 2011
1,151
530
136
I'm set for now with a 9900k (got it a week after launch) and a 2080TI, that said, I will be pointing all my friends/family exclusively to AMD for the conceivable future (already do 75%+ Ryzen builds now). If AMD keeps up this kind of momentum and gaming performance is on par with the 9900k or superior, I'll build a Zen 3 based main box when it's released which will be the first AMD build for my main gaming/productivity personal box since Core2 was released. Very impressive execution by AMD that I hope bleeds over in time to the GPU side as well; I wouldn't mind leaving team green if AMD has options.


Do we have an idea of what kind of launch quantities we're looking at? Paper launch or widespread availability? I have a few buddies that I'll be building boxes for so I hope they have decent inventory.
 

Mockingbird

Senior member
Feb 12, 2017
733
741
106
That obviously not the case there is a $330 and $380 products that use full spec dies. Inevitably they will sell more then the 12c chips. At even just $600 they get better margins just because No extra CPU package (pins/substrate/installation), no extra IO Die, no extra retail packaging. Sales would be a lot lower over the 12c and 8c chips anyways. It's probably because of several things. 1. Power usage (getting good clocks at reasonable power usage maybe be hard), 2. For that reason Binning (need to get sufficient stock of dies that can run at desired clocks within desired power) 3. ASP. AMD for whatever reason desided that they didn't want to push 8c pricing much lower then what we were already paying. They have 3 choices for a 16c chip. Push everything down to slot into $500 price point and lower ASPs. Another would be to push into Threadripper pricing with it ($600+). The final and what they seemed to have decided is to cap AM4 to ~$500. So cut out the 16c option while they build stock on the binned chips. This gives them a CPU in the pocket for Comet Lakes release or any movement on competitive pricing from Intel. When it comes out they can slide the 12c solution down to $400 and slide everything else down $30 or so.

Now personally I said and still think withholding on the 16c part is a bad idea. But there is wiggle room there. But I don't think they are in a great position to be withholding and should still be going down the "disruptive" war path. But this might be a good move for the start of 7nm when everything is at its most expensive and keep the 6c parts in better position financially (like it might be a bad idea to come out of the gate with those guys sub $200).

Anyone familiar with AMD thinking would have known that AMD would only pushed down the prices at the top.

What AMD wanted most was to increase profit margin and steal sales from Intel.

AMD was already dominating the <$200 market; Intel simply isn't very competitive in this segment.

It's the $200-$500 market that AMD has its eyes on; Intel dominate this segment.

If AMD increase the core count across the lineup, it would make the whole lineup much more attractive, which is not what AMD want.

Instead, AMD want to draw consumers attention especially to the $500 Ryzen 9, and did so by giving 50% more core count than Intel did
 
Reactions: Yeroon

Mockingbird

Senior member
Feb 12, 2017
733
741
106
Yeah, you guys are right. I found a better picture from an expo and after adjusting brightness it does appear to have a fan. I want to buy a new xX70 motherboard, and it looks like I'm going to have to choose between x570 with fan or x470 without full PCIe4 support. I hate noise with a passion and seeing how PCIe4 native support is the only selling point of x570 I'll try to find x470 on sale. Shame.

Gigabyte X570 AORUS XTREME doesn't have a fan.
 
Reactions: Yeroon

Mockingbird

Senior member
Feb 12, 2017
733
741
106
Not just annoying, but completely mind boggling. The cult of Jim had their day or reckoning 5/27/2019 and are dissolved.

Unfortunately, the AdoredTV cult is not dead.

You can still see its followers (the Adored-apologists) going around and performing mental gymnastics to explain why the prophet wasn't wrong.

There has been enough praise and condemnation
of him in this thread already. At this point, it's beating
a dead horse, and it is a thread detraction.


The Ryzen 3000 series was officially announced,
so focus on that discussion.


AT Mod Usandthem
 
Last edited by a moderator:
Reactions: krumme
Feb 4, 2009
35,238
16,707
136
I'm set for now with a 9900k (got it a week after launch) and a 2080TI, that said, I will be pointing all my friends/family exclusively to AMD for the conceivable future (already do 75%+ Ryzen builds now). If AMD keeps up this kind of momentum and gaming performance is on par with the 9900k or superior, I'll build a Zen 3 based main box when it's released which will be the first AMD build for my main gaming/productivity personal box since Core2 was released. Very impressive execution by AMD that I hope bleeds over in time to the GPU side as well; I wouldn't mind leaving team green if AMD has options.


Do we have an idea of what kind of launch quantities we're looking at? Paper launch or widespread availability? I have a few buddies that I'll be building boxes for so I hope they have decent inventory.

No link but from what I’ve read processor availability appears to be good, AMDs yields appear to be good to better than expected
With 50 plus X570 boards planned to be available it’s probably safe to assume there won’t be a shortage. Also B350, B450, x370 and X470 boards can be an option.
 
Reactions: ozzy702
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |