(16core added)AMD 3000 specs! 4.7 GHZ, R9 3950x, R7 3700x, 3800x.

[DanielMDA]

20 hours ago, Jurrunio said:

So each CCX module has 6 cores by design?

No, it still has 4. Each chiplet has 8 cores, with two four core CCX´s each

[Biggerboot]

The Ryzen 7 lineup was a nice surprise.  i9 9900k performance for 200$ less.  Sounds good.

[TahoeDust]

Just now, Biggerboot said:

The Ryzen 7 lineup was a nice surprise.  i9 9900k performance for 200$ less.  Sounds good.

We'll see.  The benchmarks will be interesting.  If it can match the 9900k in gaming and in single threaded tasks, it will be a hell of a deal.

[Biggerboot]

3 minutes ago, TahoeDust said:

We'll see.  The benchmarks will be interesting.  If it can match the 9900k in gaming and in single threaded tasks, it will be a hell of a deal.

Yeah, the original leaks I was hearing about had me worried that most of the single threaded improvements would come from higher clock speeds (the 5ghz rumors and all that).  I'm actually glad that they didn't feel the need to dish out 16 core right away and kept the clock speeds conservative.  The aggressive power delivery on the higher boards seem more reserved for future cpu's if I had to guess.

[justpoet]

9 minutes ago, Biggerboot said:

Yeah, the original leaks I was hearing about had me worried that most of the single threaded improvements would come from higher clock speeds (the 5ghz rumors and all that).  I'm actually glad that they didn't feel the need to dish out 16 core right away and kept the clock speeds conservative.  The aggressive power delivery on the higher boards seem more reserved for future cpu's if I had to guess.

Linus actually said in his video that the VMs were because the board manufacturers didn't want to be caught unable to go forward like with the first gen AM4 boards.  I would argue that better VMs are just much easier and cheaper to do now than they were then too, so it is a small cost to add for both overclocking and future compatibility options.

[Results45]

This might come in-handy.    RYZEN 3000 SERIES SPEC SHEET FROM ANANDTECH:
AnandTech		Cores Threads		Base Freq	Boost Freq	L2 Cache	L3 Cache	PCIe 4.0	DDR4	TDP	Price (SEP)
Ryzen 9	3900X	12C	24T	3.8	4.6	6 MB	64 MB	16+4+4	?	105W	$499
Ryzen 7	3800X	8C	16T	3.9	4.5	4 MB	32 MB	16+4+4	?	105W	$399
Ryzen 7	3700X	8C	16T	3.6	4.4	4 MB	32 MB	16+4+4	?	65W	$329
Ryzen 5	3600X	6C	12T	3.8	4.4	3 MB	32 MB	16+4+4	?	95W	$249
Ryzen 5	3600	6C	12T	3.6	4.2	3 MB	32 MB	16+4+4	?	65W	$199

 

[justpoet]

1 hour ago, Results45 said:

This might come in-handy.    RYZEN 3000 SERIES SPEC SHEET FROM ANANDTECH:
AnandTech		Cores Threads		Base Freq	Boost Freq	L2 Cache	L3 Cache	PCIe 4.0	DDR4	TDP	Price (SEP)
Ryzen 9	3900X	12C	24T	3.8	4.6	6 MB	64 MB	16+4+4	?	105W	$499
Ryzen 7	3800X	8C	16T	3.9	4.5	4 MB	32 MB	16+4+4	?	105W	$399
Ryzen 7	3700X	8C	16T	3.6	4.4	4 MB	32 MB	16+4+4	?	65W	$329
Ryzen 5	3600X	6C	12T	3.8	4.4	3 MB	32 MB	16+4+4	?	95W	$249
Ryzen 5	3600	6C	12T	3.6	4.2	3 MB	32 MB	16+4+4	?	65W	$199

 

That really looks a ton like they're just binning from the top down and disabling cores (maybe due to defects, maybe just due to being worse binned) for the 5.  This leads to the natural hope that the 16c version will be at least similarly clocked to the 9, or if we look at the 9 as being the same but better boost as 2 3600x's, consider that for the 16c based off of the 3800x.  My guess would be that they're just saving up binned stock for the 16c since they don't need to release it yet, to make sure that they get more press coverage later, and to make sure it can actually clock the same or maybe .1 better than the 9.  If we follow that thinking, we can look at TDPs and guess that we would move from 105W to around 120W-130W based on how the chart is moving with clocks and cores. (30w for .2, 40w for .3 when looking at base), or that when looking at doubling the cores (3600x to 3900x) we don't gain a ton of TDP (only 10w, which seems oddly low).  There will, of course, be less than a doubling of TDP because a good chunk of that is the IO chiplet, which we know with PCIe4 will be using a fair amount of power on its own (much like the motherboard chipsets), but only 10w TDP to add the 2nd 6c chiplet seems like an oddity that will need to be investigated once the chips are actually out.

[Stefan Payne]

1 hour ago, Results45 said:

This might come in-handy.    RYZEN 3000 SERIES SPEC SHEET FROM ANANDTECH:
AnandTech		Cores Threads		Base Freq	Boost Freq	L2 Cache	L3 Cache	PCIe 4.0	DDR4	TDP	Price (SEP)
Ryzen 9	3900X	12C	24T	3.8	4.6	6 MB	64 MB	16+4+4	?	105W	$499
Ryzen 7	3800X	8C	16T	3.9	4.5	4 MB	32 MB	16+4+4	?	105W	$399
Ryzen 7	3700X	8C	16T	3.6	4.4	4 MB	32 MB	16+4+4	?	65W	$329
Ryzen 5	3600X	6C	12T	3.8	4.4	3 MB	32 MB	16+4+4	?	95W	$249
Ryzen 5	3600	6C	12T	3.6	4.2	3 MB	32 MB	16+4+4	?	65W	$199

 

PCIe is not correct as 4 of those lanes are for Chipset and not usable on most consumer Boards (only on "A300" "Chipset").

And with X570 AMD claims 40 PCIe 4.0 lanes...

[Stefan Payne]

4 minutes ago, huilun02 said:

AMD's next lineup of server chips is going to bring Intel to its knees...

Oh well, only up to 160 PCIe Lanes with Rome systems...

 

1 Rome CPU still has 128 PCIe 4.0 lanes (wich might or might not support CCIX, Gen-Z or IF), 2x128- 2x48 = 160 

Because AMD reduced the amount of lanes between the CPUs from 64 to 48...

[Spotty]

1 hour ago, Results45 said:

Ryzen 7	3700X	8C	16T	3.6	4.4	4 MB	32 MB	16+4+4	?	65W	$329
Ryzen 5	3600X	6C	12T	3.8	4.4	3 MB	32 MB	16+4+4	?	95W	$249
Ryzen 5	3600	6C	12T	3.6	4.2	3 MB	32 MB	16+4+4	?	65W	$199 

Does anyone else think the TDP values for the Ryzen 5's and the Ryzen 7 look a little... Strange?
The Ryzen 5 3600X has an almost 50% higher TDP than the Ryzen 7 3700X for only a 200MHz (~5%) base clock increase, despite having 2 fewer cores?

If anyone has any thoughts or ideas as to why that might be the case I'd be interested in hearing them.

[ForeverYoung]

AMD YES

[mr moose]

7 minutes ago, Spotty said:

Does anyone else think the TDP values for the Ryzen 5's and the Ryzen 7 look a little... Strange?
The Ryzen 5 3600X has an almost 50% higher TDP than the Ryzen 7 3700X for only a 200MHz (~5%) base clock increase, despite having 2 fewer cores?

If anyone has any thoughts or ideas as to why that might be the case I'd be interested in hearing them.

 

I'm not too sure myself but my first thought is that because AMD rate their TDP using thermal and voltage limits it would have been rated closer to 4.4Ghz rather than the base clock for both the 3600X and the 3700X.  So not exactly sure but the 3600X could be doing less work at the same frequency and voltage meaning it could simply be inefficient using those metrics.

[rambi36]

Time for me to get a job lol

[Giganthrax]

Am I missing something, or is there absolutely no mention of 3200g/3400g APUs?

[The Benjamins]

3 hours ago, Results45 said:

This might come in-handy.    RYZEN 3000 SERIES SPEC SHEET FROM ANANDTECH:
AnandTech		Cores Threads		Base Freq	Boost Freq	L2 Cache	L3 Cache	PCIe 4.0	DDR4	TDP	Price (SEP)
Ryzen 9	3900X	12C	24T	3.8	4.6	6 MB	64 MB	16+4+4	?	105W	$499
Ryzen 7	3800X	8C	16T	3.9	4.5	4 MB	32 MB	16+4+4	?	105W	$399
Ryzen 7	3700X	8C	16T	3.6	4.4	4 MB	32 MB	16+4+4	?	65W	$329
Ryzen 5	3600X	6C	12T	3.8	4.4	3 MB	32 MB	16+4+4	?	95W	$249
Ryzen 5	3600	6C	12T	3.6	4.2	3 MB	32 MB	16+4+4	?	65W	$199

 

Well if you looked at AMDs spec page you would see all of them are set to 3200Mhz RAM. It also goes over what coolers it comes with.

 

https://www.amd.com/en/products/specifications/processors/11781,11756,11761,11766,11771

[Stefan Payne]

1 hour ago, Spotty said:

Does anyone else think the TDP values for the Ryzen 5's and the Ryzen 7 look a little... Strange?
The Ryzen 5 3600X has an almost 50% higher TDP than the Ryzen 7 3700X for only a 200MHz (~5%) base clock increase, despite having 2 fewer cores?

If anyone has any thoughts or ideas as to why that might be the case I'd be interested in hearing them.

Different Turbo and slightly above the Threshold that AMD specified internally.

 

Also one thing to consider:

a) the VRM are current limited, not Power!

b) Zen2 based Chips have lower Voltages -> 1,2V instead of 1,3-1,4V (I've seen that the Ryzen 7/1800X seems to go to 1,5V.

 

So that means that if a VRM is made for about 65W at 1,4V, if we now reduce the Voltage to 1,2V, the Limit drops to just ~55W (rounded down).

So the 3600X might actually consume still around 65-70W, but that's over the spec, so it has to be 95W because of that.

[Humbug]

 

 

 

 

[mr moose]

9 minutes ago, Humbug said:

 

 

 

 

He speaks like a true fanboy.

[DeScruff]

2 hours ago, Spotty said:

Does anyone else think the TDP values for the Ryzen 5's and the Ryzen 7 look a little... Strange?
The Ryzen 5 3600X has an almost 50% higher TDP than the Ryzen 7 3700X for only a 200MHz (~5%) base clock increase, despite having 2 fewer cores?

If anyone has any thoughts or ideas as to why that might be the case I'd be interested in hearing them.

I haven't read fully up on everything, and am really behind on this whole thing but the first things that come to mind:

1) TDP is more like categories rather then actual hard numbers I believe. You almost never see '75W' and '85W' CPUs - or some kinda TDP gradient even though there is a massive thermal difference between the weakest 65W, and best 95W CPUs, but not usually much between the best 65W and worst 95W.
So the 3600X unfortunately goes above what can fit in the 65W spec, and so jumps to next one which is the 95W spec, meanwhile the 3700X fits still fits within the 65W spec.
Its kinda like how you can have a PSU that is 99% efficient from 0-90% load but if at 100% load its only 84% efficient, that PSU is not getting anything better then a 80 Plus Bronze sticker. - Even though its technically better then 80 Plus Titanium PSUs most of the time.

2) Maybe it has something to do with using multiple chiplets vs single chiplet?
I am assuming the Ryzen 5 uses one 6 core chiplet, and the Ryzen 7 uses two 4 core chiplets  (please correct me if I am wrong) - If putting more cores per chiplet causes a dramatic increase in TDP, (relative to more chiplets with fewer cores) That might explain why we aren't seeing a 16 core CPU yet.

[CarlBar]

I'm not going to quote everyone who responded on the statistics thing but rather give a quick generalised response.

 

First and foremost whilst you can use information to narrow down the range of possible numbers there's still the simple fact that there at least a few hundred mhz/couple of tens of watts of leeway either way. here's isn't just 2 or 3 possible answer to pick from. I'd say depending on the skew and the item in he spec the rnage is anywhere from 5 to 10 possibble answers assuming 100mhz frequency and 5 watt TDP jumps for each step.

 

The odds with that kind of rate of getting a significant number of spec items correct, (i made a mistake last night due to exhaustion, Adored's actual hit rate so far is 40%, for some reason i got it into my head we only had one R5 revealed, blame lack of sleep okay), is incredibly low. true guesswork would be lucky IMO to hit 20% accuracy, nevermind 0%.

 

But there's another point. If your guessing at everything, even if it;s educated guesses then your right and wrong answers will be relatively evenly distributed throughout the pool, (true randomness does include some clumping and TDP's can be narrowed down from the clocks you guess but the odds of a random set of guess producing exclusively clumped results is very low). What we got, (either all right or all wrong for a given item), is statistically very unlikely. The odds on that are almost certainly north of 1 in a 100 for doing it once. Twice is even higher.

[hayden1881]

 

[CarlBar]

27 minutes ago, Sypran said:

I haven't read fully up on everything, and am really behind on this whole thing but the first things that come to mind:

1) TDP is more like categories rather then actual hard numbers I believe. You almost never see '75W' and '85W' CPUs - or some kinda TDP gradient even though there is a massive thermal difference between the weakest 65W, and best 95W CPUs, but not usually much between the best 65W and worst 95W.
So the 3600X unfortunately goes above what can fit in the 65W spec, and so jumps to next one which is the 95W spec, meanwhile the 3700X fits still fits within the 65W spec.
Its kinda like how you can have a PSU that is 99% efficient from 0-90% load but if at 100% load its only 84% efficient, that PSU is not getting anything better then a 80 Plus Bronze sticker. - Even though its technically better then 80 Plus Titanium PSUs most of the time.

2) Maybe it has something to do with using multiple chiplets vs single chiplet?
I am assuming the Ryzen 5 uses one 6 core chiplet, and the Ryzen 7 uses two 4 core chiplets  (please correct me if I am wrong) - If putting more cores per chiplet causes a dramatic increase in TDP, (relative to more chiplets with fewer cores) That might explain why we aren't seeing a 16 core CPU yet.

 

8 core is 1 chiplet.

[GoldenLag]

3 hours ago, Spotty said:

Does anyone else think the TDP values for the Ryzen 5's and the Ryzen 7 look a little... Strange?
The Ryzen 5 3600X has an almost 50% higher TDP than the Ryzen 7 3700X for only a 200MHz (~5%) base clock increase, despite having 2 fewer cores?

If anyone has any thoughts or ideas as to why that might be the case I'd be interested in hearing them.

could be the sign of an awful voltage curve after 4.3 jiggies. or that the x variants have the top boost applied to more chips than non-x. instead of increasing clockspeed they might just go for more cores at a given clockspeed. 

 

keep in mind the next node from tsmc (6nm/7nm+) apperantly wont have much of a clock increase, or a clockincrease at all. just yield improvements and density improvements. so holding back performance so their next gen wont look awful? i know that is whishful thinking, but it could be a thing. 

[Stefan Payne]

1 minute ago, CarlBar said:

8 core is 1 chiplet.

2x4 Core on 2 Chiplets are 8 Cores as well.

The difference is:
For 1x8 you need one fine die - with no defects

For 2x4 you need two crap dies - half of the die might have defects but its still salvagable.

 

I'd say two dies are more probable on the 8 Cores than 1 Die because of cost and "salvaging" half dead Cores...

Though it might be possible that the 3700X is two die while the 3800X is one. We have to wait and se.

[leadeater]

33 minutes ago, Stefan Payne said:

I'd say two dies are more probable on the 8 Cores than 1 Die because of cost and "salvaging" half dead Cores...

Edit: Nope nvm, dumb. Ignore this part.