The Radeon Technology Group (RTG) has received its first Zen 2 sample!

Joined
Feb 20, 2017
Messages
992
The RTG has just received its first Zen 2 sample (to optimize for) and it's really impressive.

8C/16T

4.0 GHz/4.5 GHz

DDR4-3600 CAS 15

Radeon RX Vega 64 LE

__________________________

The good: It's already nibbling at the Core i7-8700K.

The bad: It crashes a lot.

The ugly: It crashes all the time. Some of the tests have to be run multiple times because they crashed before finishing.
 
The magic fairy told you?

Edit: and this is not good.

Engineering samples are meant to crash a lot. but ONLY coming close to matching an 8700K? Who freaking cares? It needs to BEAT a 9800K. Intel is going to have its own 8/16 CPU that will be the definitive gaming champion, AMD needs to beat that.
 
The magic fairy told you?

mockingbird has been on point bro.
I don't care if it beats intel for gaming or not. 2700x is already on par and ahead in many situations. Epecially outside of 120p 690Hz esport pro gaming everyone always seems to do when CPUs are being discussed.
So they just need to be about as good and cheaper or with more cores. Pretty easy task they have already achieved.


Mockingbird, weirdly low clocks though... I guess this is crippled a bit to not let the cat out of the bag, or a 'rough' speed as to not incriminate your source.
 
I don't care if it beats intel for gaming or not. 2700x is already on par and ahead in many situations. Epecially outside of 120p 690Hz esport pro gaming everyone always seems to do when CPUs are being discussed.

AMD is closing the IPC gap so it looks like Intel decided to clock to the moon.
 
AMD is closing the IPC gap so it looks like Intel decided to clock to the moon.
Yup and they lost process advantage.
Their turn to have hot, inefficient cpus, they are using about 25% more power in some server configurations with less performance lol. Just like the 79whatever extreme edition which was stupidly inefficient vs the initial zen offerings.

Next part thats coming is the lack of OC headroom as they hit the 14nm limits and bump base and turbo clocks.
 
mockingbird has been on point bro.
I don't care if it beats intel for gaming or not. 2700x is already on par and ahead in many situations. Epecially outside of 120p 690Hz esport pro gaming everyone always seems to do when CPUs are being discussed.
So they just need to be about as good and cheaper or with more cores. Pretty easy task they have already achieved.


Mockingbird, weirdly low clocks though... I guess this is crippled a bit to not let the cat out of the bag, or a 'rough' speed as to not incriminate your source.

Alright, If Mokingbird has the rep, I'll respect it.

If true, I'm not happy. AMD is going to lose the core advantage. The whole argument with Ryzen has been "It's not as fast per core so it loses in games versus the i7, but it has more cores and threads so it is an overall faster CPU"

If AMD stick to an 8/16, then they lose that. Even if their IPC gets close its still an overall step backward. AMD will STILL have the slower cores, but now they won't have a thread advantage.

Not happy.
 
Alright, If Mokingbird has the rep, I'll respect it.

If true, I'm not happy. AMD is going to lose the core advantage. The whole argument with Ryzen has been "It's not as fast per core so it loses in games versus the i7, but it has more cores and threads so it is an overall faster CPU"

If AMD stick to an 8/16, then they lose that. Even if their IPC gets close its still an overall step backward. AMD will STILL have the slower cores, but now they won't have a thread advantage.

Not happy.
Yeah fair call, although being realistic, it's pretty clear with the density shrink they can do more cores. Or with speed bump they can make it faster., or a blend. With revenue increasing they may offer both as they can probably afford it now. A 16 core zen2 would impact TR sales though, which is what has me wondering... Maybe it'll be like 2990wx again with the half ram controller scenario.

I'm confident final clocks will be higher and memory controller bumps which bought most of advantage this time round will be tweaked further.
Or else why would they shrink?
Just to save a few sheckels with a density shrink with nothing else on the table?
This is sandbagging, just like the early Zen leaks at 3.2ghz or so.
 
  • Like
Reactions: ltron
like this
Apparently, there has been some changes to the "interconnect" (wherever that is) that requires RTG to make changes to the video drivers and that why RTG is getting the sample.
 
Still, look at it from a boolean logic standpoint:

Zen1 versus Core 7000:
More cores 1
Faster cores 0

Zen+ versus Core 8000
More cores 1
Faster cores 0

Zen 2 versus Core 9000
More cores 0
Faster cores 0

As a salesperson who deals a LOT with end-users and what their interests are, I'll tell them "Intel is 5-10% faster in games, but AMD is 20-50% faster in everything else". That 5-10% sells A LOT of Intel CPUs. it doesn't matter if it were 1%. It doesn't matter if it were 0.001%. The truth is that "Faster" is an absolute descriptor. I still sell a LOT of AMD CPUs too, mostly for people who are not as gaming-focused or are into streaming. But the only reason is because AMD has that thread advantage. Give Intel thread parity, and AMD looses its BIGGEST eye-catching selling point.
 
Still, look at it from a boolean logic standpoint:

Zen1 versus Core 7000:
More cores 1
Faster cores 0

Zen+ versus Core 8000
More cores 1
Faster cores 0

Zen 2 versus Core 9000
More cores 0
Faster cores 0

As a salesperson who deals a LOT with end-users and what their interests are, I'll tell them "Intel is 5-10% faster in games, but AMD is 20-50% faster in everything else". That 5-10% sells A LOT of Intel CPUs. it doesn't matter if it were 1%. It doesn't matter if it were 0.001%. The truth is that "Faster" is an absolute descriptor. I still sell a LOT of AMD CPUs too, mostly for people who are not as gaming-focused or are into streaming. But the only reason is because AMD has that thread advantage. Give Intel thread parity, and AMD looses its BIGGEST eye-catching selling point.

I never said that Zen 2 would only have 8C/16T, only that that particular sample has 8C/16T.

There may or may not be more cores. I don't know.

I know next to nothing about AMD's processor teams.
 
I never said that Zen 2 would only have 8C/16T, only that that particular sample has 8C/16T.

There may or may not be more cores. I don't know.

I know next to nothing about AMD's processor teams.

I hope for the industry's sake this is an APU model with a single CCX. AMD being competitive has been the best thing for CPUs in 10 years.
 
The magic fairy told you?

Edit: and this is not good.

Engineering samples are meant to crash a lot. but ONLY coming close to matching an 8700K? Who freaking cares? It needs to BEAT a 9800K. Intel is going to have its own 8/16 CPU that will be the definitive gaming champion, AMD needs to beat that.

Its an ES chip, they never have had final clock speeds on either amd or intel. I reckon we'll see 4.2ghz all core stock for the 3700x or whatever they call it.

And i wouldn't be suprised if Zen 3 is a 6 core ccx, ddr 5 and PCIe 4.0
 
Are we ignoring the fact that Zen will likely be more efficient than Intel, regardless of which has higher clocks?

Intel may be clocking higher to beat AMD, but there will be a price for those clocks.
 
Its an ES chip, they never have had final clock speeds on either amd or intel. I reckon we'll see 4.2ghz all core stock for the 3700x or whatever they call it.

And i wouldn't be suprised if Zen 3 is a 6 core ccx, ddr 5 and PCIe 4.0

yeah zen 3 should be ddr5 at the least since it's getting the new socket. intel's 10nm is suppose to be DDR5 as well but who knows when that'll actually being released.
 
I hope for the industry's sake this is an APU model with a single CCX. AMD being competitive has been the best thing for CPUs in 10 years.

They already have a single CCX. Although it's only four cores. The new 2300X is four cores all on the same die. AMD knows exiting the NUMA node causes penalties.

Are any of the current zen+ 8 cores on a single CCX?
 
I'll tell them "Intel is 5-10% faster in games, but AMD is 20-50% faster in everything else". That 5-10% sells A LOT of Intel CPUs.
That's a lie though, it's not always just a thread advantage. Often they are side by side especially at higher than 1080 resolutions and in some games AMD is faster than Intel. So you have to say on average Intel is a little faster than AMD in games, not all the time, but with multi threaded performance, or streaming etc, AMD is usually faster.
If you want to be honest you should be more accurate with what you say to your customers.
 
They already have a single CCX. Although it's only four cores. The new 2300X is four cores all on the same die. AMD knows exiting the NUMA node causes penalties.

Are any of the current zen+ 8 cores on a single CCX?

No all are 4 core CCX.
I think they will go 8 core in the next generation or two via active interposer, because then they can reduce latency for 8 core loads, it makes most sense and gives an easy boost. Then they can save expensive 7, 7+ or 5 nm die space..
 
For all we know it could be a Ryzen 5 8 Core zen 2 chip doing apples to apples to an 8700k. Ryzen 7 could still have 12 cores but would not draw an apples to apples comparison for IPC. Duh...
 
No all are 4 core CCX.
I think they will go 8 core in the next generation or two via active interposer, because then they can reduce latency for 8 core loads, it makes most sense and gives an easy boost. Then they can save expensive 7, 7+ or 5 nm die space..

So zen is 4 core CCX in ryzen, 8 core CCX in thread ripper?
 
I am shocked people are even trying to compare Zen 2 to the 8700k. I mean we know nothing about the chip at all. Plus it is an ES sample.....

I am looking forward to seeing what the chip can do, but I want more info!
 
Somebody from the Zen team came by and took back the development kit (processor, motherboard, and memory)
 
I can buy trends and Ghz .. what I want to see is how Zen and Vega scale married together an a APU went we hit DDR5 and bigger frame buffer as that would move it up to HD5850-7950 depending on how it scales .
 
I can buy trends and Ghz .. what I want to see is how Zen and Vega scale married together an a APU went we hit DDR5 and bigger frame buffer as that would move it up to HD5850-7950 depending on how it scales .

Really just need to get some HBM on package for, let's call them, 'higher end' APUs. Think more along the lines of what they collaborated with Intel on.
 
  • Like
Reactions: N4CR
like this
The magic fairy told you?

Edit: and this is not good.

Engineering samples are meant to crash a lot. but ONLY coming close to matching an 8700K? Who freaking cares? It needs to BEAT a 9800K. Intel is going to have its own 8/16 CPU that will be the definitive gaming champion, AMD needs to beat that.

Really? 9800k is what 4.7ghz boos rightt? If this is running 4ghz stock and 4.5ghz boost and has the rumored 10-15% IPC uplift do you really thing it will be that much behind 9800k? I am not sure how difference there will be between 8700k and 9800k in gaming. I doubt much! Plus you should also take in to consideration its an early engineering sample, things usually get much better when final chip comes out. I think we are probably still 8-12 months away from launch and they already have 7nm zen 2 samples. That is pretty damn good! Means they aren't just bullshitting about 7nm products looking good and them throwing all their eggs in 7nm bucket.
 
  • Like
Reactions: GHRTW
like this
YEah, with all this RTG insider crap, tell us some juicy deets on how AMD is planning on competing with the 3080Ti (because that will be the competition by the time AMD releases whatever product they're cooking up)
 
YEah, with all this RTG insider crap, tell us some juicy deets on how AMD is planning on competing with the 3080Ti (because that will be the competition by the time AMD releases whatever product they're cooking up)
They won't so don't worry about it. The consumers buy Nvidia if you would put a Nvidia logo on a turd it will outsell anything made by AMD...

Alright, If Mokingbird has the rep, I'll respect it.

If true, I'm not happy. AMD is going to lose the core advantage. The whole argument with Ryzen has been "It's not as fast per core so it loses in games versus the i7, but it has more cores and threads so it is an overall faster CPU"

If AMD stick to an 8/16, then they lose that. Even if their IPC gets close its still an overall step backward. AMD will STILL have the slower cores, but now they won't have a thread advantage.
Not happy.

I can redefine it for you when you buy Intel processors you are killing your own future by being faster now would kill your choice in the future and be stuck again with whatever Intel deems worthy for you to purchase.

You could claim that AMD would do the same if in the same position but that point does not matter (not relevant). Anyone with half a working brain should never in this era buy Intel.

There is no one on this forum that was screaming their heads off of excitement for the next 4c8t Intel and still people are not smart enough to take the performance loss on AMD cpu.
 
So zen is 4 core CCX in ryzen, 8 core CCX in thread ripper?
Nope, all Zen chips are same across the range pretty much(within each family, e.g. Zen, Zen+), just varying numbers of them (Threadripper/Epyc) or various cuts made to them (2x4 core CCX with a defective memory controller for Ryzen 1700-1800, 2x3 core CCX for 1600, 1x 4 core CCX for 1500 or whatever it is. There was a B2 Stepping for Epyc from memory though, with memory controller improvements, remember Epyc uses all functions on the die so is the 'top bin' as such in terms of defects at least. So Epyc got the newer dies in some cases with tweaks as it's their high value product, so can pay off quickest for this change.

So yes, this is why Intel is getting hammered. They have to make a series of larger monolithic multi core dies to go over 8 cores. They have to make multiple dies for their product line. AMD only makes one really (assuming B2 stepping becomes the updated mainstream part also), so they can use 98% of their sillicon.. Intel has a high defect rate with e.g. Xeon 8180.
They also have a lot of heat and growing efficiency issues as they are pushing clocks to stay relevant with AMD in multicore applications, let alone narrowly edge AMD on average in single core.
 
The magic fairy told you?

Edit: and this is not good.

Engineering samples are meant to crash a lot. but ONLY coming close to matching an 8700K? Who freaking cares? It needs to BEAT a 9800K. Intel is going to have its own 8/16 CPU that will be the definitive gaming champion, AMD needs to beat that.

how can you beat something that you have no performance numbers for?
 
tell me how navi and next gen chips are looking? LOL! I had to ask!

YEah, with all this RTG insider crap, tell us some juicy deets on how AMD is planning on competing with the 3080Ti (because that will be the competition by the time AMD releases whatever product they're cooking up)

This is a processors forum.

If you want to talk about GPUs, go to the video cards forum.
 
So zen is 4 core CCX in ryzen, 8 core CCX in thread ripper?
No, zen from r3 to epyc is 4 core ccx. APUs have a single CCX + the integrated graphics, CPUs have two CCXs (8 cores total) per die, with some disabled in certain models, threadripper and epyc have 2 and 4 active dies respectively (16 and 32 cores). TR2 has two additional active dies in certain models (for 4 total).
Edit: Illustrated for clarity:
IMG_20180914_115016.jpg
 
Last edited:
So, consumers should be smart enough to not get the best product for their money?

This is your argument for AMD?

:ROFLMAO:

Your best product is not really the best to begin with. My argument starts with people having a brain to begin with ...
 
Back
Top