News AMD previews Ryzen 3rd generation at CES

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Kedas

Junior Member
Dec 6, 2018
20
9
36
#51
in their EPYC demo they did run 2*28 cores against 1x64 cores
and AMD was 15% faster, doesn't that mean Zen2 probably closed the IPC gap
for that application because the EPYC had 15% more cores.
Meaning each core intel and amd worked at the same performance, we don't know the clocks though. (increasing intel to 64 cores would give them the same speed)
Intel was at 2*205W for 56 cores and AMD assume around 200W? for 64cores (180W)
 

Topweasel

Diamond Member
Oct 19, 2000
4,556
172
126
#53
I don't have issues with AMD at all.

I frequently suggest AMD CPUs to people.

In fact, I just tried to steer a poster to AMD from Intel.
I meant it solely in the way you meant this
We don't need childish games, though.

They don't make AMD look good to me.

It makes them look childish.
This one was an unfounded misunderstanding. Bu the point still stands. They aren't going to be a "here's the product, let the reviewers have at it" company. I mean even if they wanted to be, there is to much of a difference in their markets to just be another "Professional" supplier. But I don't think they want to be that company anyways. They will always be a bit of a childish company even if they have have things that miss the mark like "RIP Volta".
 

Saylick

Senior member
Sep 10, 2012
573
16
91
#54
in their EPYC demo they did run 2*28 cores against 1x64 cores
and AMD was 15% faster, doesn't that mean Zen2 probably closed the IPC gap
for that application because the EPYC had 15% more cores.
Meaning each core intel and amd worked at the same performance, we don't know the clocks though. (increasing intel to 64 cores would give them the same speed)
Intel was at 2*205W for 56 cores and AMD assume around 200W? for 64cores (180W)
I have a feeling that AMD intentionally downclocked their EPYC sample to show a clear lead over Intel in that head-to-head without tipping their hand too too much. There's no need for AMD to show off their maximum performance advantage this early on. They even said that it is a sample and that clocks aren't final.
 
Feb 23, 2017
408
246
96
#55
in their EPYC demo they did run 2*28 cores against 1x64 cores
and AMD was 15% faster, doesn't that mean Zen2 probably closed the IPC gap
for that application because the EPYC had 15% more cores.
Meaning each core intel and amd worked at the same performance, we don't know the clocks though. (increasing intel to 64 cores would give them the same speed)
Intel was at 2*205W for 56 cores and AMD assume around 200W? for 64cores (180W)
I'm fairly certain that I heard Lisa Su say that the Epyc chip was clocked pretty low. Don't hold me on that though.
 

Shlong

Diamond Member
Mar 14, 2002
3,100
1
91
#56
If they announce a 12-core or 16-core Zen2 with IPC/Frequency that is similar to a 9900K or faster, I will be upgrading from my 1700.
 

Shivansps

Platinum Member
Sep 11, 2013
2,340
132
126
#57
Seeing that empty space makes me wonder why they dont create a vega 5 gpu, pair that thing with 64 bit ddr4/ddr5 mem buses and a if link and you got the perfect GPU to reemplace the decade old RX530, fight the gt710 at ultra low end and to put on every cpu with only one CPU chiplet like the one showed today.
 
Last edited:

Saylick

Senior member
Sep 10, 2012
573
16
91
#58
I'm fairly certain that I heard Lisa Su say that the Epyc chip was clocked pretty low. Don't hold me on that though.
Yeah, there was some rumors about a 1.8 GHz LP (or Low Power) EPYC 2 sample that popped up. EPYC 2 is rumored to have final clocks in-line with EPYC 1, or about 2.2 GHz - 2.3 GHz thereabouts. If the sample they used in the head-to-head was that 1.8 GHz sample, I would not be surprised to see AMD extend that 15% lead over Intel's current fastest 2x28 Xeon 2-socket configuration by another 22-28% for a grand total of 40-47% performance gain in heavy multi-threaded applications when they ship their flagship EPYC 2 product.
 
Feb 23, 2017
408
246
96
#58
The other thing; no dummy chiplet. Leaving the space probably gives off a bigger statement than having a second dummy chiplet where it'd fit; we'd all be thinking that the 8c CPU in her hand was 2 salvaged dies, whereas the omission just says "Oh Intel, you're in trouble."
A picture speaks a thousand words. :)
 

IRobot23

Senior member
Jul 3, 2017
598
17
76
#59
So TR also double cores to 64c?
How fast will be ryzen for gaming?
 

Saylick

Senior member
Sep 10, 2012
573
16
91
#60
So TR also double cores to 64c?
How fast will be ryzen for gaming?
As I understand it, gaming is more sensitive to memory latency than some other applications. How well Ryzen 3000 does against Intel for gaming will probably be very dependent on how well designed and thought-out AMD were with respect to the new IO die. If they've worked out any potential issues regarding latency (read: competitive with Intel) then the fastest 8C will be probably as fast as the 9900K, if not faster.
 

Charlie22911

Senior member
Mar 19, 2005
519
26
116
#63
View attachment 2307
Look closely below the top die, what do you see when its titled at this angle?
:cool:

R.I.P - Shintel
Thats a really good eye! I completely missed that, it’s hard to believe that was anything but intentional; such a clever play.
Coupled with her tweet, I think we are pretty much guaranteed a 16 core part. I wonder if they’ll take a page from Apples “One more thing...” playbook and surprise drop it at some point, I sure hope so. I’m so happy I bought all that AMD stock in 2009-2011.
 

TheGiant

Senior member
Jun 12, 2017
261
16
76
#65
I still don't get the hype

for me there is big amount of fog in that presentation, in Athlon (the first one) and Athlon 64 times AMD was quite clear about performance and it demolished the p4

but here its like cherry picked demos...I have a bad feeling about that gaming

the only thing I consider pretty nice is power, so finally we don't have a pocket atomic reactor on the balcony

as for the radeon 7, I have a bad feeling for power from that presentation, it maybe as fast 2080 but with the same or more power....at 7nm, we will see

the real question is the availability of the 7nm ryzen parts...I see it as september when they sa q2-q3 ..which means too close the the icelake release if Intel finally recovers from the Brian Krzanich disaster (he should have a drink with Hector Ruiz)..
 

tamz_msc

Platinum Member
Jan 5, 2017
2,148
86
106
#66
the real question is the availability of the 7nm ryzen parts...I see it as september when they sa q2-q3 ..which means too close the the icelake release if Intel finally recovers from the Brian Krzanich disaster (he should have a drink with Hector Ruiz)..
Mid 2019 is May-June, likely Computex. Ice Lake desktop is not coming until 2020.
 

piesquared

Golden Member
Oct 16, 2006
1,587
10
136
#67
We don't need childish games, though.

They don't make AMD look good to me.

It makes them look childish.

Just put out a good product at a good price and you won't need to play such games.

Let your product do the talking.
That's about as sour grapish as it gets.
 

Sane Indian

Junior Member
Jan 9, 2018
5
0
11
#68
The space for extra chiplet is clever idea. Now, AMD can salvage all the chips. They can use 1-7 core salvaged chiplets making a full 8-core CPU.
 

TheELF

Platinum Member
Dec 22, 2012
2,637
44
106
#70
If you look at benchmarks between the 1700 and 9900k, the latter is about 44% better on average. AMD said the clocks weren't final either, so it could well end up that you can get a 50% bump over a 1700. I think that's certainly worth considering. The additional gains with Zen 3 probably won't see a similarly substantial jump.
The 2700 improved a whole bunch on cinebench over the 1700 but overall the 9900k is still 38% ahead,
one branch of computing doesn't really tell you anything about a different branch of computing.
https://cpu.userbenchmark.com/Compare/Intel-Core-i9-9900K-vs-AMD-Ryzen-7-2700/4028vs3957

I mean according to AMD the first ZEN was already supposed to blow intel out of the water,they made it look sooooooo gooooood against intel hedt in their presentation.
 

dlerious

Senior member
Mar 4, 2004
629
23
116
#71
I can totally live with a mid 2019 release... gives me a chance to stash away a bit more funds and allows my 1700 to have one last good hurrah. Seems like the PCIe 4.0 support might prompt a board upgrade though.
Part of the delay may be waiting for X570.
 

dlerious

Senior member
Mar 4, 2004
629
23
116
#72
Unbelievable that that the 9900k runs at 170Watts+... Thats near the suck down of my friggin 16 core threadripper. Jesus christ.. They literally riced out a civic (ringbus) w/ NoS on that processor :oops:... (Vttaaacchhh turboooo swwwwwuuuufffffftttt)
I believe those were system power draw, not CPU power draw.
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
16,947
359
136
#73
The 2700 improved a whole bunch on cinebench over the 1700 but overall the 9900k is still 38% ahead,
one branch of computing doesn't really tell you anything about a different branch of computing.
https://cpu.userbenchmark.com/Compare/Intel-Core-i9-9900K-vs-AMD-Ryzen-7-2700/4028vs3957

I mean according to AMD the first ZEN was already supposed to blow intel out of the water,they made it look sooooooo gooooood against intel hedt in their presentation.
So you are saying the benchmarks that were shown today at CES are fake ? That the new cores are now NOT on par with the 9900k ?
 

NostaSeronx

Platinum Member
Sep 18, 2011
2,216
32
106
#74
imho, I would rather get a μRipper.

64-core Jaguar, that is between two 7nm logic chiplets and sharing an I/O die. Yeah, no machine can rival the low-cost, easily ported Jaguar!
 

TheELF

Platinum Member
Dec 22, 2012
2,637
44
106
#75
So you are saying the benchmarks that were shown today at CES are fake ? That the new cores are now NOT on par with the 9900k ?
You can bet your shirt that they selected the best combination of hardware that made the 9900k look as weak as possible,just as they did at the original presentation of ZEN.
Being on par in cinebench doesn't mean much for userbench is what I'm saying,the 2700 especially the x was already pretty much on par,just minus 5% against the 9900k at stock.
https://www.techpowerup.com/reviews/Intel/Core_i9_9900K/9.html
 


ASK THE COMMUNITY