The Ryzen "ThreadRipper"... 16 cores of awesome

[beginner99]

Oh that video is wrong, I saw the real video on reddit. The real video is here https://youtu.be/Z1j3S6MKB_E
I figured you would need that much thermal paste as the CPU is massive.

IMHO still way, way too much paste. Paste is here to fill minor/tiny unevenness in the already flat surfaces. Meaning you only need small amount. Half of that applied will get squeezed out anyway.

 

[dooon]

Planet 3DNow!
AMD Threadripper ist nur ZP-B1 somit kein teildeaktivierter EPYC (ZP-B2)

YD1900A9U8SAE – 1900 8/16
YD190XA8U8SAE – 1900X 8/16
YD1920A9UC9AE – 1920 12/24
YD192XA8UC9AE – 1920X 12/24
YD1950A9UGAAE – 1950 16/32
YD195XA8UGAAE – 1950X 16/32

Kleine OPN Aufklärung:
YD = Ryzen
1900 = Nr
A9 = Thermal Power Dingsda
U = Sockel ID (SP3r2)
8 = Anzahl der Cores in Hex
S = Cache
AE = Revision wobei AE = ZP-B1 und AF = ZP-B2

 

[Space Tyrant]

Planet 3DNow!
AMD Threadripper ist nur ZP-B1 somit kein teildeaktivierter EPYC (ZP-B2)

Interesting, but I'm not convinced. When Threadrippers show up later with B2 stepping -- and are otherwise indistinguishable from Epyc's -- what will that mean?

1) That TR's *are* based on Epyc -- but the initial run of TR was simply based on an early Epyc (with B1 stepping) run? (but AMD chose not to or never intended to use them for Epyc)

or:

2) That TR's are indeed built from scratch as TR's but we can no longer see a difference?

Either may be the case, but I don't see this as definitive proof.

 

[The Stilt]

The stepping "difference" could be simply artificial as well, to differentiate the two products at least on paper. If the two chips use the same microcode, I'd say it is then quite obvious.

 

[tamz_msc]

The stepping "difference" could be simply artificial as well, to differentiate the two products at least on paper. If the two chips use the same microcode, I'd say it is then quite obvious.

But wasn't this so called B2 stepping supposed to fix some things, according to CPCHardware?

Personally I'd rather see fewer SKUs in the high end, though I've always hoped for a lower-TDP 16 core, just like the 1700.

 

[The Stilt]

But wasn't this so called B2 stepping supposed to fix some things, according to CPCHardware?

Personally I'd rather see fewer SKUs in the high end, though I've always hoped for a lower-TDP 16 core, just like the 1700.

Don't know.
If it has some errata fixes, it makes zero sense to use the old B1 die in Ryzen models (including TR).

 

[dooon]

ALIENWARE Area-51 Threadripper Edition
ZD1840A8UGAF4 : ZP-B1
CPU-Z and AIDA64

source
中关村在线 : 外星人配锐龙!AMD 16核CPU首发简测
HotHardware : Alienware Area-51 Ryzen Threadripper Edition Hands-On With Benchmark Annihilation

 

[Kenmitch]

I guess the case makes it look Alien. Not a fan of the look.

 

[Glo.]

https://videocardz.com/71319/amd-ryzen-threadripper-1900-1900x-1920-and-1950-spotted

What will naysayers say about Bits And Chips credibility if Fottemberg will be right about entry level 16 C Threadripper CPU price? 😉

 

[jpiniero]

https://videocardz.com/71319/amd-ryzen-threadripper-1900-1900x-1920-and-1950-spotted

What will naysayers say about Bits And Chips credibility if Fottemberg will be right about entry level 16 C Threadripper CPU price? 😉

Good guess? It's clear they don't have any legit sources.

 

[Rngwn]

Good guess? It's clear they don't have any legit sources.

Videocardz's citation:
https://www.amd.com/Documents/AMD-Product-Master.pdf

 

[Glo.]

Good guess? It's clear they don't have any legit sources.

They do, actually.

 

[mattiasnyc]

Hmmm.... Well if they release an 8 core chip I can't help but wonder if they shouldn't have included that in the first lineup/release. I'm betting I'm not the only one would rather save some money and get 8 cores but spend to get into x399, and possibly upgrade later. To me PCIe lanes and memory bandwidth is still a potential issue in the long term (on AM4). I mean, I'd expect the lowest clocked 8-core TR to be only a hair more than an 1800x.

 

[IEC]

Hmmm.... Well if they release an 8 core chip I can't help but wonder if they shouldn't have included that in the first lineup/release. I'm betting I'm not the only one would rather save some money and get 8 cores but spend to get into x399, and possibly upgrade later. To me PCIe lanes and memory bandwidth is still a potential issue in the long term (on AM4). I mean, I'd expect the lowest clocked 8-core TR to be only a hair more than an 1800x.

I would be curious to see how performance compares at the same clocks, as well. Considering it will have quad channel memory vs dual channel.

 

[IllogicalGlory]

They do, actually.

How do you know?

Half the things they've tweeted have been far from the mark, like soldered Kaby Lake-X, GP104 hitting 3GHz and price cuts on the 6900K. They were also wrong about high end Skylake dies, claiming a 32 core variant.

 

[GoNavy1776]

Im not understanding here? Is zpb1 worse than b2 stepping? Or are we just trying to see if the chips are built from scratch or failed epyc chips?

I understand that AMD themselves stated that the 2 extra dies that were delidded and discovered by Dr8Bauer or however you spell his game name was in fact spacers to gokd the lid on level to the other functional dies.

Or am I off in lala land somewhere and none of this makes sense.

 

[Glo.]

How do you know?

Half the things they've tweeted have been far from the mark, like soldered Kaby Lake-X, GP104 hitting 3GHz and price cuts on the 6900K. They were also wrong about high end Skylake dies, claiming a 32 core variant.

What about AMD information?

 

[jpiniero]

I understand that AMD themselves stated that the 2 extra dies that were delidded and discovered by Dr8Bauer or however you spell his game name was in fact spacers to gokd the lid on level to the other functional dies.

The ES that he had looked like real dies but it sounds like the final version will have spacers.

 

[maddie]

https://videocardz.com/71319/amd-ryzen-threadripper-1900-1900x-1920-and-1950-spotted

What will naysayers say about Bits And Chips credibility if Fottemberg will be right about entry level 16 C Threadripper CPU price? 😉

They still have some pretty dumb speculation however, at least in my opinion.

"Now I wonder if 8-core Threadripper is going to use only one out of four dies?"

How can you have 4 channel memory and 64 PCIe lanes with one die? Seems to defeat the whole purpose of Threadripper.

 

[dark zero]

The only way that the 8 core exist with those specs is that it can overclock over 5 Ghz, just like the legendary i5 2500K.

 

[IEC]

The only way that the 8 core exist with those specs is that it can overclock over 5 Ghz, just like the legendary i5 2500K.

I give this a 0% chance of happening with ZP-B1 dies under air cooling.

LN2, sure.

 

[Abwx]

They were also wrong about high end Skylake dies, claiming a 32 core variant.

They still can release one, but they ll have to glue two 16C chips to do so....

 

[moinmoin]

Im not understanding here? Is zpb1 worse than b2 stepping?

A new stepping usually contains minor hardware bug fixes. As only Epyc uses B2 stepping the fixes may be related to a part of Zeppelin only used in Epyc chips. To rumor further maybe early Threadripper chips were Epyc chips before they caught the bugs that make B1 stepping unusable for Epyc. Or whatever. Don't think anyone should care aside simple curiosity.

They still have some pretty dumb speculation however, at least in my opinion.

Maybe one good source and plenty of filler stuff out of their @sses.

 

[formulav8]

Z+ could do a PD like update, up to 7% higher IPC and some clock headroom. But honestly, it won't fill up the gap with Intel this coming year with with Zen+ ipc/clocks. Even if Zen+ was LPU process improvments. (Would love to be proven wrong for sure though).

 

[moinmoin]

Do we know for which products AMD will offer Zen+ versions? Raven Ridge seems a given now. Epyc I recall them saying will skip it, with the next version being based on Zen 2. What about Threadripper and iGPU-free Ryzen?