Personal computing discussed

Moderators: Flying Fox, morphine

 
chuckula
Gold subscriber
Gerbil Jedi
Topic Author
Posts: 1941
Joined: Wed Jan 23, 2008 9:18 pm
Location: Probably where I don't belong.

RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 1:09 pm

But you do get Eight... count 'em Eight Coars on 7nm Miracle Process with a 14nm northbridge in there.

https://www.anandtech.com/show/13829/am ... eight-core

And it's a HARD LAUNCH... in Q3.

But don't worry, unlike those evil slimy SOBs at Intel who hate innovation, AMD will emphatically NOT SEGMENT THE INTEGRATED GRAPHICS on these chips whatsoever!
AMD bravely decided that there's no need for integrated graphics since AMD bravely invented the APU. It took A LOT OF COURAGE for them not to slap a Navi GPU with HBM 3 on that package. Oh sure, they could easily have done so and launched it next week, but AMD has COURAGE dammit!

YOU DIE INTEL! YOU DIE SLOW!
4770K @ 4.7 GHz; 32GB DDR3-2133; GTX-1080 sold and back to hipster IGP!; 512GB 840 Pro (2x); Fractal Define XL-R2; NZXT Kraken-X60
--Many thanks to the TR Forum for advice in getting it built.
 
chuckula
Gold subscriber
Gerbil Jedi
Topic Author
Posts: 1941
Joined: Wed Jan 23, 2008 9:18 pm
Location: Probably where I don't belong.

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 1:19 pm

Incidentally, AMD's own internal published benchmark scores put their chip slightly behind the 9900K even in Cinebench.... I'm not expecting miracles if that's what AMD publishes in its own marketing propaganda.

But apparently a 2019 7nm AMD 8 core part can consume somewhat less power than an ancient obsolete Intel design. I look forward to the same people who spent the last decade claiming that power consumption on the desktop doesn't matter fall over themselves to call this a miracle when if finally launches.
4770K @ 4.7 GHz; 32GB DDR3-2133; GTX-1080 sold and back to hipster IGP!; 512GB 840 Pro (2x); Fractal Define XL-R2; NZXT Kraken-X60
--Many thanks to the TR Forum for advice in getting it built.
 
Rand
Silver subscriber
Gerbil
Posts: 54
Joined: Mon Dec 27, 2004 2:34 pm

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 2:00 pm

Not to derail the AMD is the greatest/worst company in the history of humanity but they positioned it as slightly faster then the 9900K, not slower.

Unknown 3rd gen Ryzen, presumably their fastest part
2057 CB score - 133.4 Watts

i9 9900k:
2040 CB score - 179.9 Watts.

You may now return to your regularly scheduled corporation of choice hero worship session.
 
DancinJack
Maximum Gerbil
Posts: 4259
Joined: Sat Nov 25, 2006 3:21 pm
Location: Kansas

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 2:06 pm

Rand wrote:
Not to derail the AMD is the greatest/worst company in the history of humanity but they positioned it as slightly faster then the 9900K, not slower.

Unknown 3rd gen Ryzen, presumably their fastest part
2057 CB score - 133.4 Watts

i9 9900k:
2040 CB score - 179.9 Watts.

Where'd you get those numbers? Anandtech said it was "This new 3rd Generation Ryzen processor scored 2023." Not that I care about Cinebench at all though.
i7 6700K - Z170 - 16GiB DDR4 - GTX 1080 - 512GB SSD - 256GB SSD - 500GB SSD - 3TB HDD- 27" IPS G-sync - Win10 Pro x64 - Ubuntu/Mint x64 :: 2015 13" rMBP Sierra :: Canon EOS 80D/Sony RX100
 
dragontamer5788
Gerbil Team Leader
Posts: 200
Joined: Mon May 06, 2013 8:39 am

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 2:10 pm

Cinebench doesn't care about memory latency. But AMD usually uses Cinebench in these presentations... so that's to be expected.

With that being said, the picture makes it look like 16-cores is possible...

https://images.anandtech.com/doci/13829 ... 78x452.jpg
 
chuckula
Gold subscriber
Gerbil Jedi
Topic Author
Posts: 1941
Joined: Wed Jan 23, 2008 9:18 pm
Location: Probably where I don't belong.

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 2:12 pm

Rand wrote:
Not to derail the AMD is the greatest/worst company in the history of humanity but they positioned it as slightly faster then the 9900K, not slower.

Unknown 3rd gen Ryzen, presumably their fastest part
2057 CB score - 133.4 Watts

i9 9900k:
2040 CB score - 179.9 Watts.

You may now return to your regularly scheduled corporation of choice hero worship session.


Tell ya what: When the RTX-2080 having a greater real-world power advantage delta vs the Vega 64 while having zero process advantage suddenly becomes a huge deal, then maybe just maybe I'll forget about the last decade of "nobody cares about power consumption" that I've heard used to defend AMD.

Oh... and when Icelake ends up being more efficient.. and make no mistake it will be... We get to ask what's wrong with AMD.
4770K @ 4.7 GHz; 32GB DDR3-2133; GTX-1080 sold and back to hipster IGP!; 512GB 840 Pro (2x); Fractal Define XL-R2; NZXT Kraken-X60
--Many thanks to the TR Forum for advice in getting it built.
 
DancinJack
Maximum Gerbil
Posts: 4259
Joined: Sat Nov 25, 2006 3:21 pm
Location: Kansas

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 2:12 pm

dragontamer5788 wrote:
Cinebench doesn't care about memory latency. But AMD usually uses Cinebench in these presentations... so that's to be expected.

With that being said, the picture makes it look like 16-cores is possible...

https://images.anandtech.com/doci/13829 ... 78x452.jpg

Maybe with drastically reduced clocks. I mean, we're already talking 130+W on 8 cores for Cinebench.
i7 6700K - Z170 - 16GiB DDR4 - GTX 1080 - 512GB SSD - 256GB SSD - 500GB SSD - 3TB HDD- 27" IPS G-sync - Win10 Pro x64 - Ubuntu/Mint x64 :: 2015 13" rMBP Sierra :: Canon EOS 80D/Sony RX100
 
dragontamer5788
Gerbil Team Leader
Posts: 200
Joined: Mon May 06, 2013 8:39 am

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 2:15 pm

DancinJack wrote:
dragontamer5788 wrote:
Cinebench doesn't care about memory latency. But AMD usually uses Cinebench in these presentations... so that's to be expected.

With that being said, the picture makes it look like 16-cores is possible...

https://images.anandtech.com/doci/13829 ... 78x452.jpg

Maybe with drastically reduced clocks. I mean, we're already talking 130+W on 8 cores for Cinebench.


It'd be like a Threadripper 2990wx. You up the TDP to crazy-town levels and go ham.

But lets say 180W TDP 16 cores on a normal consumer desktop. Maybe only high end motherboards with high-end VRMs would support it, but that'd basically be what the i9-9900k is anyway (since the i9-9900k can go up to 165+W...)

https://images.anandtech.com/doci/13400 ... er%202.png

There's definitely room in the market for a 180W 16-core Ryzen 3xxx series. Current motherboards aren't ready for it, but it'd be nice to have something cheaper than Threadripper (for the people who don't need all those PCIe lanes or quad-channel memory).
 
Krogoth
Gold subscriber
Gerbil Elder
Posts: 5591
Joined: Tue Apr 15, 2003 3:20 pm
Location: somewhere on Core Prime
Contact:

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 2:16 pm

It is almost what I had expected. The massing space on the packaging looks like a placeholder for future small "Navi" iGPU not a second Rome chiplet.

Zen2-based Ryzens just need to rival Coffee Lake R's performance while consuming less power and not a complete blast furnace at maximum load. It has the potential to be cheaper to make assuming 7nm TSMC process matures and yields improve.
Last edited by Krogoth on Wed Jan 09, 2019 3:17 pm, edited 2 times in total.
Gigabyte Z390 AORUS-PRO Coffee Lake R 9700K, 2x8GiB of G.Skill DDR4-3600, Sapphire RX Vega 64, Corsair CX-750M V2 and Fractal Define R4 (W)
Ivy Bridge 3570K, 2x4GiB of G.Skill RIPSAW DDR3-1600, Gigabyte Z77X-UD3H, Corsair CX-750M V2, and PC-7B
 
Rand
Silver subscriber
Gerbil
Posts: 54
Joined: Mon Dec 27, 2004 2:34 pm

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 2:29 pm

DancinJack wrote:
Rand wrote:
Not to derail the AMD is the greatest/worst company in the history of humanity but they positioned it as slightly faster then the 9900K, not slower.

Unknown 3rd gen Ryzen, presumably their fastest part
2057 CB score - 133.4 Watts

i9 9900k:
2040 CB score - 179.9 Watts.

Where'd you get those numbers? Anandtech said it was "This new 3rd Generation Ryzen processor scored 2023." Not that I care about Cinebench at all though.



AMD's presentation, which was probably a best case scenario so AnandTech's benchmarks are likely more realistic although they do say the AMD chip wasn't running at final clockspeeds which is maybe a tiny bit of hope that the release performance might be better.
Cinebench tends to love Ryzen relative to most software though, so even if final clockspeeds are higher I'm dubious that 8C Zen is going to be much different then the 9900K albeit at less power and presumably lower prices.

A little better then I expected tbh, although those dreaming of 5GHz 16C options are probably awfully disappointed or just praying AMD will add another chiplet and just decided not to mention that.
 
dragontamer5788
Gerbil Team Leader
Posts: 200
Joined: Mon May 06, 2013 8:39 am

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 2:31 pm

Rand wrote:
DancinJack wrote:
Rand wrote:
Not to derail the AMD is the greatest/worst company in the history of humanity but they positioned it as slightly faster then the 9900K, not slower.

Unknown 3rd gen Ryzen, presumably their fastest part
2057 CB score - 133.4 Watts

i9 9900k:
2040 CB score - 179.9 Watts.

Where'd you get those numbers? Anandtech said it was "This new 3rd Generation Ryzen processor scored 2023." Not that I care about Cinebench at all though.



AMD's presentation, which was probably a best case scenario so AnandTech's benchmarks are likely more realistic although they do say the AMD chip wasn't running at final clockspeeds which is maybe a tiny bit of hope that the release performance might be better.
Cinebench tends to love Ryzen relative to most software though, so even if final clockspeeds are higher I'm dubious that 8C Zen is going to be much different then the 9900K albeit at less power and presumably lower prices.

A little better then I expected tbh, although those dreaming of 5GHz 16C options are probably awfully disappointed or just praying AMD will add another chiplet and just decided not to mention that.


Remember that Zen2 will have AVX256-bit pipelines (4x 256-bit pipelines, to be precise). Its still slower than Intel's ideal 512-bit AVX512, but the old argument for SSE-based Cinebench no longer really applies.

The main issue with Cinebench is that the benchmark doesn't care about memory latency or bandwidth. Expect weaker video game experiences with the I/O die adding latency to main memory.
 
Rand
Silver subscriber
Gerbil
Posts: 54
Joined: Mon Dec 27, 2004 2:34 pm

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 2:35 pm

DancinJack wrote:
dragontamer5788 wrote:
Cinebench doesn't care about memory latency. But AMD usually uses Cinebench in these presentations... so that's to be expected.

With that being said, the picture makes it look like 16-cores is possible...

https://images.anandtech.com/doci/13829 ... 78x452.jpg

Maybe with drastically reduced clocks. I mean, we're already talking 130+W on 8 cores for Cinebench.


That's system power consumption though, not CPU alone.
They could presumably double the core count and come in at somewhere close to equal power consumption of the 9900K if they were willing to reduce clockspeed a decent amount. 12 Core at similar clocks to the part they described should be entirely doable in a similar thermal envelope to the 9900K though. I don't think they'd do it, but they did once release the FX-9590 and it's hilarious 220W TDP so who knows.
 
Rand
Silver subscriber
Gerbil
Posts: 54
Joined: Mon Dec 27, 2004 2:34 pm

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 2:42 pm

dragontamer5788 wrote:
Remember that Zen2 will have AVX256-bit pipelines (4x 256-bit pipelines, to be precise). Its still slower than Intel's ideal 512-bit AVX512, but the old argument for SSE-based Cinebench no longer really applies.

The main issue with Cinebench is that the benchmark doesn't care about memory latency or bandwidth. Expect weaker video game experiences with the I/O die adding latency to main memory.


Not sure AVX 512 matters too much, usage of it is still pretty slim and niche. So just getting into a similar range as Intel's Coffee Lake parts for AVX256 should be sufficient for most people.
I'll be interested to see what memory latency turns out to be, they must be doing something to try to minimize the impact. Zen's latency is already rather poor, adding the I/O die on is only going to make it noticeably worse.
 
NTMBK
Gerbil XP
Posts: 317
Joined: Sat Dec 21, 2013 11:21 am

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 2:43 pm

If they don't release a 16-core variant with two chiplets, I'll be amazed.
 
dragontamer5788
Gerbil Team Leader
Posts: 200
Joined: Mon May 06, 2013 8:39 am

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 2:49 pm

Rand wrote:
dragontamer5788 wrote:
Remember that Zen2 will have AVX256-bit pipelines (4x 256-bit pipelines, to be precise). Its still slower than Intel's ideal 512-bit AVX512, but the old argument for SSE-based Cinebench no longer really applies.

The main issue with Cinebench is that the benchmark doesn't care about memory latency or bandwidth. Expect weaker video game experiences with the I/O die adding latency to main memory.


Not sure AVX 512 matters too much, usage of it is still pretty slim and niche. So just getting into a similar range as Intel's Coffee Lake parts for AVX256 should be sufficient for most people.
I'll be interested to see what memory latency turns out to be, they must be doing something to try to minimize the impact. Zen's latency is already rather poor, adding the I/O die on is only going to make it noticeably worse.


Or maybe... Zen's latency is already poor, that adding the I/O die won't be noticeable? :-)

Lol, yeah, wait for benchmarks. Its definitely a weak-point that I'm worried about in the design.
 
Village
Gerbil
Posts: 80
Joined: Sat Dec 13, 2003 5:48 pm

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 2:51 pm

I'm looking forward to my future Ryzen 7 3800X 8c/16t 4.4ghz base + 5.2ghz boost. ;) I would think that the potential future of a 16c/32t chip would likely come with a new X570 chipset to handle the presumed 230 watts required and possibly quad-channel memory. I just can't see dual channel being enough to handle it effectively, but maybe.
 
chuckula
Gold subscriber
Gerbil Jedi
Topic Author
Posts: 1941
Joined: Wed Jan 23, 2008 9:18 pm
Location: Probably where I don't belong.

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 2:55 pm

Given how much larger the I/O chip is than the actual cores, and given some of the rumors surrounding Epyc, I have a controversial theory: I think that to make that chiplet smaller, AMD put the L3 cache on the I/O hub chip and that chiplet is just Cores + L2 caches + the I/O interface circuitry.

We'll see where the L3 ends up being placed.
4770K @ 4.7 GHz; 32GB DDR3-2133; GTX-1080 sold and back to hipster IGP!; 512GB 840 Pro (2x); Fractal Define XL-R2; NZXT Kraken-X60
--Many thanks to the TR Forum for advice in getting it built.
 
DancinJack
Maximum Gerbil
Posts: 4259
Joined: Sat Nov 25, 2006 3:21 pm
Location: Kansas

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 3:03 pm

chuckula wrote:
Given how much larger the I/O chip is than the actual cores, and given some of the rumors surrounding Epyc, I have a controversial theory: I think that to make that chiplet smaller, AMD put the L3 cache on the I/O hub chip and that chiplet is just Cores + L2 caches + the I/O interface circuitry.

We'll see where the L3 ends up being placed.

No way. That would be crazy.
i7 6700K - Z170 - 16GiB DDR4 - GTX 1080 - 512GB SSD - 256GB SSD - 500GB SSD - 3TB HDD- 27" IPS G-sync - Win10 Pro x64 - Ubuntu/Mint x64 :: 2015 13" rMBP Sierra :: Canon EOS 80D/Sony RX100
 
chuckula
Gold subscriber
Gerbil Jedi
Topic Author
Posts: 1941
Joined: Wed Jan 23, 2008 9:18 pm
Location: Probably where I don't belong.

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 3:10 pm

DancinJack wrote:
chuckula wrote:
Given how much larger the I/O chip is than the actual cores, and given some of the rumors surrounding Epyc, I have a controversial theory: I think that to make that chiplet smaller, AMD put the L3 cache on the I/O hub chip and that chiplet is just Cores + L2 caches + the I/O interface circuitry.

We'll see where the L3 ends up being placed.

No way. That would be crazy.


It's less crazy that it sounds when you remember that the primary function of that chiplet is to be used in a server where there are a bunch of other chiplets. Putting the L3 on that I/O hub lets you have a truly *shared* L3 cache for all of the cores that's also sitting right next to the memory controller (remember that thing that's not integrated anymore?) Data that go to and from RAM tend to pass through the L3 anyway, and it's a hell of a lot easier to have one L3 -- even if it's not integrated on die -- rather than 8 different L3 caches.

Now, if this was a part that had been designed specifically to be used in a consumer desktop, putting the L3 in with the rest of the cores would be a no brainer. But then, so would integrating the memory controller and PCIe with the rest of the chip when you don't have to worry about scaling to huge core counts. These parts are server leftovers, and AMD's server design decisions are going to have consequences on the desktop.
4770K @ 4.7 GHz; 32GB DDR3-2133; GTX-1080 sold and back to hipster IGP!; 512GB 840 Pro (2x); Fractal Define XL-R2; NZXT Kraken-X60
--Many thanks to the TR Forum for advice in getting it built.
 
DragonDaddyBear
Gerbil Elite
Posts: 803
Joined: Fri Jan 30, 2009 8:01 am

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 3:15 pm

Given it's still an 8-core chiplet on 7nm that seems like a really big cpu-only die. But having an additional cache outside of the main cores isn't unheard of. Intel did this sort of thing with an L4 in the 5th gen 'C' processors with their Iris Pro. The proposed additional cache could also help serve some feeding of postulated Vega chiplets.
 
derFunkenstein
Gold subscriber
Gerbil God
Posts: 24837
Joined: Fri Feb 21, 2003 9:13 pm
Location: Comin' to you directly from the Mothership

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 3:26 pm

NTMBK wrote:
If they don't release a 16-core variant with two chiplets, I'll be amazed.

Especially the way the package is laid out. It's practically begging for a second chiplet.

https://images.anandtech.com/doci/13829/cpu44.jpg
I do not understand what I do. For what I want to do I do not do, but what I hate I do.
Twittering away the day at @TVsBen
 
Krogoth
Gold subscriber
Gerbil Elder
Posts: 5591
Joined: Tue Apr 15, 2003 3:20 pm
Location: somewhere on Core Prime
Contact:

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 3:29 pm

derFunkenstein wrote:
NTMBK wrote:
If they don't release a 16-core variant with two chiplets, I'll be amazed.

Especially the way the package is laid out. It's practically begging for a second chiplet.

https://images.anandtech.com/doci/13829/cpu44.jpg


I think an IGP is more likely. They are probably waiting until little Navi is finalized.
Gigabyte Z390 AORUS-PRO Coffee Lake R 9700K, 2x8GiB of G.Skill DDR4-3600, Sapphire RX Vega 64, Corsair CX-750M V2 and Fractal Define R4 (W)
Ivy Bridge 3570K, 2x4GiB of G.Skill RIPSAW DDR3-1600, Gigabyte Z77X-UD3H, Corsair CX-750M V2, and PC-7B
 
Krogoth
Gold subscriber
Gerbil Elder
Posts: 5591
Joined: Tue Apr 15, 2003 3:20 pm
Location: somewhere on Core Prime
Contact:

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 3:33 pm

chuckula wrote:
Given how much larger the I/O chip is than the actual cores, and given some of the rumors surrounding Epyc, I have a controversial theory: I think that to make that chiplet smaller, AMD put the L3 cache on the I/O hub chip and that chiplet is just Cores + L2 caches + the I/O interface circuitry.

We'll see where the L3 ends up being placed.


L3 Cache is definitely on the I/O chip. It makes no fiscal sense to create a separate chiplet design for the desktop/laptop market. Zen1-based Ryzens were using Zeppelin chiplets and AMD will continue this practice. The real question is how much L3 cache there is and how it is connect to the CPU.
Last edited by Krogoth on Wed Jan 09, 2019 3:40 pm, edited 1 time in total.
Gigabyte Z390 AORUS-PRO Coffee Lake R 9700K, 2x8GiB of G.Skill DDR4-3600, Sapphire RX Vega 64, Corsair CX-750M V2 and Fractal Define R4 (W)
Ivy Bridge 3570K, 2x4GiB of G.Skill RIPSAW DDR3-1600, Gigabyte Z77X-UD3H, Corsair CX-750M V2, and PC-7B
 
derFunkenstein
Gold subscriber
Gerbil God
Posts: 24837
Joined: Fri Feb 21, 2003 9:13 pm
Location: Comin' to you directly from the Mothership

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 3:34 pm

Krogoth wrote:
I think an IGP is more likely. They are probably waiting until little Navi is finalized.

Maybe, but I'd be surprised if APUs are anything that big. But if that's the case nothing is stopping them from doing both.
I do not understand what I do. For what I want to do I do not do, but what I hate I do.
Twittering away the day at @TVsBen
 
Krogoth
Gold subscriber
Gerbil Elder
Posts: 5591
Joined: Tue Apr 15, 2003 3:20 pm
Location: somewhere on Core Prime
Contact:

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 3:39 pm

derFunkenstein wrote:
Krogoth wrote:
I think an IGP is more likely. They are probably waiting until little Navi is finalized.

Maybe, but I'd be surprised if APUs are anything that big. But if that's the case nothing is stopping them from doing both.


It would address a strategic weakness that AMD has in the OEM world and unify their laptop/desktop line-up. Socket AM4 already struggles with satisfying eight-core Zen1 SKUs. It'll end-up being memory bandwidth starved if AMD goes beyond eight cores.
Gigabyte Z390 AORUS-PRO Coffee Lake R 9700K, 2x8GiB of G.Skill DDR4-3600, Sapphire RX Vega 64, Corsair CX-750M V2 and Fractal Define R4 (W)
Ivy Bridge 3570K, 2x4GiB of G.Skill RIPSAW DDR3-1600, Gigabyte Z77X-UD3H, Corsair CX-750M V2, and PC-7B
 
Rand
Silver subscriber
Gerbil
Posts: 54
Joined: Mon Dec 27, 2004 2:34 pm

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 3:40 pm

Krogoth wrote:
derFunkenstein wrote:
NTMBK wrote:
If they don't release a 16-core variant with two chiplets, I'll be amazed.

Especially the way the package is laid out. It's practically begging for a second chiplet.

https://images.anandtech.com/doci/13829/cpu44.jpg


I think an IGP is more likely. They are probably waiting until little Navi is finalized.


I'd think an APU would be targeting lower price points though as that's what AMD has traditionally done and packaging silicon from two different companies in two different countries together is probably adding a fair bit of expense that wouldn't be ideal for low end parts. So I think any second chiplet would more likely be just packing in extra cores.
 
the
Gerbil Elite
Posts: 935
Joined: Tue Jun 29, 2010 2:26 am

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 3:47 pm

chuckula wrote:
Given how much larger the I/O chip is than the actual cores, and given some of the rumors surrounding Epyc, I have a controversial theory: I think that to make that chiplet smaller, AMD put the L3 cache on the I/O hub chip and that chiplet is just Cores + L2 caches + the I/O interface circuitry.

We'll see where the L3 ends up being placed.


L3 cache is still on the chiplets. What has likely been added to the Epyc IO die is a directory cache for multiple sockets with the possibility of supporting CCIX devices off of the PCIe lanes.

Another interesting question is if AMD put all the display IO logic into the consumer IO die for when they add GPUs to the package.
Dual Opteron 6376, 96 GB DDR3, Asus KGPE-D16, GTX 970
Mac Pro Dual Xeon E5645, 48 GB DDR3, GTX 770
Core i7 3930K@4.2 Ghz, 32 GB DDR3, GA-X79-UP5-Wifi
Core i7 2600K@4.4 Ghz, 16 GB DDR3, GTX 970, GA-X68XP-UD4
 
NTMBK
Gerbil XP
Posts: 317
Joined: Sat Dec 21, 2013 11:21 am

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 3:49 pm

I expected an IGP in the northbridge, kind of surprising they didn't add one. A 3CU GPU to run displays, without wasting a chiplet "socket" on a GPU.
 
Amiga500+
Gerbil
Posts: 63
Joined: Wed Sep 21, 2016 2:10 am

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 4:32 pm

NTMBK wrote:
I expected an IGP in the northbridge, kind of surprising they didn't add one. A 3CU GPU to run displays, without wasting a chiplet "socket" on a GPU.


Would have been very nice - always useful to have a basic backup display head to get you up and running if/when your main GPU dies. But I suppose its too niche to be sensible.
 
chuckula
Gold subscriber
Gerbil Jedi
Topic Author
Posts: 1941
Joined: Wed Jan 23, 2008 9:18 pm
Location: Probably where I don't belong.

Re: RyZen 2 Desktop CONFIRMED... to not have 16 cores

Wed Jan 09, 2019 4:42 pm

NTMBK wrote:
I expected an IGP in the northbridge, kind of surprising they didn't add one. A 3CU GPU to run displays, without wasting a chiplet "socket" on a GPU.


I expect a future quasi-IGP in the empty spacer that's being conjectured to have the other 8 cores.
4770K @ 4.7 GHz; 32GB DDR3-2133; GTX-1080 sold and back to hipster IGP!; 512GB 840 Pro (2x); Fractal Define XL-R2; NZXT Kraken-X60
--Many thanks to the TR Forum for advice in getting it built.

Who is online

Users browsing this forum: No registered users and 2 guests