AMD's Ryzen CPUs (Ryzen/TR/Epyc) & Vega/Polaris/Navi GPUs

https://www.kitguru.net/components/...000-will-offer-significant-performance-gains/
Rumours suggest AMD Ryzen 4000 will offer significant performance gains

More rumours.
However, this time, it seems that they are stating Zen 3 will still be on AM4 (not AM5) socket.


EDIT:
https://www.techcenturion.com/amd-zen-3

This article clearly states the following:
AM4 Support & Codenames
AMD’s CEO, Dr. Lisa Su, has mentioned it quite a few times about their commitment to the AM4 Platform. Here’s an official slide that confirms that AMD’s 2020 Desktop CPU will be on AM4.



So, from official AMD slides, Zen 3 is (for now) still slated to be supported by AM4 (which is of course subject to change).
Where does WccfTech even get its info?

 

Zen 3 might stick around long enough to be on both AM4 and AM5, that's one possibility.

AMD has done this before at the transition between DDR2 and DDR3 with AM3 supporting both DDR2 and DDR3:

"The AM3 socket has 941 pin contacts in a different layout while AM2+ processors have 940 pins. ... It is likely because the built-in memory controller in AM2/AM2+ processors only supports DDR2 (unlike AM3 processors, which support both DDR2 & DDR3 memory)."
https://en.wikipedia.org/wiki/Socket_AM3

Zen 3 could support both DDR4/AM4 and DDR5/AM5 CPU's - unknown is whether AM5 will have a different or more pinout than AM4.

It's still unknown whether we will have DDR5 generally available in 2020, as we were supposed to have it originally in 2018:

"DDR5 Release Date
Back in March of 2017 JEDEC announced that DDR5 was being worked on and should be released in 2018. Jumping forward to November 2018, SK Hynix finally announced the first ever DDR5 compliant RAM module, which they in turn said would be available to consumers come 2020.Dec 20, 2019"
https://www.wepc.com/reviews/ddr5-latest/

Also, will we have 5nm in time for the DDR5 push from AMD - if so then I would assume Zen 4 will be DDR5, but if 5nm isn't available for mass production in time to compete with Intel's DDR5 release then AMD could possibly extend Zen 3 use to match Intel.

"TSMC's 5nm technology is the second available EUV process technology. It showed promising imaging capability with expected good wafer yield. Our 5nm technology entered risk production in March 2019 and target for volume production in 2020."
https://www.tsmc.com/english/dedicatedFoundry/technology/5nm.htm

So far TSMC seems confident that 5nm thus Zen 4 will be available in 2020.

If everything aligns we might get Zen 4 + 5nm + DDR5 + PCIE 5 end of 2020 / beginning of 2021.

Otherwise AMD could extend Zen 3 support onto AM5 socket + DDR5 + PCIE 5 (or remain on PCIE 4?) until enough 5nm production is available... perhaps 5nm will be as constrained as 7nm initially and AMD will have to choose what new products get 5nm.

Intel's DDR5 / PCIE 5 timing could affect AMD's schedule as well, if AMD want's to match Intel's support of DDR5 / PCIE 5.

It seems we need to re-calibrate our expectations of how long our investments in hardware will be in vogue. AM4 3950x and TRX40 3970x while awesome today might play out their top status in weeks and months rather than the historical 12-18 months. Yes, I know TRX80 will kick in quicker than that, but that's in the same family as well.

We could see an AM4 Zen 3 upgrade in a short time that would allow for an upgrade in 2020, with a quintuple technology bump of DDR5 + PCIE 5 + Zen 4 + 5nm + AM5 within 6-8 months - also in 2020!

Wild.

 

5nm seems to be well on track... in fact, it was stated by TSMC that yields have exceeded their expectations and that currently, the yields are above 50% (better than 7nm)... so there's virtually no doubt that Zen 4 will be on time - or it could be released earlier (but this depends on AMD entirely given that they said Zen 3 won't be released until late 2020 - which seems a bit late if you ask me).

As for compatibility, you could be correct in saying that Zen 3 could be split between AM4 and AM5 - depending on how well it competes with Intel and AMD release schedule... but at this point, I don't think AMD will have too much to worry about Intel competing - that is, if AMD continues existing 'onslaught' of regular releases every year (but the Zen 3 uArch seems to be 'delayed' in regards to release cycle to end of 2020, as historically, we had AMD releasing new iterations in early/mid year).

Unless AMD plans on Zen 3 to be short lived? But that doesn't make too much sense.
It could be they are just moving the release schedule a bit.

 

Wow, speculations, rumors and outright guesses. Be smart and build what is available to build.

 

That's an approach people will take anyway but they might be swayed into waiting a bit if new tech is around the corner and they're not in any particular hurry.

Plus why have debate forums if you wouldn't talk about speculation and data that was posted by companies who release the he in question?

 

Everyone has known AM4 was to 2020, which includes Zen 3. Nothing new here.

You ranted about not speculating them speculate like crazy here. AM5 is in 2021. It requires a new socket. Compatibility chances with any prior gen to Zen 4 is extremely low, especially for how long the socket lived. Stop trying to feed the bull that maybe zen 3 will be compatible. We saw what they did to TR. There is a chance of needing more pins because AMD said core count increases are not ending, meaning there will likely NOT be socket compatibility. This isn't an AM3+ situation.

This isn't that simple. Anandtech has an article from the iedm conference where they discussed yields. Basically, if AMD did a direct die shrink, no changes, they could have over around 55% yield or more, which in raw number of usable dies is more than the current 7nm process. But, the defect density is really high, whereas 7nm defect rate is extremely low.

Hell, if AMD only did around a 34% die shrink, they would have over 50% yields on 5nm right now. Think of where that will be when used in 2021!

So there is some nuance that needs discussed. Also, there was a rumor that 7nm+ yields dove below 80% a couple months back, but no explanation on why given.

Really, you should take context clues. Hmscott was berating purchasers of the HEDT as speculating compatibility but here is engaged in the same frivolous behavior. That was what was being pointed out.

 

I no longer trust what comes out of their mouths. Nor do I trust those who say do not listen to speculation etc. but then give it out.

 

Here's a nice overview by BuildZoid of the TRX40 motherboards:

Round-Up: AMD Threadripper 3 TRX40 Motherboards for 3970X & 3960X
Dec 22, 2019
Gamers Nexus
This motherboard round-up looks at the AMD TRX40 boards for Threadripper 3 CPUs (3960X, 3970X), talking about what makes the best Threadripper motherboard. In this video, we'll be talking about the AMD TRX40 motherboards for Threadripper 3 CPUs, like the Threadripper 3960X and 3970X. We'll talk about the best motherboards for Threadripper in 2019 (and 2020), but also cover them from angles of workstation use, multi-GPU use, I/O or SSD-heavy use, overclocking usages, and more. We're also talking about the budget or "low-end" AMD TRX40 boards, so it's not just the high-end stuff.
01:34 | ASUS ROG Zenith II Extrem
04:15 | Gigabyte TRX40 Aorus Xtreme
06:03 | MSI Creator TRX40
08:35 | Gigabyte TRX40 Designare
11:52 | ROG Strix TRX40-E Gaming
15:58 | Gigabyte TRX40 Aorus Master
17:30 | ASRock TRX40 Taichi
21:30 | MSI TRX40 PRO10G
23:42 | MSI TRX40 Pro WiFi
25:14 | ASRock TRX40 Creator
29:40 | ASUS Prime TRX40-Pro
32:25 | Gigabyte TRX40 Aorus Pro WiFi

 

der8auer and Hardware Unboxed take a look at the Powercolor 5700XT Liquid Devil

A closer look at the MOST EXPENSIVE RX 5700 XT - PowerColor Liquid Devil for 600 €
Dec 25, 2019
der8auer


PowerColor RX 5700 XT Liquid Devil, The Fastest 5700 XT Money Can Buy
Dec 19, 2019
Hardware Unboxed


Prophecy by it's very nature is hit and miss - AdoredTV

AMD vs Intel - The Past, The Present and Near Future
Dec 17, 2019
AdoredTV
A look at AMD vs Intel from a non-political perspective


PC Hardware Loose Ends: December 2019 [END OF YEAR!]
Streamed live 22 hours ago
Moore's Law Is Dead
A lot has happened in 2019 when it comes to PC Hardware, join me for a live Q&A!
1) 2:20 Vega's Future, APU's in 2020, & AM4+
2) 13:00 OLED Owners advice
3) 17:00 E-Waste, and the the slowing of upgrades
4) 24:00 3950X Update 5) 29:40 Ice Lake vs Zen+
6) 35:00 A giant thank you to the fans of MLID
7) 45:00 Beginning of Q&A
8) 1:11:10 Big Navi Strategies
9) 1:30:00 Waiting for Zen 3, and GF 7nm?
10) 1:45:00 Doggo Cameo


Mike L 15 hours ago
"I found my that 3950x while running a power bug type all core load runs 15 deg C cooler than my 3900x while running the same load."

Moore's Law Is Dead 19 hours ago
"That's what I am saying. The 3950X is legitimately a tier above the 3900X. It's not just the cores."

@ajc9988 @TANWare both suffered a big surprise that their pages and pages (dozens of pages) of speculation didn't prevent them from suffering.

They both got caught short by AMD dropping x399 and even though there is no evidence to the contrary they believe AMD owed them either notice that x399 was ending or AMD owes them one more generation of x399 CPU's.
Neither of them appear to have the maturity to deal with AMD's decisions without lashing out against others.

They still haven't been able to deal with it. I can only imagine we'll be hearing them kvetch about AMD and how they can't trust them for years to come.

Of course AMD did what they did for technical reasons and couldn't pre-announce the change due to business reasons, but they won't accept that and feel personally slighted.

I put them both on ignore for a week or two when they start up with that BS again, like now.

 

We just called you out for the same action. Own your crap! Really, it's like you privately calling us Intel shills when you are literally an AMD promoter to a fault.

So when you make the suggestion that AMD will have socket compatibility with Zen 3 chips on AM5 on the basis of NOTHING, not even them having compatibility for other lines, you are the one being irresponsible here.

I even pointed out you encouraging the idea there might be x399 socket compatibility AFTER leaks started emerging showing no compatibility.

So don't come with that trash!

 

There were plenty of rumors suggesting x399 and whitehaven compliance for TR2, very selective memory indeed. Also as mentioned calling people out.

 

SFF 3950x... because, why not? Interesting components.

3950X Mini ITX Build...! (Z-Case P50 Review)
Dec 29, 2019
Tech YES City
Ryzen 9 3950X, air cooler, RTX 2080 Super all in a Mini-ITX Gaming PC...?! Well if you are a small form factor lover, then this is a new case called the Z-Case P50, which retails for..... $199 USD and the required Flex ATX 600W Power Supply will cost an additional $109 USD.... So at these prices this case has a lot to live up to, and today we put the biggest and beastliest hardware inside to stress it to the max.

Spoiler: Components List:

Components List: Ryzen 9 3950X - https://www.amazon.com/s?k=ryzen+9+3950x
32GB DDR4 Corsair Dominator - https://www.amazon.com/Corsair-Dominator-Platinum-2x16GB-PC4-27700/dp/B07PJXC8JS
Noctua L12S Cooler -
https://www.amazon.com/s?k=L12S+cooler
ASRock X570 TB3 Mini iTX (I had to bend the cooler both sideways and upwards to make this fit, so maybe get low profile ram etc. if you want to do it differently lol) - https://www.amazon.com/s?k=asrock+x570+itx
1TB NVMe m.2 - https://www.amazon.com/s?k=1tb+m.2
RTX 2080 Super - https://www.amazon.com/NVIDIA-GeForce-Super-Founders-Graphics/dp/B07W3P4PC2

Check out the Z-case P50 here - https://www.sfftec.com/product-page/z...
Flex ATX Power Supply - https://www.sfftec.com/product-page/e...

Happy New Year!!

 

Zen 3 can either have a very short run or an extended run due to a combination of competition or supply issues with TSMC.

AMD has done a good job meeting their own schedules, along with TSMC coming through in time for AMD production releases, but not everything is on the schedule: x399 support ending for one.

If AMD follows their current plan Zen 3 will be shortly followed by Zen 4, and that seems like too quick of a tempo - but then again AMD doesn't need to wait for anything if the TSMC planned process production can keep up and if Intel pulls a rabbit out of the hat AMD's schedule will deliver the best competition against Intel they can offer as quickly as possible.

AMD also needs to consider their customers ability to absorb change quickly without a disturbing value decline due to rapid growth. Such as already experienced with the end of x399 support.

If the rumors are true AMD plans on delivering 12nm replacements for original Ryzen 14nm parts which would continue to fill the market with an ever increasing range of Ryzen CPU options. To me, that seems to be more than the market needs given Zen 2, Zen 3, Zen 4 are being delivered in such rapid cadence according to AMD's published schedule.

Perhaps AMD thinks 12nm Ryzen 1/1+ parts will fill the low end giving protection for the Zen 2, Zen 3, Zen 4 parts as time moves forward. AMD seems to like to keep around previous generations for a long time - with some exceptions - and that can keep pressure off of the new process production by filling needs with previous generations.

Too many unknowns to accurately speculate, but there sure are a lot of options available to AMD to fill the market with Ryzen at low cost and high performance to wipe out Intel's market share - even below Intel's target of 30%, down from 90%.

Then there's the laptop market that Intel is clutching to it's chest. Will AMD get a shot at filling that market with APU's and mobile CPU's for the masses in 2020?

2020 could be an amazing year for PC builds, at least for AMD.

 

The Asrock TRX40 Creator is the lowest cost TRX40 MB with 10Gbit Ethernet + 3x M.2's (2 gen4 NVME M.2 + 1 Sata M.2)...the most expensive ASUS ROG Zenith II Extreme has 5x M.2's, but it's a big jump in price - about $500 more! I going to have to see more testing on the 3970x / 3960x to see if it's got what it takes to run fully loaded for extended periods.
https://www.asrock.com/mb/AMD/TRX40 Creator/index.asp#Specification

Asrock TRX40 Creator Motherboard Review!
Dec 31, 2019
Level1Techs


TRX40 AORUS Master Vs. ASRock TRX40 Creator - Bigger is NOT Always Better.
Nov 30, 2019
Tech YES City
Two 'Entry Level' Threadripper Zen 2 TRX40 Motherboards are up on the test bench today, with a deep dive into the ALC4050H Realtek Codec and how AMD have bridged the I2S to allow for chipset audio control via USB 2 vs traditional CPU connection. We also test the 16 phase VRM vs the 8 Phase VRM on the Creator TRX40... and well let's just say, that bigger isn't always better .


Funkykit misread the PCIE 4.0 layout for the 4 slots, there are not 4 16x, there are only 2, and 2 8x slots:

There's Only One True Creator - ASRock TRX40 Creator
Premiered Dec 18, 2019
Funkykit
Today, we'll be taking a quick look at the ASRock TRX40 Creator motherboard. It features AMD's TRX40 chipset, and suitable for all the latest Ryzen Threadripper 3000-series processors. We've already reviewed ASRock's other model, the TRX40 Taichi, which we thought was an excellent motherboard that offered a ton of features and excellent performance. And we can expect the same with the TRX40 Creator version, except this time it will have extra features that's dedicated more for content creators. The ASRock TRX40 Creator is primarily aimed and targeted for content creators, and offers some interesting features such as Aquantia 10G LAN, Drangon 2.5G LAN, Wifi 6 (802.11ax), and 3 x Hyper M.2 slots for SSDs (PCIe Gen4x4 & SATA3).


Simo Koivukoski 1 week ago
"Two of those PCIe 4.0 x16 slots have only PCI Express x8 lanes. So PCIe slot lanes are x16 x8 x16 x8."

Asrock TRX40 Creator - $459 from newegg and newegg ebay:
https://www.newegg.com/p/N82E16813157902
https://www.ebay.com/itm/ASRock-TRX...0-SATA-6Gb-s-ATX-AMD-Motherboard/383305620872

Asrock TRX40 Taichi - $499 from newegg and newegg ebay:
https://www.newegg.com/p/N82E16813157903
https://www.ebay.com/itm/ASRock-TRX...0-SATA-6Gb-s-ATX-AMD-Motherboard/303395288794

https://www.newegg.com/Product/Productcompare?CompareItemList=13-157-902,13-157-903

The Taichi has some improvements but loses the 10Gbit ethernet so that's a big drawback to the Taichi for me.

Nice chart results comparison of the Creator vs Taichi from Tech YES City:

TRX40 ASRock Taichi Review - The ULTIMATE Threadripper Solution...?
Dec 11, 2019
Tech YES City
Is this 'the one'? Well... besides the steep price tag, yes. It is a motherboard that has a massive 550 gram heatsink with 2 active 37mm fans, which makes the VRM very cool even whilst on full load. Also the aethestic, bios, onboard audio and everything else checks out on this board. Though there are a few minor (or major) differences between this and other boards, so if you are keen to purchase a TRX40 board, then this review is for you.

 

We know that AMD has a 'departure agreement' with GLOFO which says they must continue making some orders which would be made on GLOFO node until 2020-2021.
This likely means that I/O production will continue on 14-12nm node until Zen 3 or Zen 4.
The Ryzen 1600 on 12nm is likely another way to fill out those orders from GLOFO - which might be followed by Zen+ cheaper products (although I have no way of verifying this).

There's too big of a gap between Zen 2 and Zen 3. AMD for some reason delayed Zen 3 release to late 2020 (or roughly 6 months compared to previous releases), this could change of course, but right now, we don't have reason to think it will.

I think there's been a recent interview with an AMD official who mentioned a release cycle of 12-18 months.
Looks like Zen 3 fell onto the 18 months release cycle.

 

Bull. AMD did not release the initial zen 2 chips until July. Even assuming a Q4 release, that is a 16 month cadence, as I doubt they would launch passed October or early November.

Further, for Zen 1, we saw a march release, followed by April for the bulk of mainstream, then a couple lower in may, iirc. For Zen+, it was April/may time frame. So let's just say 14 months. Then July for Zen 2, or 14 months. So, late Q3 would be another 14 month cadence. By my calculation, that is NOT 6 months, as that would be a 2021 release. Just saying.

Moreover, the reason for delay was likely two things: 1) TSMC was rumored to have a yield dip on 7nm+, and 2) purchasing fab time and trying to figure out the increased demand.

For the first point, this was reported a couple months ago, that the yields on the new plus variant fell below 80%. No more information was given. That could be a chemical issue (see last summer where a bad batch of chemicals scrapped wafers), could be a technical issue that cropped up, etc. We just don't know. But that could be a contributor.

For the second point, we all know about TSMC going from a 2 month to a 6 month lead time. Many are jumping on 7nm. Nvidia even said the bulk of their GPUs will now be TSMC 7nm. So, AMD has to now place larger orders, give the last Zen2 chips enough time to clear inventory, and do a larger silicon buildup because they know demand is growing. That is in addition to a larger 505mm GPU die for Navi 21, which will eat up wafers. Also, they have to pump out the semi-custom silicon for two consoles, both releasing during 2020. Talk about a big year, lots of orders, and thereby lots of fab time. I'm sure TSMC is loving the extra work for AMD, which is showing how good their process is.

So why late 2020, especially when talking about Nvidia possibly being delayed to late 2020 on GPUs? Buildup. They will have to deliver the console chips far enough ahead that the systems can be built. Granted those are on the 7nm rather than plus lines, but they will also still have to crank out Zen 2 and Navi 10 GPUs during all year. Also, 7nm+ has fewer lines to do it than 7nm, so they may not be able to get enough chips until then or get on those lines until then.

There is a lot of speculation there, but embedded is lots of facts. That is why there may be an extra wait.

 

Which they never promised any compatibility for -- their 2020 promise was specifically about AM4.

 

Not at first but they did eventually start just mentioning AM4, purposely not mentioning TR4 at all. That is where the problem lies, AMD knew TR4 was being exluded but intentionally did not mention it to the enthusiasts. I just doo not trust them.

 

Are you daft? I literally was scolding hmscott for saying AM5 could be compatible with Zen 4 2021 chips, and you respond to me with this instead of hmscott? Really?

Also, TANWare has a point. They started by saying Ryzen, which would include threadripper, in February. By summer of 2017, they changed to saying AM4.

Here, though, you miss all context of hmscott lying about compatibility, then try acting like I'm misleading, which if you read my darn comments, you'd get I was preventing hmscott from causing any sort of confusion.

That's why I have a problem with you. You shift the conversation to fit your predetermined narrative, including focusing on tangential points, rather than staying with the main point of discussion. If you have ADHD, I do apologize. I have friends with that and it can be tough to handle sometimes. It is not a knock on your intelligence. It is a knock on focus.

If you want to rehash old conversations from new perspectives, I'm all for it. If you wasn't to explore tangents, I'm for that. But explain you are going down a rabbit hole up front.

We had a similar issue in Ryzen v Intel thread yesterday. I made my point very clear why frequency still matters for Epyc servers being used to replace systems in one's house. You, instead, tried to twist the meaning of what I was saying and why I was saying it and focused on tangential points, as if those were not in some sense merely parts of my overall discussion and part of my overall statement. I clearly said a couple years it would be better, then moved to why current Zen 2 wouldn't be enough yet, then even gave estimated Zen 3 IPS estimates and the speeds needed to get roughly what was seen on Zen 2 mainstream, yet you harped on IPS being used, which I addressed, and brought up BS counterpoints regarding leaving OCing headroom on the table when I specifically at that moment was discussing gaming performance on that setup and speeds to match, roughly, mainstream CPUs and gaming performance.

So just stop or ask directly what I'm saying if you missed it, because I write volumes when speaking. Otherwise, I'll just go with my new plan of ignoring you as you miss the very point I was making, then quote me out of context to try to make yourself seem on point.

 

That's the facts, jack

AMD never mentioned TR4 as having any specific duration or generations of upgrades moving forward.

With HEDT performance is paramount, so if there is a possibility to improve performance with a platform upgrade then the vendor moves ahead with the platform upgrade - even if that upgrade isn't backwards compatible.

Consumer focused products (AM4) are different in that the vendors try to have some semblance of long term stability. Performance is important, but the budget of the individual is harder hit by breaks in compatibility moving forward.

HEDT is generally bought for specific projects or needs that require higher than desktop computer performance and will be kept on task until depreciated - and then they are either kept on task until their service is no longer needed, or they are recycled and replaced with the next fastest - or more cost effective product.

HEDT isn't designed to be an ePeen boost for PC Master Race contenders wanting to impress their friends with their over the top benchmarks - because that won't last.

HEDT is very costly as over time the platform will be rapidly obsoleted into a non-compatible upgrade often requiring a completely new build.

AMD will continue to do this rapid obsoletion or fragmentation of their HEDT platform(s) again and again whenever possible to keep improving throughput and introduce innovation after innovation.

As will Intel, whenever Intel gets their pile together and points it in a forward direction successfully. Intel will obsolete their HEDT products sooner than later if at all possible, and have done so many times in the past.

Progress, what a concept.

If you want to play in the big leagues, you better bring a bigger stack of cash, and even better - have a *real* need that will turn a profit to fund the inevitable constant stream of platform upgrades. That way you won't have to moan and grouch over costly platform upgrades, instead you'll be excited to integrate the new technology into your projects.

And, @ajc9988 / @TANWare please quit embarrassing yourselves by dragging on and on with your whining, complaining, and rudely lashing out at others - it's silly and pointless - AMD did the right thing, get over it.

 

Fork you!

Literally, you were telling a person that Zen 3 would be compatible with AM5. Now shut it. You are trying to change topics when you are guilty of what you say we are complaining about!

 

Here is you literally telling people that Zen 3 could be on BOTH AM4 and AM5. Did AMD say that? Did they promise that? You are saying we are complaining because AMD did not promise to have TR4 compatibility. Here, you are telling people something THAT AMD HAS NOT SAID, then literally building up a potential, all so you can later say AMD never said they would do it.

This is why we are yelling and screaming at you, then you turn it into something else to distract from YOU SAYING TO A PERSON THERE WILL BE COMPATIBILITY BASED ON NOTHING.

So when we tell you shut up, you are embarrassing yourself by continuing!

 

That's what speculation is, or didn't you know that? We were discussing between ourselves - not including you btw - until you butted in for no good reason and started berating both of us.

You go on about some BS or another page after page, speculating to no end - things that haven't been said or suggested, to no end - so much so I've complained about it in the past.

A little goes a long way. Learn to reserve your wild speculations to occasional musings, instead of posting page after page of things that haven't happened and won't happen, things that haven't been announced or likely won't be announced.

Once in a while is fine, we all get the urge from time to time. Like a week before a conference where vendors might announce or give a sneak peek at what is down the road.

What irks people is when you are spewing forth daily, not posting new things or thoughts but repeating the same old BS multiple times a day.

What ticks me off is when you lash out at people that don't agree with you and have a different opinion, idea, or theory. That's when I start ignoring you.

Trying keeping a civil demeanor, kindness goes a long way, even if you don't agree with someone let them express themselves. And if you must disagree do it nicely without spite and malice in your tone.

Please take a break, maybe enjoy some holiday cheer, and then come back and try to be nice to everyone.

 

That's you jumping to conclusions again. I didn't say that Zen 3 would be compatible with AM5, here is what I did say:

" Zen 3 might stick around long enough to be on both AM4 and AM5, that's one possibility."

I then went on to show the elements that make up Zen 3 and Zen 4 and how the elements for the AM4 to AM5 transition might occur.

We don't know whether 7nm EUV or 5nm will arrive on time in enough production, and like 7nm if AMD is thinking about keeping around older products with updated processes (14nm => 12nm) then who knows how AMD will mix processes, architectures, and AM4 upgrades to AM5.

Then I discussed the AM2+ to AM3 migration with AM3 supporting DDR2 and DDR3.

All historical elements to consider for future possibilities.

In no way did I say any of those things were going to come to pass, I discussed them all as possibilities.

It's you that wants to cause problems that is trying to put words in my mouth and ascribe unintended intentions to what I said far beyond what I actually said.

What's up with you anyway? Don't you have the joy of the holiday season in you? Looking back and looking forward is in integral part of New Years.

Lighten up, drink some nog, and be happy.

 

I have yet to embarrass myself nor lash out, You need to look in the mirror for that. Since you can not possibly, without being a shill, know what TR2 would have been like on the x399 for comparison you can not say AMD did the right thing ! Nuff said!

 

CES 2020 might be fun

AMD Announces Press Conference at CES 2020
"SANTA CLARA, Calif., Dec. 31, 2019 (GLOBE NEWSWIRE) -- AMD (NASDAQ: AMD) today announced that it will hold a press conference as a part of the official CES 2020 Media Days, and hosted by AMD President and CEO Dr. Lisa Su. The event is scheduled for Monday, Jan. 6 at 2:00 p.m. PST at the Mandalay Bay.

A live stream of the event will be available on the AMD YouTube channel. An archived version of the webcast will be available approximately two hours after the event and can be found on the AMD YouTube channel."
http://ir.amd.com/news-releases/news-release-details/amd-announces-press-conference-ces-2020

AMD Announces Press Conference at CES 2020 | Advanced Micro Devices
https://www.reddit.com/r/Amd/comments/eig1t3/amd_announces_press_conference_at_ces_2020/

" AMD‏ Verified account @AMD
At #CES2020, AMD will push the envelope yet again to make 2020 an incredible year for high-performance computing. Join us for our press conference on Monday, January 6th at 2PM PT on YouTube!" 1:13 PM - 31 Dec 2019

https://twitter.com/AMD/status/1212119569487351808

I thought Intel wanted 30% share of the desktop market? It seems that's already lost as well, Intel's down to 14% in December...

Mindfactory Report December 2019
https://imgur.com/a/Xk2E763#RqoPsmR




AMD closes the year with a new record high @ mindfactory.de
https://www.reddit.com/r/Amd/comments/eilijx/amd_closes_the_year_with_a_new_record_high/

Lisa's Collection of Ryzen + more...

Lisa Su‏ Verified account @LisaSu
"Had fun organizing my collection of goodies. So proud of all of these products. Even more excited about what we have in store for 2020! Happy New Year from all of us at @AMD!!" 7:38 AM - 1 Jan 2020

 

Oh Boohoo.

Come on, at least I base my speculation on something and cite sources. You do not half the time, especially when it comes to your bull on socket compatibility.

What I'm saying is keep such bull to yourself. You even promoted TR compatibility when rumors already said no compatibility, then point to the Osborne effect to justify it. You promote AMD GPUs when it is empirically worse, etc.

People consider you a shill. Period.

So you don't have to like it, but you need to understand it.

 

First that is DIY market, OEM's are needed. Second, way too much hype and speculation.

 

And no, Intel said 30% overall of all silicon. That means an average market share stretched over CPUs, GPUs, flash, etc. And Intel's market share IS at around 82% for desktop according to Mercury, one of the analyzers of market share.

Even with current sales, until people deprecate older hardware, Intel's market share will remain high.

 

The $85 12nm Ryzen 1600AF

What will the future bring if AMD is rendering their older architecture's on new processes? Will Zen 2 find it's way onto 7nm EUV process or even 5nm? Same for Zen 3 / Zen 4, will they also get the uplift of newer processes than they were originally released?

New AMD Ryzen 5 1600 AF CPU Review & Benchmarks vs. Original, R5 2600, & 3600
Jan 2, 2020
Gamers Nexus
In this review, we're looking at AMD's new Ryzen 5 1600 AF processor, which is a 2nd Gen Ryzen CPU (with 1st Gen naming) to replace the R5 1600. AMD somewhat silently released a new CPU -- the AMD Ryzen 5 1600 AF -- which is technically a better R5 1600, but with the same name. It's a 12nm part rather than 14nm, something we confirmed with AMD, and that brings with it benefits for power efficiency, overclocking headroom, and even stock performance (in a significant fashion). In this benchmark, we're looking at the AMD R5 1600 "AE" (the original) versus the R5 1600 AF, the R5 2600 (which we need to do some refreshed testing on once we're back from CES, but it hasn't changed much), and the R5 3600. The R5 1600 AF impressed us, especially at the budget CPU price, and has established itself as one of the best budget CPUs for gaming or light production workloads.

 

Detailed facts quoting sales quantities and product dollars sold month after month from vendors is hype and speculation?

Even Intel is saying their gross margin is dropping. And installed base isn't market share, market share is what is sold each quarter or month or defined period of time. The sales of previous months or quarters - the installed base - don't count for market share in new months or quarters numbers. That's how that works.

 

So you can't avoid being rude to people when discussing things here on NBR? I've seen you do this to everyone, now it's my turn?

I've easily quoted far more factual references than you have here on NBR. Almost all of my posts include links and quotes.

You are fooling yourself if you think you are backing up what you say and I am not. Check your own posts going back - you ramble on for pages with no links or quotes.

I quoted the Osborne effect to show why no manufacturer wants to pre-announce future product changes in advance because it might kill current sales.

I'm an AMD fan, I don't work for AMD or benefit from their rapid rise to success. I also don't benefit from Intel's failure to perform or their rapid loss of market share. I am not a shill for anyone or anything.

Like so many others, even though I own and support Intel and Nvidia products, I don't like the companies and what they do, and just like millions of others I complain about what they do, while most others mostly remain silent - "what can I do about it?", many of us don't remain silent.

That's what I am doing, not remaining silent about the BS Intel and Nvidia pull.

" So you don't have to like it, but you need to understand it." - That's why I've just explained it to you, again.

Past and future Intel failures are and will be due to what I and so many other people have complained about for many years - that Intel has been a poor steward of their responsibilities for the technology they shepard, and Intel's nefarious actions driven by greed will do them in.

It's happening now, Intel is paying the price for their greed. Likely Nvidia will continue on their path following Intel's example and end up the same way, unless they turn their greed around and start treating the people their serve right, their customers.

If you two ( @ajc9988 @TANWare ) choose to keep trashing AMD while proclaiming how glorious Intel still is - even as Intel falls further and further behind, that's up to you.

I'm not buying what you are selling

 

Wtf are you talking about? I have literally said only two CPUs are currently worth buying under very specific circumstances. You make the mistake of not understanding criticizing AMD, or a shill for AMD, does NOT mean it is an endorsement of it's competitor! And you feel giving honest analysis of when the 10980XE and 9900K are worth it, backed by the tech press having similar analysis, is an attack on AMD. The only people that think that are either paid by a troll farm hired by AMD OR blinded fanboys.

You can't even understand that a critique of switching sockets on TR products is not an endorsement of Intel, rather a critique of business strategy. Any analysis further shows that the 2970WX and 2990WX are both inadequate for many workloads, shown empirically since their release. Instead, you pushed for people to buy those or but mainstream AFTER people said the PCIe lanes are insufficient and memory bandwidth is as well. That is telling people to ignore their needs and buy AMD. That is shill behavior.

You have seen me post numerous analyses while citing sources. All you do is post pro-AMD pieces and go "see, buy AMD."

You sure as hell loved my analyses when they benefited AMD over the past couple years. And as soon as I get critical of them, then you act like I'm an Intel shill. NEWSFLASH: I am critical of ALL companies!

So get over yourself. You are using standard subversion techniques to try to recast to the talking points you want.

Also, CITE YOUR SOURCE ON YOUR SOCKET COMPATIBILITY STATEMENT!!! You can't because it is a reckless statement!

 

Then AMD is at 5% server, 18% desktop, and 14% mobile under how you defined it. That means you are lying to people about market share and Zen 2 did not really significantly change market share. I already posted the link to Tom's Hardware quoting the Mercury Research numbers.

That means you intentionally mischaracterized Intel's statement on percent goals AND are trying to mislead consumers by presenting DIY purchases as market share when DIY is a sliver of the desktop market, whereas the bulk is in OEM, ODM, and other system integrators. So why are you lying to people?

 

I am not announcing this as true, or saying anyone else said it, I said it myself as the opening statement of that post in response to @Deks post:

" More rumours.
However, this time, it seems that they are stating Zen 3 will still be on AM4 (not AM5) socket."

And my response to that was:

" Zen 3 might stick around long enough to be on both AM4 and AM5, that's one possibility."

Then I went on to detail a number of thoughts with 3 links and quotes as references to help fill in the historical details as potential moving forward.

Read his post and quotes / links, and then my follow up - I was responding to his post, not conjuring up something out of nowhere. He responded again in kind.

We were having a casual conversation not having a press conference. There's no need to get so worked up about it, seriously you have gone far overboard with this, again.

What I don't like - just my feedback to you - is when you rant and are rude - including name calling - to people and you argue on and on over insignificant points that don't warrant the pages and pages of back and forth that always seems to ensue. It drowns out useful discussion and news.

If you are not in the conversation, please don't butt in and rudely berate those in that conversation like you did quoting us out of context and spinning on about that for pages and pages.

I don't like seeing it happen to others, nor do I enjoy being hooked into it myself. I don't want to ignore you, but when you rant on and on like this that's my first reaction.

Please let's end the argument and move forward. It's still the Holiday and I'd like to enjoy it.

 

We don't see much Radeon / Ryzen Mac OS content here - mostly we have Windows / Linux - and this guy is showing MacOS + Radeon 5700XT video encoding improvements as he sees them.

The test's are done on a hackintosh (Intel x99), but he also has Ryzen 3700x / 5700XT new PC video's, and showed the new Mac Pro with the 5700XT installed (Apple image) and suggests future content about all of these platforms.

[4K][HDR] AMD 5700 XT - Huge Encoding Gains
Dec 29, 2019
objrevs
The 5700XT has created quite the buzz everywhere. It's the first 7nm GPU coming in for consumers at a great price point especially if you manage to grab it during discounts, offers or promo periods. In my video, I look at the encoding / decoding performance benefits when upgrading my from my RX580 for my X99 hackintosh.

 

Noted hype and speculation as a second part and that is all the talk of non market silicon.

 

Given the recent x399 transition to TRX40, what about TRX80 / WRX80, how will compatibility play out with them and the ThreadRipper 3 CPU offerings?

In this case the rumored soon to arrive ThreadRipper 3 48 core 3980x and 64 core 3990x, will they be offered to be compatible on TRX40 - the current batch of ThreadRipper 3 motherboards?

I've seen a lot of video reviews of the current TRX40 ThreadRipper 3 3960x and 3970x mention getting ready for the 3990x as that's the one first rumored to be released, with them assuming that the 64 core ThreadRipper 3 is going to be compatible on TRX40 - and that they plan to upgrade to that 64 core ThreadRipper 3 CPU. That got me thinking...

Here's what was published in August of 2019 about the TRX80 and WRX80 motherboards at the same time as the TRX40:

AMD Readies Three HEDT Chipsets: TRX40, TRX80, and WRX80
by btarunr Aug 29th, 2019 03:45 Discuss (53 Comments)
https://www.techpowerup.com/258739/amd-readies-three-hedt-chipsets-trx40-trx80-and-wrx80

You can read the rest, but what caught my eye was the Chart listing CPU core count delineation between TRX40, TRX80, and WRX80:

Of course they got the x399 compatibility wrong, like most people did, but maybe they didn't get everything wrong.

If that Core-counts line is correct that means the TRX40 will stop with the 32 core 3970x, and the higher core count ThreadRipper 3 CPU's will be supported only by the TRX80 and WRX80 chipsets. There's no specifics on what the differentiation will be with the TRX80 and WRX80 chipsets.

All rumors, nothing is for sure, but perhaps don't count your TRX40 compatible CPU's before they ship.
---------------------------
Well, this is what I expected...AMD simply didn't order enough production for the new Zen 3 sku's:

AMD confirms that Zen 2's supply issues were not a "TSMC Issue"
AMD didn't order enough parts, it's as simple as that
Source: Anandtech | Author: Mark Campbell Published: 31st December 2019
https://www.overclock3d.net/news/cp...zen_2_s_supply_issues_were_not_a_tsmc_issue/1

"...In a recent interview with Anandtech's Ian Cutress, AMD's CTO Mark Papermaster confirmed that Ryzen 3rd Gen's early supply issues "wasn't a TSMC issue", stating that demand outstripped what AMD had expected and planned for.

The long and short of this story is that AMD didn't order enough chips to meed the early demand for its Zen 2 processors. That was why AMD couldn't keep its processors in stock. Now, AMD's Ryzen supply issues are seemingly at an end. Presumably, this is thanks to larger wafer orders from AMD after Zen 2's early success. If TSMC couldn't supply enough chips, there would still be a shortage of Ryzen 3rd Generation processors. AMD's Zen 2 launch shortages were due to AMD underestimating demand. It's as simple as that."

Hopefully AMD is increasing their orders...

 

(Possible) Relative performance of upcoming Zen 2 mobile parts from leaked Userbenchmark website:
https://www.userbenchmark.com/UserRun/23316225

Compared to Ryzen 2700:
https://cpu.userbenchmark.com/AMD-Ryzen-7-2700/Rating/3957

Compared to Ryzen 2700x:
https://cpu.userbenchmark.com/AMD-Ryzen-7-2700X/Rating/3958

Now, if I'm reading those results accurately, the 4800u is between 2700 and 2700x in performance... in a supposed 15W TDP envelope (presumably configurable to 25W - lets hope that they don't intricately limit it to 15W though and allow the APU to spread itself to 25W if necessary).

If they pair this with 5600 XT mobile (which essentially has Vega 56 performance), then it would be a decent laptop (especially if paired with 3200MhZ RAM and proper cooling).

That still leaves the H APU's which have higher TDP (35W/45W?) and presumably higher clocks.

If this is accurate, then the mobile segment will get very interesting (if OEM's decide to use Zen 2 in the first place - which they hopefully will given that Intel's been having supply issues).

 

One more reason AMD could offer support for users with the old chipset.

I don’t trust any of this. In short... Apple got the needed waffers. Once they move on to new tech and AMD will get what they need.

“Since 2016, the Taiwan Semiconductor Manufacturing Company (TSMC) has been the exclusive supplier of Apple’s A-series chips and according to DigiTimes, the orders will go to the same foundry this year as well”

Saved by Apples move to 5nm https://wccftech.com/apple-tsmc-a14-bionic-5nm-production-q2-2020/

 

From what I see so far, 3960x and 3970x are starting to show up.With this the price gouging is starting to come down too. This is because neither chip is a direct upgrade so unless you are building a new machine the CPU will not be a hot seller. The up to 3950x is an upgrade option and sells accordingly. Even the 2990x was not a great seller as AMD HEDT was more about the I/O than CPU grunt.

As an example;

 

On the first part, wholly agreed.

On the second part, I disagree to a degree. AMD knows it is the underdog. It also knows it only has a small market share. They did not think they could sell what they did. They also didn't expect the interest in their server chips to be so high, so all dies that qualify go directly to Epyc.

Either way, shows shortsighted predictions. They didn't properly order their wafers.

Now, if they move up Zen 3, that would be better than ordering more Zen 2 when supplies are so constrained already. I honk everyone would be cool with that, but not likely to happen.

 

There's no reason to imagine any other possibilities what with AMD going public and taking the responsibility for their actions.

Apple's rumored 5nm iphone 12 production is supposed to happen in September 2020, 9 months from now. Do you really think Apple's 7nm TSMC needs will disappear instantly right now?

So you think AMD was covering for TSMC by taking the responsibility themselves? Why would AMD think that's needed?

So you are saying that AMD should have said that AMD ordered up to the maximum number of dies TSMC could produce for them on 7nm? To me that would have sounded less believable.

AMD must order / need far fewer 7nm wafers to be run by TSMC than Apple does. Even if AMD ordered as many 7nm wafers from TSMC as wanted by customers, Apple's needs would far outnumber AMD's needs.

So far AMD hasn't lied to us, why go out of your way to doubt them?

TSMC could certainly rebuke their explanation, but I have't seen TSMC put out any statement one way or the other.

Until I see a statement from a reliable source refute AMD's statements, I'll accept AMD's statement as the factual truth.

 

I believe AMD underordered 7nm with the 3900x etc.. This explains why the death of x399 was so sudden. AMD could not afford the 3960x being just an upgrade.

 

FYI - AMD's CES 2020 presentation is starting in a few minutes from now...

AMD at CES 2020
Unlisted
8,070 waiting•Scheduled for Jan 6, 2020
AMD
Watch the AMD CES Media Days press conference and see how AMD will be pushing the limits of high-performance computing for 2020.
http://forum.notebookreview.com/thr...ts-video-articles.831509/page-2#post-10981079

Finally AMD is releasing CPU's and GPU's for laptops that might match the performance of their desktop products. Coming in 1H 2020, Asus G14 in February 2020, and Dell in Q2 2020:



AMD Smartshift - SOC? Optimization gives 10% performance improvement in games:

Dell G5 SE Q2 2020


ASUS ROG Zephyrus G14 hands-on: Ryzen power in a light-up laptop
Jan 6, 2020
Engadget
We've liked previous Zephyrus laptops well enough, but with the G14, Asus is working hard to stand out from the crowd. The new G14 packs a Ryzen 7 into a slim gaming chassis (albeit a down-clocked version of the 45 watt chip), and adds a unique LED grid to the laptops cover. This doesn't serve any performance purpose, but you can customize the lights with text, images, or even animated GIF. If you're looking to stand out at your next gaming get together, the Zephyrus G14 is sure to attract attention.

 

Lisa Su came out with 1 more thing... the 3990x


(image upload limit is 5 per post, continuation of previous post)
http://forum.notebookreview.com/thr...laris-navi-gpus.799348/page-725#post-10981131

AMD left out TRX80 / WRX80 and the rumored 3980x 48 core. Maybe AMD needed to shorten the time on stage and will deal with it later. Feb 7th is a ways off...

 

64 cores
128 threads
+25k in CBR20

 

I like the idea of 5600M and 5700M in mobile.
5700M is likely the one to shoot for as AMD could probably easily undervolt the desktop 5700 without actually reducing clocks - but even if they do drop performance by 5-10% from desktop (similar to V56 in Helios 500), then its fine too.

What I would like to see is quality components with smart/proper execution of cooling that results in a cool and quiet unit which can run both CPU and GPU at maximum indefinitely (think Helios 500 but possibly scaled down a bit to drop the thickness - doesn't need to be as quiet or cool, but similarly so is probably more than doable).

And of course, better OEM software/BIOS support for their laptops with easy access to internals.

 

Doing the math, and figuring 100% scale, if the 3970x was running 3.8 GHz then the 3990x was at 2.9 GHz.

 

This in spades.

All I'm thinking of is the Hackintosh build that smashes the top end Mac Pro with the 3175x at a fraction of the cost.