Sunday, May 5th 2019

AMD's Zen 2 Threadripper Conspicuously Absent From Company's Latest Roadmaps

We've all taken a look at AMD's March 2019 product roadmap, which showed us the upcoming 2019 tech the company would be bringing to the table in its "non-stop product momentum". However, it seems that this non-stop product momentum might be coming to an unexpected twist of fate that might delay it from entering the last station - the Zen 2-based Threadripper. In the company's latest May earnings call roadmap, the company silently removed the Zen 2 Threadripper from its product roadmap - where it used to sit right after the launch of Zen 2-based Ryzen products for consumers, is now just a big crop of the space it occupied.

This might mean many things, and a mistake on someone's part while cropping the PowerPoint slide could be the only thing going on here. However, the best and most plausible speculation that can be entertained when one considers this is simple - a supply problem. With the 7 nm node being the newest, most dense fabrication process possible, and with AMD having to share TSMC's 7 nm wafer production with a number of high profile companies - such as Qualcomm, for instance - may mean that supply is simply too tight to support Zen 2-based products across so many product stacks - Ryzen and Epyc - at the same time.
Perhaps AMD has delayed manufacturing of Threadripper Zen 2 products until after the initial bout of sales from their upcoming Ryzen 3000 series CPUs fades away, while also allowing for yields to mature and production wrinkles to be ironed out. It would definitely make sense, as Threadripper is definitely the product to cut in such a case - it doesn't carry as much of an impact on AMD's financials and outlook as a deficient Epyc production would bring, and doesn't get in the way of Ryzen 3000 series production to keep on AMD's momentum against Intel's offerings. This definitely does seem like a smart path for AMD to undertake, though it really is a shame that we seemingly won't be looking at a halo product such as the latest gen Threadripper as soon as we thought we might.
Add your own comment

63 Comments on AMD's Zen 2 Threadripper Conspicuously Absent From Company's Latest Roadmaps

#26
Caring1
Who really cares if TR is cut from the line up, we're getting increased core count and faster clocks, be happy folks.
Posted on Reply
#27
Vayra86
NdMk2o1oWow what happened in here, its like an anti amd sausage fest... The usual suspects plus a few new faces all stroking each others Intel epeens... Bye
Manu_PTMore like the whole forum is an anti Intel + Nvidia sausage fest. That´s how TpU is known nowadays on every forum. Go check every Intel or Nvidia article comments section and surprise yourself.
This sounds like a very balanced forum to me, if you can both read your own truth on here, we really can't get much better can we.

Let's all be anti everything so we can hate in equal measure. Fun!
Posted on Reply
#28
95Viper
Stay on topic.
Quit bickering.
Trolls leave.
Only and last warning.

Thank You, have a nice, clean, civil discussion on the topic.
Posted on Reply
#29
InVasMani
ZhangirDuysekeSo when Nvidia or Intel make mistakes you f*cking hate them but when shitty AMD does this bullshit you "it's no problem"
I don't see the problem with this Intel and Nvidia don't need to be cut any slack where they are situated at present. It's much easier to give a AMD a bit more leeway for the time being. Neither Intel nor Nvidia deserves any real leeway at this point in time so in relative terms "It's no problem." not to mention AMD didn't do much wrong here they missed their road map a bit, but hey no worries Intel's 10nm is right on schedule!?
Manu_PTMore like the whole forum is an anti Intel + Nvidia sausage fest. That´s how TpU is known nowadays on every forum. Go check every Intel or Nvidia article comments section and surprise yourself.
As it should be in a David vs Goliath tech battle. Besides when Goliath is a price gouging sob he'll receive no sympathy or condolences when brought back down to size.
Posted on Reply
#30
NdMk2o1o
Manu_PTMore like the whole forum is an anti Intel + Nvidia sausage fest. That´s how TpU is known nowadays on every forum. Go check every Intel or Nvidia article comments section and surprise yourself.
Your kinda proving my point, but please carry on :clap:
Posted on Reply
#31
EarthDog
biffzinkerOr Ryzen 3x00 is packing 16/12 cores so Threadripper isn't needed right away?
There you go... surely it's a combo of a lot of things, but I'd imagine its covered with existing TR, and the crossover of so many c/t on mainstream Zen 2 anyway.

Edit: Lol at TPU being TPU... :(
Posted on Reply
#32
Wavetrex
HEDT will become unnecessary very soon.

It only appeared in the past because Intel has offered maximum 4 cores on the mainstream for so long... and they found a marketing way of selling massively overpriced extra cores by calling it "High-End Desktop".
They could have easily made 6-core or 8-core on the mainstream since Broadwell (which is also 14nm, like presend day Coffee Lake) if they really wanted... but with a monopoly and AMD nowhere, they invented this "HEDT" concept instead.

If Zen2 is capable of 16 cores on the mainstream, TR (or AMD's HEDT) is simply unnecessary for anyone but the most extreme workstation users, who still have the choice of the 32-core existing TR.
It could also (almost) kill Intel's HEDT, simply with (much) lower platform price.
Posted on Reply
#33
oxidized
TR doesn't really makes sense anymore if Ryzen 3000 series lineup will include 12 and 16 cores SKUs
Posted on Reply
#34
InVasMani
InVasManiI don't see the problem with this Intel and Nvidia don't need to be cut any slack where they are situated at present. It's much easier to give a AMD a bit more leeway for the time being. Neither Intel nor Nvidia deserves any real leeway at this point in time so in relative terms "It's no problem." not to mention AMD didn't do much wrong here they missed their road map a bit, but hey no worries Intel's 10nm is right on schedule!?

As it should be in a David vs Goliath tech battle. Besides when Goliath is a price gouging sob he'll receive no sympathy or condolences when brought back down to size.
Vya DomusOr maybe Zen 2 TR isn't going to come out this year ? Stop looking too much into this.
Exactly beside maybe AMD will cherry pick the better high frequency Epyc dies for TR and the delay just reflects better binning for whatever Zen 2 TR ends up becoming. At this point I'm not at all surprised by a possible TR delay information on it has been scarce for over a year now while we've seen quite a bit on Ryzen and Epyc. I don't think that's just coincidence go figure.
Posted on Reply
#35
Splinterdog
If nothing else, at least I now know what HEDT stands for.
Posted on Reply
#36
Hardware Geek
InVasManiExactly beside maybe AMD will cherry pick the better high frequency Epyc dies for TR and the delay just reflects better binning for whatever Zen 2 TR ends up becoming. At this point I'm not at all surprised by a possible TR delay information on it has been scarce for over a year now while we've seen quite a bit on Ryzen and Epyc. I don't think that's just coincidence go figure.
Considering the amount of demand for the epyc processors, which can't be overclocked, it only makes sense to delay threadripper and save the dies that clock the fastest until they have a sufficient supply when they do launch it. I can wait until next year to upgrade
Posted on Reply
#37
kings
InVasManiAs it should be in a David vs Goliath tech battle.
I don´t agree with that, doing that would turn out to be just cheap criticism.

Companies should be praised or criticized for their products (quality/performance/utility/features/price, etc) and not because they are big or small companies!

Or are we going to start praising bad products, just because they are from a small or smaller company? (and before there is confusion, I'm not talking about AMD, I'm talking generically about any company).
Posted on Reply
#38
Valantar
kingsI don´t agree with that, doing that would turn out to be just cheap criticism.

Companies should be praised or criticized for their products (quality/performance/utility/features/price, etc) and not because they are big or small companies!

Or are we going to start praising bad products, just because they are from a small or smaller company? (and before there is confusion, I'm not talking about AMD, I'm talking generically about any company).
There's nuance to be added to this: larger companies (particularly those with dominant/near-monopolistic market positions) deserve and require more scrutiny as they have more power to abuse and are more likely to do so. Likewise, it's easier to cut the underdog some slack - for the same reasons, mostly, as having less power not only lessens the ability to abuse it, but puts concrete roadblocks in the way of success. This is why a minor delay for AMD is much less of a big deal than a minor delay for Intel, let alone a major delay like 10nm. Gargantuan market leaders have every opportunity to deliver beyond expectations, and thus failing to do so is disappointing - hence how people quickly grew tired of Intel's 4c8t +~7% performance/year cadence. Small-fry competitors impress when they stay competitive despite having a fraction of the resources - like AMD's blockbuster success with Ryzen. Put rather bluntly, it's the same logic as praising a runner with a limp for finishing the marathon at all, while criticizing the world-class athlete for performing below par, even if the athlete's time is half of the other. Different standards for different entities, even when operating in the same field.

This doesn't mean that AMD should be excused for bad products (I don't see many people touting the VII as the second coming of Raptor Jesus, so I don't think this is actually happening), or that they're somehow "a good corporation" (I'd argue that those don't exist, as the goal of any profit-oriented business is to extract money from the general public), but that they deserve more praise for (even moderate) success and less criticism for not quite making deadlines than Intel.
Posted on Reply
#39
Valantar
eidairaman1You have been reported
I did the same, and @95Viper did step in above. Doesn't seem to have made that much of a difference, sadly.
Posted on Reply
#40
eidairaman1
The Exiled Airman
ValantarI did the same, and @95Viper did step in above. Doesn't seem to have made that much of a difference, sadly.
He may not have credentials to remove posts here.

That persons comments do not belong here.
Posted on Reply
#41
Hardware Geek
eidairaman1He may not have credentials to remove posts here.

That persons comments do not belong here.
Sadly, civility seems to be the exception nowadays.
I'm not a fanboy of either company, but I'm glad to see AMD doing well. Competition is beneficial to consumers. I doubt we'd have seen Intel increase core counts had AMD not introduced Ryzen.
Posted on Reply
#42
SL2
I don't understand comments like "16 cores ought to be enough for anybody". Sure, most buyers wants 16 cores or less,
but the next Threadripper is supposed to have 64 cores, that's on a whole different level, even with a lower clock speed.

I don't think the average TPU member is the target audience, at least I'm not.
Posted on Reply
#43
Xuper
Perhaps 2020? How about Putting 32GB HBM on TR 16 Cores?
Posted on Reply
#44
lexluthermiester
XuperPerhaps 2020? How about Putting 32GB HBM on TR 16 Cores?
THAT would be a hell of a thing! Pricey, but awesome!
Posted on Reply
#45
Hardware Geek
MatsI don't understand comments like "16 cores ought to be enough for anybody". Sure, most buyers wants 16 cores or less,
but the next Threadripper is supposed to have 64 cores, that's on a whole different level, even with a lower clock speed.

I don't think the average TPU member is the target audience, at least I'm not.
Agreed. These kinds of products are mostly for creative professionals and geeks like me who want all the pcie lanes. I'm planning on building a home server with mine when it does get launched.
Posted on Reply
#46
IceShroom
Looks like lots of paid shill with no grey-matter in this news/rumor.
That road map is for 2019. AMD could to release TR 3000 Series in 2020 when they have sufficient amount of Chiplet for TR.
And TR release 5-6 month after mainstream(Zen/Zen+) and server(Zen). So if mainstream release in Q3( June 6th month, July 7th month, August 8th month), 6 month after that will be January 2020 - March 2020.
Posted on Reply
#47
HD64G
As some already wrote, it will be launched in 2020. They need more highly-binned chips for TR in order to get the perofrmance and wattage needed. That is also the only reason for the delay of the 16C Ryzen 3000 chip. They need the best clocks possible this time in order to get at least equal or win Intel in single-threaded apps as MT will be their field eitherway.
Posted on Reply
#48
Valantar
IceShroomLooks like lots of paid shill with no grey-matter in this news/rumor.
That road map is for 2019. AMD could to release TR 3000 Series in 2020 when they have sufficient amount of Chiplet for TR.
And TR release 5-6 month after mainstream(Zen/Zen+) and server(Zen). So if mainstream release in Q3( June 6th month, July 7th month, August 8th month), 6 month after that will be January 2020 - March 2020.
"I disagree slightly with the judgement of the person writing the report, so they obviously are a brainless paid shill." Yeah, that's not how a constructive discussion works. Not to mention that your arguments have been presented previously in this thread by people capable of not being an ass while doing so.
Posted on Reply
#49
EarthDog
Hardware GeekSadly, civility seems to be the exception nowadays.
I'm not a fanboy of either company, but I'm glad to see AMD doing well. Competition is beneficial to consumers. I doubt we'd have seen Intel increase core counts had AMD not introduced Ryzen.
I SOOOOOO agree with your first two lines...and then we diverge. :p

I wouldn't have minded if core counts continued to go up slowly... it was a perfect market segmentation to reality ratio... mainstream had up to 6c/12t while HEDT was that and a lot more. Now, most consumers will think it has 'moar corez sew et must bee bet0r'.... and for those who can use them, that is correct. The problem is most users can't and software is still severely behind as well. As I said earlier, at least back in the day with clock speed wars and IPC, that improved everyone's experience regardless. Here, its really an only if they are used type situation.
Posted on Reply
#50
efikkan
I believe Threadripper 3 was expected in Q4 or later, so I wouldn't worry to much for it lacking in the short term roadmap.
notbLimited 7nm supply, lack of demand, extremely low-volume product.

TR is something they should make from surplus EPYC dies, but I doubt there are many. The 7nm supply may not be enough to make all the EPYCs they'd like to sell.
Demand is certainly not the problem, but Threadripper requires higher bins than Ryzen, so supply is probably an issue for 7nm.
notbAlso, why not unify under the EPYC brand?
Well, they are different markets, so I guess some segmentation makes sense?
At least it's not as bad as Intel's double HEDT lineup: Skylake-X and Skylake-W.
londisteIs there a need for Threadripper? According to current state of leaks and assumptions, Ryzen 3000 will go up to 16 cores which would cut deeply into the already niche Threadripper. Yes, there will always be uses cases for 4 memory channels and 128 PCI-e lanes but will that justify keeping another platform?
I would ask the opposite question; why would anyone need 16 cores (for now) without other "HEDT features"? Except for market hype, why does a mainstream system need 16 cores?

The purpose of HEDT is CPUs for workstations, like content creators and developers, which needs a balance between core count and core speed, along with good memory bandwidth etc. Threadripper 2 largely fails as a HEDT platform due to the latency issues which becomes substantial for 2970WX and 2990WX.
Posted on Reply
Add your own comment
Mar 15th, 2025 22:26 EDT change timezone

New Forum Posts

Popular Reviews

Controversial News Posts