Speculation: i9-9900K is Intel's last hurrah in gaming

Page 14 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Will Intel lose it's gaming CPU lead in 2019?


  • Total voters
    184
  • Poll closed .

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
If cinebench alone were a metric for overall performance then FX would have already killed off intel...

Sorry, but that is ridiculous. Until Summit Ridge came out, AMD consistently lost every Cinebench out there. R10, R11.5, and R15 crowns all belonged to Intel. The 6700k and 7700k were faster in Cinebench than every FX ever released. Example:

https://www.anandtech.com/bench/product/1289?vs=1543

Here the 6700k wins ST and MT Cinebench R10 and R11.5 against a 9590.

All of this makes sense to me. I even think an eventual 16C/32T AM4 part is possible, but at lower clocks,

Case in point: AMD should be able to sell a 16c/32t AM4 part with base and boost clocks similar to those of the 1800x with a TDP similar to that of an 1800x. The only reason why I don't think the 16c part would be stuck with such a narrow margin between base and max turbo is that 14LPP had such an issue with power consumption close to 4 GHz. TSMC 7HP shouldn't have that problem.
 
  • Like
Reactions: Tlh97 and moinmoin

OTG

Member
Aug 12, 2016
101
175
116
But a Zen2 where it wins literally everything (superior IPC + 5Ghz range + moaaaar cores), I think they would price that at a level the market would support instead of roughly half price across the board.

This is where I'm convinced you're mistaken.
Intel isn't getting $500 for a 9900k because it costs that much to produce, it's because A) people will pay it, and B) supply problems. As they get more 14nm production lines running B goes away, and when Zen 2 launches A does too.
IDK how low Intel can sell a 9900k and make a profit, but it's probably under $200.
So, you have a $500 AMD 12-core competing not against an absurdly overpriced Intel 8-core, but a (probably still overpriced, but in the same range) 10 or 12-core, and a (let's say) $250 9900k. That's a much different picture, and if AMD doesn't have an *undeniable* advantage, Intel's mindshare will win 9/10 times.

AMD simply CANNOT afford to waste months playing catch-up in a tit-for-tat price battle, they need to *WIN*.
If they make buying a 9900k a stupid decision, THAT is how they'll gain marketshare, THEN they can start charging higher prices. Probably never as crazy as Intel during the Bulldozer years, but premium, competitive, prices.
Intel isn't going away, and they didn't become Chipzilla by designing bad products.
AMD will only have a few years before Intel has an answer to Zen 2/3/4, they gotta go for the throat.
 

ondma

Platinum Member
Mar 18, 2018
2,718
1,278
136
Don't be so sure about that. Intercore latency is going to be a possible issue thanks to the chiplet design. At least on 16c chips anyway.



Hector Ruiz isn't running the company anymore. Lisa Su has either directly or indirectly provided up to two major advantages over AMD's position in 2004-2005: she has increased CPU supply, and she has reduced desktop prices for AMD's product. AMD may be on the cusp of regaining the performance lead in nearly every possible segment - something they haven't accomplished since the x2 days. The 1800x came close (it was competitive with both the 7700k and 6800k), but it will take Zen2 to get the job done. And Intel will have few answers until 2021. So AMD can pull a Hector and try to increase margins, which leaves Intel an opening to counterattack at a comfortable margin position (something Intel's investors love) with 7nm products. Or they can repeat what they did in 2017, keep prices capped at the $500 and below range, and soak up market share as Intel flounders.

Hector Ruiz honestly believed that Intel would be stuck in Netburst hell forever (they wouldn't). I think Su is smarter than that. Intel will be back. AMD can't afford to charge $1k for a top-end Zen3 sku when competing with next-gen Intel CPUs.
Lots of "ifs" here. But "if" AMD takes the lead in both number of cores and lightly threaded performance, and "if" one assumes intel will take the lead again at some point, then why would AMD *not* want to charge the highest possible price when Zen 2 comes out in order to maximize profits? As far as taking market share, I understand that, but even assuming they price the chips very cheaply, do they have sufficient capacity? Again if they are constrained by capacity, it also does not make sense to sell the chip at a bargin bin price.
 
  • Like
Reactions: CHADBOGA

Arkaign

Lifer
Oct 27, 2006
20,736
1,377
126
This is where I'm convinced you're mistaken.
Intel isn't getting $500 for a 9900k because it costs that much to produce, it's because A) people will pay it, and B) supply problems. As they get more 14nm production lines running B goes away, and when Zen 2 launches A does too.
IDK how low Intel can sell a 9900k and make a profit, but it's probably under $200.
So, you have a $500 AMD 12-core competing not against an absurdly overpriced Intel 8-core, but a (probably still overpriced, but in the same range) 10 or 12-core, and a (let's say) $250 9900k. That's a much different picture, and if AMD doesn't have an *undeniable* advantage, Intel's mindshare will win 9/10 times.

AMD simply CANNOT afford to waste months playing catch-up in a tit-for-tat price battle, they need to *WIN*.
If they make buying a 9900k a stupid decision, THAT is how they'll gain marketshare, THEN they can start charging higher prices. Probably never as crazy as Intel during the Bulldozer years, but premium, competitive, prices.
Intel isn't going away, and they didn't become Chipzilla by designing bad products.
AMD will only have a few years before Intel has an answer to Zen 2/3/4, they gotta go for the throat.

"Winning" isn't something that shareholders of western corporations are directly interested in. It's profits. It always has been. Now, one can argue with some merit that such a narrow strategy, of which shows limited long-term vision, is a wise one overall, but it's definitely the standard.

We don't really have any evidence that AMD would be interested in a severe price war, or sacrificing enormous amounts of margins to any real extreme.

When 1800X came out, it was 3.6Ghz and MSRP'd @ $499. It was a faster MT, slower ST product. The rest of the Zen 1xxx family was similarly scaled. Better deal overall, but nothing that was extreme, although it did pants the silly overpriced workstation offerings from Intel in a lot of ways if your workload was a match.

Sales were GOOD, prices came down later.

When 2700X came out, it was 3.7/4.3 and $329. Again a faster MT, slower ST product that was competitively priced. Gaming was still solidly on Intel, but MT was again a better deal with AMD. And AMD's option was a wide open road for 1xxx buyers, unlike Intel and their silly chipset games.

Sales were GOOD, prices came down later, though not as much (reputation and market share increasing).

When Polaris came out, it was a pretty compelling bargain, but not insanely so. Sadly, it got hijacked by the mining crisis and gains with gamers and general PC enthusiasts were muted because they were suddenly selling for 200-400% of MSRP.

When Vega56/64 came out, MSRP was roughly in line with Nvidia's options. Of course, this got also hijacked by continued mining mania, so it didn't get much play.

If we go by history, then they will be a better buy than Intel for a lot of people, but not to the point of cutting their own margins more than absolutely necessary. Their CPU reputation and hype is very high, the end of the mining craze and old console gen is feeding a very resurgent PC gaming scene, and quality Zen2 products won't need to be priced at cut rates to sell through. Especially if 7nm volume isn't enough to supply it anyway.

Going further back in history to the time of outright AMD supremacy : the AMD Socket 939/754/AM2 products of 2004 through Conroe were priced as complete flagships from top to bottom. $1k 'FX', $300-$850 X2s, and on and on. They were better, and they charged absolute top dollar.

Leadership and overall direction may change, but publicly traded corporate strategy virtually never wavers : make every penny possible this quarter. The only thing that can drive down prices is : lack of demand, and competition. That's it. If Zen2 is extremely good, then it's going to make AMD a ton of money and sell really well. But we're not going to get charity, or a self-sacrificing price war for indeterminate reasons. Like, what would be the goal to going ham with price wars? Intel wouldn't go under, they have enough cash to sit on their hands for donkey's years. Their reputation wouldn't get any better. The quality of their products has been speaking for itself, and their Ryzen stuff has been a home run even if it's not 100% the best at every task. Market share? That lasts only for as long as you have the price/performance and tiers to support. If Intel came out with 10nm/whatever products that reclaimed the throne at competitive prices, then buyers would simply go there for that timeframe. Outside of a few brand loyalist goofs, people generally buy what they perceive as the best choice for their money. And if they think they need more brand awareness, then actual marketing to the wider world would be the way to go.

Shower me in $250 9900k killers, but man, it seems wildly optimistic to me.

I'd put solidly better than even odds that Zen2 SKUs can match or beat Intel's CL lineup in most things (stock v stock).

I'd put extremely low odds that pricing will be outside of the norms we've seen for ages now. Beating Intel on price/performance, unless they gain a full IPC win, then we'll see outright flagship level pricing and profiteering. It's capitalism, and their SH will demand so.
 

TheELF

Diamond Member
Dec 22, 2012
3,967
720
126
Sorry, but that is ridiculous. Until Summit Ridge came out, AMD consistently lost every Cinebench out there. R10, R11.5, and R15 crowns all belonged to Intel. The 6700k and 7700k were faster in Cinebench than every FX ever released. Example:

https://www.anandtech.com/bench/product/1289?vs=1543

Here the 6700k wins ST and MT Cinebench R10 and R11.5 against a 9590.
Yeah because that's when bulldozer came out...along the 6700k and 7700k...
 

moinmoin

Diamond Member
Jun 1, 2017
4,934
7,619
136
As far as taking market share, I understand that, but even assuming they price the chips very cheaply, do they have sufficient capacity?
TSMC is fabbing 7nm chips for Apple, Huawei and Mediatek. Apple alone amounts to around 200 million iPhones per year. There is no way AMD's capacity is in any way limited at TSMC. (It may be at GloFo, but that'd be just another reason to scratch the WSA for GloFo's inability to fulfill its side of obligations and move somewhere else.)

"Winning" isn't something that shareholders of western corporations are directly interested in. It's profits. It always has been. Now, one can argue with some merit that such a narrow strategy, of which shows limited long-term vision, is a wise one overall, but it's definitely the standard.

We don't really have any evidence that AMD would be interested in a severe price war, or sacrificing enormous amounts of margins to any real extreme.
You appear to miss one important point: The actual margin AMD has on product. The whole of Ryzen and Epyc product range is AMD's biggest margin business, none of their other products compare. Yes, even with all the price drops and the original pricing being this competitive to comparable Intel pricing, it's by far their biggest margin.

The cause is two-fold: Nowadays the biggest expense is not the manufacturing cost of chips but the cost for designing and creating the mask. This upfront cost needs to be recuperated by all the chips produced with it, so the more chips the better, the higher the yield the faster the break even point can be reached and true profit achieved. AMD is achieving great margins with its CPUs right now thanks to
  1. limiting the amount of unique dies thus reducing overall upfront costs
  2. keeping dies small and creating bigger chips using MCM which helps achieving high yields from the start
  3. the whole product design and range is optimized to salvage as many "defect" dies as possible, increasing effective yield further
  4. as a bonus, since few dies are used and focus is on small dies with redundant designs for a high yield, a move to new process nodes is not as expensive in total cost than with the traditional monolithic die approach
With Zen chips AMD has been pretty aggressive at pursuing new process nodes as far as the x86 market is concerned. This creates kind of a catch-22 for them, since to break even with new dies they have to produce a lot of them, while the time window available for doing so is limited by the fast succession of new nodes.

This is why AMD will continue pricing their product comparatively low. Even and especially margin wise AMD literally can't afford limiting its market reach by high prices that could limit the scale of its chip production.
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
why would AMD *not* want to charge the highest possible price when Zen 2 comes out in order to maximize profits?

They need marketshare and they need mindshare. You're thinking like Hector. They need to be ready for when Intel hits back.

As far as taking market share, I understand that, but even assuming they price the chips very cheaply, do they have sufficient capacity? Again if they are constrained by capacity, it also does not make sense to sell the chip at a bargin bin price.

TSMC has no problems supplying them with wafers. Especially with flagging demand from the smartphone/tablet market.

Yeah because that's when bulldozer came out...along the 6700k and 7700k...

Look at Haswell, it still stomps the 9590 in Cinebench:

https://www.anandtech.com/bench/product/1289?vs=836

Intel had been beating AMD at Cinebench since R10 first came out. I think the only thing that was ever relatively competitive was Thuban, but no not really:

https://www.anandtech.com/bench/product/185?vs=45

Nehalem won. Oops.

The idea that Cinebench is, or always has been, an "AMD benchmark" is a complete farce.
 

Asterox

Golden Member
May 15, 2012
1,026
1,775
136
They need marketshare and they need mindshare. You're thinking like Hector. They need to be ready for when Intel hits back.



TSMC has no problems supplying them with wafers. Especially with flagging demand from the smartphone/tablet market.



Look at Haswell, it still stomps the 9590 in Cinebench:

https://www.anandtech.com/bench/product/1289?vs=836

Intel had been beating AMD at Cinebench since R10 first came out. I think the only thing that was ever relatively competitive was Thuban, but no not really:

https://www.anandtech.com/bench/product/185?vs=45

Nehalem won. Oops.

The idea that Cinebench is, or always has been, an "AMD benchmark" is a complete farce.

"Actually it was approximately a reverse situation", but then "AMD Ryzen was launched and a scandal happened".


2017-02-10-171948.jpg
 

Timorous

Golden Member
Oct 27, 2008
1,532
2,535
136
I'm not sure you completely understand me here :)

I would not be surprised in any way if AMD is able to notch ahead here. The catch is :

Total IPC win. Ok, possible.
Total Clock Speed equal or win. Again, possible.
16C/32T variant on AMD. Again, possible.

Where it falls apart is IF all three of those come together, choosing to sell a 16C/32T capable of ~5Ghz @ $499. Just on pure common sense, it doesn't mesh with reality. Because that would mean that a MUCH smaller 8C/16T would be already capable of being far superior to the 9990k. Thus, if these 'leaks' were correct in all manner, then the pricing would perhaps undercut Intel at every level, but not to that degree. Extreme buyers (the 2080ti/Titan types) would happily eat up 16C/32T 5Ghz @ $999, or maybe cut them at a break and go $899.

Similarly, Intel has a ton of 2C and 4C items that sell widely to OEMs. Ryzen 2 '3' level products being 6C/12T to start seems like a waste, when a 2C/4T and 4C/8T offering @ 7nm with these types of gains would be massively effective at cutting Intel out of every price point yet maximizing sales. Now the argument against this may simply be timing, if 7nm capacity can't be sourced to run that many types of Zen 7nm yet, so they're focusing on larger chiplets to start, with smaller monodies to follow? Nearly impossible to say until we get more confirmation on the lineup.

Finally, how often have we seen gigantic leaps with a revision? Intel went from 32nm to 14nm on Core (Sandy to Coffee), and we didn't see that much movement in raw clock speed potential. Now that can be due to balancing pipelines for IPC and all other types of considerations, but the range between ~4 to 5+ seems particularly sticky. One could observe that the iGPU portion of these Intel gens kept growing in space, thus using up the potential improvements in core count/etc. Which I totally agree with in so many ways. The iGPU on Intel CPUs is pretty darned big in die space. But it doesn't seem to count much in relation to the cores that actually ARE on the SKUs. On the flip side, we've seen Vega 64 to Vega 7, 14nm to 7nm, nearly identical cores, very modest clock boost, yet it actually increased in power consumption. The performance gains seemed more closely tuned to the improved HBM and memory performance overall. And, how was such a thing priced? Right around its competition, more or less.

So you see, if the list had leaked with no mention of Ryzen 3 6C/12T, and no prices listed, I'd be much more trusting towards it. Applying business-sense logic to it with those things in consideration however, makes me look at it as probably fanfic.

AMDs biggest issue is getting into OEM builds. Coming up with an offering that is almost too good for OEMs to refuse gets them in the door which means support systems to handle AMD machines are built and expanded upon. Going forward, as these are in place it is easier for OEMs to justify having a nice selection of AMD based products even when Intel eventually get on top of their current issues.

I think AMD realize how great a chance this is for longer term prosperity so seizing it while it exists is more beneficial for their future than squeezing more profits from the DIY crowd.

AMD are after a much bigger fish but to get OEMs to move they almost need to be forced into it.
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
AMDs biggest issue is getting into OEM builds. .

I have noticed that a few big OEMs have started selling Ryzen R3 base systems with cheap Radeon graphics cards (like the 550 or whatever) instead of using APUs. That was possibly a response to Intel's shortages at the low end when they were having more 14nm capacity constraints than they have now, but still, it seemed like it was easier for OEMs to adopt the R3 than the 2400g for whatever reason. Maybe Picasso or Renoir will turn around that situation.
 

OTG

Member
Aug 12, 2016
101
175
116
"Winning" isn't something that shareholders of western corporations are directly interested in. It's profits. It always has been. Now, one can argue with some merit that such a narrow strategy, of which shows limited long-term vision, is a wise one overall, but it's definitely the standard.

We don't really have any evidence that AMD would be interested in a severe price war, or sacrificing enormous amounts of margins to any real extreme.



Profits ARE my point. Long-term profits.
AMD doesn't care about "winning" per se (no such thing, unless Intel just stops existing), they care about marketshare.
Intel's had so many products, that were so much better, for so long, that when Ryzen launched, AMD had single-digit market share. They've gained ground since then, but it's almost certainly still under 20%.

Let's say for discussion that AMD has 20% of the desktop CPU market. They have new chips coming out, that look to match Intel in single-threaded performance, and completely dominate in multithreaded tasks.
You seem to be arguing that, since they now have a premium product, they should price their chips competitively with the similarly performing Intel product, and start pulling in some of that sweet, high-margin goodness.
I'm saying that if AMD prices their chips anywhere near what a comparably performing Intel product, 9/10 consumers will buy Intel without a second thought. AMD may as well not exist for most people, and if they've heard the name it was probably a POS $400 tower from walmart.
Competitive pricing will get AMD bigger margins, but their marketshare will barely grow.
Remember, AMD started the price war when they sold the 1800x for less than half the price of the 6900k. This forum was full of people who were absolutely convinced that if AMD had a good product, they'd go for margins and price it high, $800+.
They didn't, and now Intel's 8-core is $500(ish), and the 2700x is priced like the 8600k. Seriously, they're both $280.
Intel cut their huge margins, dropped prices, and limited their marketshare losses.

Zen 2 will almost certainly be similar. Marketshare, marketshare, marketshare. If they can get above 35% or 40% marketshare, then I absolutely agree that we'll see their prices start to rise. We'll never see a $1000 8-core again, though. Those days are gone, and both AMD and Intel will have to settle for "reasonable" margins.

Leadership and overall direction may change, but publicly traded corporate strategy virtually never wavers : make every penny possible this quarter.

You clearly haven't been paying attention to AMD.
From Zen, to paying down debt early, to boosting R&D for the GPU division, they've been taking a long-term view for years now.
Lisa Su isn't some fly-by-night CEO out for a quick boost in stock price followed by selling assets and a golden parachute, she's clearly willing to do the groundwork to make AMD a major player in every market they serve.
 
Last edited:
  • Like
Reactions: Tlh97

VirtualLarry

No Lifer
Aug 25, 2001
56,227
9,990
126
Don't be so sure about that. Intercore latency is going to be a possible issue thanks to the chiplet design. At least on 16c chips anyway.
You know what the dual-chiplet core configuration reminds me of? Core2Quad's MCM design. Things have apparently, come full circle. Here's hoping that Zen2 will "out-Conroe" Intel and the rest of the industry. :)
 
  • Like
Reactions: Tlh97 and Markfw

Arkaign

Lifer
Oct 27, 2006
20,736
1,377
126
Profits ARE my point. Long-term profits.
AMD doesn't care about "winning" per se (no such thing, unless Intel just stops existing), they care about marketshare.
Intel's had so many products, that were so much better, for so long, that when Ryzen launched, AMD had single-digit market share. They've gained ground since then, but it's almost certainly still under 20%.

Let's say for discussion that AMD has 20% of the desktop CPU market. They have new chips coming out, that look to match Intel in single-threaded performance, and completely dominate in multithreaded tasks.
You seem to be arguing that, since they now have a premium product, they should price their chips competitively with the similarly performing Intel product, and start pulling in some of that sweet, high-margin goodness.
I'm saying that if AMD prices their chips anywhere near what a comparably performing Intel product, 9/10 consumers will buy Intel without a second thought. AMD may as well not exist for most people, and if they've heard the name it was probably a POS $400 tower from walmart.
Competitive pricing will get AMD bigger margins, but their marketshare will barely grow.
Remember, AMD started the price war when they sold the 1800x for less than half the price of the 6900k. This forum was full of people who were absolutely convinced that if AMD had a good product, they'd go for margins and price it high, $800+.
They didn't, and now Intel's 8-core is $500(ish), and the 2700x is priced like the 8600k. Seriously, they're both $280.
Intel cut their huge margins, dropped prices, and limited their marketshare losses.

Zen 2 will almost certainly be similar. Marketshare, marketshare, marketshare. If they can get above 35% or 40% marketshare, then I absolutely agree that we'll see their prices start to rise. We'll never see a $1000 8-core again, though. Those days are gone, and both AMD and Intel will have to settle for "reasonable" margins.

All of that is fine, yes, but you're not quite getting what I'm saying.

The (unverified, and probably over optimistic given credible engineering leaks) rumors of :

Superior IPC
5ghz+ clocks reachable across :
8C/16T, 12C/24T, 16C/32T products

Does not make sense with the reported prices of 8C/16T for $250.

Not only would that item immediately be better than the 9900k, but pricing it literally half of Intel is leaving waaaay too much on the table. Their shareholders would revolt.

Undercutting them a bit at say $399-$449 would more than make sense. Ryzen 2000 stuff is already widely popular, they'd have people beating their doors down to buy these at that price.

Not to mention the stuff north of there.

12C would beat Intel's reported 10-core handily. Thus it would exist in its own realm, say $600-$700. 16C would be utter flagship territory with that level of IPC and clock, and command $750-$1000. And they would absolutely sell at those prices.

You might say 'but threadripper!'. However, that doesn't really apply. TR is mass cores, but not elite IPC, nor 5ghz all core capability, and it requires expensive motherboards and serious cooling. Zen2 16C/32T as rumored would instantly obsolete TR and nearly all Intel mass core products instantly. You wouldn't give that away at $500. It would be utterly ridiculous to do that. There is a point where you devalue your own product/brand.

Undercutting Intel a bit at every competitive tier is possible while also having halo products that Intel couldn't touch, which is a strategy that would reach the maximum margin and brand strength.
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
You know what the dual-chiplet core configuration reminds me of? Core2Quad's MCM design.

Not . . . exactly.

C2Q MCMs had, what, a DMI link between the cores, and then each core had its own access to the FSB? The dual-chiplet CPUs are sort of set up like this (I think):

X-X
|
I/O
|
X-X

Where '-' and '|' are IF links, and X is a CCX. Each CCX appears to have equal access to the I/O die. But basically, if you are trying to communicate between a CCX in one chiplet and a CCX in another, you are now two hops away. Which is worse than what we had in 8c (single die) Zen/Zen+. It's also potentially worse than, say, a 2950x since there was no intermediary I/O die. The real question is whether the two hops are that big of a deal, or if most of the IF latency is related to coherency.
 

B-Riz

Golden Member
Feb 15, 2011
1,482
612
136
I think the "leak" that so many are giving credence to is mostly nonsense.

"Superior" IPC is a bit of a misnomer in this case, all AMD has to do optimize what was laid out in Zen, which is what Zen2 is.

Think of Zen as the 80% good enough, the real deal is on 7nm.

A $600 - $700 12c on the mainstream socket is a non-starter. The whole Ryzen deal has been more for less, more cores for less money. Look at Threadripper, more cores less money than Intel.

IF AM4 gets 16c, then Threadripper 3 is starting at 18c +, because it is a bigger package.

I really don't see why this is so hard to fathom. We were told / shown in January the extra space on the die.

Epyc2 is 64c single socket; that is a huge deal in the world of the virtualized server, I cannot emphasize enough how big a deal that is because software licensing is based on server socket.

And remember, they are going to need to offer chips that can be run in the TDP of all the boards under X570; X570 is going to be for highest clocked X series.
 
  • Like
Reactions: Tlh97 and OTG

itsmydamnation

Platinum Member
Feb 6, 2011
2,744
3,077
136
remember consoles are getting 8 zen 2 cores probably high 2ghz clock maybe even low 3ghz. 8 strong cores is the new peasant level , to be real PC master race its 12 or up. I expect Ryzen1 release pricing but being 16/12/8 instead of 8/6/4
 

OTG

Member
Aug 12, 2016
101
175
116
All of that is fine, yes, but you're not quite getting what I'm saying.

The (unverified, and probably over optimistic given credible engineering leaks) rumors of :

Superior IPC
5ghz+ clocks reachable across :
8C/16T, 12C/24T, 16C/32T products

Does not make sense with the reported prices of 8C/16T for $250.

Not only would that item immediately be better than the 9900k, but pricing it literally half of Intel is leaving waaaay too much on the table. Their shareholders would revolt.

Undercutting them a bit at say $399-$449 would more than make sense. Ryzen 2000 stuff is already widely popular, they'd have people beating their doors down to buy these at that price.

The 9900k's current price means NOTHING. It's $200 overpriced, minimum, due to Intel's much publicized supply problems, and the fact that AMD still can't beat it for gaming.
A similar chip (because realistically, AMD's 8-core won't be significantly faster, if at all) at $400 is a price that will IMMEDIATELY be irrelevant.
Pricing it at $250 means that Intel will have to slash prices to match, or they get ZERO sales. This cuts into Intel's margins, and more importantly, gets more AM4 boards into consumer desktops.
Again, sacrifice a bit on margins to gain (literally) millions more sales, and potentially billions more in revenue a few years down the line. Just because most companies can't look past the next quarter, doesn't mean none of them can.

12C would beat Intel's reported 10-core handily. Thus it would exist in its own realm, say $600-$700. 16C would be utter flagship territory with that level of IPC and clock, and command $750-$1000. And they would absolutely sell at those prices.

You're right, they would absolutely sell. Slowly. Because the number of people who buy $700 CPUs is absolutely miniscule. Hell, $500 is a LOT of money for a CPU, Intel got away with pricing like that because of Bulldozer. We're seeing prices regress to something that actually makes sense, unlike 10 years of $340 4-cores.
Remember, they're only producing ONE 7nm die, and they're using it in EVERYTHING. They're gonna crank those little guys out by the hundreds of millions, and the 14nm IO die is both cheap to produce, and helps AMD by using wafers that count towards their WSA with GloFo.
Even with packaging costs, these CPUs should be super cheap to make.
AMD can AFFORD to drop prices, because the most important thing to them is that every Ryzen sale is one less for Intel, and again, MARKETSHARE is the goal, not profit margin per CPU sold.

You might say 'but threadripper!'. However, that doesn't really apply. TR is mass cores, but not elite IPC, nor 5ghz all core capability, and it requires expensive motherboards and serious cooling. Zen2 16C/32T as rumored would instantly obsolete TR and nearly all Intel mass core products instantly. You wouldn't give that away at $500. It would be utterly ridiculous to do that. There is a point where you devalue your own product/brand.

Fine, I'll do it: But Threadripper.
TR is mass cores, but the 2950x is actually AMD's highest clocking chip (single core), has way more PCIE lanes, twice as many memory channels, an enormous socket and surface area for heat dissipation and a whole buncha chiplets, and extremely robust board specs.
16-core AM4 can't match Threadripper in anything but core count (at the low end of TR).
I expect the 12-core Ryzen to be the fastest AM4 chip in everything but very multithreaded workloads.
Because: you mentioned cooling. 16-core Ryzen is gonna be a huge headache to cool, there's a good chance a 16-core TR will clock significantly higher, in addition to the other benefits of the platform.
A 16-core R9 may obsolete *current* Threadripper and Intel HEDT, but not in every task, and not once Zen2 TR releases with more cores, higher clocks (remember, TR gets top 5% of Ryzen dies), on top of the pcie lanes etc.
 

jpiniero

Lifer
Oct 1, 2010
14,510
5,159
136
remember consoles are getting 8 zen 2 cores probably high 2ghz clock maybe even low 3ghz

Thing is, you're probably AT LEAST a year from either console's launch, and probably 15-18 months. And then, (yeah) especially because of the backwards compatibility that they are pushing, the non-exclusive launch titles will for the most part still be designed around the PS4/XB1.
 

ElFenix

Elite Member
Super Moderator
Mar 20, 2000
102,414
8,356
126
And I'm not a millionaire only because I don't have a million dollars....
Don't be so sure about that.....
On latency: IF got upgraded to IF 2, so, I am guessing that fixes a lot of latency issues....
Did you miss the original Ryzen launch? ...
I agree....
All of this makes sense to me. ...
Lots of "ifs" here. ...
TSMC is fabbing 7nm chips for Apple, Huawei and Mediatek. ...
AMDs biggest issue is getting into OEM builds....
Profits ARE my point. Long-term profits....
You know what the dual-chiplet core configuration reminds me of?...


Can y'all keep this vaguely about the 9900k and its gaming performance? We've got plenty of Zen2 threads for y'all to discuss it in.

TIA

AT Moderator ElFenix
 

Arkaign

Lifer
Oct 27, 2006
20,736
1,377
126
Can y'all keep this vaguely about the 9900k and its gaming performance? We've got plenty of Zen2 threads for y'all to discuss it in.

TIA

AT Moderator ElFenix

Respectfully, Zen2 is referenced both in the OP poll, as well as the basis of the thread's subject. We cannot discuss the merits of whether 9900k/Intel will keep/lose gaming crown without discussing Zen2 and future Intel releases. As 9900k is out, reviewed, priced, and known, there is not a lot to talk about with it in isolation.

If this is unacceptable, I understand, but perhaps the thing should just be locked in that case, as it makes the possibility of discussing the OP/thread basis impossible.
 

ElFenix

Elite Member
Super Moderator
Mar 20, 2000
102,414
8,356
126
Respectfully, Zen2 is referenced both in the OP poll, as well as the basis of the thread's subject. We cannot discuss the merits of whether 9900k/Intel will keep/lose gaming crown without discussing Zen2 and future Intel releases. As 9900k is out, reviewed, priced, and known, there is not a lot to talk about with it in isolation.

If this is unacceptable, I understand, but perhaps the thing should just be locked in that case, as it makes the possibility of discussing the OP/thread basis impossible.
Understood, but lots of posts lately look like they're about nothing other than Zen2, or, worse, discussion of processors from 10 years ago. It is understood that we cannot discuss 9900k's place in the market without discussing the market, but there's a difference between discussing X with reference to Y and discussing X by itself. The conversation lately seems to be far more of the latter than the former.

AT Moderator ElFenix
 
  • Like
Reactions: Tlh97

Arkaign

Lifer
Oct 27, 2006
20,736
1,377
126
Gotcha.

In that event, I guess we should try to limit discussion in this thread directly to 9900k/Zen Family/Intel 10th gen gaming only.

In which case, I think the best chance of that would be if an 8C/16T single chiplet model can reach rough parity @ IPC and ~5Ghz+ air clocks.

Certainly doable. And if so, I don't think Core 10 model i9 10 core would be any better unless they somehow managed to ramp clocks further, which seems exceedingly unlikely.

Not a whole heck of a lot to do now but wait for the final silicon and SKUs to get into the proper hands :)
 

TheELF

Diamond Member
Dec 22, 2012
3,967
720
126
Gotcha.

In that event, I guess we should try to limit discussion in this thread directly to 9900k/Zen Family/Intel 10th gen gaming only.

In which case, I think the best chance of that would be if an 8C/16T single chiplet model can reach rough parity @ IPC and ~5Ghz+ air clocks.

Certainly doable. And if so, I don't think Core 10 model i9 10 core would be any better unless they somehow managed to ramp clocks further, which seems exceedingly unlikely.

Not a whole heck of a lot to do now but wait for the final silicon and SKUs to get into the proper hands :)
If it's about gaming alone then no they don't need that, neither AMD nor 10th gen,right now anything from a 7700k and upwards has roughly the same results in gaming because we just lack the GPU that would show any difference and no benchmark shows the actual usage of all threads to see how much performance there is left on the table.
Zen 2 or 10th gen can match clocks and "IPC" (cine results) all they want they won't be able to make GPUs any faster than they are right now.
 
  • Like
Reactions: Tlh97 and Arkaign