Speculation: AMD's response to Intel's 8-core i9-9900K

[Charlie22911]

I don’t think this pricing trend is going to change, sadly, it’s the new norm. Look at pricing of pretty much every consumer good, the new goal is to wring as much money from consumers as the consumer can bear; no longer is it about charging a fair price.

 

[mattiasnyc]

Charlie,

I'm not so sure I agree entirely. I don't think what you describe is particularly "new", it seems a lot of people felt relieved when AMD showed up again with Zen since the perception was that Intel had been too relaxed for too long keeping prices higher than what people liked (I know less is better, but I think you know what I mean). So it doesn't feel that new to me.

The other thing is that if Zen 2 in 2019 is what some are hoping it might be AMD may actually not only become more competitive, but even outdo Intel in some segments. Now, if that happens we'll see more pressure downwards as far as price goes. I think Intel is slow to change, but it's also had little reason to change until now. They've had the best CPUs every year (though maybe not throughout the entire year) and they've had a very large market share. I think the future might look less easy for them.

Just my speculation though...

PS: I'm not trying to absolve AMD from anything either. The 2990wx from what I can see is priced "correctly" in the market, and it's a pretty expensive product looking at price/performance. But for those that needed that performance it may have been worth it and Intel's competing CPUs are in the same range.

 

[rbk123]

Definitely not new as Intel has been gouging for years. However don't think AMD would do any differently if they could. Their initial Zen pricing wasn't as attractive as what it is now - is just 1 example.

 

[Arkaign]

Definitely not new as Intel has been gouging for years. However don't think AMD would do any differently if they could. Their initial Zen pricing wasn't as attractive as what it is now - is just 1 example.

Weirdly, I feel like Intel's pricing has gotten worse rather than better since Ryzen.

I mean during the Sandy-Haswellish time frame, the i5/i7 were hands down the fastest consumer CPUs, yet the pricing remained pretty steady and affordable. The lack of innovation or real performance increases sucked, but you know : $200 + or - for a good i5, $300 + or - for a good i7.

Now prices have continued to climb, dumb vanity SKU 8086k for a higher fee, climbing ASP for i3 and up. New i9 for a stack of cash + it needs $$$ cooling.

I remember when AMD figured out they could raise prices after the X2s came about, suddenly we had $1k chips, with $300 2006 money being the 'value' X2 lol. And they STAYED with that a fair ways after Conroe came and crushed everything.

Idk, it's weird. Intel went from moderate pricing to MORE expensive after Ryzen. Now of course at least they gave us some actual improvements vs 7th gen finally, but man. Not really much of a bargain imho.

 

[Shivansps]

Weirdly, I feel like Intel's pricing has gotten worse rather than better since Ryzen.

I mean during the Sandy-Haswellish time frame, the i5/i7 were hands down the fastest consumer CPUs, yet the pricing remained pretty steady and affordable. The lack of innovation or real performance increases sucked, but you know : $200 + or - for a good i5, $300 + or - for a good i7.

Now prices have continued to climb, dumb vanity SKU 8086k for a higher fee, climbing ASP for i3 and up. New i9 for a stack of cash + it needs $$$ cooling.

I remember when AMD figured out they could raise prices after the X2s came about, suddenly we had $1k chips, with $300 2006 money being the 'value' X2 lol. And they STAYED with that a fair ways after Conroe came and crushed everything.

Idk, it's weird. Intel went from moderate pricing to MORE expensive after Ryzen. Now of course at least they gave us some actual improvements vs 7th gen finally, but man. Not really much of a bargain imho.

Yup prices went up, first the 1800X playing with the idea of having a $500 sku in desktop, then Intel HEDT went from $1000 for the top sku to $2000, then AMD also went to $2000 HEDT with the 2990WX and Intel petty much cemented the $500 desktop SKU with the 9900K.

We are a lot better in price/perf NOW, but this is going to hurt in the long run.

 

[mattiasnyc]

I think that's missing the point a bit though.

I could be wrong, but I feel like I saw Intel putting out higher priced CPUs that were actually better chips. There's zero reason for AMD for example to price the 2990WX lower than what they can in relation to Intel.

However, we're not just "extending" the range upwards, we're getting better CPUs overall. What was the core count before Ryzen came out? For how much money? I was considering buying a hexacore CPU before Ryzen, and it was a pricey proposition on Intel's HEDT platform. Then there was Ryzen. More for less.

I think we're seeing progress in general.

 

[dlerious]

Yup prices went up, first the 1800X playing with the idea of having a $500 sku in desktop, then Intel HEDT went from $1000 for the top sku to $2000, then AMD also went to $2000 HEDT with the 2990WX and Intel petty much cemented the $500 desktop SKU with the 9900K.

We are a lot better in price/perf NOW, but this is going to hurt in the long run.

The 7700K was released 1/2017 for $340, so the 1800X wasn't too bad at $500, especially considering it was double the cores/threads and the 6900K was over $1000 at the time - 6950X was $1700. At the $2000 price point, you have 32-cores on AMD and 18-cores for Intel. AMD has dropped prices on the equivalent Zen+ CPUs compared to 1st gen.

 

[Arkaign]

The 7700K was released 1/2017 for $340, so the 1800X wasn't too bad at $500, especially considering it was double the cores/threads and the 6900K was over $1000 at the time - 6950X was $1700. At the $2000 price point, you have 32-cores on AMD and 18-cores for Intel. AMD has dropped prices on the equivalent Zen+ CPUs compared to 1st gen.

This is true, though it's mostly concerning products that are ultra niche as far as consumer S2011 / $1k+ stuff.

The regular consumer CPUs climbed on Intel's side notably lately :/ Feels a little like boiling the

 

[coercitiv]

The 1800X screamed of being born to bring the best possible results in press reviews. Besides that one could get almost the same performance by overclocking the 1700. Imagine how that would look in the current Intel offering, with 9600K, 9700 and 9900K all being 8c/16t chips separated by binning. Nobody would care if Intel asked $500 for a 5Ghz 8c/16t if they also offered 4.5Ghz 8c/16t for $300 or lower. We would be swimming in value.

It's not about how expensive flagship is, it's all about how close the cheaper SKUs are to the flagship in terms of performance.

 

[TheGiant]

The 1800X screamed of being born to bring the best possible results in press reviews. Besides that one could get almost the same performance by overclocking the 1700. Imagine how that would look in the current Intel offering, with 9600K, 9700 and 9900K all being 8c/16t chips separated by binning. Nobody would care if Intel asked $500 for a 5Ghz 8c/16t if they also offered 4.5Ghz 8c/16t for $300 or lower. We would be swimming in value.

It's not about how expensive flagship is, it's all about how close the cheaper SKUs are to the flagship in terms of performance.

QFT
we like that AMD approach but it is not that good for business
I remember buying c2d e6300 (1,86GHz default, instant oced to 3,4GHz) and I got better performance that the 1000EUR c2extreme. The same with Athlon XP TB 1700+ 1466MHz which I oced to 2200MHz on nforce2 with 200MHz "fsb".
At this time we at least get for 1000+ EUR moar coarz, not the pretty much the same (cache, fsb) just oced chips like AMD FX or the P4EE at that time.
From my point of view, intel with 9900K made the 7900X except avx512 workloads obsolete and also the 12C SKL-X and TR too.
The first chip I would consider above 9900K is the 2950X.

 

[coercitiv]

we like that AMD approach but it is not that good for business

In the context of "good for business" the 9900K is actually quite the compromise. It may be a complete win performance wise, and it will certainly be needed in the not so distant future to face Ryzen 3000 series, but purely from a profit perspective the larger 8c/16t die only exacerbated Intel's capacity constraints (9000 series i5 SKUs use the larger die as well), hence their window for maximized profits for this year. And if you count each missed Intel buy as an AMD sale, the loss is two fold.

Now, granted, being in a place where you can't make enough product as the market is buying is certainly not a bad spot to be in, but that's the thing with investors: they don't look at how much money you're making them, they look at how much money you should be making for them

 

[mattiasnyc]

From my point of view, intel with 9900K made the 7900X except avx512 workloads obsolete and also the 12C SKL-X and TR too.
The first chip I would consider above 9900K is the 2950X.

You have to take the platform into account though. There are a fair amount of professionals that look at those CPUs for their capacity but also need better I/o options than the non-x-series can provide. All it takes is massive video files or massive audio-loads into software and 3xNVME drives straight into the CPU seems more appealing (without stealing PCIe I/o from slots).

 

[rbk123]

Weirdly, I feel like Intel's pricing has gotten worse rather than better since Ryzen.

Probably more on the high end due to availability/capacity. Since their capacity is well below demand on the high end, those prices will be high. When they get their capacity problems fixed, things will level as they attempt to sell beyond the wave of those that have to have the best chip.

 

[Arkaign]

Probably more on the high end due to availability/capacity. Since their capacity is well below demand on the high end, those prices will be high. When they get their capacity problems fixed, things will level as they attempt to sell beyond the wave of those that have to have the best chip.

I hope so. I fear that similar to GpUs, they're attempting to raise the bar to a new normal. It's the kind of thing that can be brutal to crack. Might take AMD having another Athlon 64/X2 moment to cause Intel to totally reset back to reasonable pricing. Merely being 'good' seems to be creating an expensive proposition for gamers.

 

[TheGiant]

You have to take the platform into account though. There are a fair amount of professionals that look at those CPUs for their capacity but also need better I/o options than the non-x-series can provide. All it takes is massive video files or massive audio-loads into software and 3xNVME drives straight into the CPU seems more appealing (without stealing PCIe I/o from slots).

Well the argumentation here is about the moar coarzolol performance whole day running cinebench..../sarcasm off
What you are saying is to me not desktop but profi workload. No problem buying the TR2950X. But that is not the reaction to 9900K.
As I said it is too good and at 5GHz all core is more than too good for next like 2 years.
It will take a lot of bloatware from MS to use the 16T by default in next 2 years...

 

[TheELF]

I mean during the Sandy-Haswellish time frame, the i5/i7 were hands down the fastest consumer CPUs, yet the pricing remained pretty steady and affordable. The lack of innovation or real performance increases sucked, but you know : $200 + or - for a good i5, $300 + or - for a good i7.

The quad 8350k is now priced $180 while the 4/8 got replaced by 6 real cores and now costs $260 for the 8600k
that is cheaper for faster and more.
Yes now they brought acme/mickey mouse level of absurd CPUs to the mainstream and they ask a high price for it,good for them.

 

[Markfw]

The 7700K was released 1/2017 for $340, so the 1800X wasn't too bad at $500, especially considering it was double the cores/threads and the 6900K was over $1000 at the time - 6950X was $1700. At the $2000 price point, you have 32-cores on AMD and 18-cores for Intel. AMD has dropped prices on the equivalent Zen+ CPUs compared to 1st gen.

Rounding up from $1800 to $2000 is quite a stretch on the AMD side. But the Intel really is $2000 (MSRP) Retail @Newegg right now is $1729 for the 2990wx and $2050 for the 7980XE

 

[piokos]

Yeah I guess it doesn't really matter since that Intel 8 core will cost as much as an Intel 8 core...which is a crap ton of money that no one wants to spend. That's a shame. Its a shame the CPU will cost at least $500. Why even release it? Intel would be better off simply telling us, "We could release an 8 core on mainstream, but NAH". Might as well not release it with how high they will price it. lol. Its fine though. Intel just does Intel.

It's hard to grasp why 9900K price is such a shock. Just 1.5 year ago AMD launched 1800X at $500 and it was called good value for money ("great" being reserved for 1700).
Quite a few early adopters accepted that price.

Intel can go the same way and lower the price after the initial batch. But since the sales may not be an issue, you can't really expect a drop like with Ryzen (which fairly quickly moved to ~$350).

You have to consider what you're paying for. When Ryzen 7 launched, it was the highest core count available for consumer platforms. However, it had some compatibility issues and wasn't really that quick in general - an 4C/8T i7 was still the CPU of choice for some tasks.

9900K launches as the fastest (by good margin) consumer CPU available - regardless of use scenario. Sure, it's not the most cost effective, but that's not what such a product is about.
For a particular use case - strongly multi-thread or strongly single-thread computing, gaming, media tasks etc - there are more optimal, cheaper alternatives (from both manufacturers).

 

[moonbogg]

It's hard to grasp why 9900K price is such a shock. Just 1.5 year ago AMD launched 1800X at $500 and it was called good value for money ("great" being reserved for 1700).
Quite a few early adopters accepted that price.

Intel can go the same way and lower the price after the initial batch. But since the sales may not be an issue, you can't really expect a drop like with Ryzen (which fairly quickly moved to ~$350).

You have to consider what you're paying for. When Ryzen 7 launched, it was the highest core count available for consumer platforms. However, it had some compatibility issues and wasn't really that quick in general - an 4C/8T i7 was still the CPU of choice for some tasks.

9900K launches as the fastest (by good margin) consumer CPU available - regardless of use scenario. Sure, it's not the most cost effective, but that's not what such a product is about.
For a particular use case - strongly multi-thread or strongly single-thread computing, gaming, media tasks etc - there are more optimal, cheaper alternatives (from both manufacturers).

I predicted that back in June. Not too bad, eh? And yes, it's TOO expensive. The CPU landscape has been upset radically, so it's still changing and the dust hasn't settled yet, therefore prices aren't reliable yet. This means that it doesn't matter what you used to get for your money back when intel was charging almost $800 for a 6 core, $1200 for an 8 core or that an 1800X was $500.

What matters now is that the 2700X is $309 on Newegg right now. That's insane. So $500 for the 9900K is ridiculous just like many people knew it would be. It's also insane because we all know 7nm Zen 3 is going to rock the damn house and lots of people are waiting for those chips, so Intel should try to incentivize the purchase by lowering the price, but we all know that won't happen because Intel just does Intel. Meanwhile Zen 3 approaches like a giant ROCK FROM SPACE.

 

[piokos]

I predicted that back in June. Not too bad, eh? And yes, it's TOO expensive. The CPU landscape has been upset radically, so it's still changing and the dust hasn't settled yet, therefore prices aren't reliable yet. This means that it doesn't matter what you used to get for your money back when intel was charging almost $800 for a 6 core, $1200 for an 8 core or that an 1800X was $500.

You can't compare mainstream and HEDT platforms that way. There's no smooth upgrade path from one to another, the socket is huge and with marginal popularity, the features promote particular use cases (i.e. not so many HEDT CPUs with iGPU).

Assuming similar performance and features, an 8 core mainstream CPU can (and should!) cost more than an 8 core HEDT one - just based on the fact that it's on a mainstream socket.

What matters now is that the 2700X is $309 on Newegg right now. That's insane. So $500 for the 9900K is ridiculous just like many people knew it would be.

You're paying the premium for having the fastest mainstream CPU. If something better comes out (from either AMD or Intel), the price will go down ($400 seems feasible). Until then, there's really no reason for Intel to sell this CPU cheaper, so why would they?
There's only 2 ways to improve revenue: selling more or selling for more. And Intel can't improve their market share at this point.

It's also insane because we all know 7nm Zen 3 is going to rock the damn house and lots of people are waiting for those chips, so Intel should try to incentivize the purchase by lowering the price, but we all know that won't happen because Intel just does Intel. Meanwhile Zen 3 approaches like a giant ROCK FROM SPACE.

I assume you mean Zen 2. Of course, this will be a big performance jump for AMD. But 7nm consumer CPUs are months away.
Intel sells 40 mln CPUs each month. Let's say they're going to add a $100 premium for as long as they can. That's 4 bln USD extra revenue with no additional costs, resulting in over 3 bln USD profit (after tax). MONTHLY.

 

[moonbogg]

OK so what are we arguing about again? I'm lost now.

 

[PotatoWithEarsOnSide]

They aren't selling 40m 9900Ks per month, that's for sure. I'm willing to bet that there aren't even 4k of them out in the wild yet.
Everone loves hyperbole.
Time to return from Cuckooland.

 

[PotatoWithEarsOnSide]

I love also how revenue now also equals profit according to the Cuckooland post.
If Intel were making $3bn profit per month, after tax, from just their CPUs, even I'd have shares in them.

I'm going to pluck some figures out of my ass now, and use them to tell you how phenomenal Pacman is.
Nope, Cuckooland is overpopulated and there's no way that common sense gets past border control anyway.

 

[epsilon84]

The 1800X screamed of being born to bring the best possible results in press reviews. Besides that one could get almost the same performance by overclocking the 1700. Imagine how that would look in the current Intel offering, with 9600K, 9700 and 9900K all being 8c/16t chips separated by binning. Nobody would care if Intel asked $500 for a 5Ghz 8c/16t if they also offered 4.5Ghz 8c/16t for $300 or lower. We would be swimming in value.

It's not about how expensive flagship is, it's all about how close the cheaper SKUs are to the flagship in terms of performance.

Hmmm, I see your point but the 9700K actually isn't *that* far off a 9900K in 'overall CPU performance', for example:
https://www.techpowerup.com/reviews/Intel/Core_i7_9700K/19.html

Compared to the 9900K its about 14% slower across the CPU benchmark suite and equal in gaming. That's not a terrible step down in performance relative to the 9900K IMO. Of course, highly MT workloads will see a bigger performance disparity, but conversely lightly threaded workloads will be virtually the same.

If the 9700K was a 4.5GHz 8C/16T chip like you suggested almost nobody would buy the 9900K, since people will simply overclock it to the same clocks and save $200. It would end up being exactly the same situation as the 1700 vs 1800X like you said, which isn't actually a 'good' thing if you are Intel trying to convince people to buy your top of the line flagship SKU. You need more than a simple clockspeed boost to differentiate between an i7 and i9.

I do think Intel has potentially one last 'trick' up its sleeve WR to competing against Zen 2, even if they are stuck at 14nm for the entirety of 2019. Assuming 10nm is a total write off and won't be seen until 2020, how can Intel refresh its CPU lineup next year to stay relevant with Zen 2 performance? By re-enabling HT on the lower end i5/i7 SKUs.

Imagine a '10600K' being 6C/12T and '10700K' being 8C/16T, it won't cost Intel anything extra in terms of costs to simply enable HT, but it allows their entire lineup to receive a ~25% boost in MT performance and thus stay relevant with Zen 2 at least on a price/performance level. Of course, this would basically render the 9900K irrelevant, but that is still better than being steamrolled by Zen 2 in performance across the entire i5/i7 lineup.

 

[coercitiv]

If the 9700K was a 4.5GHz 8C/16T chip like you suggested almost nobody would buy the 9900K, since people will simply overclock it to the same clocks and save $200. It would end up being exactly the same situation as the 1700 vs 1800X like you said, which isn't actually a 'good' thing if you are Intel trying to convince people to buy your top of the line flagship SKU. You need more than a simple clockspeed boost to differentiate between an i7 and i9.

Agreed, however my post was written as a remark to those who say 1800X == 9900K in terms of proposal to consumers (relative to their respective timeline), which it was not.

I do think Intel has potentially one last 'trick' up its sleeve WR to competing against Zen 2, even if they are stuck at 14nm for the entirety of 2019. Assuming 10nm is a total write off and won't be seen until 2020, how can Intel refresh its CPU lineup next year to stay relevant with Zen 2 performance? By re-enabling HT on the lower end i5/i7 SKUs.

Agreed again, in fact this reasoning was exactly what made me say the 9900K is likely going to be eclipsed by Intel themselves in 2019.