Intel Reportedly Puts Up 5GHz Core i9-9990XE CPU For Auction

[Guest]

An alleged document reveals Intel's plan to auction the unannounced Core i9-9990XE 14-core processor.

Intel Reportedly Puts Up 5GHz Core i9-9990XE CPU For Auction : Read more

 

[TechyInAZ]

I'm dumbfounded. Intel is now beyond unpredicable.

 

[hotaru251]

if only these companies would band together and lowball the sale price just to give Intel the shaft...

What most companies in tech industry need to do.

Stop letting this type of crap happen :|

 

[spentshells]

What's your best guess? 3500 I'd say

 

[InvalidError]

I'm dumbfounded. Intel is now beyond unpredicable.

The ultimate form of pricing parts up to whatever the market will bear when Intel's own pricing and shortages show that Intel's own pricing is too low to maintain stocks of its more popular SKUs: let the market decide the final price via auction. May the best wallets win!

 

[rabbit4me1]

this is your chance AMD. to stay out of the stupid race!!!

 

[Jim90]

Desperate measures indeed, Intel.
I'm guessing this pretty closely validates AMD's upcoming CPU tech e.g. Ryzen 9 3850x 'mainstream' processor at 5.1 boost TDP 135W - if this is not far off then we shouldn't be surprised with Intel's response.
Now imagine what Threadripper will be like.
Good times ahead!

 

[rantoc]

If indeed true (it's intel after-all so wouldn't surprise me, the greed seems to have no bounds with them) the price will be a kidney and perhaps a liver or 5...

 

[justin.m.beauvais]

So, Intel is taking their absolute best cores and auctioning them off as these monstrosities. Forget about overclocking them, the vast majority of that headroom will be taken up by Intel pushing their silicon to the very edge. Heck, these specialty systems are probably going to need the absolute best water cooling that money can buy... that or their own chillers. AMD's demo must have really gotten under Intel's skin. This will be an astoundingly fast CPU... at an astoundingly high price... with astoundingly high temps and power draw.

It isn't the worst idea Intel has ever had... performance for the sake of performance isn't a bad thing... but this feels like AMD and their high end 9000 series FX CPUs. Push the clocks higher, just do it, wring everything you can out of the chip... except instead of trying to compete with CPUs simply beyond their means, this is Intel throwing everything they have at remaining top dog.

I hope AMD really kicks them in the teeth with Zen 2. AMD might just have what it takes to break Intel's hold on the speed crown. That was what it took last time to get Intel to REALLY innovate and what brought us the Core architecture. We can only hope that things go so well for the consumer this time around.

 

[atomicWAR]

So this is what it feels like to laugh and cry at the same time. Thanks for that Intel!!!

 

[Supahos]

This I actually don't have a problem with anyone looking for this cpu for basically any purpose basically has a gouge me sign. I'm actually surprised they didn't make it an open auction to run prices up.

My guess is that even top of.the line aios won't come close to cooling it so they sell it to system designers to put the job of cooling it on them

 

[redgarl]

Just keep buying Intel, It works! It works for everyone...

 

[redgarl]

So, Intel is taking their absolute best cores and auctioning them off as these monstrosities. Forget about overclocking them, the vast majority of that headroom will be taken up by Intel pushing their silicon to the very edge. Heck, these specialty systems are probably going to need the absolute best water cooling that money can buy... that or their own chillers. AMD's demo must have really gotten under Intel's skin. This will be an astoundingly fast CPU... at an astoundingly high price... with astoundingly high temps and power draw.

It isn't the worst idea Intel has ever had... performance for the sake of performance isn't a bad thing... but this feels like AMD and their high end 9000 series FX CPUs. Push the clocks higher, just do it, wring everything you can out of the chip... except instead of trying to compete with CPUs simply beyond their means, this is Intel throwing everything they have at remaining top dog.

I hope AMD really kicks them in the teeth with Zen 2. AMD might just have what it takes to break Intel's hold on the speed crown. That was what it took last time to get Intel to REALLY innovate and what brought us the Core architecture. We can only hope that things go so well for the consumer this time around.

I don't want them to innovate, I want them to be crush beyond their pseudo-monopoly recovery. I want the market to be disrupt for good.

 

[Strider_X]

I'm always amazed at the angry posts when product pricing is discussed. Intel is operating in a free market economy. They saw a potential demand and chose a unique way to supply it. They have an available product and hope to fill a niche. If consumers like the value proposition then they will pay auction prices, if not then the chips will go unsold. These are luxury items and the market will determine if they are successful.
Everyone should just sit back and watch the show, both Intel and their competitors will learn from the exercise.

 

[Olle P]

This could get interesting...
How many CPUs would there be in a lot on the auction?
How much does the computer builders think their customers are willing to cough up for a computer with this CPU?

My guess is that a computer builder would be interested to pay fairly big money on a few of these, just for the bragging rights to offer it in their product portfolio.
So a combined total sale of one or two dozen CPUs of this SKU seems within reach...

 

[JamesSneed]

This is kind of fun having Intel sell binned CPU's. Albeit I do wonder what is going to happen to everything up to the 9960X when AMD brings 16 cores mainstream this summer.

 

[AgentLozen]

I don't want them to innovate, I want them to be crush beyond their pseudo-monopoly recovery. I want the market to be disrupt for good.

If AMD hypothetically crushed Intel beyond recovery, then AMD would become the monopoly and we'd be back to the 2015 trend of 2% performance improvements each generation.

The market is healthiest when there is competition. We NEED AMD and Intel both at their best.

 

[JamesSneed]

I don't want them to innovate, I want them to be crush beyond their pseudo-monopoly recovery. I want the market to be disrupt for good.

If AMD hypothetically crushed Intel beyond recovery, then AMD would become the monopoly and we'd be back to the 2015 trend of 2% performance improvements each generation.

The market is healthiest when there is competition. We NEED AMD and Intel both at their best.

I agree. Oddly enough there are rumors floating around today that Intel is thinking of buying AMD. Seems very unlikely unless the rumor has a little truth to it and Intel is thinking of buying the GPU division.

https://www.extremetech.com/computing/283865-rumors-once-again-surface-that-intel-could-purchase-amd

 

[InvalidError]

This is kind of fun having Intel sell binned CPU's. Albeit I do wonder what is going to happen to everything up to the 9960X when AMD brings 16 cores mainstream this summer.

If you really need 12+ cores right now, chances are you also could use either quad-channel memory or extra PCIe lanes and would still be better served by LGA2xxx or TR4 CPUs than anything AMD could put on AM4.

16C32T on AM4 sounds nice in theory and for bragging rights, but I am skeptical it'll be useful in practice beyond a limited number of particular cases that fit within L3 caches due to limited memory bandwidth per core and relatively limited IO. It'll make more sense with DDR5 pushing baseline bandwidth to 4GT/s and beyond in 2020-2021.

 

[InvalidError]

If AMD hypothetically crushed Intel beyond recovery, then AMD would become the monopoly and we'd be back to the 2015 trend of 2% performance improvements each generation.

We are getting ~5%/year incremental IPC increases for Intel because Intel has squeezed the amount of instruction-level parallelism it can extract from typical x86 code almost dry years ago and being stuck on 14nm doesn't afford much extra room to push it any further beyond small process and architecture tweaks. The main reason AMD could achieve greater IPC gains per generation is because it was so far behind Intel. Now that AMD's Zen 2 appears to have caught up with Intel's seemingly insurmountable lead, AMD's IPC gains are going to get much smaller too since they'll be bashing their skulls against the same level of extracting more ILP from dry x86 rocks as Intel has been for the past several years.

 

[kinggremlin]

It's been a common practice for years to bin golden chips with higher factory clocks than any publicly available chips and sell them directly to companies with the needs and deep enough pockets. AMD undoubtedly does this too. The only slightly unusual part about this is the auction format of selling them.

 

[alexlf]

Intel has became pathetic. What a joke!

 

[TCA_ChinChin]

For people saying 16/32 cores/threads on AM4 is not worth it for a variety of reasons, my argument is that if the price is right, even if such a chip offers only slightly better productivity performance than current core count products, it'll be worth it. I don't care what it costs AMD, or how bandwidth limited, or "useless", that many cores are, if AMD can put out 16/32 core/thread chips at the same price as Intel's 8/16 core/thread chips, it'll be better, especially if matches performance in single threaded ares. As for IPC gain, again, it doesn't matter if AMD has reached the same plateau as Intel, as there are other methods to extract performance that can be and should have been (by Intel) used to do so. A good example is what Intel is doing NOW in 10nm with (rumored) increased cache, and architecture redesign, 3d 'Foveros' technology. Instead of banging their heads on the x86 rocks to extract 5-10% IPC gains each year, Intel should have invested that time and effort in smarter ways to tackle these problems as they are doing now when AMD has finally caught up.

 

[AgentLozen]

As for IPC gain, again, it doesn't matter if AMD has reached the same plateau as Intel, as there are other methods to extract performance that can be and should have been (by Intel) used to do so. A good example is what Intel is doing NOW in 10nm with (rumored) increased cache, and architecture redesign, 3d 'Foveros' technology. Instead of banging their heads on the x86 rocks to extract 5-10% IPC gains each year, Intel should have invested that time and effort in smarter ways to tackle these problems as they are doing now when AMD has finally caught up.

Well said.

In 2012, at the Ivy Bridge launch, I don't think it was an IPC plateau that caused Intel to cripple their CPUs with low quality TIM between the IHS and CPU. Ivy Bridge represented a 22nm shink. How much better was it than Sandy Bridge? 5%?

Then a year later, I saw the Haswell review on Tomshardware "Haswell is Faster, Desktop Enthusiasts Yawn". On the conclusion page, the writer states that, "The Core i7-4770K, specifically, is a bit faster than the -3770K it replaces—but only because of IPC improvements. It runs at the same 3.5 GHz and sports the same four cores otherwise." Haswell was only a small improvement over Ivy Bridge and only a small improvement over Sandy Bridge for that matter.

I can keep going about how Broadwell, Sky Lake, and Kaby Lake were all fairly unimpressive despite the 14nm shrink. What's important is that when AMD put the pressure on Intel in 2017, we IMMEDIATELY saw 6 core Coffee lake. Then a year later, we get 8 core Coffee Lake with a soldered IHS. InvalidError you are ABSOLUTELY correct that IPC is getting harder and harder to improve on. That doesn't mean there aren't other ways to improve performance.

 

[InvalidError]

As for IPC gain, again, it doesn't matter if AMD has reached the same plateau as Intel, as there are other methods to extract performance that can be and should have been (by Intel) used to do so. A good example is what Intel is doing NOW in 10nm with (rumored) increased cache, and architecture redesign

Increasing caches is EXTREMELY expensive in terms of die area and performance cost, which is why cache size bumps are so few and far between. Intel's architecture "changes" are relatively minor tweaks and Intel has been doing similar minor tweaks with most generation, nothing new there. Intel hasn't had a fundamentally new architecture in ~10 years, largely because they'd gain little to nothing from such an effort at this point. Massive architecture overhaul are only relevant while searching for the most optimal way of implementing an ISA. Once you have found a general architecture that is close to theoretically optimal, there is nothing left to do beyond that other than incrementally tweak it to get closer to theoretical limits. That's where AMD's Zen 2 and Intel are on x86.

There is no magical "smarter way" of getting more performance out of x86. ILP/IPC is effectively maxed out, clock frequencies are about as high as process technology can manage, the only thing left is increasing core counts up to whatever makes sense for available memory bandwidth and IO.

Having more cores than what the memory bandwidth and latency can bear is detrimental to performance: more active cores mean less L3 cache per active core, more cache lines getting thrashed more often resulting in a massive increase in memory controller load, cores wasting even more time having to re-fetch stuff from RAM, wasting even more memory controller and CPU time. While AMD may be able to cram 16C32T on AM4, I suspect most heavily threaded desktop workloads will benefit from parking cores or restricting the number of cores/threads applications can use because of this.