Anandtech: Intel Announces 8 Core i9-9900KS: Every Core at 5.0 GHz, All The Time

Is everyone forgetting about Sunny Cove? A bit premature to be taking a victory lap when this has the familiarity of the A64/Core2 timeline all over again.
And yes, Jim Keller's work on Zen is still being fleshed out - I'm as excited about the Zen 3000 series as anyone else.

But AMD hasn't demonstrated they can maintain performance increases without his engineering acumen at the helm.
 
I'd be surprised if it had any more headroom than a standard 9900K. This is probably just binned 9900K chips that they know can reliably run at 5Ghz all core.

I would too, but we won't actually know until we get these in hand.
 
Neither a tic nor a toc. Impressive iNtel, very impressive.
The least you could do is toss in a free game or 2, ball cap, t-shirt and sign the damn die.
 
This is a joke, right? It is nothing but a 300mhz factory overclock, and they are obviously binning from the regular 9900k pool. Wow has Intel become desperate.

Well it is. Its had nothing but negative PR lately with all the security vulnerabilities that have been exposed. Fixes mitigate much of Intel's IPC advantage over AMD. This brings them into closer parity. If Zen 2 clocks the way its rumored to, then Intel will lose much of its clock speed advantage as well. Intel's almost never had a core count advantage over AMD, so they don't have that either. So yeah, I would expect Intel to start throwing out all kinds of crazy models to see what sticks in terms of sales.
 
Can’t the 7700k, 8700k, 9700k and 9900k all OC to 5ghz all-core with a proper mobo and duel fan AIO watercooling?

This will need the same thing to get 5ghz on all cores right?
 
Well it is. Its had nothing but negative PR lately with all the security vulnerabilities that have been exposed. Fixes mitigate much of Intel's IPC advantage over AMD. This brings them into closer parity. If Zen 2 clocks the way its rumored to, then Intel will lose much of its clock speed advantage as well. Intel's almost never had a core count advantage over AMD, so they don't have that either. So yeah, I would expect Intel to start throwing out all kinds of crazy models to see what sticks in terms of sales.

Ideally this panic from Intel will mean good things for consumers. Not just in terms of crazy products but, hopefully, in terms of pricing as well. It would be nice for Intel and AMD to actually compete for once without Intel pulling illegal tactics to keep their lead.
 
Can’t the 7700k, 8700k, 9700k and 9900k all OC to 5ghz all-core with a proper mobo and duel fan AIO watercooling?

This will need the same thing to get 5ghz on all cores right?

I don't know if it can be done on an AIO as I've never tried one on those CPU's. However, I've had a lot of CPU's that can clock to 5GHz. I think my 8700K couldn't quite hit that. I think I only ever got it up to 4.9GHz. My 7600K, 7700K, 7740X, and 9600K all certainly could.
 
I don't know if it can be done on an AIO as I've never tried one on those CPU's. However, I've had a lot of CPU's that can clock to 5GHz. I think my 8700K couldn't quite hit that. I think I only ever got it up to 4.9GHz. My 7600K, 7700K, 7740X, and 9600K all certainly could.

Roger that.
 
Can’t the 7700k, 8700k, 9700k and 9900k all OC to 5ghz all-core with a proper mobo and duel fan AIO watercooling?

This will need the same thing to get 5ghz on all cores right?

The 9900k is the one in that group known to have temperature and power consumption issues when pushed to 5ghz or more, beyond what the others do. So yes, for a typical 5ghz all core OC the use of a 240mm AIO cooler or a custom water loop is more or less compulsory.

However, that's why I made the comment about this new SKU being binned. A really good 9900k can do 5.2-5.3ghz with that same 240mm AIO cooler. Intel is going to take those golden CPUs and sell them as this new 9900ks SKU and advertise 5ghz, which they'll be able to do at a low enough TPD to not require that much cooling. I expect they'll still run hot.
 
Is that before or after you have to disable Hyperthreading due to security issues?

the same security issues AMD present due SMT? (aka hyperthreading) AMD even state that in their security page and then it was patched both for intel and AMD via OS mitigation.

spectre2.PNG
 
...no one wants your midget porn and MMO gaming data

Well.... what cha go for that midget porn? :)


Or it may stand for "stupid" as in "Thanks for the money and support, STUUUUPID!"

Maybe, maybe not. Upgrading from a 6700K or newer? Not a good idea. Coming from a Core 2 Duo or a whole new build or whatever? Sure, I can see it being a decent purchase. Buy the best you can afford.
 
Is everyone forgetting about Sunny Cove?

Yeah, I am finding it interesting that people would rather make pointless fun of Intel, than consider new technology.

I posted this story, which is just an 8086K redux, and bit of binning an nothing more, and an Ice Lake (Sunny Cove plus next gen graphic) story about the same time.

The Ice Lake story, with new benchmarks for it's next gen IGPU, got 3 replies, this one with nothing new got 50+ replies (at time of posting).

So interest in new technology seems negligible compared to hurling insults at Intel.
 
Yeah, I am finding it interesting that people would rather make pointless fun of Intel, than consider new technology.

I posted this story, which is just an 8086K redux, and bit of binning an nothing more, and an Ice Lake (Sunny Cove plus next gen graphic) story about the same time.

The Ice Lake story, with new benchmarks for it's next gen IGPU, got 3 replies, this one with nothing new got 50+ replies (at time of posting).

So interest in new technology seems negligible compared to hurling insults at Intel.
Because most here aren't interested in incremental iGPU gains.
 
All I know is that Pentium 4 processors hit 3.8ghz in 2005. Up to that point we had been seeing faster and faster processors. For the past nearly 15 years we have been stuck in the same range of speeds. The biggest things have been more cores, often going backwards in speeds. While enterprise has sometimes been able to take better advantage of the additional cores for the most part consumer applications are still mostly not multi-threaded.

Of course there have been other big advances with new instruction sets, efficiency improvements, etc but to me it seems like they hit some massive brick wall 15 years ago when it comes to just raw single thread clock speeds while everything else around this most crucial component of computers kept moving forward with more and more speed. Storage technology was still IDE back then, buses were still PCI, graphics were in AGP slots, memory, networking, etc. Everything else including processors prior to then all sort of moved along together with new speed improvements. Then BAM, processors just stagnated and everything else kept moving forward with new flat our raw speed increases. But the speed of the most basic single computation (in many respects anyways) has just been dead stagnant.

I'm sure it will not be cheap but I am happy that over the past couple of years we are finally seeing stock raw clock speeds getting over 4ghz and seeing 5 is even better.
 
  • Like
Reactions: Meeho
like this
All I know is that Pentium 4 processors hit 3.8ghz in 2005. Up to that point we had been seeing faster and faster processors. For the past nearly 15 years we have been stuck in the same range of speeds. The biggest things have been more cores, often going backwards in speeds. While enterprise has sometimes been able to take better advantage of the additional cores for the most part consumer applications are still mostly not multi-threaded.

Of course there have been other big advances with new instruction sets, efficiency improvements, etc but to me it seems like they hit some massive brick wall 15 years ago when it comes to just raw single thread clock speeds while everything else around this most crucial component of computers kept moving forward with more and more speed. Storage technology was still IDE back then, buses were still PCI, graphics were in AGP slots, memory, networking, etc. Everything else including processors prior to then all sort of moved along together with new speed improvements. Then BAM, processors just stagnated and everything else kept moving forward with new flat our raw speed increases. But the speed of the most basic single computation (in many respects anyways) has just been dead stagnant.

I'm sure it will not be cheap but I am happy that over the past couple of years we are finally seeing stock raw clock speeds getting over 4ghz and seeing 5 is even better.

I still think we are stuck. The x5690 could do 5ghz on its 6 cores back in 2011 all they do now is eat up the overclocking headroom and push out chips already at there limit. All we have gained since then is 2 cores and some ipc gains. server chips keep moving because there is still room to move as they can reinvest any efficiency gains into more cores but that of course only helps when programs can actually use the cores.
 
  • Like
Reactions: Meeho
like this
I still think we are stuck. The x5690 could do 5ghz on its 6 cores back in 2011 all they do now is eat up the overclocking headroom and push out chips already at there limit. All we have gained since then is 2 cores and some ipc gains. server chips keep moving because there is still room to move as they can reinvest any efficiency gains into more cores but that of course only helps when programs can actually use the cores.

I agree to an extent but that was overclocked speeds on a non-server/pro workstation board. Native speeds are what I'm after as there is no guarantee of stability when overclocking. Even the server chips have hardly moved. They have actually for the most part regressed to slower clock speeds in order to pack more cores in is all. At their fastest clock speeds pushing up towards the desktop parts speeds they are down to the same number of cores as desktop parts.

When it comes to raw basic single core speeds Moore's Law fell flat on its face 15 years ago.
 
For that there is {M}ediocreNOocp.com

Come all of my iGPU enthusiast friends to my ALL NEW iGPU softcore gaming site SoftOCP.com

Or for our over 65 crowd that still has a hard-time getting a hard-on for gaming of any kind I've introduced a new site just for you! Just head on over to FlaccidOCP.com and get the love back into your gaming life!
 
I agree to an extent but that was overclocked speeds on a non-server/pro workstation board. Native speeds are what I'm after as there is no guarantee of stability when overclocking. Even the server chips have hardly moved. They have actually for the most part regressed to slower clock speeds in order to pack more cores in is all. At their fastest clock speeds pushing up towards the desktop parts speeds they are down to the same number of cores as desktop parts.

When it comes to raw basic single core speeds Moore's Law fell flat on its face 15 years ago.

the speeds we see now are just overclocked chips. I dought there will be any overclocking headroom at all on a 5ghz 9900k. There was nothing stoping intel from binning westmere chips farther and putting out a completly stable 5ghz 6core chip they just had no reason to play games like that when people could just buy and overclock the chips. we know its possible as it wasnt all that hard to make one of those chips run at that speed. Just about all chips are limited by power consumption and it is and has always been a trade off between cores and clock speed as a result. we have made small gains in efficiency that have helped but really the only reason we have these large many many core chips is we put them on a socket that can push 250w+ with a large enough die that they can stay cool. Unforcuantly that makes the chips more expencive to produce so we wont get mainstream chips capable of pushing and despiating enough power to get up there in clock speed. If intel wanted to put a quad core on the 3647 socket using 250w they could get the clock speeds waaayyyy up (if they actually engineer the chip for such) but there isnt realy a market for that
 
now when amd will be tapped out on node shrink. nothing is said past zen 2 as far as i know. it will be intel playing around for another round? consider what they can do on 14 nm still 10 nm would probs be a killing blow against amd, silicon is nearing it's end in terms of shrinks that much i do know.
 
we live too much in speculation and uncertaintiy for a year or two instead of waiting for the facts instead hehe. while if u take the middle ground in the speculation it usually is somewhat correct :p
 
we live too much in speculation and uncertaintiy for a year or two instead of waiting for the facts instead hehe. while if u take the middle ground in the speculation it usually is somewhat correct :p

Facts are that for gaming, the most strenuous common desktop workload, six cores top out what makes a difference for performance; eight is the right kind of overkill. After that, individual core performance matters.

This CPU gets the very best of both available, and we still have people complaining.
 
Awwwe
Imagine a company that has nothing to offer except for pre-binning parts and charging more all the while destroying your chance of winning the silicon lottery.
Welcome to the new iNtel.

Nothing to offer except the fastest consumer CPUs one can get?

Or would you also complain about the turn-key overclocks on AMD's X CPUs?
 
Nothing to offer except the fastest consumer CPUs one can get?

Or would you also complain about the turn-key overclocks on AMD's X CPUs?

Intel is in a world of hurt, at least for the moment. Competition is driving innovation and everything but Intel's profit margin is benefiting from it. Also, how much slower are Intel processors going to become since all these security holes are being found?
 
Intel is in a world of hurt, at least for the moment. Competition is driving innovation and everything but Intel's profit margin is benefiting from it.

So you believe that Intel has stopped innovating?

Also, how much slower are Intel processors going to become since all these security holes are being found?

For the consumer who will buy these CPUs? None. Enterprise hardware is in a completely different class.

intel wasn't even the first to bin 5ghz 9900k's https://siliconlottery.com/collections/coffeelake-r/products/9900k50g. or as they state 38% of regular 9900k's can already hit 5ghz

...and this makes it 100%. Are you going to claim that as a bad thing?
 
So you believe that Intel has stopped innovating?



For the consumer who will buy these CPUs? None. Enterprise hardware is in a completely different class.



...and this makes it 100%. Are you going to claim that as a bad thing?

good thing or bad thing depends on what intel feels like charging to bin their own chips. All im saying is this isnt any advance at all for intel its not impressive either. Its just chip binning
 
...and this makes it 100%. Are you going to claim that as a bad thing?

No, 9900k's will still be the same. This is Intel merely doing the same thing as silicon lottery and taking that 38% and slapping another letter on it (along with a price premium). Somehow I really doubt they've changed the chip itself.

There's absolutely zero to be enthusiastic about, except I guess you get to give Intel that extra cash instead of silicon lottery.
 
good thing or bad thing depends on what intel feels like charging to bin their own chips. All im saying is this isnt any advance at all for intel its not impressive either. Its just chip binning

They didn't advance the model number- they added a letter to the end. Why would you complain about it not being an 'advance'?

This is Intel merely doing the same thing as silicon lottery and taking that 38% and slapping another letter on it (along with a price premium). Somehow I really doubt they've changed the chip itself.

But why would you consider that in the first place?
 
No, 9900k's will still be the same. This is Intel merely doing the same thing as silicon lottery and taking that 38% and slapping another letter on it (along with a price premium). Somehow I really doubt they've changed the chip itself.

There's absolutely zero to be enthusiastic about, except I guess you get to give Intel that extra cash instead of silicon lottery.

depending on how the 9900ks are stocked and how intel pulled the chips for the binning the overclock abilities of the 9900k might actually decrease on adverage.
 
They didn't advance the model number- they added a letter to the end. Why would you complain about it not being an 'advance'?



But why would you consider that in the first place?

why even bother announce chip binning without announcing the price when that is the sole factor of if its worth it or not. Intel is trying to show a chip that can do 5ghz at 8 cores as if that is something worth announcing and its not when its just a binned chip.
 
why even bother announce chip binning without announcing the price when that is the sole factor of if its worth it or not. Intel is trying to show a chip that can do 5ghz at 8 cores as if that is something worth announcing and its not when its just a binned chip.

Well, guaranteed 5GHz is something new. Remember that most CPU SKUs are 'bins'. As in nearly all. You going to belittle AMD for their X series? That's less of a bump than this CPU.
 
Back
Top