Are Intel getting Worried? - 9900KS

[trivik12]

if this is AMD's conroe moment why are there no samples for tech sites to review. What made Conroe moment so huge was reveal around spring IDF and anandtech That had 2.66 ghz conroe beating AMD's best chip.

it would have been good to see the Zen 2 against 9900K or whatever is the best Intel chip.

 

[DrMrLordX]

Intel had Conroe samples in the wild at least 6 months before the actual product launch. AMD is taking a different approach here, actively controlling leaks wherever possible and doling out information very slowly.

 

[abc40]

If the 9900KS is a highly binned 9900K, does that mean we will no longer see stock 9900K that can overclock at or past 5GHz? (basically Intel playing the role of Silicon Lottery). Or am I missing something...?

If so, that would be quite annoying.

 

[PotatoWithEarsOnSide]

If the 9900KS is a highly binned 9900K, does that mean we will no longer see stock 9900K that can overclock at or past 5GHz? (basically Intel playing the role of Silicon Lottery). Or am I missing something...?

If so, that would be quite annoying.

Bingo.
Reading between the lines it is pretty obvious that is what they'll do. They're essentially obsoleting the 9900K as an overclockers CPU moving forward, but you know for certain they won't price it accordingly.

When is a K not a K?

 

[Zucker2k]

If the 9900KS is a highly binned 9900K, does that mean we will no longer see stock 9900K that can overclock at or past 5GHz? (basically Intel playing the role of Silicon Lottery). Or am I missing something...?

If so, that would be quite annoying.

Thanks to AMD, this is going to be the trend moving forward.

 

[DrMrLordX]

If the 9900KS is a highly binned 9900K, does that mean we will no longer see stock 9900K that can overclock at or past 5GHz? (basically Intel playing the role of Silicon Lottery). Or am I missing something...?

If so, that would be quite annoying.

Why would you buy a 9900k at this point, unless you maybe have a slower CPU in a Z390 board and want an upgrade that isn't a 9900KS?

Thanks to AMD, this is going to be the trend moving forward.

Yes, let's blame AMD for Intel overbinning their CPUs.

 

[Zucker2k]

Why would you buy a 9900k at this point, unless you maybe have a slower CPU in a Z390 board and want an upgrade that isn't a 9900KS?



Yes, let's blame AMD for Intel overbinning their CPUs.

Are you serious? The 9900KS is more binned than the 2700x, which can't even overclock to single core turbo speeds on highend air-cooling? The 1800x also suffered from the same disease; 4.1GHz, if you're lucky. More 9900Ks are hitting 5GHz+ so yeah, Intel stole this from AMD's playbook. No ifs or buts about this.

 

[Markfw]

Are you serious? The 9900KS is more binned than the 2700x, which can't even overclock to single core turbo speeds on highend air-cooling? The 1800x also suffered from the same disease; 4.1GHz, if you're lucky. More 9900Ks are hitting 5GHz+ so yeah, Intel stole this from AMD's playbook. No ifs or buts about this.

But at this time (a month away) it sure looks like the 3800x or 3900x will beat a 9900ks, and use a lot less power, and require a lot less cooling, so why not wait a month ? Availabiliy 7/7.

 

[DrMrLordX]

Are you serious? The 9900KS is more binned than the 2700x, which can't even overclock to single core turbo speeds on highend air-cooling? The 1800x also suffered from the same disease; 4.1GHz, if you're lucky. More 9900Ks are hitting 5GHz+ so yeah, Intel stole this from AMD's playbook. No ifs or buts about this.

Stole it from their playbook?

Intel is rereleasing the exact same CPU overbinned because they have nothing else (except 10c Comet Lake which is curiously MiA). AMD did not do that. Intel is doing it because they're completely stuck. AMD did it because of the process they used for the 1800x and 2700x (both 14LPP and 12nm have harsh voltage curves close to the design FMAX, leaving very little OC headroom period. In fact, AMD has been releasing CPUs that exhibit that behavior since . . . Kaveri).

If Intel were releasing a 4.5 GHz all-core-turbo IceLake in Q4 2019 then you might have a point. But they aren't, so you don't. But hey, go ahead and blame AMD since you're going to do that anyway.

 

[abc40]

Why would you buy a 9900k at this point, unless you maybe have a slower CPU in a Z390 board and want an upgrade that isn't a 9900KS?

Because in an impetus of folly, I bought a brand new Z390 mobo 2 months ago (on sale) and am stuck with it. If I manage to sell it at a not huge loss on eBay, I know what to do on 7/7...

 

[DrMrLordX]

Because in an impetus of folly, I bought a brand new Z390 mobo 2 months ago (on sale) and am stuck with it. If I manage to sell it at a not huge loss on eBay, I know what to do on 7/7...

I guess it would make sense to get one in that case. 9900ks might sell for pretty cheap on eBay in July if you can't get rid of the board.

 

[maddie]

Stole it from their playbook?

Intel is rereleasing the exact same CPU overbinned because they have nothing else (except 10c Comet Lake which is curiously MiA). AMD did not do that. Intel is doing it because they're completely stuck. AMD did it because of the process they used for the 1800x and 2700x (both 14LPP and 12nm have harsh voltage curves close to the design FMAX, leaving very little OC headroom period. In fact, AMD has been releasing CPUs that exhibit that behavior since . . . Kaveri).

If Intel were releasing a 4.5 GHz all-core-turbo IceLake in Q4 2019 then you might have a point. But they aren't, so you don't. But hey, go ahead and blame AMD since you're going to do that anyway.

Been wondering why that is and keep coming back to their resonant clock mesh tech.

One thing about resonant effects. They work very well when the operating frequency allow the resonant effect to occur and very badly, even negatively, when you leave the resonant range. Could very well explain the huge difficulty in deviating from the optimum frequency. There might very well start to occur out of phase destruction needing much higher voltages to keep the signal strength.

Pure speculation of course.

 

[TheELF]

If the 9900KS is a highly binned 9900K, does that mean we will no longer see stock 9900K that can overclock at or past 5GHz? (basically Intel playing the role of Silicon Lottery). Or am I missing something...?

If so, that would be quite annoying.

Depends on what exactly they are binning for,they could just be looking for the exact opposite of what all the fanbois are thinking...bin the 9900ks searching for the best performance/watt pieces to reach 5Ghz all core on lower power.
It's the easiest way to get 9900k overclocked results in benchmarks with as low an TDP as possible,so no articles can say that it's too hot and/or consumes too much power.
It will make them look pretty good without hurting the 9900k in any way.

 

[OTG]

Are you serious? The 9900KS is more binned than the 2700x, which can't even overclock to single core turbo speeds on highend air-cooling?

The answer to this is obviously "yes". Ryzen's voltage wall has absolutely nothing to do with Intel desperately pushing any product that can even appear to compete with Zen 2.

If it were common or easy to find a 9900k that can hit 5ghz on all 8 cores*, they wouldn't be launching these chips in *December*.


*Yes, I know lots people do it. Those people don't have to sell CPUs in volume.

 

[abc40]

I guess it would make sense to get one in that case. 9900ks might sell for pretty cheap on eBay in July if you can't get rid of the board.

Except that the 9900KS is scheduled for holiday season (Q4) unless I'm wrong. Pfff

 

[PotatoWithEarsOnSide]

Depends on what exactly they are binning for,they could just be looking for the exact opposite of what all the fanbois are thinking...bin the 9900ks searching for the best performance/watt pieces to reach 5Ghz all core on lower power.
It's the easiest way to get 9900k overclocked results in benchmarks with as low an TDP as possible,so no articles can say that it's too hot and/or consumes too much power.
It will make them look pretty good without hurting the 9900k in any way.

How low is low enough?
They'll be eating close to 180w for 8c at 5GHz even with the most efficient CPUs. That'll be twice what AMD will be matching it with the 3800X.

 

[jpiniero]

Except that the 9900KS is scheduled for holiday season (Q4) unless I'm wrong. Pfff

Yeah. Remember that the Coffee Lake Refresh was released in October. AT's review is dated October 19th. Since we don't know when Comet Lake S is going to be released (but we can be sure now it won't be in 2019) it would make some sense if their original intention was to do a small clock tweak then.

I'm still interested in seeing if Intel does source R0 for the reviewers.

 

[BigDaveX]

if this is AMD's conroe moment why are there no samples for tech sites to review. What made Conroe moment so huge was reveal around spring IDF and anandtech That had 2.66 ghz conroe beating AMD's best chip.

it would have been good to see the Zen 2 against 9900K or whatever is the best Intel chip.

Historically, AMD have never bothered doing that, aside from one occasion where they let out a prototype Athlon 64 system that got hamstrung by using some garbage VIA chipset and posted really underwhelming numbers. Intel have a knack for pulling Emergency Edition-type products out of the hat whenever the need arises, so AMD want to avoid tipping their hand.

Even Intel have only really done it for two major releases - Conroe, when their entire product line-up was getting completely annihilated by the Athlon 64, and Nehalem, where they had regained the lead on the desktop and mobile, but were still getting their backsides handed to them in the server arena.

 

[Zucker2k]

Stole it from their playbook?

Intel is rereleasing the exact same CPU overbinned because they have nothing else (except 10c Comet Lake which is curiously MiA). AMD did not do that. Intel is doing it because they're completely stuck. AMD did it because of the process they used for the 1800x and 2700x (both 14LPP and 12nm have harsh voltage curves close to the design FMAX, leaving very little OC headroom period. In fact, AMD has been releasing CPUs that exhibit that behavior since . . . Kaveri).

If Intel were releasing a 4.5 GHz all-core-turbo IceLake in Q4 2019 then you might have a point. But they aren't, so you don't. But hey, go ahead and blame AMD since you're going to do that anyway.

Are you making an overbinning argument or the motivation for doing so? AMD had to clock both Zen and Zen+ to the gills in order to be competitive against Kabylake and Coffeelake in lightly threaded loads. Now Intel has to overclock the 9900k to the gills in order to be competitive against the upcoming Zen2. What's the confusion here?

 

[DrMrLordX]

What's the confusion here?

There is no confusion. Thanks for confirming that AMD isn't responsible for anything Intel is doing internally.

Except that the 9900KS is scheduled for holiday season (Q4) unless I'm wrong. Pfff

Yeah better to just get a standard 9900K in July after prices come down, that's what I meant. I would not recommend the KS (sorry, I was trying to pluralize 9900k) since it will come with a big price tag. Older 9900k chips on eBay should sell for $300-$400 once Matisse is on the streets.

How low is low enough?
They'll be eating close to 180w for 8c at 5GHz even with the most efficient CPUs. That'll be twice what AMD will be matching it with the 3800X.

I think they'll be lucky to get as low as 180W for any sustained benchmark load if all-core turbo is really gonna hit 5 GHz.

 

[extra]

Can't believe my login still works, it's been ages. LOL.

The key thing I took away from the announcement was this: "Q4".

So...it'll be around for the Christmas season? I would have thought it would have been pushed up to next month honestly. Q4 is pretty disappointing. It says to me that these may be some nicely binned chips and that maybe the power won't be so bad, they may be building up inventory of 'em. By then maybe prices on the Intel CPUs will have come down in response to Ryzen 3000 as well.

The other thing though is the platform. I know there's lots of great Intel boards and that PCIe 4 doesn't *really* matter that much but...let's be honest here--enthusiasts--we like the latest stuff. People are going to want the AMD x570 platform just for that. Even if it's just to get a new PCIe4 ssd and see those huge numbers. I don't think the 5ghz on all cores is enough unless the price is also great. People will build an AMD box instead. What might change that? If these are binned like crazy and can hit 5.2 at a reasonable voltage or something--that would do it.

As for myself: finally feel like it's going to be worth upgrading from the 4770k. Whether that be to a new Intel cpu or Ryzen I haven't decided. My 4770k works good. I had good luck building some Ryzen 2000 series computers for others. So..shrug. And no I don't think the security bugs that have hit Intel are that big of deal for the regular person.

 

[DrMrLordX]

@extra

Now's a good time to be watching the PC scene, since it may not exist in its current form in 5-10 years (we'll see). Once we hit either 3nm or 5nm - whichever is the last process anyone is willing to sell to us plebs - it'll be sideways for awhile and then . . . who knows.

Q4 could be as early as October. Intel officially launched the 9900k in October of 2018 and the 8700k in October of 2017 (availability on 8700k was really short until November of the same year; 9900k was hard to get until December 2018). I would not be surprised to see at least a paper launch then, with broad availability in Nov or Dec. We'll see.

I wouldn't expect too much from the 9900KS, but hey, you never know.

 

[chrisjames61]

The point is that intel is making even more money now that AMD is competing because intel could just add cores to old tech and sell them as new gens and sell them as well as a new gen would sell and better since previous gens where so lackluster,they increased their PC-centric income as well,they made more money across the board ever since AMD is competing because they don't have to fear that a good SKU is going to put amd out of the business.

You have it all wrong, and if you think it is right do you think it is any kind of sustainable business model?

 

[Zucker2k]

There is no confusion. Thanks for confirming that AMD isn't responsible for anything Intel is doing internally.

Responsible? I smell confusion. You should be talking "overbinning," which is my whole point - clocking chips to within 100MHz of their lives!

 

[OTG]

Responsible? I smell confusion. You should be talking "overbinning," which is my whole point - clocking chips to within 100MHz of their lives!

The 2700x being binned (duh) doesn't mean the 9900ks isn't binned even more. AMD needed to work within the relatively tight constraints of a brand new architecture, and a 12nm node that hasn't had 4 years of increasingly desperate optimization. They designed a smart turbo scheme that gets damn near every last bit of performance out of the silicon, put it on sale, and moved on.
That last point is important. AMD is already on 7nm. They don't have to spend multiple years and hundreds of millions squeezing every drop of performance and binning for the top 0.01% of chips, because why on earth would they?
Intel can't move on, because their next node ain't working, so they're on the third release of exactly the same chip.
Only this one isn't even available until 6 months from "launch", 5 months after AMD releases the chip it's supposed to compete with.
Because, IMHO, they just don't have the stock they need.
I wouldn't even be surprised if they quietly drop it before launch, because after 5 months of reviews for the 3800x and 3900x, a 200w 8 core silicon unicorn is going to look a whole lot less appealing.