Confirmed - i9 9900k will have soldered IHS, no more toothpaste TIM

[Dayman1225]

Pretty sure Matisse hits in early Q2 2019? Should be April.

Doubting that, Lisa said that Ryzen comes after Rome, and Rome will probably launch in Q2/Q3.

 

[DrMrLordX]

Eh? I thought Rome was earlier than that . . .

 

[Dayman1225]

Eh? I thought Rome was earlier than that . . .

AMD says "2019", I am assuming that is after Q1.

 

[DrMrLordX]

AMD says "2019", I am assuming that is after Q1.

For their sake, let's hope not.

But hey this is getting off-topic so, wait and see.

 

[Borealis7]

great Intel i9-9900K conversation going on here.

 

[DrMrLordX]

great Intel i9-9900K conversation going on here.

Well, Matisse is going to be the obvious competition. Doesn't have a lot to do with the solder issue, though.

 

[krumme]

Doubting that, Lisa said that Ryzen comes after Rome, and Rome will probably launch in Q2/Q3.

Probably depends on the server reception? If its a huge success there consumers is certainly shafted and will get few late and expensive parts. If server reception is lukewarm its opposite.

 

[Topweasel]

I think the assumption on Ryzen replacement coming after Rome was due at the time to AMD dual homing the Zen 3 dies with TR and EPYC getting TSMC dies and AMD getting GF dies and the projection being that GF wasn't going to be ready until after TSMC. If all dies are going to be the same from company (TSMC). We will probably see EPYC first because they will be lower volume and not need to clock as high (like AMD did when they first released the A64, the first releases were Opteron's that ran at about 80% the speed of the first A64's that came out a month or so later). But it shouldn't take the time it was oing to when GF was still going to be a manufacturer for them.

 

[Dayman1225]

I think the assumption on Ryzen replacement coming after Rome was due at the time to AMD dual homing the Zen 3 dies with TR and EPYC getting TSMC dies

Nope, Lisa Su said that Ryzen is after EPYC, no assumption here.

 

[PeterScott]

Nope, Lisa Su said that Ryzen is after EPYC, no assumption here.

Plus it makes sense to use the dies to build the most expensive, lowest volume part to pioneer 7nm. Once the yields/volumes are up and costs are down, then you can migrate those dies to consumers.

 

[Topweasel]

Nope, Lisa Su said that Ryzen is after EPYC, no assumption here.

When did she say it? Before or after GF announced they were backing out of 7nm? If you read my post I still suggested this might still hold true, but won't be months like originally projected.

Also I probably used the wrong terminology. The understanding is probably a better word, that being on our part not Lisa's.

 

[LTC8K6]

great Intel i9-9900K conversation going on here.

Oh well...

 

[Dayman1225]

Before or after GF announced they were backing out of 7nm?

Before. Here's the quote:

on the timing of the 7-nanometer Ryzen, I would just keep it as, it's after the 7-nanometer EPYC. So we'll launch 7-nanometer EPYC first. I wouldn't say, it's very far out, but I would say it's after.

 

[Markfw]

Guys, lets get back on topic, this is not an AMD thread either.

Posting as a moderator

 

[DrMrLordX]

Okay, so here is the real question:

9900k has solder, might hit 5.3 GHz. Conservatively it shows up in substantial quantities in November (October if prospective buyers are lucky).

So where does Intel go after that? Are they going to re-respin the 9900k for higher clocks a year out? Can they? It's gonna be a long, cold trudge to Icelake-S or whatever they're gonna call the desktop parts. And once they get there, it might not even be as fast as the 9900k.

 

[jpiniero]

So where does Intel go after that? Are they going to re-respin the 9900k for higher clocks a year out? Can they? It's gonna be a long, cold trudge to Icelake-S or whatever they're gonna call the desktop parts. And once they get there, it might not even be as fast as the 9900k.

Yeah, re-spin again. Intel made it sound like they are going to make more improvements to transistor quality with Cooper Lake so perhaps there's room for even higher clocks. Maybe a 10+0 model.

 

[Topweasel]

Okay, so here is the real question:

9900k has solder, might hit 5.3 GHz. Conservatively it shows up in substantial quantities in November (October if prospective buyers are lucky).

So where does Intel go after that? Are they going to re-respin the 9900k for higher clocks a year out? Can they? It's gonna be a long, cold trudge to Icelake-S or whatever they're gonna call the desktop parts. And once they get there, it might not even be as fast as the 9900k.

I mean Intel has been known to occasionally push the outer edge of their process before, the 8086k and the PIII 1.13 come to mind. But I doubt they will push it that hard. If you mean overclocked then yeah probably on the same level as a 7740x or 8700k delid-ded. I would suggest tempering hopes based on the core increase. But it's always been thermals and heat dissipation that has really held up clocks on Intel's 14nm chips.

 

[maddie]

Okay, so here is the real question:

9900k has solder, might hit 5.3 GHz. Conservatively it shows up in substantial quantities in November (October if prospective buyers are lucky).

So where does Intel go after that? Are they going to re-respin the 9900k for higher clocks a year out? Can they? It's gonna be a long, cold trudge to Icelake-S or whatever they're gonna call the desktop parts. And once they get there, it might not even be as fast as the 9900k.

Staff have to do stuff. keep sharp. Can't lose those skills can we.

14nm+++++ leads to a further reduction in density resulting in even tighter supplies of 14nm fab output. I guess it will balance out if they continue to lose some market share, so less dies needed

 

[PeterScott]

Okay, so here is the real question:

9900k has solder, might hit 5.3 GHz. Conservatively it shows up in substantial quantities in November (October if prospective buyers are lucky).

So where does Intel go after that? Are they going to re-respin the 9900k for higher clocks a year out? Can they? It's gonna be a long, cold trudge to Icelake-S or whatever they're gonna call the desktop parts. And once they get there, it might not even be as fast as the 9900k.

I said it before, but I think Intels super tweaked 14nm process will make their switch to 10nm that much harder. I expect it will take a couple of years before 10nm can run as fast 14nm. I am not sure where they can go next for a performance bump.

 

[StinkyPinky]

I seem to recall reading an article somewhere where Intel admitted that themselves. That their first 10nm cpus will in fact be slower in some instances.

 

[jpiniero]

14nm+++++ leads to a further reduction in density resulting in even tighter supplies of 14nm fab output. I guess it will balance out if they continue to lose some market share, so less dies needed

I don't know if we should assume that any transistor performance gains will come at a density loss. Plus some of the volume in late 2019 will be 10 nm, that will help greatly. OTOH, Cooper Lake is going to be a die hog so we'll see.

 

[PeterScott]

I don't know if we should assume that any transistor performance gains will come at a density loss. Plus some of the volume in late 2019 will be 10 nm, that will help greatly. OTOH, Cooper Lake is going to be a die hog so we'll see.

Why is Cooper lake a die hog? Aren't all the *Lake chips minor tweaks.

 

[jpiniero]

Why is Cooper lake a die hog? Aren't all the *Lake chips minor tweaks.

Probably going to be 24 cores x 2 at the very least.

 

[moonbogg]

The 9900K looks to be a seriously badass CPU. It's a shame I don't need a CPU though and it's a shame 10nm is starting to lick the inside of my ear and whisper "wait for meeee".

 

[arandomguy]

Okay, so here is the real question:

9900k has solder, might hit 5.3 GHz. Conservatively it shows up in substantial quantities in November (October if prospective buyers are lucky).

So where does Intel go after that? Are they going to re-respin the 9900k for higher clocks a year out? Can they? It's gonna be a long, cold trudge to Icelake-S or whatever they're gonna call the desktop parts. And once they get there, it might not even be as fast as the 9900k.

The interesting thing is that they can possible gain (well recoup) 3-10% via hardware fixes for spectre/meltdown compared to the current firmware fixes.

Another low hanging fruit would be a more intelligent turbo design, this would boost effective clock rates for non OCed builds (ala GPUs or Zen+ now).

Their IMC obviously can handle higher DDR4 speeds so they could choose to spec the "stock" speed higher which is another few % points (Zen+ stock DDR4 speed is higher than Coffeelake's).

You combine that with say Icelake's uarch brings 5-15% IPC all the above could go a long way to mitigating any clock speed decreases.

An interesting thing is on the consumer gaming side the most interesting addition Intel could possibly do is actually support variable refresh in Icelake's GPU and directly supporting pass through.