i7-9700k 8/16 core and others leaked? Is this legit?

[LTC8K6]

I think I need to revise my opinion about an 8c/16t 9700K chip. I think the only way it is practical is for it to be on the 10nm+ process.
I will be surprised if Intel has an 8c/16t 14nm++ chip that performs well enough to be called 9700K and fits within the TDP envelope of the 8700K.

 

[gOJDO_n]

14nm or not, this should give Intel the outright desktop performance crown back, even in MT performance. Performance/watt would be it's biggest weakness, but that is of less concern on a desktop chip. Intel could negate the higher power consumption with the use of solder rather than TIM as another poster above said, I'm not holding my breath on that one though.

The usage of TIM will not reduce the power usage, but will reduce the cores temperature. As for the performance/watt, CoffeLake CPUs already have 15~20% better performance/watt than Ryzen (8600K vs 1600X, 8700K vs 1800X). And I believe that Intel will introduce new Turbo-Boost with a smarter and more precise boost algorithm which will control individually every core's frequency and voltage. So, 2018 will be very interesting. 🙂

https://www.techpowerup.com/reviews/Intel/Core_i7_8700K/18.html

 

[frozentundra123456]

Actually, the current situation will not shift as much in Intel's favor as you might think. If it's on 14nm++ like the 8700k, I doubt that it's top clockspeed will go up by much, and all they get is a max 33% increase in MT performance assuming perfect scaling.

In contrast, AMD is going to get a 10% increase in performance across the board with Pinnacle Ridge, plus who knows what from optimized IF/IMC performance. Should be interesting to see who comes out on top.

Huh? Intel has about a 20 to 25% lead (after overclocking) in clockspeed and a small IPC advantage. So if they increase CL to 8 cores, and even *if* AMD gains 10%, of which I have seen no proof, my math tells me Intel would still have a 10% or more lead. That is assuming of course that an 8 core CL will still overclock to close to 5 ghz.

 

[Hulk]

The usage of TIM will not reduce the power usage, but will reduce the cores temperature.

I have been thinking about this. If better TIM increases heat transfer and therefore lowers core temperatures wouldn't that allow for lower voltage and therefore lower power consumption of the CPU?

I'm no expert, just wondering?

 

[Dufus]

Higher temperatures for same clocks and workload can dissipate more power. However isn't reducing temperatures going to encourage people to overclock higher and draw even more power resulting in even higher current draw and possibly faster degradation over time?

 

[beginner99]

The only places the mainstream platform will be lacking is with the PCIe lanes (the only place I care about PCIe lanes is with NVMe being routed through the PHC)

Agree. That is the biggest issue. They should double the lane count of DMI or move to pcie4 ASAP. X370 from AMD is actually superior because you get 4xlanes for NVMe ssd from the CPU.

 

[eek2121]

Yes. Intel is floundering on 10nm, and they need something to compete with amd, so they're releasing a mainstream 8 core on 14nm.

They are floundering on quite a bit more than 10nm. It is clear that they were completely caught off guard by Ryzen. Despite what many would argue, the i7 8700k wasn't meant to bee seen by the public eye. The same with Intel's HEDT parts. This is evidenced by poor motherboard/platform support (Chipsets only lasting 1 generation? Odd CPU caveats within the same platform? NVME raid limits?)

EDIT: To be clear here, Intel absolutely has 6 and even 8 core parts in house. At any time, large companies like these keep multiple in-house prototypes such as this. After years of AMD catering to the budget segment, they didn't expect to have to actually USE one of those prototypes.

If the 9700k clocks nicely I could see it being the next Sandybridge. Will be interesting what this means for Intel's HEDT lineup, I don't see more than 8 cores being necessary for anyone except people doing HPC or video professionals for a long time. The only places the mainstream platform will be lacking is with the PCIe lanes (the only place I care about PCIe lanes is with NVMe being routed through the PHC) and memory bandwidth (though this can be alleviated with high speed DDR4).

The other place where Intel's new lineup will be very interesting is with iMacs. An iMac could become a legit AIO gaming machine using bootcamp plus an external GPU with the next gen Thunderbolt 3 controller (enables DP 1.4, which has DSC thus there should be enough bandwidth for both a 5k video signal and the CPU-GPU traffic). An 8-core option for the iMac could really put a damper on the Mac Pro lineup.

I have a handful of games that utilize all 16 cores of my 1950x. I am currently developing a game that scales across cores. Never say never. Intel needs to get in the core game.

Intel have many reasons not to let AMD to overtake a CPU crown in any segment. They already have learned the lesson 15 years ago with Athlon/Athlon64 and Opteron.
Such 8 core desktop CPU like the i7 9700K would be a double-edge sword for Intel. It will compete with Ryzen 7 and ThreadRipper with 8 to 12 cores, but it will also compete with the X299 platform. It will cannibalize the sales of all the Skylake-X CPUs with LCC(Low Core Count) of up to 10 cores because it is going to outperform them badly in most of the apps at half the price for the mainboard and the CPU. Clock for clock, core for core and thread for thread the i7 9700k will outperform any 8 core CPU x86 that will exist when it will be released. I hope that Intel will finally replace the TIM with a solder, so it can achieve higher clocks (perhaps 5GHz 1-core / 4.5GHz 8-core boost). I expect Intel to improve their Turbo-Boost algorithm with a more sophisticated one which will control each core frequency and voltage individually. Maybe they will add 8 more PCI-e 3.0 lanes to the CPU so the new platform will have the Intel X299 and AMD X399 platforms premium possibility to run two VGAs @PCIe 3.0 x8 + two SSDs @PCIe 3.0 x4.

By the time this chip comes out, Pinnacle Ridge chips will be out and old news. AMD hopes to have 7nm by Q2 2019, so Intel, if they even manage to establish a solid performance lead over Pinnacle Ridge (doubtful) will lose it again when Zen 7nm comes out.

Wait, that says that ninth-gen is Cannonlake. I thought we weren't going to get Cannonlake desktop parts at all?

Anandtech stated that 9th gen will possibly be 14nm. All indications show that we are stuck on 14nm for a while longer for Intel.

14nm or not, this should give Intel the outright desktop performance crown back, even in MT performance. Performance/watt would be it's biggest weakness, but that is of less concern on a desktop chip. Intel could negate the higher power consumption with the use of solder rather than TIM as another poster above said, I'm not holding my breath on that one though.

Time marches forward, AMD isn't resting on it's laurels either. We will happily see what 2018, 2019, and beyond deliver.

What are the chances of this being compatible with existing Z370 mobos?

Also, the fact that HT will be enabled on all chips will be a welcome change if true, and has been a long time coming IMO. Even a 4C/8T i3 will be a very capable general desktop and gaming chip.

 

[moonbogg]

Also, this will have the ring bus advantage. It matters a lot. That's why people prefer the 8700K over anything Ryzen or even X299 has to offer. That mesh stuff slows things down. Its not good. 8/16 on old school ring bus with high clocks will be really good.

 

[LTC8K6]

I don't think Intel is floundering much, if at all.
I recall reading that Intel's 14nm is basically equivalent to the competition's 10nm, and Intel's 10nm is equivalent to the competition's 7nm.

Intel's 10nm and AMD's 7nm are basically equivalent.

And we know that Intel has already moved on to 10nm+, which would really be what everyone else calls 7nm+.

 

[Dayman1225]

Anandtech stated that 9th gen will possibly be 14nm. All indications show that we are stuck on 14nm for a while longer for Intel.

That's called a guess. 9th gen is ICL according to @FanlessTech and ONLY ICL, plus yesterday HWInfo announced preliminary support for Icelake Y/U. It's coming, but when is the question, Also according to Intel's own graphs there is about a 6-9 month gap between 10nm and 10nm+(just before 2019). So late 2018 and early 2019 are my bets.


https://twitter.com/FanlessTech/status/934109005693284357

https://twitter.com/FanlessTech/status/934580364575756294


Intel's slide:

and full slide:

Source of the slides, Manufacturing day in Beijing in September:

https://newsroom.intel.com/newsroom...017/09/mark-bohr-on-continuing-moores-law.pdf

https://newsroom.intel.com/newsroom...mark-bohr-on-intels-technology-leadership.pdf

 

[Dayman1225]

I can't see this happening. It's all TIM from now on very likely. I mean, they use TIM on Xeon Phi "Knights Landing" CPUs.

They don't use solder on Skylake SP either.

It's all TIM now and I don't see them going back.

 

[Insert_Nickname]

Agree. That is the biggest issue. They should double the lane count of DMI or move to pcie4 ASAP. X370 from AMD is actually superior because you get 4xlanes for NVMe ssd from the CPU.

Actually, Intel could solve this easily by adding a second PCI x16 (or x8 if cheap) root complex with bifurcation support. Then you could have a x16 slot for graphics, 3x (!) x4 (or 1x x8 and 1x x4) links for storage and a x4 PCH link.

But that'd cut into the HEDT market.

 

[FIVR]

I think I need to revise my opinion about an 8c/16t 9700K chip. I think the only way it is practical is for it to be on the 10nm+ process.
I will be surprised if Intel has an 8c/16t 14nm++ chip that performs well enough to be called 9700K and fits within the TDP envelope of the 8700K.

I think it will do everything but the last thing. You won't get it into 95W but they might do some aspirational math and get it into <120W. The TDP doesn't really matter because intel users don't care and they will use a new socket anyway.

 

[FIVR]

The real question i have about this 8 core is how will this affect margins? I guess an 8 core can't be that much bigger than a 6 core, but you would think it would cost them more and they would have to validate all 8 cores for extremely high clocks. Intel is good about managing costs though, so I bet they have done the math and know they can afford it.

 

[Dayman1225]

The real question i have about this 8 core is how will this affect margins? I guess an 8 core can't be that much bigger than a 6 core, but you would think it would cost them more and they would have to validate all 8 cores for extremely high clocks. Intel is good about managing costs though, so I bet they have done the math and know they can afford it.

I'd like to think they can afford it! They make $60 billion~(revenue not profit) YoY for godsakes!

 

[StinkyPinky]

That's called a guess. 9th gen is ICL according to @FanlessTech and ONLY ICL, plus yesterday HWInfo announced preliminary support for Icelake Y/U. It's coming, but when is the question, Also according to Intel's own graphs there is about a 6-9 month gap between 10nm and 10nm+(just before 2019). So late 2018 and early 2019 are my bets.


https://twitter.com/FanlessTech/status/934109005693284357

https://twitter.com/FanlessTech/status/934580364575756294


Intel's slide:

and full slide:

Source of the slides, Manufacturing day in Beijing in September:

https://newsroom.intel.com/newsroom...017/09/mark-bohr-on-continuing-moores-law.pdf

https://newsroom.intel.com/newsroom...mark-bohr-on-intels-technology-leadership.pdf

So Intel have all but admitted that 10nm is a regression in performance compared to 14++. Heck, even 10+ looks a slight regression. 10++ is due to 2020, which in Intel speak really means 2021. That's very poor.

 

[imported_bman]

I have a handful of games that utilize all 16 cores of my 1950x. I am currently developing a game that scales across cores. Never say never. Intel needs to get in the core game.

Any games that actually give a performance boost by using 16 cores vs 8 cores? Parking threads across 16 cores, where 10 of those cores don't hit above 20% utilization is not a ringing endorsement of needing more than 8 cores. The only games that I have seen that give a marginal advantage for having more than 8 cores are Civ 6 and Gears 4 (while running at 720p, play at 1080p or higher and those advantages evaporate even when compared to a 6 core CPUs).

Next gen consoles are most likely going to have 8 cores with SMT, so I suspect 8 cores will become the sweet spot for a long time. Also some point you can't split your code into more threads so the frame rate is limited by the slowest completing thread, so having more cores will do little to improve performance.

 

[Zucker2k]

That's called a guess. 9th gen is ICL according to @FanlessTech and ONLY ICL, plus yesterday HWInfo announced preliminary support for Icelake Y/U. It's coming, but when is the question, Also according to Intel's own graphs there is about a 6-9 month gap between 10nm and 10nm+(just before 2019). So late 2018 and early 2019 are my bets.


https://twitter.com/FanlessTech/status/934109005693284357

https://twitter.com/FanlessTech/status/934580364575756294


Intel's slide:

According to this slide, 14nm++ is as good as "two full nodes" better than 14nm if one compares according to previous performance jumps on the graph. So is it the case that 14nm++ turned out to be way better than Intel projected and forced them to raise the bar on 10nm, or 14++ turned out to be just that good while 10nm turned out to be ordinary? Because if you look at where 14nm sits against 10nm, on the performace scale, the gap is twice as big compared to previous node to node performance jumps, yet 14nm++ sits even higher still than 10nm.
Edit: Edited for clarity.

 

[blublub]

Sounds great for us. Thanks AMD for the competition. I'll pick up a 9700k now.

It took Ryzen for intel to have to match AMD core for core.

This CPU competition trickledown effect is hilarious.

Now, i5s in laptops are almost on par with a high end haswell CPU. This is just ridiculous.

Remember when people were recommending dual cores......
Ya... 2 threads will definitely have a LOT of longevity to them now /sarcasm

Oh this competition is FUN!

If you want to thank AMD and make sure this continue,s you should buy an AMD CPU - because if AMD can't catch up financially you wont have competition in 3-5 years and are stuck with Intel once and for all because AMD will not survive another downfall

 

[Ken g6]

I think I need to revise my opinion about an 8c/16t 9700K chip. I think the only way it is practical is for it to be on the 10nm+ process.
I will be surprised if Intel has an 8c/16t 14nm++ chip that performs well enough to be called 9700K and fits within the TDP envelope of the 8700K.

What if there's no IGP on board? And why does it have to fit a TDP? It's a K processor on a new board. That can include lots of caveats.

 

[VirtualLarry]

The usage of TIM will not reduce the power usage, but will reduce the cores temperature.

Not quite true. Our esteemed member IDontCare had some extremely-detailed threads with charts and graphs, showing the relationship between temperature and power-usage.

The long and the short of it is, if you can keep a CPU cooler, it will use less power, for the same frequency.

 

[TigerMonsoonDragon]

Good thing I'm not going coffee lake.

It'll be an interesting year next year.

 

[frozentundra123456]

They are floundering on quite a bit more than 10nm. It is clear that they were completely caught off guard by Ryzen. Despite what many would argue, the i7 8700k wasn't meant to bee seen by the public eye. The same with Intel's HEDT parts. This is evidenced by poor motherboard/platform support (Chipsets only lasting 1 generation? Odd CPU caveats within the same platform? NVME raid limits?)

EDIT: To be clear here, Intel absolutely has 6 and even 8 core parts in house. At any time, large companies like these keep multiple in-house prototypes such as this. After years of AMD catering to the budget segment, they didn't expect to have to actually USE one of those prototypes.



I have a handful of games that utilize all 16 cores of my 1950x. I am currently developing a game that scales across cores. Never say never. Intel needs to get in the core game.



By the time this chip comes out, Pinnacle Ridge chips will be out and old news. AMD hopes to have 7nm by Q2 2019, so Intel, if they even manage to establish a solid performance lead over Pinnacle Ridge (doubtful) will lose it again when Zen 7nm comes out.



Anandtech stated that 9th gen will possibly be 14nm. All indications show that we are stuck on 14nm for a while longer for Intel.

Seriously, you think intel was never going release a mainstream hex core without AMD? They may have rushed the release, but Coffee Lake has been in the works a long time, and is actually a response to the 10 nm delays. I also would like to see some benchmarks showing all those games that utilize16 cores, or more precisely, that are faster on 16 cores than on fewer but faster ones.

 

[DooKey]

Depending on the Zen refresh the 9700K might be replacing my 1700X.

 

[IntelUser2000]

https://www.anandtech.com/show/12087/intel-xeon-e3-1200-v7-xeon-entry-xeon-e

Xeon-E is now Xeon Entry.

Specifically the models I quoted were indeed mobile versions of it. It seems Anandtech is saying its based on Coffeelake, though that picture from ChipWiki makes it clear it would be based on Skylake-SP. We'll see.