Intel Core i9-9900k 8c/16t, i7-9700K 8c/8t, i7-9600k 6c/6t 2nd Gen Coffee Lake CPU's + Z390

I'm saying this to make an important point.
People are arguing relentlessly about their chips not being stable and what the default stable voltage is for their processors, and now they're arguing that Asus overvolts by 100mv or something.
If they knew the default VID for their highest per core turbo multiplier, then they would know what they need to be stable (based on their loadline calibration / LLC setting).

Also VID is essential when using adaptive voltages, as the voltage given is based on the VID.

The problem with laptops (especially MSI laptops) is due to a BUG that was initially on the Z370 desktop boards but was later fixed after people were complaining about *VCORE* being over 100mv over what they set in the bios.

On the MSI laptops, the cpu's voltage, when overridden with static voltage, is influenced and boosted by the internal IA AC DC Loadline setting. That means, before any vdroop (MSI uses some sort of hardwired vdroop compensation also, or LLC), if you put 1.25v into a MSIbook, you would be feeding the chip about 1.35v+ at full load, because IA AC DC Loadline is still linked to manual CPU VCORE. Setting IA AC DC loadline to "1" fixes that.

The initial 8700K desktop boards had the exact same problem...the 2.10 mOhms IA AC DC loadline reference value was being applied to *BOTH* CPU VID and CPU VCORE incorrectly, when static vcore was being set, causing both to be set way higher than they were supposed to be. A bios update fixed this (I forgot which boards were affected but I remember Asrock boards were).

 

Show me a Cinebench run at 5Ghz, with HWinfo64 open during run, and after, and CPU-Z open as well. Thanks.

 

Here's your screenshots.

1) 5 ghz, IA AC DC Loadline=1. VCORE=1.270v. CB active. The second onboard sensor is far more accurate for vcore than the first. VID reported pretty close to my stable required vcore at full load



CB run finished (VID reported too low at idle).


To show that the VID shown has no relation to CPU vcore, here is setting CPU to 4.5 ghz in throttlestop and running it again. VCORE is still 1.270v.




CB run with IA AC DC loadline set to default (Auto). Clearly VID has no relation to what is pre-programmed thanks to the 1.60 (?) mOhms of vid boost now.


CB run finished with IA AC DC Loadline=auto.

 

Aww man come on I was just kidding. What game did you lose?

Mate we need power draw, package CPU power etc.

 

Did you look in Throttlestop? It's there.

Notice that power draw is reported *HIGHER* with IA AC DC Loadline=auto than with it 0.01 mOhms, yet temps are the same.

 

Sorry no your first picture didn't have it friend. I just refreshed and saw your additional photos.

Yep calling BS on that hidden 100mV on Asus boards.

 

I'm willing to bet "$100 dollars"--$1 dollar for each mv, that the Asus voltage is being "boosted" by the default IA AC DC Loadline setting of 1.60 mOhms, since in that long "german" thread linked by that guy on OCN, he set 1.14v in the Bios and the default VID is reading as 1.22v.

Old bugs are new again.

 

My voltage in BIOS is 1.270v and my VID is showing 1.259v under load with 1.2v at idle. My Vcore shows 1.252v at idle and 1.217v under load. This is all at default. I want my $100.

 

You need loadline calibration enabled. If your vcore is dropping by 50mv from idle to load, that's from not enough LLC.
Did you set internal IA AC DC loadline=1 in your bios?

 

O trust me it's enabled, the vdroop on this board is horrible. I'm hoping Asus eventually fixes it like last year. LLC7, of maximum 8 on my board is enabled. And no like I said, my this is default option on IA AC DC Loadline.

 

Yeah I went with Gigabyte this time.
GB came a long way from their "boot loop" P67 and Z68 days. Asus has just gone downhill on everything except their "Extreme" boards.

 

Gigabyte VRM this time around definitely seems to be superior in all regards, but at 5Ghz, even 5.1 or 5.2Ghz the VRM isn't going to be limiting on the Asus board. I measured a max of 58C overnight with Prime95 at 5Ghz and not a single issue with stability.

That said I also prefer the BIOS of Asus as it's what I'm used to using. It's near identical to my Z370 board with more goodies as it's their tier higher board. At the end of the day, chip quality is the more important factor in any build. You could buy a $600 motherboard and still get **** results with a low binned chip.

How far have you been able to push that chip so far?

 

Haven't gone past 5 ghz with HT on because temps get unmanageable.

 

https://imgur.com/a/1DNHd7I

 

Excellent chip.

 

Got my 2080 installed. Stock on both video and 9700K running COD Block Ops 4 for a few hours is around 64 C Processor / 50 C Video card. Stock on the card seems to boost to about 2050. Will do some more testing overclocked soon but heading to New Orleans for a week so will have to hold off with tweaking and measuring temps.

 

Nice build, looks good doesn't it!

 

https://www.gamersnexus.net/guides/3398-best-cpus-of-2018-gaming-photoshop-production-streaming



9900K named best well rounded CPU by GamersNexus.


“Where the R5 2600 holds a well-deserved award for best overall value, the i9-9900K – even with its somewhat absurd pricing – does deserve to nab the award for Most Well-Rounded.

To be fair, this one is still a toss-up between the 9900K and the 8700K, the latter of which received this award last year. To us, this award category means one thing: This is a CPU that can play games at the highest framerate and with the most consistent performance, but can also execute in common production applications with top-class performance.”

https://www.newegg.com/Product/Prod...tion=9900k&cm_re=9900k-_-19-117-957-_-Product

Been in stock now for well over a full day. Looks like the shortage has been resolved, if only temporarily.

 

The Best Z390 Motherboards for VRMs, 10Gb LAN, Mini-ITX, Micro-ATX (2018)
Gamers Nexus
Published on Nov 22, 2018
Z390 motherboards are new for 2018, and we've got a list of the best ones. Analyzed by expert motherboard reviewer Buildzoid, we pick the top 9th Gen boards. Buildzoid looks at the best Z390 motherboard in different price and use case categories, including budget Z390, extreme overclocking, motherboards with more PCIe 'lanes' via PLX, boards with 10-gigabit ethernet, and best mini-ITX and micro-ATX motherboards. This is particularly relevant with Black Friday and Cyber Monday sales on motherboards.

Spoiler: Links to all the products are below!

Listen to the relevant sections to understand why we chose them.

BEST ULTRA HIGH-END Z390 MOTHERBOARDS

Gigabyte Z390 Aorus Extreme (Newegg) - http://geni.us/0pAFyXo

Z390 MOTHERBOARDS WITH PLX CHIP

ASUS WS Z390 Pro (Amazon) - http://geni.us/581O6A
SuperMicro C9Z390-PGW (Amazon) - http://geni.us/TBB6

BEST Z390 MOTHERBOARD WITH 10GbE LAN

ASRock Z390 Taichi Ultimate (Amazon) - http://geni.us/c0A2ZB

BEST MINI-ITX Z390 BOARD

ASRock Z390 Phantom Gaming-ITX (Amazon) - http://geni.us/E9SckzE

BEST MICRO-ATX Z390 MOTHERBOARD

ASUS ROG Maximus XI Gene (Amazon) - http://geni.us/aSvbA
Gigabyte Z390 M Gaming (Amazon) - http://geni.us/DCA8d

BEST VRM FOR THE DOLLAR

Gigabyte Z390 Aorus Master (Amazon) - http://geni.us/jNpH
Gigabyte Z390 Aorus Ultra (Amazon) - http://geni.us/UjAb0Yb
Gigabyte Z390 Aorus Elite (Amazon) - http://geni.us/fie5
Gigabyte Z390 UD (Amazon) - http://geni.us/qGU7fJ

mobo PCB Breakdown: ASUS TUF B450M-Pro Gaming
Actually Hardcore Overclocking
Published on Nov 22, 2018

Spoiler: Click to See Video

Asus ROG Ryuo and Ryujin CPU Cooler Review
OC3D TV
Published on Nov 22, 2018

Spoiler: Click to See Video

 

https://hothardware.com/news/intels...ktop-14nm-cpus-rumored-to-pack-up-to-10-cores

10 core Comet Lake-S inbound. Seems Intel is not done with their 14nm consumer line just yet and this will be Intel's answer to Zen 2 until they launch 10nm. We all know they've done the 10 core ring bus with the 6950x on 14nm before, but this would probably be 14nm++ tech and would give them an opportunity to tweak for much higher base and turbo clocks, and refine their solder application.

Edit: A dual ring bus? This sounds interesting....

https://wccftech.com/intel-comet-lake-10-core-processors-rumor/

 

It is pathetic, IMO. They have two 10c chips on 9th series HEDT. Dual ring bus used to be used on HCC chips by Intel and there is a penalty going from one ring to the other.

This is more from fear of Zen2, and it doesn't even make sense with power consumption. If the 8 core can't fully be used on some Z390 boards because Intel lies about TDP in such a way, that means gigabyte 12 phase boards might be the primary one's handling a chip like this, or needing a while new board. In a laptop, forget about it.

So you seem excited over something that doesn't serve much purpose. If the 9900K is $500 ($600 for many looking due to shortage), just get a damn 10-core HEDT! More lanes and memory bandwidth. Or find the 1950X on sale for $440-600 right now (yes, I've seen that 16-core priced that low recently).



Sent from my SM-G900P using Tapatalk

 

no need to lose any nerves about it buddy, the masses are still gonna rush and get that mainstream 10 core part and complain about it "only" reaching 5.0 ghz max. OC instead of 5.3/5.2 as seen on 4 and 6 core parts

Sent from my Xiaomi Mi Max 2 (Oxygen) using Tapatalk

 

B360 + i9-9900K = DISASTER. - Worse Than i5
UFD Tech
Published on Nov 26, 2018
Have you ever tried to pair a high end CPU with a mid-tier board?


Battlefield V Multiplayer CPU Benchmark, Ryzen 7 2700X vs. Core i9-9900K
Hardware Unboxed
Published on Nov 26, 2018

 

Nice. Looks like nothing changed from the beta aligns with reports online.

 

BGA Killer is doing its job.

 

https://imgur.com/a/Ug7cSC4

2300CB is proving elusive for now. Getting closer, but not quite there yet.

 

Core i9-9900K Achieves 5.50 GHz Overclock on a Z170 Chipset Motherboard Techpowerup.com Today, 07:27
It is already established that the incompatibility between Intel's 8th and 9th generation Core socket LGA1151 processors and Intel's 100-series and 200-series chipset motherboards is artificial, and that with the right BIOS modding, you can get the newer processors to work on the older motherboards. Finnish overclocker "Luumi" made a video demonstration of an i9-9900K successfully overclocking to 5.50 GHz all-core on an ASRock Z170M OC Formula micro-ATX motherboard. The overclocked processor is CineBench and Prime95 stable.

 

yessss, THATS what i want to hear/read! the good old Z170 baby!

 

That's pretty cool, but his VRM temperatures were over 90 degC when running non-AVX Prime, which isn't workable long term. He also had lots of active air flow around the VRM heatsinks too. I suppose if you run an undervolt & overclock with a 9900K then a Z170 board could be ok, of course you need a modded BIOS too so that it can even boot. 8700K is probably the max recommended CPU for Z170 in my eyes, and of course you need a modded BIOS for that one too. It is impressive that his board could pull over 200W and still not blow up though, I've had 148W out of my motherboard with my 6700K in OCCT @4.7Ghz and 1.4V, but that's still a long way off 200W+.

 

Beating the Intel 9900k Liquid Helium CPU World Records using Liquid Nitrogen.. Possible? Lets Try..
Bearded Hardware
Streamed live 5 hours ago, Starts @ ~30:00
My buddy Jason aka (shadyreaper) was blessed by the Viking Tech Gods, he found a chip so good that we will come close to beating the Liquid Helium cooled CPU World Records. Liquid Nitrogen goes to -196C, Liquid Helium goes to -230C.

 

Wccftech: AMD Dominates Retail CPU Sales in November, Outsells Intel 2 to 1 at Germany’s Largest E-Tailer.
https://wccftech.com/amd-dominates-retail-cpu-sales-outselling-intel-2-to-1/



For all the screaming about the pricing of the CFL-R 8 cores, they are holding up Intel's revenue figures if mindfactory is representative of wider sales trends. It will be interesting if these numbers continue once the hype dies down.



Its actually quite bizarre to look at that data, the 2700x was nowhere near the 8700K for 1H, now sales by revenue are about equal meaning loads more units of 2700X are actually being sold.

Also noteworthy is how three TRs have their own chunks on the bar, yet no X299 cpu sells enough to get its own chunk, and the 7980xe makes less sales by revenue than pentiums and doesn't even get mentioned

I hope Clevo are being smart and developing mobile AM4 gaming systems and a TR workstation right now

 

wow! that is pretty sweet yep, the 2700x sales are based on the users having held out to check and see what intel's 9th gen would bring to the table. then they saw initial pricings and went the AMD route without looking back haha

Sent from my Xiaomi Mi Max 2 (Oxygen) using Tapatalk

 

Finally... Good news if it's correct. @Mr. Fox

Intel is working on Core i9 9900 KF Series that lacks iGPU, also B365 chipset at 22nm Guru3d.com | on: 12/02/2018 | source: gamer.com.tw

A couple of leaked slides show some new stuff from Intel, first off with their 14nm production shortage, they will be moving what is called the B365 chipset towards a 22nm node. The chipset would replace B360.

More interesting news, however, is that Intel seeks to plan additional processor SKUs. Intel would be releasing 'KF' models of the ninth generation processors. Listed are the Core i9-9900KF, i7-9700KF, i5-9600KF, i5-9400F, i3-9350KF and i3-8100F.

 

So, CPUs without iGPUs equal cheaper to make & buy? I was gonna say less power consumption, but iGPU is totally inactive when dGPU is being used? If power consumption less, then more performance possibly?

 

Not sure if the GPU module still can use some minor power if voltage rails is disconnected. But at least this means the prices should be lower out from Intel.

 

I highly doubt they'll be fabbing a separate die just to remove igpu given the lead time and expense that involves. The igpu is still about 30% of the area of the 8-core die (down from about half on Skylake-S) so this theoretically saves about that proportion of chips that would otherwise be dud due to a critical manufacturing defect only in the igpu area, and I guess this matters more now that supply is constrained. It sounds like a way to get a few extra chips out there to market.

Perfect for the LGA Clevos that don't have the igpu wired up and is completely unpowered and unusable anyway

 

What happens if you set the iGPU voltage to 0v if the iGPU is disabled?
Will the CPU (LGA or BGA) even boot? Or is supply voltage (sliced, unsliced, etc) not used if the iGPU is inactive?

 

the difference is probably so minor its not even worth measuring. micro chips have been developed for so long im sure they already have good control over it when its not being used. as for if it'll actually have any effect, i have 0 clue.

 

Would this provide for a cheaper CPU, easier to manufacture, and or better overclocks?

 

by removing the igpu, the comet lake 10c seems more likely now. basically a refresh with possible 10 cores next year.

 

cheaper cpu - only if intel actually decides to let the consumers profit from their lowered costs

easier to manufacture - nope, same manufacturing process, just that intel can use more dies that would otherwise be classified as "defective"

better overclocks - very unlikely. the best clocking chips tend to be the ones closer to the center of a wafer and that goes hand in hand with lowered risk of defective silicon.

so the ones with the disabled igpu are very likely the ones from the outer wafer perimeter, thus id actually expect WORSE binning from these KF SKUs....

so dont get too excited about this guys, this smells like trashy silicon recycling for higher profit margins (as opposed to doing something "nice" for the consumer)

Sent from my Xiaomi Mi Max 2 (Oxygen) using Tapatalk

 

Overclocks have zero to do with the iGPU and everything to do with binning. CPU's with a lower default VID will overclock better or require less voltage for a certain speed. Maybe the bins will be better for that respin? No one knows.

As others said, price should drop as its cheaper to manufacture. Should also give lower temps as well.

 

- no doubt it'll prob be like $10 cheaper at best rofl, no way intel will let this go. saving money from iGPU and count it towards it's own profit gain.

- possibly easier to manufacturer, it is a monolithic die afterall with IGP included. remove the iGP and theres less chance of an 8 core cpu can't be sold as 9900k because of a defective igp.

- better overclock imho is a maybe, if it comes out in 2019 then they would have a year to make a true 14nm+++, or they wont.

- silicon recycle it is

 

It seems the wccftech readers agree, AMD in 2019:

Which 2019 mainstream processor platform are you most excited about?

https://wccftech.com/amd-ryzen-3000-cpus-x570-chipset-and-pcie-gen-4-support/

The video to the wccftech article about AMD's lead in sales over Intel:

AMD Dominates Retail CPU Sales in Germany’s Largest E-Tailer
WccftechTV
Published on Dec 2, 2018
AMD Dominates Retail CPU Sales in Germany’s Largest E Tailer Where It Outsold Intel 2 to 1.

 

A lot more K processors from same waffer means a larger selection and more manufactured devices. = Lower average prices from the stores on all affected models. They are to high now because of the shortage. Flooded market = lower prices The stores can't screw up prices as before.

 

Maybe, I suppose it comes down to how much stock of defective silicon they have due to IGP defects, if they have a lot then what you say would make sense, otherwise they would be able to make more dies per wafer if they redesigned the chips without the IGP included, which would lower production costs meaning potentially larger profits and/or better value consumer products. Yes, I suppose there is some cost associated with "re-tooling" if that's a thing in this process (don't know enough about it), redesign & seperate production runs, but you'd think redesign would be quite simple on a theoretical level as they're just cutting out the IGP and it's connections to the rest of the CPU.

 

I think easier way is to disable iGPU via microcode/firmware patches and save some cost and to level the pricing whilst compared to Ryzen w/o iGPU.

 

naw bruh, its intel we are talking about here. they'll somehow charge extra $$ and justify it.

 

Kind of like what they did with the 8086K, taking well-binned CPUs that would otherwise have been 8700Ks and adding another $50-$75.