I9 9900k

Distracted · 21 Oct 2018 at 21:50

It will be interesting to see what sort of numbers places like silicon lottery are seeing for overclocks.

NinjaCool · 22 Oct 2018 at 10:23

Hardware Unboxed explains the temperature discrepancies with Linus and OC3D's 9900K review.

DarrenM343 · 22 Oct 2018 at 10:33

Distracted said:
Agreed. No way would they have used solder TIM if they could have used regular TIM.

That said, not every 9 series is an inferno.

Who knows. THings are as much marketing days. Enthusiasts complain about toothpaste TIM, give them enthusiast CPU's with solder, even if temps no better

. I suspect paste would have been just as good especially if they just used a premium brand, but, hell I'm no expert.
Would be surprised if reviewers didn't get cherry picked samples. I reckon Jay was too busy cracking jokes to notice the temps

Silent_Scone · 22 Oct 2018 at 10:39

[Death] said:
not defensive, just think its mad some reviewers are getting mad temps, must be somthing wrong or bad samples, max ive seen now is 84, which for all core 5ghz i think is good tbh

There's a bit of a variance between samples from what I've seen here. 5.1 for non AVX is a very good chip.

Distracted · 22 Oct 2018 at 10:43

NinjaCool said:
Hardware Unboxed explains the temperature discrepancies with Linus and OC3D's 9900K review.

Wow. Intel really needs to do something about their TDP figures.

Maybe have a base TDP and a boost TDP if they insist on marketing the lower number.

Then have boards rated for either base or boost accordingly.

Thx for sharing this.

RavenXXX2 · 22 Oct 2018 at 11:21

Distracted said:
Were you running other things while doing that CB run? Looking at benches online from reviews, 5.1 should be scoring 2200 ish.

Looks like his board/chip is throttling, would explain low temps. See vid.

Distracted · 22 Oct 2018 at 11:30

RavenXXX2 said:
Looks like his board/chip is throttling, would explain low temps. See vid.

Yeah i saw that video. Commented on it in the post above yours.

Death is using the z390 Hero though, and said his chip does near 2200 with 5ghz. Seems to be an issue with ram rather than vrm throttling.

RavenXXX2 · 22 Oct 2018 at 11:41

Well the hero is the board being reported as throttling with a 4 phase vrm, and his R15 score is very low for 5.1Ghz, 5Ghz is 2200 points.

Distracted · 22 Oct 2018 at 12:26

RavenXXX2 said:
Well the hero is the board being reported as throttling with a 4 phase vrm, and his R15 score is very low for 5.1Ghz, 5Ghz is 2200 points.

I thought the z390 hero was 8+2?

RavenXXX2 · 22 Oct 2018 at 12:34

Steve at hardware unboxed says it's a fake 8 phase.

I'll leave the original comment below but right now we're 99% sure the ROG Maximus XI Hero has a big fat 4-phase VRM designed to look like an 8-phase. So fake 8-phase that's really a 4-phase VRM, but we will continue to update this post as we get more info.

Distracted · 22 Oct 2018 at 12:38

RavenXXX2 said:
Steve at hardware unboxed says it's a fake 8 phase.

Ty

TrixP10 · 22 Oct 2018 at 12:38

Intel's 10nm is DEAD?

https://semiaccurate.com/2018/10/22/intel-kills-off-the-10nm-process/

Distracted · 22 Oct 2018 at 12:38

Can you link the source so i can see the rest please? Interested in reading it.

RavenXXX2 · 22 Oct 2018 at 12:40

[Death] · 22 Oct 2018 at 13:56

No throttling from what I can see on any temp sensor read outs using multiple software to check, at work now until tonight so will look again then, 5ghz runs fine now ram is sorted getting over my score at 5.1, will put chip back to 5.1 after work and rerun it, chip defo not going into throttle tho no where near max for that and vrm ain’t too hot so should not be throttling

BluD · 22 Oct 2018 at 21:01

Distracted said:
Buying a CPU has always been a lottery.

You go in hoping to win or at least get your moneys worth, but sometimes you lose. Seems like there were a few losers amongst the reviewers this time round, which is unusual.

Jayz2cents and gamersnexus had no difficulty hitting 5ghz. Called it easy.

Others saw crazy temps to get there it seems.

Yeah its always a big lottery, i normally get toastie chips that are hopeless without delidding, this is the first non soldered chip ive owned that i havent had to delid from intel or AMD.

Distracted · 22 Oct 2018 at 21:17

BluD said:
Yeah its always a big lottery, i normally get toastie chips that are hopeless without delidding, this is the first non soldered chip ive owned that i havent had to delid from intel or AMD.

Well the 8086k is a binned 8700k so it makes sense that it would be decent. Glad you finally got a good one!

tostitobandito · 22 Oct 2018 at 21:51

RavenXXX2 said:
Well the hero is the board being reported as throttling with a 4 phase vrm, and his R15 score is very low for 5.1Ghz, 5Ghz is 2200 points.

That is completely wrong. This guy doesn't know what he's talking about. The new Maximus boards are 8 phase on the core (4x2 with doublers I believe), for 8+2 total. The issue is that different motherboards have different default behaviors with multi-core enhancement (MCE). If this guy thought that his Asus board wasn't allowing to get enough power, he needs to go make sure MCE is enabled. If you leave MCE in "auto", the Asus boards at least seem to treat that as disabled. Other MB manufacturers seem to enable it by default, probably so they benchmark faster. Disabling it by default is actually the "proper" thing for them to do since MCE can dramatically increase core voltage far past where it needs to be.

This is proven by the fact that GN and others have easily OC'd that board to 5.3 GHz and beyond with the 9900K, drawing twice as much power as Linus and this guy claimed they were limited to. The VRM's are fine, being 8 phases of 50 amps each.

TLDR - With MCE disabled (or auto on Asus) you won't exceed the Intel power specs and will clock even lower than stock boost across all cores. If you want to uncork these CPU's, you need to either enable MCE or go and set your voltages and frequencies manually. A lot of these 9900K benchmarks would probably be better in terms of heat/power if people had taken the time to find the minimum stable core voltage for whatever clocks they're running.

lovingbenji · 22 Oct 2018 at 21:51

Hello there, just a quick question before I order a 9900k OEM : those processors are brand new, never tested ? Apart from no box and 1 year warranty, really nothing else ?

Distracted · 22 Oct 2018 at 22:06

lovingbenji said:
Hello there, just a quick question before I order a 9900k OEM : those processors are brand new, never tested ? Apart from no box and 1 year warranty, really nothing else ?

Seems that way. The processors that get binned are separate stock according to Gibbo. So after binning even those that don't reach high clocks remain with the system builder people and don't get put back into the warehouse stock for selling on.

That is as much info as we have had about it.

So when you buy an oem from Ocuk it should be new, plain box, 1yr warranty, but with the intel starter software pack including killing floor 2, some adobe stuff and COD 4.