CPU NEW 10900k! (with bonus weird marketing points)

That was a first gen board. AM4 is now on the 500 series, which is third gen. 320 is about the lamest board you could have got. The only thing limiting that 320 board is the phases. Note the max is 65w for the CPU. However, a £400 that wasn't.





Hits 5.4ghz @ 1.35v. No mention of heat or power draw, still slower than 3900x.

Which kinda makes it pointless, even for "But they are faster" gaming fanatics. Hear me out. If you want the fastest gaming CPU then that will quite probably still be the 8 core. If you were doing anything else? the 3900x will be better, and you still have the option of 3950x. Both of which I foretell are going to become much cheaper, too. And you can run either of those easily (and overclock them) in a £180 motherboard.

BTW going back to PCIE4? yeah, it's a waste of time. So quite why they are pushing that is bizarre. There really was no need for that, apart from "Small willy syndrome".

 

The single thread is the more interesting thing on this intel chip at 5.4Ghz, it is no surprise less threads produces lower number in cinebench multi, who really cares about that, just a nice benchmark for most.

I would use PCIE 4, its not a question of willy waving, it is there why wouldn't you use it if you have it, it would be silly not to surely?

I mean you don't sit there and put your GPU on PCIE2.0 do you?

Certainly on these boards with low lane count if you have pcie 4 there would be no performance penalty on gpu dropping down to 8 lanes when you use more add ins as there is in a PCIe 3 system at the moment, sure it is small but it is there.

 

I'm wondering if the extra 3-400mhz will make much of a difference. I really expected it to be faster than a 3900x. Like, with such a massive clock gain I really did think it would be faster. Even at 5.4ghz it's around 5% slower.

There's also no word on how it was cooled, how hot it got and how long it could sustain those clocks. CB15 especially on modern fast CPUs really doesn't take very long at all. It also (IIRC) doesn't have the massive AVX clock defacit that CB 20 does.

That's why I think it's pretty pointless. Because the 8 core will probably overclock the same, and be every bit as good at gaming. As for GPUs and PCIE 2/3? I left my X79 rig on 2.0 because installing the patch made 0 difference (as you would expect). There are only two PCIE 4 things ATM. Firstly the higher end 5000 series AMD GPUs (and it's a waste of time) and storage. And you can't even use it for storage because the 10900k doesn't support PCIE 4.

So it's literally like they have just added it as a selling point, even though you can't use it.

 

No one knows, it's a leak. An actual leak, not the usual advertising rubbish. What is interesting is that it says it had a maximum core temp of 62. Which is quite clearly either impossible, or that was chilled. What's also very strange is the minimum temp was 0.

Even more telling? look at the 3950x scores above it.

@5.3ghz. lmao, there is absolutely no way in heck that chip is on air or standard water.

 

Thanks dude. So that means that 5.3 is going to be absolutely impossible on all cores and threads on even high end water cooling.

 

Sounds like really there is not a lot more to be had out of these designs now in terms of clkspeed.

 

Indeed. You'll probably find the 10 core has lower sustained clock speeds than the 9900k and the forthcoming 8 core. So this won't even be better for gaming.

Not looking good when it can't even match a 3900x stock.

 

https://videocardz.com/newz/intel-core-i9-10900k-tested-and-compared-with-ryzen-9-3900x-3950x

Exactly what was expected really. Eats an awful lot of power.

 

Ooooof! But I can see why they dropped the prices, seems to perform exactly where it should! If only those motherboards weren't so expensive! £500 for a CPU I can stomach, £500 for a motherboard seems mad!

 

It's worrying. I watched Linus earlier and it's odd. The video just had a really odd vibe about it. I mean yeah, he doesn't want to lose face with Intel, but he seems actually worried about this CPU being horrible.

He did drop a few important things that made it worth watching. Firstly that even at stock clocks this CPU will reach the 80s under water. Noting the word under water. Then he went on to say how bad the Pentium D was.

Thing is? he knows. He knows because he will have the stuff in hand now and has probably tested it, so it just seemed like damage limitation to me and "Oh Intel every one is going to slag you off so I will get in first".

The rest? all BS. How he goes on about the nice guys at Intel letting him into secret rooms to see secret gear yet won't let him film it and he can't tell any one. Why would I care about that? that's Linus' brown nose perk.

 

As he said in the video, its all about the executives. I've used a fair few of Intels other industry products, and they are utterly fantastic and the support is better than any other company I've come across. Had problems getting a neural compute stick 2 to work with my CNN model for work and they have paid support on the Open Vino support forums who basically give free help to anyone! I've never seen hat before! (Plus he was called Jesus, which made it doubly amusing). And I think thats the truth of it. We're used to Intels main focus being CPU's, but for the last few years its been on a lot of other things. They were the only ones beating back Quallcom in the cellular modem market for a long time! And Quallcom is a truely awful company, but you don't see people boycotting or being outraged by them, even though it affects our lives far more! (sorry, I appear to have gone on an early morning ramble) I think the long and short of it in the Intel higher ups are quite content with a 'good enough' CPU that has far lower R&D costs because they'd so busy pumping money into other industries. Although the Movidius (The AI processor) has a £500k SDK cost!!!!

 

Same with Nvidia (except their chips also happen to be wiping the floor with the competition): they put an utterly massive amount of effort into developer support. Driver-side optimisation, code snippets, outright entire libraries, will literally assign your their developers to help with your code, etc. Not just on the gaming side, but on the GPGPU side and RT sides too.

Which is why when you see the normal internet whinging about "but X is proprietary and Y is Open Source! They must have been paid to use X!": X has a large support network dedicated to helping you use X effectively and easily, Y has some code comments and a single support guy whose job is to tell you to read the code comments.

 

Exactly, its why Open CV is actually so good as Intel does a lot of support for it! Opensource only works well if there's someone actively developing it, if its stagnant its basically worthless. The amount of times I've come to use a really useful API and there litterally have been programming errors I have to fix on open source code are to many!

 

Once again LTT doing in 10 minutes what everyone else does in 25! So doesn't actually look too bad, if the motherbaords were cheaper it'd be very compelling!

 

Watching Debauers video:


He discusses how you basically can't overclock them on water because of the heat density, so this is actually a strong argument for a chiplet style design, as then you're spreading out your thermal density a bit.

 

Mostly because they're already overclocked out the wazoo from the factory. Any additional overlooking is by definition extreme overclocking, because all the garden variety slap-in-a-jumbo-cooler overclocking was done before you got your hands on it.

 

Not really true as it can do 7GHz on all cores, and that is very extreme

 

Bunch of big subs over on hwbot.org today for the release, 7G for an multicore bench is just mental.

 

Just as a warning. It seems that the shills received very highly binned 10900k. I thought it was odd that shill videos like this started popping up.



I wouldn't mind, but it's just so predictable. This is obviously Intel twisting nipples for damage control, whilst back here in reality things are not quite what Jay there would have you believe.

Three days ago a forumite buddy of mine on another forum bought a retail 10900k. Here is what happened...

so I decided to grab a 10900k for my new build judge all you want

Im busy playing with it and have found it very good for overclocking but at 1.345V 5.2ghz all cores if I run prime 95 avx small fit's it hits over 100 degrees instantly. Anyone else with one of these experiencing this just want to know its normal and not my waterblock mount???

Also I've noticed a core to core delta of a whopping 20 degrees

Is this normal?

Just want to judge my experience against someone else to make sure everything is normal and I've not got any loop issues

Running Aida64 cpu, FPU, cache stress test 1 hr with Valley in background
Cpu package 93 degrees, core temps 74 74 78 76 86 93 89 80 81 73

He was then advised to lower the voltage, as it was really high. After doing so he posted this.

I dont like using an offset as a lot of games nowadays use avx instructions so might as well run at the most stable avx clock I can get this cpu non avx stable at 5.3 no problem at these voltages.

The thing that threw me was watching a few videos of people running 1.45v 5.1ghz avx on a single 360 rad and getting lower temps in prime 95 I cant even run prime 95 without hitting 108 degrees at over a 1 volt less

Stablity wise I've done 12hrs occt non avx, 2hrs aida64 with valley running in the background no errors

The thread rolled on, and no matter what he tried nothing worked. This is what happened.

Ok guys so I thought YOLO and went for it and I delidded my 10900k and applied liquid metal and what a differnece!!!!!!

Same test as before aida64 with valley in the background my temps are now average:
Cpu package 76 core temps 65 65 68 67 68 70 66 68 62 64

Previous

Cpu package 93 core temps 74 74 78 76 86 93 89 80 81 73
No more core to core delta and roughly a 20 degree improvement in temps!!!!!

This is without letting my liquid metal and kryonaught cure so temps should improve further

He had to delid it. Which is in a completely different league to delidding a CPU before. This required a delidding tool, heat gun and god knows what else.

Yeah it was a squeaky butt moment!!

I used my heat gun to heat the cpu up whilst in the delid tool so the indium was nice and soft and it came off very easily.

The solder is very soft and got most off with my finger nails and plastic scraper.

I then used some 99 percent alcohol and 3000 grit sand paper to sand the die to a nice shine

I removed all the sealant used clear nail varnish on the exposed points and then applied liquid metal.

Its sitting on there loose but I'm not fussed as I can seal it later when I can gran some high temp sealant

So I would be very careful if you are considering one of these. Because in the cold light of day they really are not nice to live with.