Intel Core i9-10900

[W1zzard]

and you may lose the ability to increase power limits

You won't. Intel doesn't segment the power limit controls. Maybe the BIOS won't have options, you can always use Throttlestop.

 

[sumolDeLaranja]

It would be interesting to know what's the minimum budget required to achieve comparable results to this test's pseudo-unlocked 10900.
I mean, the difference between an ASUS Z490 Maximus XII Extreme and an ASRock X570 Taichi is around 600€ per my local price aggregator.

 

[blu3dragon]

It would be interesting to know what's the minimum budget required to achieve comparable results to this test's pseudo-unlocked 10900.
I mean, the difference between an ASUS Z490 Maximus XII Extreme and an ASRock X570 Taichi is around 600€ per my local price aggregator.

You don't need something that extreme Look up the price of an Asus Prime z490-p.

 

[W1zzard]

It would be interesting to know what's the minimum budget required to achieve comparable results to this test's pseudo-unlocked 10900.
I mean, the difference between an ASUS Z490 Maximus XII Extreme and an ASRock X570 Taichi is around 600€ per my local price aggregator.

Yeah that pricing is crazy. Other than features, don't expect any significant performance difference between cheapest and most expensive. VRMs might be hotter, will be fine. Plenty of decent Z490 boards out there below 200.

 

[yeeeeman]

The most impressive thing in my opinion is that at stock it consumes similar power to 3600x which is a 6 core baked on 7nm, while performing similar to a 3800x. Not bad.

 

[cucker tarlson]

i7 and i9 offi spec is DDR4-2933, below i7 only 2666

running 4133 cl16-16-16-31 on i5 10500 no problem.

The most impressive thing in my opinion is that at stock it consumes similar power to 3600x which is a 6 core baked on 7nm, while performing similar to a 3800x. Not bad.

yes it also throttles like crazy.
max turbo is what matters here,with PLs removed.
I dont think anyone buying an i9 or r9 should really consider power efficiency on 10/12 cores.what matters here is performance and it's really good.10900 beats 3900XT in cpu tests,not to mention gaming.

It would be interesting to know what's the minimum budget required to achieve comparable results to this test's pseudo-unlocked 10900.

a basic z490 mobo except for a couple of asrock ones.
same price as x570 probably.

 

[W1zzard]

running 4133 cl16-16-16-31 on i5 10500 no problem.

Z490 is required for "memory overclocking"
@_Flare listed the default spec, everything above that is considered mem oc by intel

 

[HD64G]

You need to look at the blue bar

So, you suggest anyone to get an relatively expensive non-K Intel CPU, pair it with a very expensive motheboard to withstand the 250W power consumption when unlocked? Not a very sensible choice me thinks. A R9 3900X is much better all-round. For gamers with 1080P, high refresh rate screens having a 2080Ti, even a 9700K is better choice. This CPU is as useless to exist as the R7 3800X(T) from AMD.

 

[W1zzard]

So, you suggest anyone to get an relatively expensive non-K Intel CPU, pair it with a very expensive motheboard to withstand the 250W power consumption when unlocked?

That ASUS motherboard is stupid, but it's what I have, and have been using for all Comet Lake reviews. So not switching until next rebench, to ensure fair comparison. I'm actually suggesting you buy the cheapest LGA1200 board with the features you want. Cheapest Z490 is fine, cheap H410, too.

A R9 3900X is much better all-round

Not according to my data. 3900X is slightly slower on average in CPU tests, 7% slower at 1080p, $10 more expensive, only 20% better power and 2°C cooler. I can give you "similar if not fully focused on gaming"

 

[HD64G]

That ASUS motherboard is stupid, but it's what I have, and have been using for all Comet Lake reviews. So not switching until next rebench, to ensure fair comparison. I'm actually suggesting you buy the cheapest LGA1200 board with the features you want. Cheapest Z490 is fine, cheap H410, too.

I beg to differ and will repeat that any mediocre board wont withstand for long a CPU that uses 250W when pushed. So, it is by no means economical to get that 10900 and unlock it over a stock 3900X that can work properly on a $100 board. My opinion ofc.

 

[W1zzard]

that any mediocre board wont withstand for long a CPU that uses 250W when pushed.

Not sure about that. The 3900X is 200 W full system, 10900 is 240 W, so your argument becomes "$100 boards can handle 200 W, but not 240 W" ? Could be a possibility, not convinced, but I don't have any good data.

Edit: Did a quick test just for you, 10900K @ 5 GHz manual, Prime95 = 225 W. 10900K because I was too lazy to uninstall the 10900 from the ASUS board
MSI Z490 Gaming Plus (which I bought for € 155 in retail), VRMs just warm, plenty of headroom, cooling the CPU is almost impossible though, with that Intel TS cooler

 

[navjack27]

I'm using this CPU on the new stock cooler on a asrock b460m pro4. Doing a review of my own. It does handle this CPU completely with a raised power limit of around 110w but I've tested 65 to 125 PL1 overrides. Also tested using my memory kits 3466 timings at the 2933 max. You'd be surprised what you can get away with hd64g

Edit: one fun thing I'll drop in here is that undervolting exists. Max this Mobo can do is -0.100v and that's what I've got it set at for my 24/7. 110w PL1 with that undervolt. I tested every voltage interval in 0.005v steps with 6 runs of realbench sans the opencl test with hwinfo logs. It's a measurable difference. I also did my test suite at "stock" 65w PL1 and jedec memory timings, 110w and 125w. Then I did em again at my memory kits 3466 timings at 2933. The review will be up on https://thechipcollective.com/ "soon" but the results are over here

https://twitter.com/i/web/status/1282369498935418880

 

[HD64G]

Not sure about that. The 3900X is 200 W full system, 10900 is 240 W, so your argument becomes "$100 boards can handle 200 W, but not 240 W" ? Could be a possibility, not convinced, but I don't have any good data.

Edit: Did a quick test just for you, 10900K @ 5 GHz manual, Prime95 = 225 W. 10900K because I was too lazy to uninstall the 10900 from the ASUS board
MSI Z490 Gaming Plus (which I bought for € 155 in retail), VRMs just warm, plenty of headroom, cooling the CPU is almost impossible though, with that Intel TS cooler
View attachment 162195

Thanks for the effort to provide more info @W1z . You wrote in the review that the PL2 limit is 224W. So, how the heck a normal board of $150 will stand the test of time if you combine that with this CPU if used with that power consumption? The 3900X uses 140W at stock. And we should leave temps of the CPUs aside as most can afford a $30-40 cooler when building a system of over $1000. But I would buy a very good and expensive board to use with a 225W CPU. No point to own that vs a 3900X for productivity or a 10700K if gaming is the target.

 

[W1zzard]

i7 and i9 offi spec is DDR4-2933, below i7 only 2666

finished retesting with 2933, new charts are up

PL2 limit is 224W

It's just the limit. I don't think it can reach that in normal circumstances, so consider it "infinite, no limit". In theory you could increase the voltage, but makes no sense because multiplier locked

 

[navjack27]

It's just the limit. I don't think it can reach that in normal circumstances, so consider it "infinite, no limit". In theory you could increase the voltage, but makes no sense because multiplier locked

Yeah that PL2 is very generous

 

[blu3dragon]

Thanks for the effort to provide more info @W1z . You wrote in the review that the PL2 limit is 224W. So, how the heck a normal board of $150 will stand the test of time if you combine that with this CPU if used with that power consumption? The 3900X uses 140W at stock. And we should leave temps of the CPUs aside as most can afford a $30-40 cooler when building a system of over $1000. But I would buy a very good and expensive board to use with a 225W CPU. No point to own that vs a 3900X for productivity or a 10700K if gaming is the target.

In general, Asus, Gigabyte and MSI all did a good job speccing up the vrms on their z490 boards to cope with the power requirements of comet lake. Look up some more reviews if you like. The Asus Prime Z490-P at $160 can handle the power just fine and W1zzard just showed that the MSI Z490 Gaming Plus can too. It seems you need to be more careful with ASRock (or rather just avoid ASRock for comet lake).

The more expensive boards basically give you additional io, more power capability for overclocking (not needed for this chip, and generally will require pretty extreme cooling), and slightly better memory oc capability (at least for Asus, have not researched the others).

 

[Poul-erik]

First this.

And so this, to me, it makes no sense.

 

[W1zzard]

And so this, to me, it makes no sense.

Have you read the review? At least the conclusion text?

 

[HD64G]

A messed-up product that has no purpose in market other than to gain traction for Intel CPU product line. It is exactly the same as the XT line of Ryzen 3000. No need to make it a bigger thing that it is. Over and out of this discussion. Everyone can have an opinion. Arguments always matter though.

 

[navjack27]

The place in the market for this is for the masses of people who have bought K series CPUs in the past and DO NOT overclock or use ANY of the features that K offers. Turbo is getting good. Power limits are finally at ranges where they won't be hit, at least for the PL2. There are people who will not ever be in the market for AMD. These are enthusiasts too and not just some newbie. That's the market segment but the public has been almost brainwashed into thinking that if you are getting a Intel CPU you NEED TO get a K. It's wrong and the public needs to be educated on how good these things are.

 

[Bee9]

The stock cooler is a joke with 200W.
Really tempting to get one of these and build a gaming rig for summer. It's getting up to 36C now in my place.

 

[Searing]

Ok I've now tested it with two mobos. There is a Gigabyte B460M DS3H (don't confuse it with the other B460 model, D3H is better) that is total garbage. The VRM hit 110 degrees and it throttled at stock settings (Gigabyte used 85W at stock, not 65W, not Intel's guidance) and I had to lower the limit all the way to 75W for it to work properly with that mobo. Returned!

I also used an Asus H410M and it is great (except the speckled white, yuck, I prefer all dark colours on my mobos)! The stock cooler can only cool 125W at maximum, so I set the motherboard max power draw to 110W and it hit 98 degrees for the CPU temperature, which is fine. Drawing 110W on the Asus H410M, the VRM is fine (note: unlike Gigabyte, Asus hides the VRM temp, but I can see no throttling), the cooler is fine, and you get 10 cores at about 4ghz and 100 percent load. Goes back up to 5ghz under 100 percent load of course. Neat!

Asus H410M, i9-10900, and stock cooler at 110W (set it in the BIOS), it works!

I'd really like Intel to release another all black cooler (the new one looks great, too bad it only comes with the i7 or i9, would be perfect for the 6 core CPUs) that is twice as thick and can cool 150W instead of 110W. I hate aftermarket coolers, they are overpriced or junk. If AMD put better bearings in their stock coolers so they'd be quiet at low RPM I'd never buy another aftermarket cooler either. I can't stand anything cheaper than the whopping $95 CAD Noctua U12S all black, that's expensive, but nice. The stock Intel black cooler is pretty neat, just not strong enough. I'll use the stock cooler though, I like the $100 cost savings and don't need more than 4ghz on 10 cores for 100 percent load.

 

[navjack27]

got some questions from twitter about this.

https://twitter.com/i/web/status/1283185165670461443

https://twitter.com/i/web/status/1283184685724643328

 

[Berfs1]

I'm gonna be honest, something is up with some of the results. I don't really understand how a 10900K at 5.2 GHz loses to a i9-10900K at 2.5 GHz. I made these weird numbers up because you could probably understand when running a CPU much faster, it SHOULD perform faster right? Take a look at your 10900K numbers. I don't really believe that a 10900K which clocks higher, loses to a stock 10900. Oh by the way, how does an i5-10500 beat a 10900K performance wise? That really doesn't make any sense. You could argue that maybe there are extra core to core latencies involved, but then you need to explain why a 10900 beats a 10500, but a 10900K LOSES to the 10500. That makes zero sense. Check the clocks. I guarantee you something isn't right about that.

 

[Bee9]

I'm gonna be honest, something is up with some of the results. I don't really understand how a 10900K at 5.2 GHz loses to a i9-10900K at 2.5 GHz. I made these weird numbers up because you could probably understand when running a CPU much faster, it SHOULD perform faster right? Take a look at your 10900K numbers. I don't really believe that a 10900K which clocks higher, loses to a stock 10900. Oh by the way, how does an i5-10500 beat a 10900K performance wise? That really doesn't make any sense. You could argue that maybe there are extra core to core latencies involved, but then you need to explain why a 10900 beats a 10500, but a 10900K LOSES to the 10500. That makes zero sense. Check the clocks. I guarantee you something isn't right about that.

What application are you referring to? Some applications just don't use all cores.
In many cases, the unleashed , overclocked to the max 10900 (in the red bar) has higher clock speed than the K variants chilling at default speed. With Intel sketchy boosty thingy, you will get a lot of variable clock speed between runs and that may cause variation in the result.
I don't see anything wrong with the chart.