10th gen i9 test results show disappointing performance

[mr moose]

1 hour ago, x-eagleeye-x said:

Oh how the mighty have fallen! Should serve as a hard lesson for them not to rest on your laurels. I hope they put out something amazing though. Competition is always good because consumers win. I don't want AMD to become what Intel is now.

What happened to Intel was not the result of complacency or resting on laurels.  Despite their best efforts they had issues with 10nm and AMD used their team of professionals to build a competitive product. The outcome is what one should expect when AMD create a product that is great in it's own right and Intel run into unexpected problems. 

[realpetertdm]

10 hours ago, GoldenLag said:

So next year.......

 

Will they continue their masterplan by releasing a 10900k with 2 more cores?

The Intel i9-11900k, now with 12 cores and 24 threads. Avaliable 2021

[Bombastinator]

4 hours ago, OlympicAssEater said:

10900k is going to be more hotter than 9900k. Good luck cooling it with air because big boy air coolers are not enough for 9900k. 

It’s base non overclocked TDP is higher anyway.  That may be to accommodate the new features.  It also may not.  We shall see what we shall see.  As to the air v water thing any cpu if pushed hard enough eventually requires liquid nitrogen to remain stable.  The claimed 220w is within range of a large air cooler.  It’s at or near max though.  

[it_dont_work]

Probably about time there's (likely) some hardware stagnation, last few years have seen some large improvements 

[GoldenLag]

48 minutes ago, it_dont_work said:

last few years have seen some large improvements 

You mean the "minor" node improvements and the great idea of elongating the ringbus to add more cores?

 

If that is improvements, then intel can continue untill the platter cant fit the really long die

 

(Old iphone memes aside, the die has really just gotten longer)

[RejZoR]

Everyone complaining over 14++++++++++++++nm and I realized I'm still on 22nm 4.6 GHz all core on 5820K at 1.2V. It was the 6800K that jumped to 14nm. Hehe. Then again that has been a very long time ago, especially given they are still on 14nm today.

 

Realistically speaking though, Intel can't really do much at this point. 5GHz clocks territory is already so high they have very little space to work with in regards of clock itself. Only way for them to boost single thread performance is to increase IPC. That has been done several times through history when clocks reached their maximum. Remember 4+ GHz Pentiums and then suddenly clocks dropped dramatically with Core processors yet they beat them in all scenarios? Similar with Bulldozer, although we know its flaws. 5GHz and then drop to 3.5+ GHz with speeds beyond anything Bulldozer could offer. I think Intel is gearing towards that, but they don't have anything to show yet. AMD has some room left up to around 5GHz and then they'll have to bump up IPC as well. 5GHz seems to be the ceiling where you're hitting certain electrical problems as well as heat from switching all those billions of transistors really quickly.

[Kilrah]

7 hours ago, OlympicAssEater said:

Good luck cooling it with air because big boy air coolers are not enough for 9900k. 

Really? I can cool mine to all core boost with a puny L9i. And the NH-D14 in my other rig happily dissipates 100W more than what a 5GHz all core overclocked 9900K puts out.

[it_dont_work]

30 minutes ago, GoldenLag said:

You mean the "minor" node improvements and the great idea of elongating the ringbus to add more cores?

 

If that is improvements, then intel can continue untill the platter cant fit the really long die

 

(Old iphone memes aside, the die has really just gotten longer)

My chronology may be slightly out of whack with yours, I'm more thinking decade 

[GoldenLag]

19 minutes ago, it_dont_work said:

My chronology may be slightly out of whack with yours, I'm more thinking decade 

I mean sandy bridge from nehalem was a pretty neat jump. 

 

Which was early last decade. 

 

And at that point in the consumerspace the die remained fairly consistant. 

 

Serverspace is another story. 

[Flying Sausages]

5 hours ago, Bombastinator said:

It’s base non overclocked TDP is higher anyway.  That may be to accommodate the new features.  It also may not.  We shall see what we shall see.  As to the air v water thing any cpu if pushed hard enough eventually requires liquid nitrogen to remain stable.  The claimed 220w is within range of a large air cooler.  It’s at or near max though.  

9900k is already hot so the 10900k is going to be more hotter than 9900k with 2 more additional core in place. It is still on 14nm so literary it is going to more hotter when you start adding more core. 125watts tdp is the base clock and boost clock or overclock is going to be 250 watts or more.

 

3 hours ago, Kilrah said:

Really? I can cool mine to all core boost with a puny L9i. And the NH-D14 in my other rig happily dissipates 100W more than what a 5GHz all core overclocked 9900K puts out.

I don't believe L9i can cool the 9900k at all core boost. You probably has a golden 9900k because a lot of users are not happy with the temp in 5ghz all core oc.

[MageTank]

4 minutes ago, OlympicAssEater said:

9900k is already hot so the 10900k is going to be more hotter than 9900k with 2 more additional core in place. It is still on 14nm so literary it is going to more hotter when you start adding more core. 125watts tdp is the base clock and boost clock or overclock is going to be 250 watts or more.

 

I don't believe L9i can cool the 9900k at all core boost. You probably has a golden 9900k because a lot of users are not happy with the temp in 5ghz all core oc.

It's not as simple as more cores = harder to cool though. With additional cores, we might get a larger die which can provide a larger surface area to transfer heat to the IHS. The die size of the 9900k is small, especially compared to the LCC 10 core X299 chips. With Intel going with LGA 1200, I am hoping the slight increase in size allows them to use a larger die. If it's the same relative size as the 9900k, then yeah, it might be difficult to cool.

[GoldenLag]

9 minutes ago, MageTank said:

With Intel going with LGA 1200, I am hoping the slight increase in size allows them to use a larger die

Would require node tuning to lower density. And with supply issues its unlikely at best they would want to create bigger dies. 

 

Also that is the size of the substrate. And not even that. Its defines the pin count. 

 

Also the heat putput during an allcore boost is likely higher. As such the total heat needed to dissapate is higher. 

 

 

Its probably not harder to cool than the 9900k at stock.

[jones177]

16 hours ago, Caroline said:

They're just going for the shit 5GHz crap no matter what to keep the gamer fanboys happy because apparently that number means something to them

It only doses as far as I know with CPUs like the i7 8700k and i7 8086k going from stock. The next performance plato is at 5.35hz on these chips so a chip with 5.3 could be interesting.  

 

I use the 5ghz in my heavily modded bethesda games. It removes the CPU lows in those games. With a i7 2600k the lows were in the 30s. An i7 6700k put the lows into the 40s. An i7 8700k at 4.7ghz on all cores put the lows into the 50s and a i7 8086k puts them above 60. Since these game are vsynced to 60 the lows no longer exist. 

It was one of my goals to remove these lows that I first experienced in 2007. It took until 2018 to do it.  

 

In my vanilla games the overclock did not matter at all until I upgraded from a 1080 ti to a 2080 ti.  The 2080 ti likes 5ghs and I would only recommend a 5ghz Intel chip if you are running one. With anything less get a Ryzen.  

 

11 hours ago, OlympicAssEater said:

10900k is going to be more hotter than 9900k. Good luck cooling it with air because big boy air coolers are not enough for 9900k. 

An i9 9900k can run cool with a Noctua NH D15 using a single fan even with all cores set to 5ghz. 

They just need a good motherboard like an Aorus Ultra.

 

[Nanners]

I got my 9900k at 5.1 all core with NH-D15 at 30C over ambient during games in a Cooler Master H500 and fans at 50% ish (inaudible with headphones) with a 2080 Ti preducing fair amount of heat at 45C over ambient at it's max stable OC with similar noise production.

[Kilrah]

6 hours ago, OlympicAssEater said:

I don't believe L9i can cool the 9900k at all core boost. You probably has a golden 9900k because a lot of users are not happy with the temp in 5ghz all core oc.

It's between 4.7 and 4.5 depending on ambient temp on long loads. All core 5GHz draws an additional 30-40W for just 300MHz, so yes you draw 30% more for just 5% more performance and heat goes through the roof. Not worth the hassle IMO, all core boost is good enough for me (and I have no choice, can't do better in my 4.7L SFF case). 3.6 base clock would certainly not be satisfying though.

 

6 hours ago, MageTank said:

With additional cores, we might get a larger die which can provide a larger surface area to transfer heat to the IHS.

Absolutely, people really tend to underestimate the infuence of die size and die-IHS-cooler interfaces.

When I upgraded from 7700K to 9900K I did some tests on both with the same cooler, on the 7700K it would hit Tjmax with 100W, the same cooler with the 9900K could dissipate 30W more with the same temp just because of better transfer from die to cooler.

[Bombastinator]

2 minutes ago, Kilrah said:

It's between 4.7 and 4.5 depending on ambient temp on long loads. All core 5GHz draws an additional 30-40W for just 300MHz, so yes you draw 30% more for just 5% more performance and heat goes through the roof. Not worth the hassle IMO, all core boost is good enough for me (and I have no choice, can't do better in my 4.7L SFF case). 3.6 base clock would certainly not be satisfying though.

 

Absolutely, people really tend to underestimate the infuence of die size and die-IHS-cooler interfaces.

When I upgraded from 7700K to 9900K I did some tests on both with the same cooler, on the 7700K it would hit Tjmax with 100W, the same cooler with the 9900K could dissipate 30W more with the same temp just because of better transfer from die to cooler.

If we’re lucky then and the 10 series 8 core i7s are 10 core CPUs with two cores locked off, the  series 10 8/16 will run cooler than the series 9 8/16 stuff.  Unlikely.  One can hope though.

[Flying Sausages]

18 hours ago, Kilrah said:

It's between 4.7 and 4.5 depending on ambient temp on long loads. All core 5GHz draws an additional 30-40W for just 300MHz, so yes you draw 30% more for just 5% more performance and heat goes through the roof. Not worth the hassle IMO, all core boost is good enough for me (and I have no choice, can't do better in my 4.7L SFF case). 3.6 base clock would certainly not be satisfying though.

 

Absolutely, people really tend to underestimate the infuence of die size and die-IHS-cooler interfaces.

When I upgraded from 7700K to 9900K I did some tests on both with the same cooler, on the 7700K it would hit Tjmax with 100W, the same cooler with the 9900K could dissipate 30W more with the same temp just because of better transfer from die to cooler.

Your d14 probably can't handle all core 5 ghz oc. Even der8auer had a  bad time with the 9900k too. 9900k is already hot at stock. 84c @ stock on air with 8 core. Imagine what the stock temp will be on 10900k. 10900k to me is 9900k with 2 more additional core and some tuning.

 

[Kilrah]

32 minutes ago, Harry P. Ness said:

Your d14 probably can't handle all core 5 ghz oc.

My D14 is on an overclocked 5960X putting out 280W. My 9900K puts out about 160W at 5GHz all core. But yeah at this point the limit isn't the cooler, it's what's below it as the graph you posted confirms.

[Bombastinator]

1 hour ago, Harry P. Ness said:

Your d14 probably can't handle all core 5 ghz oc. Even der8auer had a  bad time with the 9900k too. 9900k is already hot at stock. 84c @ stock on air with 8 core. Imagine what the stock temp will be on 10900k. 10900k to me is 9900k with 2 more additional core and some tuning.

 

The testing seems to me to show that AIO did no better though and custom water made barely any difference at all. It was lower at 5.1 but higher at 5.0.  Could be within great variance though.

Assuming AIO could handle a higher thermal load than air additional thermal load capacity doesn’t help.  The bottleneck doesn’t seem to be the cooler’s ability to remove heat so much as the CPU’s ability to transfer it.  It’s not a question of how big the cooler is.  There appears to be a maximum amount of useful cooling and that level may actually be below the level of the d14 rather than above it.