Intel 9900k Owners Club

[BLUE-ANGEL]

Ok, so I will invite myself and say hi to all the other 9900k owners,

Hi everyone!

Specs:
-i9 9900K (Currently at stock)
-H150i PRO
-ROG MAXIMUS XI HERO Z390 BO4 (WI-FI)
-32GB GSKILL TRIDENT Z NEO RGB 3600MHz
-STRIX 2080 OC
-512GB M.2
-512GB SSD
-2TB & 2x4TB HDD
-CORSAIR HX1200i PSU
-RAZER BLACKWIDOW V2
-RAZER MAMBA WIRELESS
-PHANTEKS PRO M ACRYLIC
-LL SERIES FANS
-COMMANDER PRO
-LG 34UM68-P
-DELL SE2717H
-ASTRO A50
-LOGITECH Z906

3DMark results.

Result not found

www.3dmark.com

Okay, not to burst your bubble or anything, just to give you an IDEA....I also run a i9 9900K at stock with an RTX 2080 but NOT the OC model, but did a bit of OC myself to the card only. CPU stays stock and my Timespy score is 11870. Also, I only have 16GB 2666 memory.
You have a higher CPU score than I have, but lower Graphics score so I am sure you can do better on that score bud....just my 2 cents.

 

[HNO3ZA]

Okay, not to burst your bubble or anything, just to give you an IDEA....I also run a i9 9900K at stock with an RTX 2080 but NOT the OC model, but did a bit of OC myself to the card only. CPU stays stock and my Timespy score is 11870. Also, I only have 16GB 2666 memory.
You have a higher CPU score than I have, but lower Graphics score so I am sure you can do better on that score bud....just my 2 cents.

I have never OC a GPU, maybe I should attempt that a bit? Just a bit, to improve score?

 

[BLUE-ANGEL]

I have never OC a GPU, maybe I should attempt that a bit? Just a bit, to improve score?

Yes you can....its not difficult with Afterburner...definitely worth the try. I run a silly 212 cooler on my CPU at the moment so I obviously might get a bit of CPU throttling but you should definitely be hitting close to my score or even better who knows. Give it a try bud.

 

[baasgene]

Guys could one or two of you please do me a small favor? Run Realbench stress test just for 15 minutes and check the cpu load for me? Please run at least 5ghz all cores.

My cpu load isn't 100 constant (as reported by Realbench), it sometimes drops to 40 for a second or 2 and jumps back to 100, and it isn't stuck on 100 it ranges from 99-100%.

Scores on all cpu benchmarks are in line, and no stuttering during gameplay or anything.

This happens on the Asus Tuf Z390-Pro gaming and my current Auros Z390 Ultra.

Tonight when I get home I will check if it happens when the cpu is at stock too. Maybe it's a sign of instability, currently at 1.29v 5ghz. Passes stress testing with AVX at 0 offset with max temps of 86c.

 

[HNO3ZA]

Guys could one or two of you please do me a small favor? Run Realbench just for 15 minutes and check the cpu load for me? Please run at least 5ghz all cores.

My cpu load isn't 100 constant (as reported by Realbench), it sometimes drops to 40 for a second or 2 and jumps back to 100, and it isn't stuck on 100 it ranges from 99-100%.

Scores on all cpu benchmarks are in line, and no stuttering during gameplay or anything.

This happens on the Asus Tuf Z390-Pro gaming and my current Auros Z390 Ultra.

Tonight when I get home I will check if it happens when the cpu is at stock too. Maybe it's a sign of instability, currently at 1.29v 5ghz. Passes stress testing with AVX at 0 offset with max temps of 86c.

I dont know much about OC myself, but would love to see if my baby can handle 5.0GHz on all cores... But Im too scared to pickle with them settings... LOL

 

[BLUE-ANGEL]

Guys could one or two of you please do me a small favor? Run Realbench just for 15 minutes and check the cpu load for me? Please run at least 5ghz all cores.

My cpu load isn't 100 constant (as reported by Realbench), it sometimes drops to 40 for a second or 2 and jumps back to 100, and it isn't stuck on 100 it ranges from 97-100%.

Scores on all cpu benchmarks are in line, and no stuttering during gameplay or anything.

This happens on the Asus Tuf Z390-Pro gaming and my current Auros Z390 Ultra.

Tonight when I get home I will check if it happens when the cpu is at stock too. Maybe it's a sign of instability, currently at 1.29v 5ghz. Passes stress testing with AVX at 0 offset with max temps of 86c.

You probably have a decent cooler with that...I can't run 1.29v on 5GHz my system crashes when I overclock and play games due to having a 212 cooler. And 15min is a bit short, I think any decent stress test should not take less than 1 hour. If I remember correctly when retail companies assembles a gaming PC for you they do a 24hour stress test, I might be wrong.

 

[HNO3ZA]

So who here is willing to guide n nub like myself that has some experience with this OC stuff....

 

[baasgene]

I dont know much about OC myself, but would love to see if my baby can handle 5.0GHz on all cores... But Im too scared to pickle with them settings... LOL

Set LLC to 6, manual vcore to 1.3v, set XMP profile. Should set you up, oh and cpu core to 50x, sync all cores.

 

[baasgene]

You probably have a decent cooler with that...I can't run 1.29v on 5GHz my system crashes when I overclock and play games due to having a 212 cooler. And 15min is a bit short, I think any decent stress test should not take less than 1 hour. If I remember correctly when retail companies assembles a gaming PC for you they do a 24hour stress test, I might be wrong.

I passed an hour, I just mean somebody can run it for me even for 10 minutes to check the cpu load

 

[HNO3ZA]

Set LLC to 6, manual vcore to 1.3v, set XMP profile. Should set you up

I speak most International languages known to man, except Greek, and that looks like Greek to me... Hahaha.

I will check it out tonight and report back if I struggle, thnx man.

 

[BLUE-ANGEL]

Set LLC to 6, manual vcore to 1.3v, set XMP profile. Should set you up

1.3 volts might be just too little depending on the temps and load. For the i9 9900K they normally advise about 1.32 volts, but that's to test.

 

[baasgene]

1.3 volts might be just too little depending on the temps and load. For the i9 9900K they normally advise about 1.32 volts, but that's to test.

That depends on chip lottery overall, but I will definitely increase Vcore and see what happens. Normally Realbench would either report instability, or pc would blue screen if the Vcore was insufficient. So this is a bit of a strange case.

 

[Nyt Ryda]

I passed an hour, I just mean somebody can run it for me even for 10 minutes to check the cpu load

Done. No dips to 40%, even between the handbrake tests.


Maybe its the monitoring software lagging ? I had to detach the msi Afterburner window and click on it to be in focus to get it to update (else it just stays frozen, but hwmonitor updates fine).

 

[Nyt Ryda]

Passes stress testing with AVX at 0 offset with max temps of 86c.

Which test are you using for AVX ? Is that your Realbench temp ?

Also, have you considered changing out the H150i's stock fans for ML120 Pro's ?
I see there's a new H150i RGB Pro XT out that comes with 2400rpm, 75cfm, 4.2mmH2O static pressure fans versus the 1600rpm, 47.3cfm, 1.78mmH20 static pressure fans that come with the H150i Pro.

 

[baasgene]

Which test are you using for AVX ? Is that your Realbench temp ?

Also, have you considered changing out the H150i's stock fans for ML120 Pro's ?
I see there's a new H150i RGB Pro XT out that comes with 2400rpm, 75cfm, 4.2mmH2O static pressure fans versus the 1600rpm, 47.3cfm, 1.78mmH20 static pressure fans that come with the H150i Pro.

Fan speed doesn't do much when it comes to 360mm rads, even 1600rpm will have similar performance to 2400mm.

The big draw of the XT is better RGB, and a supposedly quieter pump, but thermals will not even differ by 2 degrees you'll see. There's not much you can do to improve upon a 360mm rad, this H150i is so inaudible already and I don't do rgb actually, only my ram is rgb lol. You can compare reviews between the H115i Plat and Pro and see that they perform the same, with the platinum losing on occasion for some reason, the only worthy draw is RGB imo. On a 360mm rad the difference is even less.

Thanks for your feedback! I have managed to fix this issue by upping my vcore to 1.31v. It was running out of steam for some reason. I see your thermals are the same as mine, I have managed to improve my airflow in the case, I now top out at 87c for half an hour of realbench. And that is at 1.31v so I'm going to settle for this now.

 

[baasgene]

Jirre I'm sitting here wondering what the point is of running this at 5ghz guys? Obviously as you've experienced the chip is far too hot actually for conventional AIO systems and Air.

We're talking 1.32v for example, which is not bad, but the thermals are running into the high 80s on a proper avx load. On gaming it's low especially on 1440p or upwards, but that's obvious because it's running at like 30% maybe 40% at best.

at 4.8ghz 1.25 or less vcore, temps drop like 15c... You're losing 4% performance against an almost 20% improvement in temps (in fact if you run 1440p or 4k like me you don't lose performance). And not even mentioning power draw.

I've decided to run 1.23v at 4.8ghz. Max temp on AVX stress testing is now a measly 73c, games are not even worth mentioning it's in the low 50s.

Most of us are pushed by our overclocking nature to actually run 5.0ghz, and the number is a golden one, but even from an overclocker's point of view I can't justify a cpu that runs over 80c especially 85c at full load . It just doesn't make me happy

Stock is also just fine I bet, but I like to keep my cores all running the same. This chip is just too hot for me.

 

[Nyt Ryda]

Jirre I'm sitting here wondering what the point is of running this at 5ghz guys? Obviously as you've experienced the chip is far too hot actually for conventional AIO systems and Air.

We're talking 1.32v for example, which is not bad, but the thermals are running into the high 80s on a proper avx load. On gaming it's low especially on 1440p or upwards, but that's obvious because it's running at like 30% maybe 40% at best.

at 4.8ghz 1.25 or less vcore, temps drop like 15c... You're losing 4% performance against an almost 20% improvement in temps (in fact if you run 1440p or 4k like me you don't lose performance). And not even mentioning power draw.

I've decided to run 1.23v at 4.8ghz. Max temp on AVX stress testing is now a measly 73c, games are not even worth mentioning it's in the low 50s.

Most of us are pushed by our overclocking nature to actually run 5.0ghz, and the number is a golden one, but even from an overclocker's point of view I can't justify a cpu that runs over 80c especially 85c at full load . It just doesn't make me happy

Stock is also just fine I bet, but I like to keep my cores all running the same. This chip is just too hot for me.

It's all down to personal preference.

I did a test the other day with 4.8GHz @ 1.24v vs 5GHz @1.285v, temps dropped nicely and there was hardly a performance drop - like 1-2fps in AC Odyssey at 1440p which is hardly anything on one of the most CPU intensive games out there and is unnoticeable especially with G-Sync.

But given that the new i7 10700k is looking more and more like a rebadged 9900k with a 200mhz increase in clockspeed I figured I'd just keep it at 5Ghz so the performance is on par.

 

[BLUE-ANGEL]

Jirre I'm sitting here wondering what the point is of running this at 5ghz guys? Obviously as you've experienced the chip is far too hot actually for conventional AIO systems and Air.

We're talking 1.32v for example, which is not bad, but the thermals are running into the high 80s on a proper avx load. On gaming it's low especially on 1440p or upwards, but that's obvious because it's running at like 30% maybe 40% at best.

at 4.8ghz 1.25 or less vcore, temps drop like 15c... You're losing 4% performance against an almost 20% improvement in temps (in fact if you run 1440p or 4k like me you don't lose performance). And not even mentioning power draw.

I've decided to run 1.23v at 4.8ghz. Max temp on AVX stress testing is now a measly 73c, games are not even worth mentioning it's in the low 50s.

Most of us are pushed by our overclocking nature to actually run 5.0ghz, and the number is a golden one, but even from an overclocker's point of view I can't justify a cpu that runs over 80c especially 85c at full load . It just doesn't make me happy

Stock is also just fine I bet, but I like to keep my cores all running the same. This chip is just too hot for me.

I myself run at 4.8GHz with nice stable clocks and good temps on air. I don't need anything more as I can see during gaming that I barely use 50% cpu utilization on most demanding scenery and I average at about 30-40% during gameplay so I myself am happy to run at 4.8GHz and keep them geyser element on the cooler side until winter opens the door then I can do with a heater at my feet. 😊😬

 

[HNO3ZA]

Okay, not to burst your bubble or anything, just to give you an IDEA....I also run a i9 9900K at stock with an RTX 2080 but NOT the OC model, but did a bit of OC myself to the card only. CPU stays stock and my Timespy score is 11870. Also, I only have 16GB 2666 memory.
You have a higher CPU score than I have, but lower Graphics score so I am sure you can do better on that score bud....just my 2 cents.

So I played around with the GPU a bit. Not my best score with this GPU, but there are some improvements...

Result not found

www.3dmark.com

 

[el_guapo3291]

For those of you complaining about temps I have a solution which has worked on 3 friends 9900k's... Disclaimer I know my cpu is above average and that with two 480mm rads your first thought is going be it's the overkill cooling (you would be technically wrong in assuming larger rad surface area = automatically cooler temps by default but that's a topic for another day).

So basically what I noticed with my own cpu is I can do 5Ghz all core no AVX offset @ 1.258v (Solid 1 hour plus on Aida64 FPU, Prime95, Realbench, XTU etc.) Max Temps between 70-75c - Do note I live on the KZN North Coast with very high ambient temps year round, my room is sometimes 35c air temp.

Then I tried 5.1 Ghz with no AVX offset and needed 1.32v with temps now averaging closer to the low 80s - Note the large jump in voltage for just 100 mhz.

Then I tried 5.2 Ghz with no AVX offset and needed 1.38v with temps now averaging closer to the high 80s - Note the large jump in voltage for just 100 mhz again.

Then I tried 5.3 Ghz with no AVX offset and needed 1.42v with temps now averaging closer to the mid 90s - Note the large jump in voltage for just 100 mhz again, this test I considered a fail due to temps.

So I now have a rough idea of what the thermal limits were so experimented with using an avx offset.

First test was 5.1 Ghz @1.26v and 100mhz offset - Passed easy with good temps ( 70-75c).

Second test 5.2 Ghz @ 1.35v and 200mhz offset - Passed easy when no AVX load but AVX load temps too high @80+

So now we get to the main section while my frequency between both those runs was 5Ghz for AVX the second run was getting a lot hotter because it was now using a higher voltage than needed for 5Ghz AVX, Once I noticed this I bumped down AVX to 4.9ghz and problem solved. Just by reducing the AVX speed by 100mhz to 4.9ghz I could now run 5.1, 5.2 or even 5.3ghz non-avx all year round.

This is perhaps a slight compromise but you have to take into consideration that South Africa is hot, unless you can afford 24/7 AC running its likely you have high ambient temps for much of the year, during winter I can get away with 5.2 Ghz no offset but during summer I have to turn down the AVX a bit. So to summarise don't chase a 1:1 ratio with AVX, Intel included the offset in Skylake and upward for a reason, sacrificing as little 100mhz AVX can make a difference to your max oc and significant impact on temps.


Other things that influence temps:

VCCSA and VCCIO - Don't run these on Auto I find the bios pumps way too much voltage into them aim for 1-1.2v max (Yes I know these are for RAM, no shit sherlock but they do have a slight influence on cpu temps)

LLC - Set only enough to cover vdroop (on asus bios that is usually Level 5-6)

IGPU - On models that have an igpu disable if not using it

Choose your stress tests carefully - Some tests are simply unrealistic workloads that draw way more than any real workload will ever get close to, I'm not going to preach about this because I'm sure some condescending "expert" will have his own choices based on subjective and anecdotal data (if any at all) so instead search on facebook for "GPU Mafia - PC Enthusiast Group" in there you will find a pinned temperature guide for intel that goes in-depth and has lots of useful info especially about which tests to use and how to use them.

 

[baasgene]

So I played around with the GPU a bit. Not my best score with this GPU, but there are some improvements...

Result not found

www.3dmark.com

Mmmm honestly those scores are lacking for your setup? On 5ghz I got a cpu score of 12000-12100. And my 1080 Ti gets 10550 graphics with only a slight overclock.

On 4.8ghz I get 11500+

Your 2080 should be reaching over 11000 with those clocks.

It’s hard to pin point the reason for these things, but there’s definitely something holding you back here.

4.8ghz - I scored 10 685 in Time Spy
5.0ghz - I scored 10 684 in Time Spy

On the 5ghz run I actually dropped the 1080 ti core with 20mhz just to see the difference so that's why it's less.

 

[BLUE-ANGEL]

Are you both running AIO coolers on your CPU's?
I have very shitty CPU score because I have a high ambient room temp, if I had to guess I think average about 30degrees and I run an Air cooler, but my GPU score is higher than both or yours. I will get a AIO in the near future.

@HNO3 attached is my RTX2080 score for your interest.
@baasgene you will see my CPU score is much lower, BUT your 1080Ti should be running more or less same GPU score than a RTX2080.

I am just waiting on winter to come so I can see how I can push the score a bit more to get past the 12000 score bar but for now I am happy at 4.8Ghz eventhough I know I can get a higher CPU score even on 4.8GHz. I know for a fact I am held back by the thermal side of my cooling with my ambient being so high relating to bad cooling at high stress on the CPU.

 

[BLUE-ANGEL]

Here is my Firestrike score just to give you an idea and to see what you get @HNO3 ....I know I can do better once I have better cooling.

 

[Joe Banker]

@BLUE-ANGEL the 1080ti's no longer run with the 2080's due to new drivers. Maybe at launch yes. 11k is about it for mine and my card is pretty srong looking at other scores around.

 

[baasgene]

Are you both running AIO coolers on your CPU's?
I have very shitty CPU score because I have a high ambient room temp, if I had to guess I think average about 30degrees and I run an Air cooler, but my GPU score is higher than both or yours. I will get a AIO in the near future.

@HNO3 attached is my RTX2080 score for your interest.
@baasgene you will see my CPU score is much lower, BUT your 1080Ti should be running more or less same GPU score than a RTX2080.

I am just waiting on winter to come so I can see how I can push the score a bit more to get past the 12000 score bar but for now I am happy at 4.8Ghz eventhough I know I can get a higher CPU score even on 4.8GHz. I know for a fact I am held back by the thermal side of my cooling with my ambient being so high relating to bad cooling at high stress on the CPU.

View attachment 59645

A 1080 Ti is 10% slower than a 2080 overall stock vs stock and OC vs OC, if I overclock it to the max I can reach close to 11000. But I don't want to push it that far.

Firestrike 4.8ghz - I scored 25 310 in Fire Strike

This was my 1080 Ti maxed out basically on an older system : I scored 10 102 in Time Spy

That was the limit of my core, and if I pushed the memory further I got some artifacts.

 

[BLUE-ANGEL]

To be honest, I used 1080Ti for my benchmarking comparison to see how close I can get to the 1080Ti manual overclock scores and sadly but true, I have found many 1080Ti's that compare equal to my score or deviate slightly with 200 points, but generaly I found them to be very much same bracket in score.

See example below as this is 1 of the few I use as target for graphics score. Whether the guy is running waterblock or not I can't recall exactly but I do believe he is not I might be wrong, but still the score is 30 000+ for a 1080Ti.

 

[baasgene]

To be honest, I used 1080Ti for my benchmarking comparison to see how close I can get to the 1080Ti manual overclock scores and sadly but true, I have found many 1080Ti's that compare equal to my score or deviate slightly with 200 points, but generaly I found them to be very much same bracket in score.

See example below as this is 1 of the few I use as target for graphics score. Whether the guy is running waterblock or not I can't recall exactly but I do believe he is not I might be wrong, but still the score is 30 000+ for a 1080Ti.

View attachment 59647

Firestrike shouldn't even be used anymore, only focus on Time Spy. It is very very accurate, go and check Red Dead Redemption 2 for example. Go and see how the 2080 and 2080 Ti is running away from the 1080 Ti. Newer games, especially DX12 is a lot better on RTX 2xxx.

Even my 1080 Ti gets 30600 on Fire Strike with a measly overclock, check the link I posted above, here it is again : I scored 25 310 in Fire Strike

 

[HNO3ZA]

A 1080 Ti is 10% slower than a 2080 overall stock vs stock and OC vs OC, if I overclock it to the max I can reach close to 11000. But I don't want to push it that far.

Firestrike 4.8ghz - I scored 25 310 in Fire Strike

This was my 1080 Ti maxed out basically on an older system : I scored 10 102 in Time Spy

That was the limit of my core, and if I pushed the memory further I got some artifacts.

Would you say my score of 10253 is on par? AS I think I got a better score on the graphics side with my 8700k, byt that might have been on a clean install as well.... Not sure if that would make such a big difference....

9900k + 2080 Score

Result not found

www.3dmark.com

8700k + 2080 Score

Result not found

www.3dmark.com

 

[Joe Banker]

Also, that is 2017

 

[baasgene]

Would you say my score of 10253 is on par? AS I think I got a better score on the graphics side with my 8700k, byt that might have been on a clean install as well.... Not sure if that would make such a big difference....

9900k + 2080 Score

Result not found

www.3dmark.com

8700k + 2080 Score

Result not found

www.3dmark.com

That's a massive difference. As I said something on your new system is holding you back. Redo Windows and have another crack at it, something ain't right.