CPU/GPU usage not maxed FPS limit not reached

[Exeon]

Specs:

 

I5-9600k

RX 6900XT

DDR4 16GB RAM 3000mhz

EVGA G2 850W PSU

 

I'm currently playing on a 2.5K resolution (2K ultrawide)

I've set a FPS limit to 144 and noticed the following.

In several games I'll get somewhere between 100-120FPS at max settings (which is more than fine by me on this resolution)

However I've noticed that neither my GPU nor my CPU are reaching usage limits, both sit around 80% usage, and the GPU doesn't seem to go above 250W usage either.

 

I'm probably going to upgrade my CPU regardless (probably a 9900K) However I'm wondering if I'm doing something wrong, I'd expect to either reach 144FPS or get 99% usage out of either the GPU or CPU, depending on what bottlenecks me.

Or can the CPU bottleneck me without it reaching 100% usage?

I doubt the RAM is the limiting factor here, though it could be faster (bought it around the time DDR4 released) I can't believe it could impact me as much.

 

None of my components are overclocked, other than boost clocks.

[SavageNeo]

2 minutes ago, Exeon said:

CPU bottleneck me without it reaching 100% usage?

Yes. If single cores hit 100% it is a bottleneck. total cpu usage does not mean that much. look at single core usages.

 

 

[Somerandomtechyboi]

7 minutes ago, Exeon said:

None of my components are overclocked, other than boost clocks.

You are really gonna want to oc that 9600k

Set to 4.8 or 4.9 at 1.3v or lower for efficiency and if your cooler can handle it mess with dynamic oc so you can get really high single or 2 core boost clocks

 

Though keep in mind the safe volt for these cpus is around 1.5v and dont go over 95c

[LIGISTX]

28 minutes ago, Somerandomtechyboi said:

You are really gonna want to oc that 9600k

Set to 4.8 or 4.9 at 1.3v or lower for efficiency and if your cooler can handle it mess with dynamic oc so you can get really high single or 2 core boost clocks

 

Though keep in mind the safe volt for these cpus is around 1.5v and dont go over 95c

1.5v is very, very high... I wouldn't consider this safe for 24/7 use. I personally don't go over 1.35 and im on customer water. Even if 1.5v is safe, it would be extremely hard to cool that. My 9900k @ 1.32 VRVOUT and 5 GHz would hit 99c almost instantly on custom water - it was a pretty bad chip thermally, but, still. 1.5v is not something most would consider a safe voltage at all.

[Somerandomtechyboi]

6 minutes ago, LIGISTX said:

1.5v is not something most would consider a safe voltage at all.

And i consider 1.7v safe on my freaking 45nm core cpus, after all they can run 2v and even boot to windows at 2v without dying and i know that 1.9v~ is a death zone voltage aka put a load on the cpu and it dies

 

Idk im an oc nut that will push my hardware to its limits cause haha 6.9ghz cpu go brrrrrrr

 

Btw on newer cpus intel spec is 1.52v so 1.5v is prob a safe voltage

Personally id go to 1.6v cause you can go over the safe volt by around 0.1v in most instances

 

 

Then again i work with bulletproof hardware and ive abused it quite severely in some cases including running 2v though my cpus xD

[Exeon]

1 hour ago, SavageNeo said:

Yes. If single cores hit 100% it is a bottleneck. total cpu usage does not mean that much. look at single core usages.

 

Good point, I just checked, seems to be a CPU bottleneck, it does switch between the GPU and the CPU, but I'm definitly CPU bottlenecked here.

 

1 hour ago, Somerandomtechyboi said:

You are really gonna want to oc that 9600k

Set to 4.8 or 4.9 at 1.3v or lower for efficiency and if your cooler can handle it mess with dynamic oc so you can get really high single or 2 core boost clocks

 

Though keep in mind the safe volt for these cpus is around 1.5v and dont go over 95c

 

I actually OC'd my old 6700K to 4.5Ghz on a NH-D15

It ended up killing the motherboard prematurely (an Asus Maximus ROG)

I wasn't amused that I had to ditch my CPU for a 9600K after 3 years of use and I quit OCing after that.

 

1 hour ago, LIGISTX said:

1.5v is very, very high... I wouldn't consider this safe for 24/7 use. I personally don't go over 1.35 and im on customer water. Even if 1.5v is safe, it would be extremely hard to cool that. My 9900k @ 1.32 VRVOUT and 5 GHz would hit 99c almost instantly on custom water - it was a pretty bad chip thermally, but, still. 1.5v is not something most would consider a safe voltage at all.

 

Any reason why you switched from the 9900K to the 10700K? I'm considering stepping up to the 9900K, I do feel like it's tuned quite highly and it runs hot, but so are it's boost clocks though.

[LIGISTX]

4 hours ago, Somerandomtechyboi said:

And i consider 1.7v safe on my freaking 45nm core cpus, after all they can run 2v and even boot to windows at 2v without dying and i know that 1.9v~ is a death zone voltage aka put a load on the cpu and it dies

 

Idk im an oc nut that will push my hardware to its limits cause haha 6.9ghz cpu go brrrrrrr

 

Btw on newer cpus intel spec is 1.52v so 1.5v is prob a safe voltage

Personally id go to 1.6v cause you can go over the safe volt by around 0.1v in most instances

 

 

Then again i work with bulletproof hardware and ive abused it quite severely in some cases including running 2v though my cpus xD

I ran 1.18 or 1.28v on my i7 920 if memory serves…. At 3.82ghz. It was a great chip voltage wise… but 1.7 is insane. And 1.5 for 9000 series isn’t safe for 24/7, and even if it was, your not keeping it cool.

 

And if you needed 1.5v for a standard 5ghz OC, you are pushing your chip way beyond what it wants. My 10700k is 5.1 ghz with VRVOUT of 1.35 I think, somewhere around there. And with my custom loop in a stress test it’s touching high 80’s. When I pushed for 5.2 ghz with more voltages it was hitting mid 90’s and not stable. So even trying to validate stability at 1.5v would be impossible, even on a 9600k I’d imagine.

 

3 hours ago, Exeon said:

I actually OC'd my old 6700K to 4.5Ghz on a NH-D15

It ended up killing the motherboard prematurely (an Asus Maximus ROG)

I wasn't amused that I had to ditch my CPU for a 9600K after 3 years of use and I quit OCing after that.

 

 

Any reason why you switched from the 9900K to the 10700K? I'm considering stepping up to the 9900K, I do feel like it's tuned quite highly and it runs hot, but so are it's boost clocks though.

You almost can’t kill a mobo from overclocking…. Unless it’s a horrible VRM. Most Asus ROG boards are pretty decent though, so I doubt it was the OC, or it was a much to high of voltage OC…. I currently have a 6700k @4.6 ghz on an asus board, with a hyper 212 evo cooling it just fine.

 

I switched to 10700k cuz a buddy of mine upgraded to a 10900k, and I was able to move over to the 10700k for effectively free after I sold my 9900k and mobo. An extra .1 GHz, and extra multiplier of ring ratio, and ~10-15c cooler temps to boot! Like I said, my 9900k was a very hot running chip. Performance wise, there is no difference. Even gaining the extra 100mhz, it doesn’t really mean much for gaming since I am 3440x1440…. I need more GPU horsepower if anything. I just OC cuz it’s fun. A stock 10700k is plenty fast, so is a 9900k. 

[Somerandomtechyboi]

2 hours ago, LIGISTX said:

 And 1.5 for 9000 series isn’t safe for 24/7, and even if it was, your not keeping it cool.

And if you needed 1.5v for a standard 5ghz OC, you are pushing your chip way beyond what it wants.

Build a custom loop and stick a waterchiller in it and set to 20c or if you wanna insulate the socket you can prob acheive subzero temps but thats kinda pointless unless you are an oc nut like me and have a powerful water chiller

 

Hmm, i guess 1.5v for 5.5ghz then?

Intel spec says 5.2 requires around 1.5v but im pretty sure thats just needless overvolting, no wonder why these things run hot when not tuned for a proper overclock or undervolt

[Somerandomtechyboi]

And by looking at some hwbot of the nehalem 45nm cpus (x34xx xeon) seems like the nehalems require as much voltage as the core cpus

 

So i guess the process node defines the safe volt?

 

Well then i guess ill have the same standard on all 45nm cpu +-0.1v

 

6 hours ago, Exeon said:

I actually OC'd my old 6700K to 4.5Ghz on a NH-D15

It ended up killing the motherboard prematurely (an Asus Maximus ROG)

I wasn't amused that I had to ditch my CPU for a 9600K after 3 years of use and I quit OCing after that.

Yea its very hard to kill a board unless you are running your cpu at 7 or 8ghz or something

 

So prob just became faulty for whatever reason, i usually dont give up on my hardware that quick, ill try cleaning with 99% isopropyl alcohol, then ill start dunking the socket in isopropyl, then ill just dunk the entire thing in a tub of isopropyl alcohol and leave it for 24h

 

Usually this works for all the issues related to filth accumulating somewhere and causing the board to malfunction

[LIGISTX]

4 hours ago, Somerandomtechyboi said:

Build a custom loop and stick a waterchiller in it and set to 20c or if you wanna insulate the socket you can prob acheive subzero temps but thats kinda pointless unless you are an oc nut like me and have a powerful water chiller

 

Hmm, i guess 1.5v for 5.5ghz then?

Intel spec says 5.2 requires around 1.5v but im pretty sure thats just needless overvolting, no wonder why these things run hot when not tuned for a proper overclock or undervolt

So, yes.... Its not that its not possible to do this, but shouldn't advice "random" internet folks to push voltages to 1.5v and call it a day. Can it be done with sub-ambient, sure... but that isn't exactly safe, nor can any normal cooling solution cool that. I have a custom loop with plenty of rad space, IHS's just can't get that much concentrated heat through them from the actual die itself, so while I can effectively cool things during game load with very low fan RPM, even with 100% fan speed the core temps still get pretty toasty when under synthetic load. Point of this is........ unless your running exotic cooling beyond a custom loop with ambient air temp cooling, you can't keep a 1.5v OC cool, and its not "safe" due to this (not to mention 1.5v will definitely lead to premature chip death - that really is a lot of volts). Again, not that you can't and you personally shouldn't do this, but for most folks online, 1.4v is the safe upper limit, with 1.35 being much more reasonable.

[Somerandomtechyboi]

31 minutes ago, LIGISTX said:

 (not to mention 1.5v will definitely lead to premature chip death - that really is a lot of volts).

*laughs with 2v on my 45nm cpu

 

It will tell you whether or not the volts are safe, when i xoc my p4 the board did not like going over 1.9v so i had to disable that and look what happened, the p4 is now degraded

 

Seems like some boards have an overvoltage or overcurrent protection

 

One way or another it will tell you if it doesnt like the voltages, heck even if you remove these safeguards by the board, it will still tell you if it doesnt like the voltages by either degrading or outright dying

[LIGISTX]

30 minutes ago, Somerandomtechyboi said:

*laughs with 2v on my 45nm cpu

 

It will tell you whether or not the volts are safe, when i xoc my p4 the board did not like going over 1.9v so i had to disable that and look what happened, the p4 is now degraded

 

Seems like some boards have an overvoltage or overcurrent protection

 

One way or another it will tell you if it doesnt like the voltages, heck even if you remove these safeguards by the board, it will still tell you if it doesnt like the voltages by either degrading or outright dying

That isn't true.... the vcore input will just yellow or red or purple, it won't really tell you much beyond that. So if someone unwillingly just inputs a voltage they read on a forum, the board will do it just fine. Will it be way to hot under load, likely yes, but most people don't test for stability correctly and would never even know, they would just see high 70's or 80's in games and potentially not second guess it - all the while they are running way too many volts through the chip and degrading it for no reason at all.

 

What I am saying is, as more experienced OCers, its our responsibility to not throw around votlages and settings that don't make sense for normal people. We have experience to understand what settings we are changing, but LOTS of folks these days just go #nofearfullsend into the BIOS and don't understand what they are doing and can really cause issues for themselves this way. We need to be cognizant of the audience, thats all

[Somerandomtechyboi]

56 minutes ago, LIGISTX said:

 #nofearfullsend into the BIOS and don't understand what they are doing and can really cause issues for themselves this way. We need to be cognizant of the audience, thats all

ppl will be too afraid to even set 1.5v in the bios even if intel spec says its fine, plus the cooling issues xD

 

Like come on lets be real here no one is stupid enough to just set it at 1.5v right off the bat without searching for other ppl oc results, 5.3 is supposedly acheivable with 1.4v on these newer cpus but i havent seen anyone use 1.5v other than linus testing out peltier cooling and got to 5.7ghz single core

 

Even if they set it at 1.5v 5.5ghz or something theyll prob do a benchmark or stability test and notice that their temps are wayyy to high and even throttling so theyll back down the clocks and volts

 

If they have exotic cooling then yea 1.5v is totally safe and actually doable without setting the cpu on fire

 

 

And yes i have no fear most of the time when ocing my cpus cause another 45nm pentium e5x00 cpu is like 1$ at most

 

So if one dies ill just buy 3 more and also kill them eventually xD

[LIGISTX]

18 hours ago, Somerandomtechyboi said:

ppl will be too afraid to even set 1.5v in the bios even if intel spec says its fine, plus the cooling issues xD

 

Like come on lets be real here no one is stupid enough to just set it at 1.5v right off the bat without searching for other ppl oc results

That isn't true... I have seen plenty of examples where people just see someone who seems knowledgable, and just tries those settings and then comes asking in a forum post why things are not working right... The mantle of responsibility (<- halo quote) falls on us as the experienced OCers. Too many folks these days just apply settings without understanding what they are doing, hell, every few forum posts on this forum alone are related to that.

[Somerandomtechyboi]

47 minutes ago, LIGISTX said:

That isn't true... I have seen plenty of examples where people just see someone who seems knowledgable, and just tries those settings and then comes asking in a forum post why things are not working right... The mantle of responsibility (<- halo quote) falls on us as the experienced OCers. Too many folks these days just apply settings without understanding what they are doing, hell, every few forum posts on this forum alone are related to that.

ive seen many more examples of idiots buying incredibly overpriced rgb ram such as rgb pro or trident z rather than idiots that go copying settings

 

I havent really seen that many ppl that do ocing, most of the newbies never oc anyways so i never really encounter that many ocers on this forum

 

Heck its rarer than the occasional rainbow idiot that just uses a literal ass cpu and gpu for their budget and focus it all on rgb rainbow puke, aka the occasional ricer pc

[LIGISTX]

1 hour ago, Somerandomtechyboi said:

ive seen many more examples of idiots buying incredibly overpriced rgb ram such as rgb pro or trident z rather than idiots that go copying settings

 

I havent really seen that many ppl that do ocing, most of the newbies never oc anyways so i never really encounter that many ocers on this forum

 

Heck its rarer than the occasional rainbow idiot that just uses a literal ass cpu and gpu for their budget and focus it all on rgb rainbow puke, aka the occasional ricer pc

I answer multiple questions per week about why peoples RAM OC didn't work right.... cuz they just see a YouTube video and away they go. Just saying, don't post settings that don't make sense just for the sake of it. Give context if nothing else... Its just the right thing to do - teaching is important.

[Somerandomtechyboi]

1 hour ago, LIGISTX said:

I answer multiple questions per week about why peoples RAM OC didn't work right.... cuz they just see a YouTube video and away they go. Just saying, don't post settings that don't make sense just for the sake of it. Give context if nothing else... Its just the right thing to do - teaching is important.

Even if its just a simple setting that requires 1 or 2 things to be set in the bios it might still cause problems?

 

not really sure on how to teach, i just learned how to oc my system by just looking up what the bios settings did and away i go, though yes theres a ton of trial and error involved

[LIGISTX]

1 hour ago, Somerandomtechyboi said:

Even if its just a simple setting that requires 1 or 2 things to be set in the bios it might still cause problems?

 

Yes. Setting 1.5v on a CPU is 1 setting, 1 setting that could fry a chip at worst, or at best seriously degrade it’s life. Without context of what 1.5v actually means, this isn’t helpful to anyone, it’s just one more possible way for someone to damage hardware, or get stuck in a boot loop and get frustrated and make a bad choice and change more settings incorrectly. Obviously if that happened, they should reset CMOS…. You and I and other experienced folks know this and can easily work through the issue. But others may not, they may be confused and upset and make another choice to change more settings and get themselves into worse shape.

 

Thats all my point is. Just have to approach the problems from a state of someone with less experience, and give them the info surrounding your advice. 1.5v is great advice for someone on exotic cooling, not for someone asking if their 9600k is bottlenecking their GPU and considering upgrading to a 9900k. If anything, start at 1.30v, see what the chip does and how it responds and what the silicon quality looks like, and then dial it in from there and explain how to go about dialing it in; watch temps, test for ~15 minutes, if it passes up the speed by 100 MHz, rinse and repeats till it crashes, then either you the volts if cooling allows or drop the clocks, rinse repeat till your at a speed and temp you like and then dial in the volts via 8 hour heavy CPU tests like AIDA FPU and Asus Realbench. If it passes 8 hours of both of those, your pretty dang stable (lots of folks just run blender to test stability, so this is a good place to explain why blender doesn’t actually test for stability and a chip that can “pass some blender rendering” isn’t actually necessarily stable, and can end up crashing in games or other programs. Then if you want to explain more, explain why VRM transient response is very important and why setting LLC is vital to stability and actually reduced temps… lots of potential to teach, just takes some time

[Somerandomtechyboi]

47 minutes ago, LIGISTX said:

Yes. Setting 1.5v on a CPU is 1 setting, 1 setting that could fry a chip at worst, or at best seriously degrade it’s life.

Nah just degrades in like 3-6 months

I bet running my cpu at 1.7v would yeild the same result but itll prob run 1.5v for a long time before it actually degrades

 

1.5v is intel spec for newer cpus, i bet you are talking all core, in which case thats ultra dumb, youd have to dynamic oc to reach 1.5v without frying your cooler or something, 1.5v all core is only dangerous cause no one has enough cooling for it, otherwise its fine, if you are talking death zone volt then maybe around 1.7-1.8v for these 14nm cpus

 

My hardware is bulletproof so short runs to 2v is ok, for the 14nm cpus thats literal insta death

 

Only reason setting 1.5v would fry your chip is if your board set a ridicolous volt like 1.8v

 

Otherwise 1.5v is just gonna have accelerated degradation if you are running high temps (80c+) or running that on all cores, exotic coolers will get away with it for much longer since they can prob run their cpu at subzero or near ambient so its more like 1.55 or 1.6v is actually when it starts to get dangerous cause degradation is even faster here

 

The degradation is prob gonna increase exponentialy at a certain volt and up so unless you surpass that volt you are safe

 

Cause if i can run my cpu at 1.7v but the death zone is around 1.9v then degradation increases exponentially with volt and heat i guess

¯\_ (ツ) _/¯ 

[Somerandomtechyboi]

1 hour ago, LIGISTX said:

setting LLC is vital to stability

Are you talking about setting it to a lower value?

As ive found that llc when i set it to my lowest stable volt it actually crashes so i prefer just yeeting llc and let vdroop do its thing and see where it lands

 

But then again i run bulletproof decade old hardware so things might be different for newer platforms

[LIGISTX]

12 hours ago, Somerandomtechyboi said:

1.5v is intel spec for newer cpus, i bet you are talking all core, in which case thats ultra dumb, youd have to dynamic oc to reach 1.5v without frying your cooler or something, 1.5v all core is only dangerous cause no one has enough cooling for it,

 

The degradation is prob gonna increase exponentialy at a certain volt and up so unless you surpass that volt you are safe

Exactly why explanation is important… most folks talk about all core vcore, not per core. This is why I was saying 1.5v is not possible without exotic cooling, and most folks who would try this would end up with problems, because 1.5v all core is a lot. 

 

12 hours ago, Somerandomtechyboi said:

Are you talking about setting it to a lower value?

As ive found that llc when i set it to my lowest stable volt it actually crashes so i prefer just yeeting llc and let vdroop do its thing and see where it lands

 

But then again i run bulletproof decade old hardware so things might be different for newer platforms

You want vdroop. Vdroop helps protect against transient spikes which are what really hurt a CPU, and cause excess heat. Again, this is above a novice level OC, but my point is it’s a good place for explanation vs just saying “set 1.5v and go for gold”. That’s all I’m getting at. 

[Heliian]

Use Intel xtu to do a basic multiplier OC, don't touch the voltage and try an increase of  0.1 ghz at a time.  Xtu has some good monitoring and recording of cpu telemetry also so you can monitor temps with it running in the background. 

 

Edit: your GPU is the more powerful component and is being "bottlenecked" by the cpu.  Mmmm beer.