Why does my new system suck compared to my old one?

[Mark Kaine]

35 minutes ago, porina said:

If I remember correctly you run very particular versions and settings for things. I'm amazed you don't have more problems. I trust nvidia to know what they're doing more than you know what you're doing.

 

It does have problems but more in not recognising some games for whatever reason. Presentmon is probably best in class at the moment, although RTSS remains popular. RTSS is based on presentmon anyway.

what's wrong with hwinfo64? 

 

it shows latency for each component....

 

 

35 minutes ago, porina said:

. I trust nvidia to know what they're doing more than you know what you're doing.

yeah, well... i don't.... the overlay specifically is terrible and doesn't do anything that other monitoring software doesn't do better lol.

 

 

ps: i dont have specific settings,  im on win 1809 (you know the version that's used for enterprise editions) and i don't really have many issues,  probably less than most for sure ( that's because im on the last known good windows version,  indeed )

 

so what's your proposal to fix OPs issue then? 

 

im still fairly convinced mine would fix it, we'll probably never know tho lol 

 

 

[porina]

1 hour ago, Mark Kaine said:

what's wrong with hwinfo64? 

 

it shows latency for each component....

Maybe I missed it, I don't recall seeing that. Can you clarify?

 

I did just have another look. I see hwinfo64 now has integration for presentmon. Feels like everyone uses presentmon as a basis.

 

1 hour ago, Mark Kaine said:

so what's your proposal to fix OPs issue then? 

Pending clarification on if it is a single game or every game. If single game it is more likely a game setting. 

 

1 hour ago, Mark Kaine said:

im still fairly convinced mine would fix it, we'll probably never know tho lol 

Your fix is basically reinstall driver. It's a shot in the dark.

[HayuPhukmiWife]

On 1/7/2024 at 11:22 PM, Mark Kaine said:

no, this actually helps, these chips are a mega fail design afterall,  but its on an application basis afaik, so good luck micromanaging this... would be way too annoying for me, and why i would never buy a fail design cpu.

 

but, besides all the "monitoring" do you actually experience real issues when using this system..?

 you didn't answer that - you don't have to, i guess.

 

 

I do notice it. IT's a quick stutter. not much, and  I can hear it in my GPU as it overcomes it or whatever. I only notice it when it hits like 40ms. its a quick hitch like a frame drop would feel, except I don't see the frame drop. For example. playing fortnite with my kid tonight i was hitting 40ms and noticed it as a quick hitch and then lets go. split second. but noticeable and annoying AF.  

 

I agree disabling and re-enablinging ecores should not be the solution. but its worth trying out to see if it solves the issue....  

On 1/8/2024 at 12:59 AM, problemsolver said:

My understanding was you had some minor micro stutter (that wasn't there on your old system) & that's what we're actually trying to solve. Or maybe I misunderstood? 

Yessir, ad noted above and before. you have the gist of it. I haven't had a chance to do anything about it. It wasn't doing it last night. I noticed I had armoury crate performance lighting mode turned on, and tried disabling it. didn't make a difference.

On 1/8/2024 at 1:42 AM, porina said:

To be clear, as you getting low or high latency in other games? If it is high in all games, it is still open as to the cause. If it is only high in one game, then it is more likely a setting for the one game.

It seems to be game specific, but its weird. in some games I would expect it more (graphically intense games) but it doesn't seem to be the case. example above... forking fortnite..... SHOULD NOT stress my system. yet i was hitting 40ms today. used to only hit 10 to my knowledge.... Same settings. aways maxed.

 

On 1/8/2024 at 7:30 AM, Mark Kaine said:

ok... but... why trust this "overlay" (it's extremely buggy in my experience, it also makes my pc laggy as hell... ironically)

 

that's why i keep asking if they actually experience issues, and since i never get an answer i will draw my own conclusions. 

 

so how to fix this?

 

ddu gpu driver, *do never use nvidia overlay again* use hwinfo64 instead.  Should be simple enough. 

I've tried clean installs with the gpu driver through nvidia but not ddu. how can i monitor latency in hwinfo? What subheading is it in. Sorry been a busy work week. haven't had time to delve into this. 

On 1/8/2024 at 7:35 AM, porina said:

If I remember correctly you run very particular versions and settings for things. I'm amazed you don't have more problems. I trust nvidia to know what they're doing more than you know what you're doing.

 

It does have problems but more in not recognising some games for whatever reason. Presentmon is probably best in class at the moment, although RTSS remains popular. RTSS is based on presentmon anyway.

presentmon latencymon same thing? I was gonna explore latnecymon, but then thought things got better. Seems as though they havent tho...

[Mark Kaine]

12 minutes ago, HayuPhukmiWife said:

do notice it. IT's a quick stutter. not much, and  I can hear it in my GPU as it overcomes it or whatever. I only notice it when it hits like 40ms. its a quick hitch like a frame drop would feel, except I don't see the frame drop. For example. playing fortnite with my kid tonight i was hitting 40ms and noticed it as a quick hitch and then lets go. split second. but noticeable and annoying AF.  

 

I agree disabling and re-enablinging ecores should not be the solution. but its worth trying out to see if it solves the issue....  

ok, now you explained the actual issue,  that's good - but yeah, that's tricky. now i wonder if it's just fortnite or other games too, with the *same* issue? 

 

and yes, i agree disabling the ecores is definitely worth a try!

[HayuPhukmiWife]

16 minutes ago, Mark Kaine said:

ok, now you explained the actual issue,  that's good - but yeah, that's tricky. now i wonder if it's just fortnite or other games too, with the *same* issue? 

 

and yes, i agree disabling the ecores is definitely worth a try!

I was pretty sure i mentioned that in the first few replies, but maybe i didn't. there was a bit of a comment overload there right quick. I will look into this when i have a few minutes and try this out to see if It makes a deference. If it does, it might not be a path I'm willing to go down as a permanent solution. What do the ecores even do? hlep with tasking? would i see the benefit of them in most standard office (light office - no rendering) use or in gaming at max settings? i went overkill with this i9 just because. I probably should have stuck with the i7. it never caused me issues.

 

Edit. So far fortnite, Hogwarts legacy, and cyberpunk all see 30-40ms. Starfield does not at +/- 10. updating my bios tonight or tomorrow from 15 to 18 to see if it helps... 

[HayuPhukmiWife]

Anyone think this could be due to my setting intel power limits to stop my CPU from being a 103 degree fireball?

I was seeing throttling when dialing the system in and disabled it to set intel limits vs mobo limits. I worry with all the constant throttling and temps seeing 101, 102, even 103 at times under load. So i set the intel limits and see 88-90 usually - 93 max. which feels WAYYYY safer and below tjmax.  

 

Am I limping the CPU potential? I'm not running any oc settings (click OC in bios or custom settings) could this have any merit? 360 AIO should be enough to keep it cool, but I thought nonstop throttling was a poor choice, and i was seeing it on specific cores and afraid of bending. That said I bought a contact frame from thermalright when diagnosing the throttling issues, but after I found out i could disable mobo settings I never installed it. Fear of bending is REAL... Thoughts?

[Mark Kaine]

On 1/8/2024 at 5:38 PM, porina said:

Your fix is basically reinstall driver. It's a shot in the dark.

yes, and do not [DO NOT] use the Nvidia overlay....strategic testing instead of repeating the same thing over and over.

not the first person having potential issues with this "overlay", and hardly the last.

 

6 hours ago, HayuPhukmiWife said:

Edit. So far fortnite, Hogwarts legacy, and cyberpunk all see 30-40ms. Starfield does not at +/- 10. updating my bios tonight or tomorrow from 15 to 18 to see if it helps... 

ok, but that implies more game specific issues rather than hw/sw

 

again,  please ddu and use hwinfo64 ,  or even MSI afterburner, instead of "Nvidia overlay"

 

6 hours ago, HayuPhukmiWife said:

updating my bios tonight or tomorrow from 15 to 18 to see if it helps

well, that's one of the first things you should do ,yes.

 

 

5 hours ago, HayuPhukmiWife said:

Anyone think this could be due to my setting intel power limits to stop my CPU from being a 103 degree fireball?

I was seeing throttling when dialing the system in and disabled it to set intel limits vs mobo limits. I worry with all the constant throttling and temps seeing 101, 102, even 103 at times under load. So i set the intel limits and see 88-90 usually - 93 max. which feels WAYYYY safer and below tjmax.  

 

Am I limping the CPU potential? I'm not running any oc settings (click OC in bios or custom settings) could this have any merit? 360 AIO should be enough to keep it cool, but I thought nonstop throttling was a poor choice, and i was seeing it on specific cores and afraid of bending. That said I bought a contact frame from thermalright when diagnosing the throttling issues, but after I found out i could disable mobo settings I never installed it. Fear of bending is REAL... Thoughts?

eh, umm, ofc all of that can cause issues in all or in some applications,  yes. 

 

not sure why it's overheating,  but generally you should run things at stock, especially while troubleshooting  - i understand that might not be possible with your cooling situation (new hardware would probably fix that) 

[porina]

1 hour ago, Mark Kaine said:

yes, and do not [DO NOT] use the Nvidia overlay....strategic testing instead of repeating the same thing over and over.

not the first person having potential issues with this "overlay", and hardly the last.

What issues? It is a reporting tool like any other. It does not cause problems. 

 

1 hour ago, Mark Kaine said:

again,  please ddu and use hwinfo64 ,  or even MSI afterburner, instead of "Nvidia overlay"

You still haven't said exactly what to look for in hwinfo, after I asked previously. Unless I missed it. I do not see any system latency type info in there. At best frame time, but it is not the same thing.

 

6 hours ago, HayuPhukmiWife said:

Am I limping the CPU potential?

It depends on how much impact it is having when it is limiting. BTW limiting is normal, the question is how much is too much. Are you doing a thermal limit or power limit? Normally a power limit will indirectly be a thermal limit, but I haven't ever used a thermal limit on CPU to control it.

 

What is the CPU usage while gaming?

 

Given it only affects some games than others, I have a feeling somehow you might have additional buffering enabled somehow. If it is in game, driver or elsewhere I'm not sure. The DDU/reinstall would certainly reset any driver settings, maybe try a manual "restore defaults" first as it is quicker. I presume you're using game defaults which should be ok.

 

Some details I don't recall if it was discussed, but may help as info. What is the Monitor max refresh rate, and does it support variable refresh like G-sync? Is V-sync on or off in game?

[Mark Kaine]

1 hour ago, porina said:

You still haven't said exactly what to look for in hwinfo,

no, i did... it shows latency for each component, which is probably important here!!!

 

1 hour ago, porina said:

It does not cause problems. 

ah, yeah, I'll take your word for it i suppose...

 

tons of issues reported and the time it almost fried my gpu aren't issues.  

 

 

[porina]

48 minutes ago, Mark Kaine said:

no, i did... it shows latency for each component, which is probably important here!!!

Exactly where? I've tried looking. Can you be more specific? Use a screenshot if it helps.

 

48 minutes ago, Mark Kaine said:

ah, yeah, I'll take your word for it i suppose...

 

tons of issues reported and the time it almost fried my gpu aren't issues.  

I've not gone looking but I've never heard of it before your comment. I regularly use it on all my systems over the years and never had a problem with it. Not saying no one ever had problems, but it probably isn't the root cause of those problems.

[HayuPhukmiWife]

3 hours ago, porina said:

It depends on how much impact it is having when it is limiting. BTW limiting is normal, the question is how much is too much. Are you doing a thermal limit or power limit? Normally a power limit will indirectly be a thermal limit, but I haven't ever used a thermal limit on CPU to control it.

 

What is the CPU usage while gaming?

 

Given it only affects some games than others, I have a feeling somehow you might have additional buffering enabled somehow. If it is in game, driver or elsewhere I'm not sure. The DDU/reinstall would certainly reset any driver settings, maybe try a manual "restore defaults" first as it is quicker. I presume you're using game defaults which should be ok.

 

Some details I don't recall if it was discussed, but may help as info. What is the Monitor max refresh rate, and does it support variable refresh like G-sync? Is V-sync on or off in game?

Asus multicore enhancement was getting the cpu to 103 at the ring in hwinfo, avg 100. some cores hitting tjmax and thermal thorttling. I have it now set to 'disabled - enforce all limits'

to my knowledge this and statements at initial startup with the first start after bios update this enforces intels chip settings and negates the mobo voltage settings... it keeps the fireball cool.   If its safe to have it hit 100-101 sometimes 103 for a split second as the cpu starts tasks then I'll run it and see if it makes a diferance, but those temps scare me. I mean... it doesnt sustain them unless im running a p95, and it staggers it across cores as they throttle. Ive heard this is norm for these 14th and 13th gen chips. but it was unheard of unless i forked up my thermal solution with my 10th gen.  In overclocking i would strive to keep it cooler than 90 degress...  I can install the cpu contact frame to reduce the fear of bending, but I dont want to run it if my mobo is trying to kill my chip.

 

Cpu usage is low. gpu usage is 98%.

 

I'll look into the ddu to frag my drivers and reinstall them. I just havent had the time. I've done clean installs through geforce experiance every time tho. I assume its a half measure comparatively?

 

Monitors are freesync (i know. i miss my gsync monitors too) usually vsync enabled. I've tested with it on and off to no change. Framerate always stays high. The frame generation isnt an issue here. theyre always 140-160.  Monitor is 165

[porina]

10 minutes ago, HayuPhukmiWife said:

Asus multicore enhancement was getting the cpu to 103 at the ring in hwinfo, avg 100. some cores hitting tjmax and thermal thorttling. I have it now set to 'disabled - enforce all limits'

It's a bit of a grey zone what Asus MCE actually does. On much older gen motherboards it was basically an easy overclock to give all cores the max turbo speed, but it could be generous on voltage to get there. With more modern CPUs I don't know what it does but personally I'd look at explicitly disabling it for normal running because it rules out overclocking on their part.

 

You can still adjust power limits as much as you like as it isn't considered by Intel to be overclocking. Intel have a suggested setting but you can go over that for more perf if your cooling is up to the job. CPU will thermally throttle if it gets too hot. In theory it is safe when not overclocking but I'd prefer to avoid it too. Prime95 is still a near worst case load and ideally you should tame it with a power limit according to your cooler potential.

 

I don't know how good the nvidia "clean install" option is, as I rarely use it myself. They used to keep old drivers and that option would clean them out, freeing up disk space. I don't know if that still applies today. DDU is the nuke option.

[HayuPhukmiWife]

8 minutes ago, porina said:

It's a bit of a grey zone what Asus MCE actually does. On much older gen motherboards it was basically an easy overclock to give all cores the max turbo speed, but it could be generous on voltage to get there. With more modern CPUs I don't know what it does but personally I'd look at explicitly disabling it for normal running because it rules out overclocking on their part.

 

You can still adjust power limits as much as you like as it isn't considered by Intel to be overclocking. Intel have a suggested setting but you can go over that for more perf if your cooling is up to the job. CPU will thermally throttle if it gets too hot. In theory it is safe when not overclocking but I'd prefer to avoid it too. Prime95 is still a near worst case load and ideally you should tame it with a power limit according to your cooler potential.

 

I don't know how good the nvidia "clean install" option is, as I rarely use it myself. They used to keep old drivers and that option would clean them out, freeing up disk space. I don't know if that still applies today. DDU is the nuke option.

Thoughts on the thermalright contact frame or the temps at +/- 100 for scant seconds?

I've heard of guys going to a 420 AIO or custom loop for these chips, but also hear that a 360 should be capable of taming it, and that a brief moment at +/-100 degrees is par for the course....

 

The ASUS MCE 'DEFINITELY' put me into a safe zone. I've seen nothing near those temps. never went within 10 degrees of tjmax since then. I havent overclocked in years, and didnt want to mess with voltages yet. so it was a nice one click solution. but the CPU is definetly underpreforming.  I mean it says my max clock is 6000mhz in hwinfo, but i usually dont see it go beyond 5700. 

[porina]

2 hours ago, HayuPhukmiWife said:

Thoughts on the thermalright contact frame or the temps at +/- 100 for scant seconds?

I don't have experience on contact frames beyond what's reported elsewhere so don't feel I can add much there.

Temps hitting 100 is possible even with a lower power limit if the load is low thread count, and a handful of cores boost up. Again, if not overclocked I wouldn't be worried about it.

 

2 hours ago, HayuPhukmiWife said:

I've heard of guys going to a 420 AIO or custom loop for these chips, but also hear that a 360 should be capable of taming it, and that a brief moment at +/-100 degrees is par for the course....

Bigger rad might help with sustained loads, but wont make so much difference for shorter peak bursts since that's probably more transfer limited.

 

2 hours ago, HayuPhukmiWife said:

but the CPU is definetly underpreforming.  I mean it says my max clock is 6000mhz in hwinfo, but i usually dont see it go beyond 5700. 

Sounds about right actually. 6.0 GHz is opportunistic and unlikely to be sustainable in heavier loads for any amount of time. Look at the 14900k (presuming that's what you have) listings. Going past 5.6 is good.

 

Max Turbo Frequency 6 GHz

Intel® Thermal Velocity Boost Frequency 6 GHz

Intel® Turbo Boost Max Technology 3.0 Frequency ‡ 5.8 GHz

Performance-core Max Turbo Frequency 5.6 GHz

[problemsolver]

14 hours ago, HayuPhukmiWife said:

Anyone think this could be due to my setting intel power limits to stop my CPU from being a 103 degree fireball?

I was seeing throttling when dialing the system in and disabled it to set intel limits vs mobo limits. I worry with all the constant throttling and temps seeing 101, 102, even 103 at times under load. So i set the intel limits and see 88-90 usually - 93 max. which feels WAYYYY safer and below tjmax.  

 

Am I limping the CPU potential? I'm not running any oc settings (click OC in bios or custom settings) could this have any merit? 360 AIO should be enough to keep it cool, but I thought nonstop throttling was a poor choice, and i was seeing it on specific cores and afraid of bending. That said I bought a contact frame from thermalright when diagnosing the throttling issues, but after I found out i could disable mobo settings I never installed it. Fear of bending is REAL... Thoughts?

Maybe... need more details on what kind of test you're running: is this all-core workloads such as Cinebench? Or single core? Either way you'll likely hit the thermal limits eventually, (without absolutely insane cooling, you pretty much ALWAYS hit the thermal limit somewhere with the top-tier consumer CPUs), but you will hit the thermal limits instantly if it's an all-core workload.

 

It's easy to test though... turn off your thermal throttling and see if there's a latency difference.

 

A 360mm AIO is not likely to keep it (14900K) from thermal throttling instantly in all-core workloads. (assuming normal pumps and fans)

I also agree with @porina that if not overclocked, leaving it without any artificial throttling is fine.

 

If you're trying to get max performance from your AIO:

• Set your pump fixed to max speed (this is actually where pumps are generally designed to operate most efficiently and used to be the normal setup in liquid-cooled PCs, until AIOs became popular with the general public. At that point, people complained about the noise, so low-speed/quiet modes were added and set as the default).
• Set your radiator fans to max speed.
   

Additional (Irrelevant) Info:

• MSI wrote a great FAQ about the 12900K explaining the cooling principles involved. (Don't get hung up on temps or voltages they cite.)
• Intel-specified Reference Heat Sink specification for proper operation this isn't of much use as I can't find any references to coolers which meet or don't meet this spec. I guess you could do a very flawed test where you limit the wattage of the CPU to 125W and see if you go above ~62 C. Again, this is super flawed, but I guess it could show if something is horribly wrong... which in your case we know it's not.

[HayuPhukmiWife]

On 1/7/2024 at 6:01 PM, problemsolver said:

TL;DR: Are you running the latest UEFI? You need to use Nvidia FrameView with Nvidia Overlay turned off if you actually want any accurate results on render latency. See guide at bottom of this post to collect data in Nvidia FrameView. Experiment with both the Balanced and Performance power plans.

 

Make sure you have the latest UEFI.

 

RAM timings, C-states, power settings and a myriad of other variables affect system latency. In addition, I do not trust a single video I've watched from Tech Yes City because they over-simplify complex problems, and do not follow the scientific method so that others can repeat their exact tests. In addition, some of the stuff I've heard the presenter say indicates a lack of fundamental understanding of core concepts in computing systems. The number of configurable variables in a modern PC is incredible, both in software and hardware. Tech Yes City, may reach correct conclusions, but the way they go about tests are flawed (in my own, useless opinion).

 

In addition, you have an Intel processor with P & E cores, so I have even less confidence in current tools, such as Latency Mon, in being able to measure system latency accurately. There is a lot of very complex stuff happening with Intel Thread Director that I have no doubt could affect some of the tools that measure latency. I would experiment with both the Balanced and Performance power plans to see if that affects Nvidia FrameView latency at all. This is another reason I recommend Nvidia FrameView and not some other tool. Nvidia has the resources, expertise, (and very strong incentives) to verify their tools are actually accurate. It's also built on PresentMon made by Intel

 

For Nvidia FrameView, see the guide here, specifically the collecting data portion.

 

I can confirm frameview reads similar values. UEFI updates. all firmwares updated etc... Should I pull some data from frameview and post it up? I haven't gotten far enough into the underperforming gpu guide to clear shaders and stuff yet. Just havent had time.

[HayuPhukmiWife]

On 1/10/2024 at 11:34 AM, porina said:

Look at the 14900k (presuming that's what you have) listings. Going past 5.6 is good.

I have the KF. But i assume similar values. should be slightly better with no integrated  graphics

 

[HayuPhukmiWife]

So, slightly unrelated to the issue with latency, I threw on the ASUS limits vs the INTEL ones and ran the AI overclock settings then ran a few games and Cinebench to take some readings. Latency didnt seem to change, I'll try all stock stuff when I have time (previous issues havent been tested with XMP disabled, but everything else was, and intel limits)

 

I dont know how cinebench stuff stacks up, anyone make heads or tails of this? does this thing look like its running fine?

 

 

 

[problemsolver]

13 hours ago, HayuPhukmiWife said:

I can confirm frameview reads similar values. UEFI updates. all firmwares updated etc... Should I pull some data from frameview and post it up? I haven't gotten far enough into the underperforming gpu guide to clear shaders and stuff yet. Just havent had time.

Posting some FrameView data is always welcome. If you record some data in HW Monitor with an all-core Cinebench running, that would be helpful too.
Sensors (thermometer Icon) >>> Start Logging >> let it run for a several complete image renders in CineBench (i.e. let it draw several complete images) probably just a minute or so

[HayuPhukmiWife]

On 1/12/2024 at 11:35 AM, problemsolver said:

Posting some FrameView data is always welcome. If you record some data in HW Monitor with an all-core Cinebench running, that would be helpful too.
Sensors (thermometer Icon) >>> Start Logging >> let it run for a several complete image renders in CineBench (i.e. let it draw several complete images) probably just a minute or so

So after re-enabling ASUS multi-core enhancement and running the danger of seeing thermal throttling, it seems to have resolved the render latency. In tooling around, this latency is expressed the same values in frameview and nvidia overlay, as well as in HWINFO under presentmon as Frame time. It seems to have been reduced by letting the mainboard try and melt my CPU. She's about 10 degrees hotter on cores 4 and 7 than the rest. But it still seems to run fine.

 

I enabled ASUS' AI overclocking, and have seen some interesting things. Like lots of heat one day, and not as much the next day. I believe this may be due to the way the "AI" OC system works (to my understanding) Feels like its adjusting the overclocking settings each time I reboot the system. It was hot the first day, and better the next. Now its seemed to balance things out pretty well. 

 

I have yet to try turning off the OC and leaving the multi-core enhancement on to see if the frame time stays reduced, and alternatively reducing the temp settings to INTELs settings and turing the OC on. I haven't really had time to check out other games to see if this trend continues, but I'm definitely seeing better latency (+/- 10ms) in some of the games I was playing. 

 

I am also seeming to experience microstutters in Hogwarts legacy still, but not drops in frame rate or increase in latency.  I have to do some more testing. And still, admittedly, have not gotten past the shader clearing in your microstuttering write up. I just thought 'magic' was working in my favor. lol  

 

I have a contact frame that I'm considering throwing on with some cable comb work, and re-applying my thermal paste/checking its coverage to see if it reduces the throttling and high temps on specific cores. maybe these are in-line with the bend area that can occur with the rectangle intel chips. What are your thoughts on the Thermalright contact frame install? I see mixed reviews. They seem to benefit more once the cpu is bent, and not have impact when its not (likely more my case, unless i'm in between those worlds.) I'm also wondering if I should install my heatsink looser. I only tightened it to what a 3 finger grip could do. and I'm not a big guy, so it should be fine. but I understand that overtightening can be an issue with overheating. IT seems like I'm running hotter than I should be though...

 

Thoughts? 

[problemsolver]

1 hour ago, HayuPhukmiWife said:

So after re-enabling ASUS multi-core enhancement and running the danger of seeing thermal throttling, it seems to have resolved the render latency. In tooling around, this latency is expressed the same values in frameview and nvidia overlay, as well as in HWINFO under presentmon as Frame time. It seems to have been reduced by letting the mainboard try and melt my CPU. She's about 10 degrees hotter on cores 4 and 7 than the rest. But it still seems to run fine.

 

I enabled ASUS' AI overclocking, and have seen some interesting things. Like lots of heat one day, and not as much the next day. I believe this may be due to the way the "AI" OC system works (to my understanding) Feels like its adjusting the overclocking settings each time I reboot the system. It was hot the first day, and better the next. Now its seemed to balance things out pretty well. 

 

I have yet to try turning off the OC and leaving the multi-core enhancement on to see if the frame time stays reduced, and alternatively reducing the temp settings to INTELs settings and turing the OC on. I haven't really had time to check out other games to see if this trend continues, but I'm definitely seeing better latency (+/- 10ms) in some of the games I was playing. 

 

I am also seeming to experience microstutters in Hogwarts legacy still, but not drops in frame rate or increase in latency.  I have to do some more testing. And still, admittedly, have not gotten past the shader clearing in your microstuttering write up. I just thought 'magic' was working in my favor. lol  

 

I have a contact frame that I'm considering throwing on with some cable comb work, and re-applying my thermal paste/checking its coverage to see if it reduces the throttling and high temps on specific cores. maybe these are in-line with the bend area that can occur with the rectangle intel chips. What are your thoughts on the Thermalright contact frame install? I see mixed reviews. They seem to benefit more once the cpu is bent, and not have impact when its not (likely more my case, unless i'm in between those worlds.) I'm also wondering if I should install my heatsink looser. I only tightened it to what a 3 finger grip could do. and I'm not a big guy, so it should be fine. but I understand that overtightening can be an issue with overheating. IT seems like I'm running hotter than I should be though...

 

Thoughts? 

I have no knowledge on the heatsink-mounting questions for the new Intel chips.

 

It's hard to judge whether something is running 'too' hot or not as most all of the high-end CPUs hit their thermal limit and stay there (as designed). The easiest way to determine if a CPU is 'too hot' is relate that to a workload. I.e. if someone has a 360mm AIO, and gets a score of 10K all-core in Cinebench R23, and their 5950X is @ 90 Celsius the entire time, something is not right. But if they scored 31K in the same exact scenario, then they've mostly likely got things setup pretty well .

 

You're speaking in a lot of general terms, and so it's really difficult impossible to be able to answer the questions you're asking.

[HayuPhukmiWife]

16 hours ago, problemsolver said:

I have no knowledge on the heatsink-mounting questions for the new Intel chips.

 

It's hard to judge whether something is running 'too' hot or not as most all of the high-end CPUs hit their thermal limit and stay there (as designed). The easiest way to determine if a CPU is 'too hot' is relate that to a workload. I.e. if someone has a 360mm AIO, and gets a score of 10K all-core in Cinebench R23, and their 5950X is @ 90 Celsius the entire time, something is not right. But if they scored 31K in the same exact scenario, then they've mostly likely got things setup pretty well .

 

You're speaking in a lot of general terms, and so it's really difficult impossible to be able to answer the questions you're asking.

So if the thermal limits and safeties are within design, and potentially function as they should, should I be looking at cpu temp vs individual core temps?

 

Maybe I'm looking at this from the wrong perspective. What should I be considering my temps in HWINFO? When I watch the CPU under the 'CPU [#0]: Intel Core i9-14900KF: DTS' and 'CPU [#0]: Intel Core i9-14900KF: Enhanced' areas, it hits 100 as a max, although not sustained. In cores individually, and Package/ring, however, when I look at the 'ASUS ROG STRIX Z790-E GAMING WIFI (Nuvoton NCT6798D)' readings (which seems to be what my mobo display and fans read off of) I dont go over 80. +/-.

 

Maybe I'm looking at the wrong things and its a non issue?  I've seen zero issues with heating or shutdowns. Maybe just perceived issues? 

[unclewebb]

1 hour ago, HayuPhukmiWife said:

should I be looking at cpu temp vs individual core temps?

The DTS value is what triggers thermal throttling. As a core gets hotter, the DTS value counts down. When the DTS reaches 0, the core is at 100°C. That is what triggers thermal throttling. That reduces maximum performance. The Nuvoton readings are not as important. The DTS readings are coming directly from the individual CPU core thermal sensors. 

 

 

On 1/10/2024 at 9:19 AM, HayuPhukmiWife said:

Thoughts on the Thermalright contact frame

That is always a good idea for a 14900K. When any core reaches 100°C, that reduces maximum performance. Anything you can do to keep all of the cores under 100°C is recommended. 

 

Here is an example of a 14900K scoring 41506 points in Cinebench R23. Your 100°C temperatures are reducing maximum performance.  

 

https://www.reddit.com/r/overclocking/comments/17mimwb/14900k_stock_cinebench_r23_scores/

 

[HayuPhukmiWife]

1 hour ago, unclewebb said:

The DTS value is what triggers thermal throttling. As a core gets hotter, the DTS value counts down. When the DTS reaches 0, the core is at 100°C. That is what triggers thermal throttling. That reduces maximum performance. The Nuvoton readings are not as important. The DTS readings are coming directly from the individual CPU core thermal sensors. 

 

That is always a good idea for a 14900K. When any core reaches 100°C, that reduces maximum performance. Anything you can do to keep all of the cores under 100°C is recommended. 

 

Here is an example of a 14900K scoring 41506 points in Cinebench R23. Your 100°C temperatures are reducing maximum performance.  

 

https://www.reddit.com/r/overclocking/comments/17mimwb/14900k_stock_cinebench_r23_scores/

 

So I set the core limits and package thermal limits in bios to not exceed 95. Although I'm not sure if that will stop the thermal throttling or just trigger it at 95 degrees.  I spent enough time messing with this while  I should be working today.  I'll do some checking into it when I get into some games with the kid tonight.

 

I feel safer with the limits set to 95, but as mentioned above. not sure if it reduced my throttle limit or gave me a buffer. It will essentially reduce clock speeds as it approaches my set limit, so I'm probably limping the system the same way. I see a max clock speed of 6200, but my average sits at 5050 like this. vs the 4500 it sees when i disable asus MCE

 

This is the area I have been watching for temps - maybe as a habit from my old system attempts at manual overclocking. HWINFO states that i should use the CPU package reading below this for actual cpu temps, but I understand that the cores will throttle respectively of that reading, and I'd like to dial it in. God damn I may try and learn how to delid this thing if the contact frame doesn't help.

 

The other workaround I found that solves it 100% is disabling ASUS multicore enhancement and letting the system run the INTEL spec voltages. But I lose 500mhz of  processing power across the board and more commonly see an average speed of 4500 than 5100.