• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Intel Core i9-12900K Alder Lake Tested at Power Limits between 50 W and 241 W

Joined
Nov 8, 2022
Messages
34 (0.05/day)
As far as I am aware of, the Intel-suggested fan control behavior is to ramp up fan speed linearly from 80 °C to 100 °C, perhaps it's what happens in your case.
If so, keeping temperatures below 80°C or not much above that should help.

Check out here: https://edc.intel.com/content/www/u...1-of-2/009/fan-speed-control-scheme-with-dts/

I don't know if HP uses 90 as TDP in this scenario, or 125 (sticker on the fan unit) or 241 (PL2 for i9). The lower the value they use, the steeper the slope and that means the fan would cycle from zero to full speed under load very quickly? Which kind of tracks with the reviews?

Or to put it another way, the system passes along the slope from zero to the max fan speed very quickly from idle to load, because the PL2 is (at i9k stock setting of 241w) so much higher than the base TDP=PL1=90.

In a perfect world, you would want the cooling capacity of the system to match the TDP envelope of the PL2. That's why crazy fancy liquid cooling exists. In the real world, with more mixed workloads, that isn't really necessary — at least not for the cost trade-off. But laptops and SFF boxes are obviously very hostile environments for hot processors and cooling systems.

I suspect, for my quiet running preference, I will have to set PL2 to something between 125 and 190. Even at 175, it would keep 95% performance at a much lower temp. It will be an experimentation to find the right number, I guess.

Intel-Core-i9-12900K-2.png
 
Joined
Jun 22, 2012
Messages
301 (0.07/day)
Processor Intel i7-12700K
Motherboard MSI PRO Z690-A WIFI
Cooling Noctua NH-D15S
Memory Corsair Vengeance 4x16 GB (64GB) DDR4-3600 C18
Video Card(s) MSI GeForce RTX 3090 GAMING X TRIO 24G
Storage Samsung 980 Pro 1TB, SK hynix Platinum P41 2TB
Case Fractal Define C
Power Supply Corsair RM850x
Mouse Logitech G203
Software openSUSE Tumbleweed
I don't know if HP uses 90 as TDP in this scenario, or 125 (sticker on the fan unit) or 241 (PL2 for i9). The lower the value they use, the steeper the slope and that means the fan would cycle from zero to full speed under load very quickly? Which kind of tracks with the reviews?

According to Intel datasheets, CPU case (IHS / integrated heat spreader) temperature on the i9-12900K should not exceed 61.9 °C. This is not a temperature end-users can easily measure as it needs a thermocouple installed on a groove machined at the geometrical center of the IHS.

It can be assumed that the manufacturer of your system (HP) will have calibrated fan control so that at a sustained power of 90W the CPU will not exceed 61.9 °C at the IHS under worst-case scenarios, and that under transient conditions at higher power this will not be exceeded (not too much, at least). This constraint is perhaps the main reason why the system is noisy.

Core temperatures will be higher than IHS temperatures (possibly in the order of 20–25 °C higher under sustained conditions at the thermal limit), and have much lower (almost no) thermal inertia compared to IHS temperatures.

In a perfect world, you would want the cooling capacity of the system to match the TDP envelope of the PL2. That's why crazy fancy liquid cooling exists. In the real world, with more mixed workloads, that isn't really necessary — at least not for the cost trade-off. But laptops and SFF boxes are obviously very hostile environments for hot processors and cooling systems.

In principle, coolers designed for a specific TDP can temporarily dissipate a higher power, which is why PL1 and PL2 exist. That was the the basic reasoning behind the turbo boost algorithm, although it's mainly valid for thermally-constrained systems like laptops, and over the years it has been exploited for obtaining better benchmark scores in reviews.

On desktop parts with much larger coolers I see this mostly useful to keep the system at suitable noise and efficiency levels. My Noctua NH-D15s could dissipate 230W or more at unreasonable noise levels, so under load I keep my i7-12700k CPU around 150W on average on the short term (dynamically variable due to the low 85 °C thermal limit I configured) and 125W on the long term.

I suspect, for my quiet running preference, I will have to set PL2 to something between 125 and 190. Even at 175, it would keep 95% performance at a much lower temp. It will be an experimentation to find the right number, I guess.

Note that those values will be mostly valid for a properly cooled i9-12900k. With a cooler designed for dissipating continuously at most 90/125W, average power under load will likely already be much lower than the 241W PL2. So, I expect you will need to set PL2 closer to 125W to actually see differences.
 
Last edited:
Joined
Nov 8, 2022
Messages
34 (0.05/day)
It can be assumed that the manufacturer of your system (HP) will have calibrated fan control so that at a sustained power of 90W the CPU will not exceed 61.9 °C at the IHS under worst-case scenarios, and that under transient conditions at higher power this will not be exceeded (not too much, at least). This constraint is perhaps this is the main reason why the system is noisy.

Note that those values will be mostly valid for a properly cooled i9-12900k. With a cooler designed for dissipating continuously at most 90/125W, average power under load will likely already be much lower than the 241W PL2. So, I expect you will need to set PL2 closer to 125W to actually see differences.

So, in the HP with stock 90, 241 settings, a reviewer or user puts the system under load. The boost is allowed to ping 241, the system gets warmer, the fan ramps to max over some amount of time (a minute? two? less?) then system is thermally and/or power throttled back to 90 steady state, while the fan is still at max, because thats the thermal envelope of the (cooling) system.

With 90, [125-190] settings, the same thing happens, just… slower? I need to find the PL2 that will keep the system quiet at all times (perhaps 125 as you suggest), or possibly something slightly higher based on medium (but not zero) fan speed and/or my possibly less sustained workloads.

Do you think this might be what the 'quiet mode' in the HP bios does? Just set PL2 to a lower (125 or whatever) value? I guess I can find out once I have it...
 

bug

Joined
May 22, 2015
Messages
13,717 (3.97/day)
Processor Intel i5-12600k
Motherboard Asus H670 TUF
Cooling Arctic Freezer 34
Memory 2x16GB DDR4 3600 G.Skill Ripjaws V
Video Card(s) EVGA GTX 1060 SC
Storage 500GB Samsung 970 EVO, 500GB Samsung 850 EVO, 1TB Crucial MX300 and 2TB Crucial MX500
Display(s) Dell U3219Q + HP ZR24w
Case Raijintek Thetis
Audio Device(s) Audioquest Dragonfly Red :D
Power Supply Seasonic 620W M12
Mouse Logitech G502 Proteus Core
Keyboard G.Skill KM780R
Software Arch Linux + Win10
So, in the HP with stock 90, 241 settings, a reviewer or user puts the system under load. The boost is allowed to ping 241, the system gets warmer, the fan ramps to max over some amount of time (a minute? two? less?) then system is thermally and/or power throttled back to 90 steady state, while the fan is still at max, because thats the thermal envelope of the (cooling) system.

With 90, [125-190] settings, the same thing happens, just… slower? I need to find the PL2 that will keep the system quiet at all times (perhaps 125 as you suggest), or possibly something slightly higher based on medium (but not zero) fan speed and/or my possibly less sustained workloads.

Do you think this might be what the 'quiet mode' in the HP bios does? Just set PL2 to a lower (125 or whatever) value? I guess I can find out once I have it...
Basically yes, maybe.

PL1 is what the system can sustain indefinitely. PL2 is kind of boost of a boost. It will go beyond what your cooler can handle, but stop after a while, allowing the cooler to get the temperature back in check.

That said, if you don't have a top-notch cooler, you probably won't see any difference because both 241 and 190 will be above what the cooler can handle. This part varies from one user to another. Even if two users have the same cooler, there can still be differences because of different airflows. So it's pretty hard to guess what you will. Better try it and see how it goes.
 
Joined
Jun 22, 2012
Messages
301 (0.07/day)
Processor Intel i7-12700K
Motherboard MSI PRO Z690-A WIFI
Cooling Noctua NH-D15S
Memory Corsair Vengeance 4x16 GB (64GB) DDR4-3600 C18
Video Card(s) MSI GeForce RTX 3090 GAMING X TRIO 24G
Storage Samsung 980 Pro 1TB, SK hynix Platinum P41 2TB
Case Fractal Define C
Power Supply Corsair RM850x
Mouse Logitech G203
Software openSUSE Tumbleweed
So, in the HP with stock 90, 241 settings, a reviewer or user puts the system under load. The boost is allowed to ping 241, the system gets warmer, the fan ramps to max over some amount of time (a minute? two? less?) then system is thermally and/or power throttled back to 90 steady state, while the fan is still at max, because thats the thermal envelope of the (cooling) system.

At 241W (if the CPU is even allowed enough current to reach this level, but let's assume it can) CPU core temperature will likely hit 100 °C very quickly, almost immediately. Fan speed will then increase accordingly, probably reaching the maximum speed within seconds depending on how the manufacturer configured its behavior. At the same time, package power will start decreasing from 241W to lower levels. Eventually, PL1 is engaged and fans should decrease to the minimum speed required to keep the CPU case (IHS) below specification temperature.

Just to have an idea, below is data from stress test I made a while back with my system, with the CPU temperature limit at 90 °C and PL2 at 190W (click to enlarge). You can see that the 190W package power is only reached for a brief while due to the temperature limit, then decreases. When PL1 is engaged (set at 125W), core temperature jumps down significantly. You might expect a similar behavior on your system, only with power decreasing even faster due to smaller cooler.

1668009903220.png

With 90, [125-190] settings, the same thing happens, just… slower? I need to find the PL2 that will keep the system quiet at all times (perhaps 125 as you suggest), or possibly something slightly higher based on medium (but not zero) fan speed and/or my possibly less sustained workloads.

If you set PL2 to a lower value the maximum temperature reached should be lower or reached slower. Difficult to say how much because it's system-dependent.

Do you think this might be what the 'quiet mode' in the HP bios does? Just set PL2 to a lower (125 or whatever) value? I guess I can find out once I have it...

Could be. Many GPUs also have a "quiet mode" which decreases performance somewhat and lowers fan speed/noise.
 
Joined
Nov 8, 2022
Messages
34 (0.05/day)
That said, if you don't have a top-notch cooler, you probably won't see any difference because both 241 and 190 will be above what the cooler can handle. This part varies from one user to another. Even if two users have the same cooler, there can still be differences because of different airflows. So it's pretty hard to guess what you will. Better try it and see how it goes.

I thinks it's pretty fair to say the HP oem doesn't have a top notch cooler. It may not be trash, but i'm not expecting anything impressive let alone good. If it was, there would be less issues with noise in these tests. I'd be happy to spend a little money to improve it, but I think this system is too new and almost certainly too corporate to have any aftermarket cooling solutions.

I'm feeling somewhat encouraged, honestly. If I can manually set the PL2 and Tau — hopefully — then I'm sure there is a combination of the two numbers that will keep things acceptably quiet. If I can't, then i'll use the bios 'quiet mode' and hope for the best.

The price I (somehow?) got is too good to cancel, I think?

Just to have an idea, below is data from stress test I made a while back with my system, with the CPU temperature limit at 90 °C and PL2 at 190W (click to enlarge). You can see that the 190W package power is only reached for a brief while due to the temperature limit, then decreases. When PL1 is engaged (set at 125W), core temperature jumps down significantly. You might expect a similar behavior on your system, only with power decreasing even faster due to smaller cooler.

Setting a max temp would be just as good (a proxy really) for setting a max fan speed, neither of which I can do on this system, I'm pretty sure. Adjusting the PL2 and Tau is really the more specific way of achieving that goal, though it will require experimentation to find the correct values for me.
 

bug

Joined
May 22, 2015
Messages
13,717 (3.97/day)
Processor Intel i5-12600k
Motherboard Asus H670 TUF
Cooling Arctic Freezer 34
Memory 2x16GB DDR4 3600 G.Skill Ripjaws V
Video Card(s) EVGA GTX 1060 SC
Storage 500GB Samsung 970 EVO, 500GB Samsung 850 EVO, 1TB Crucial MX300 and 2TB Crucial MX500
Display(s) Dell U3219Q + HP ZR24w
Case Raijintek Thetis
Audio Device(s) Audioquest Dragonfly Red :D
Power Supply Seasonic 620W M12
Mouse Logitech G502 Proteus Core
Keyboard G.Skill KM780R
Software Arch Linux + Win10
I thinks it's pretty fair to say the HP oem doesn't have a top notch cooler. It may not be trash, but i'm not expecting anything impressive let alone good. If it was, there would be less issues with noise in these tests. I'd be happy to spend a little money to improve it, but I think this system is too new and almost certainly too corporate to have any aftermarket cooling solutions.

I'm feeling somewhat encouraged, honestly. If I can manually set the PL2 and Tau — hopefully — then I'm sure there is a combination of the two numbers that will keep things acceptably quiet. If I can't, then i'll use the bios 'quiet mode' and hope for the best.

The price I (somehow?) got is too good to cancel, I think?
Like I said, I did it on 12600k with a Freezer i34 (good, but not top-notch) cooler. You will just need to lower PL2 well under 190W and maybe lower PL1 as well. Just try different levels, don't be phased if you lower PL2 a little and don't see a change, that's all I'm saying.
 
Joined
Nov 8, 2022
Messages
34 (0.05/day)
Like I said, I did it on 12600k with a Freezer i34 (good, but not top-notch) cooler. You will just need to lower PL2 well under 190W and maybe lower PL1 as well. Just try different levels, don't be phased if you lower PL2 a little and don't see a change, that's all I'm saying.

Gotcha. That's good to know. Solid State Brain suggested something closer to 125 would probably be necessary. So, between the two of you, my expectations are calibrated pretty low!

It's fascinating to me that reviewers seem to be getting very impressive results from this system in benchmarks. Even with the fans maxed and loud AF, they are still significantly thermally and power constrained.

So, with that in mind, taking a very small performance hit (limiting PL2 to 125 or thereabouts) wouldn't be the worst thing in the world for a super quiet system. See this article that speaks to this scenario.
 
Last edited:
Joined
Oct 26, 2022
Messages
57 (0.08/day)
13900KS @6Ghz should push the power consumption to 370-390W in total usage.
 
Joined
Jun 22, 2012
Messages
301 (0.07/day)
Processor Intel i7-12700K
Motherboard MSI PRO Z690-A WIFI
Cooling Noctua NH-D15S
Memory Corsair Vengeance 4x16 GB (64GB) DDR4-3600 C18
Video Card(s) MSI GeForce RTX 3090 GAMING X TRIO 24G
Storage Samsung 980 Pro 1TB, SK hynix Platinum P41 2TB
Case Fractal Define C
Power Supply Corsair RM850x
Mouse Logitech G203
Software openSUSE Tumbleweed
Setting a max temp would be just as good (a proxy really) for setting a max fan speed, neither of which I can do on this system, I'm pretty sure. Adjusting the PL2 and Tau is really the more specific way of achieving that goal, though it will require experimentation to find the correct values for me.

The problem with PL2 in this regard is that lowering it won't affect high core temperatures due to intense single- or few-threaded workloads. So, the fan could still spin up rapidly while the CPU is at a relatively low power, since it will react to the highest reported temperature inside the package (generally speaking; could be different on some configurations).
 
Last edited:
Joined
Nov 8, 2022
Messages
34 (0.05/day)
The problem with PL2 in this regard is that lowering it won't affect high core temperatures due to intense single- or few-threaded workloads. So, the fan could still spin up rapidly while the CPU is at a relatively low power, since it will react to the highest reported temperature inside the package (generally speaking; could be different on some configurations).

I see what you mean. But I don't have a specific solution for that. Hopefully that's not how it behaves.

I have seen some of the reviews imply the fan lags a bit, or perhaps doesn't scale gradually with load, but then spins up rapidly. So maybe — if I'm lucky — by holding down the overall system temp by reducing PL2 and/or Tau, the fan hopefully won't ever get crazy. I mean, its the only lever I have to pull, so...

In your scenario, I'd have to set PL2 to something *really* low like 90? And it STILL might not solve the problem.

I don't see reviews obsessing over the fan noise, just mentioning it is quite loud at load and as a notable 'con'. In the world of corporate provided review units, it's hard to calibrate how big of a factor that really is. I'm assuming it will be an issue (for me), but if was a major major fiasco, the reviews would say so?
 
Last edited:
Joined
Jun 22, 2012
Messages
301 (0.07/day)
Processor Intel i7-12700K
Motherboard MSI PRO Z690-A WIFI
Cooling Noctua NH-D15S
Memory Corsair Vengeance 4x16 GB (64GB) DDR4-3600 C18
Video Card(s) MSI GeForce RTX 3090 GAMING X TRIO 24G
Storage Samsung 980 Pro 1TB, SK hynix Platinum P41 2TB
Case Fractal Define C
Power Supply Corsair RM850x
Mouse Logitech G203
Software openSUSE Tumbleweed
In your scenario, I'd have to set PL2 to something *really* low like 90? And it STILL might not solve the problem.

Probably to something like 40W or less to have an effect on single-core power draw, which is not desirable. A better alternative would be decreasing the maximum CPU frequency for single- or few-threaded workloads from 5.2 to 4.9 GHz or so. But do try the system in practice first, before thinking of doing this.
 

Mussels

Freshwater Moderator
Joined
Oct 6, 2004
Messages
58,413 (7.97/day)
Location
Oystralia
System Name Rainbow Sparkles (Power efficient, <350W gaming load)
Processor Ryzen R7 5800x3D (Undervolted, 4.45GHz all core)
Motherboard Asus x570-F (BIOS Modded)
Cooling Alphacool Apex UV - Alphacool Eisblock XPX Aurora + EK Quantum ARGB 3090 w/ active backplate
Memory 2x32GB DDR4 3600 Corsair Vengeance RGB @3866 C18-22-22-22-42 TRFC704 (1.4V Hynix MJR - SoC 1.15V)
Video Card(s) Galax RTX 3090 SG 24GB: Underclocked to 1700Mhz 0.750v (375W down to 250W))
Storage 2TB WD SN850 NVME + 1TB Sasmsung 970 Pro NVME + 1TB Intel 6000P NVME USB 3.2
Display(s) Phillips 32 32M1N5800A (4k144), LG 32" (4K60) | Gigabyte G32QC (2k165) | Phillips 328m6fjrmb (2K144)
Case Fractal Design R6
Audio Device(s) Logitech G560 | Corsair Void pro RGB |Blue Yeti mic
Power Supply Fractal Ion+ 2 860W (Platinum) (This thing is God-tier. Silent and TINY)
Mouse Logitech G Pro wireless + Steelseries Prisma XL
Keyboard Razer Huntsman TE ( Sexy white keycaps)
VR HMD Oculus Rift S + Quest 2
Software Windows 11 pro x64 (Yes, it's genuinely a good OS) OpenRGB - ditch the branded bloatware!
Benchmark Scores Nyooom.
Solid state brain:

You've got the handle on some stupidly complex math there. I'd just follow what you're saying.

(Lower the settings, undervolt if you can)

I don't see reviews obsessing over the fan noise, just mentioning it is quite loud at load and as a notable 'con'
Reviews are always really really prone to avoiding cons, you gotta mention them but unless the products trash tier garbage they'll never bluntly say something is outright shite
"Oh its noisy but that's acceptable in this form factor" "People know this before buying" etc

(And hence why my SFF systems are dual/quad cores, the ONLY way to reduce fan noise is by lower total wattages)
 

bug

Joined
May 22, 2015
Messages
13,717 (3.97/day)
Processor Intel i5-12600k
Motherboard Asus H670 TUF
Cooling Arctic Freezer 34
Memory 2x16GB DDR4 3600 G.Skill Ripjaws V
Video Card(s) EVGA GTX 1060 SC
Storage 500GB Samsung 970 EVO, 500GB Samsung 850 EVO, 1TB Crucial MX300 and 2TB Crucial MX500
Display(s) Dell U3219Q + HP ZR24w
Case Raijintek Thetis
Audio Device(s) Audioquest Dragonfly Red :D
Power Supply Seasonic 620W M12
Mouse Logitech G502 Proteus Core
Keyboard G.Skill KM780R
Software Arch Linux + Win10
(And hence why my SFF systems are dual/quad cores, the ONLY way to reduce fan noise is by lower total wattages)
I mean, if you're too cheap to throw some LN2 in there... :p
 
Joined
Nov 8, 2022
Messages
34 (0.05/day)
Reviews are always really really prone to avoiding cons, you gotta mention them but unless the products trash tier garbage they'll never bluntly say something is outright shite
"Oh its noisy but that's acceptable in this form factor" "People know this before buying" etc

Yes, exactly this. So, I know this SFF can get loud. However, I also know that reviews skew heavily to max load and game testing. I won't be gaming, and I am extremely unlikely to be doing any rendering or video work. So I shouldn't be putting this machine under the conditions most likely to produce those noisy conditions.

(And hence why my SFF systems are dual/quad cores, the ONLY way to reduce fan noise is by lower total wattages)

And also why I've been thinking and discussing how to use lower wattage settings to limit any thermal / fan / noise spikes, even in my more modest usage patterns. I may even lower PL1 a tad depending on how things go; its on the table.

It's also possible that the bios setting for 'quiet mode' does exactly what I want. It sure would be nice if someone tried THAT in a review. But no one seems to care about quiet running?

I'm hopeful that I can find a sweet spot where this machine will be my goldilocks: small, quiet, and fast for my typical usage.
 
Last edited:
Joined
Mar 10, 2010
Messages
11,878 (2.22/day)
Location
Manchester uk
System Name RyzenGtEvo/ Asus strix scar II
Processor Amd R5 5900X/ Intel 8750H
Motherboard Crosshair hero8 impact/Asus
Cooling 360EK extreme rad+ 360$EK slim all push, cpu ek suprim Gpu full cover all EK
Memory Corsair Vengeance Rgb pro 3600cas14 16Gb in four sticks./16Gb/16GB
Video Card(s) Powercolour RX7900XT Reference/Rtx 2060
Storage Silicon power 2TB nvme/8Tb external/1Tb samsung Evo nvme 2Tb sata ssd/1Tb nvme
Display(s) Samsung UAE28"850R 4k freesync.dell shiter
Case Lianli 011 dynamic/strix scar2
Audio Device(s) Xfi creative 7.1 on board ,Yamaha dts av setup, corsair void pro headset
Power Supply corsair 1200Hxi/Asus stock
Mouse Roccat Kova/ Logitech G wireless
Keyboard Roccat Aimo 120
VR HMD Oculus rift
Software Win 10 Pro
Benchmark Scores 8726 vega 3dmark timespy/ laptop Timespy 6506
Yes, exactly this. So, I know this SFF can get loud. However, I also know that reviews skew heavily to max load and game testing. I won't be gaming, and I am extremely unlikely to be doing any rendering or video work. So I shouldn't be putting this machine under the conditions most likely to produce those noisy conditions.



And also why I've been thinking and discussing how to use lower wattage settings to limit any thermal / fan / noise spikes, even in my more modest usage patterns. I may even lower PL1 a tad depending on how things go; its on the table.

It's also possible that the bios setting for 'quiet mode' does exactly what I want. It sure would be nice if someone tried THAT in a review. But no one seems to care about quiet running?

I'm hopeful that I can find a sweet spot where this machine will be my goldilocks: small, quiet, and fast for my typical usage.
Why buy an i9 in the first place, it sounds like a i5 would have suited your use case and profile considerably better than what you bought.

Disable the E core's, manually set 90 watts max usage and undervolt it, you don't need the performance so just throw that potential in the bin by configuring it as a lower spec chip.
 
Joined
Nov 8, 2022
Messages
34 (0.05/day)
New question: Why are there no 5k displays available that are made for PCs (and not just targeted to macs)?

I have an LG 34" 5k2k that is pretty nice, but it's not full 5k. I was also considering using either two Studio Display (5k) or Ultrafine (5k). I can drive (1) up-to-8k panel via the i9 built-in 770 graphics and displayport 1.4. (I think I could just use a cable to adapt the full size displayport to a usb-c end to go into the SD or Ultrafine?)

But to get a second (or third) display at 5k+ I'd have to aggregate two of the mini-displayport together on the back of the rtx a2000 (which has four - which could theoretically drive two more 5k+ displays?). EXCEPT that I can't find ANY 5k+ displays with two video inputs!? Is this a situation that requires a docking port to do the aggregation — is that even a thing?

And just for more funsies, I ordered my machine with the optional thunderbolt 3 port. So, maybe that could drive a second 5k display? BUT, I can't find ANY information that clarifies whether that port outputs from the 770, (two streams) directly from the a2000 gpu, neither, or if its somehow selectable?

I mean, this isn't that important — the 34" display is really nice, and will serve my needs well. But, I'd love to know what's possible so I could at least consider multiple 5k — that would be a significant improvement over a single 5k2k!

Why buy an i9 in the first place, it sounds like a i5 would have suited your use case and profile considerably better than what you bought.

Disable the E core's, manually set 90 watts max usage and undervolt it, you don't need the performance so just throw that potential in the bin by configuring it as a lower spec chip.

I'm a mac guy, and I'm pretty ignorant about PC stuff. I need a PC workstation for WFH for CAD. Cad on mac is OK, but not full feature parity. Also, Cad doesn't run on arm/apple silicon, so i'm not sure there is a viable future runway for CAD on mac.

I really like the the z2 mini g9 small footprint; easy to open etc. Basically, I was dumb and foolishly ordered the top bin chip, not realizing how crippling the enclosure would be and the consequences for thermals and (potential) noise. I got an incredible price (I think?) on my order: i9k, 64gb ecc, 1 tb fast ssd, rtx a2000 for $1495 incl. tax, direct from HP with 3 yr warranty. However, now they won't allow me to downgrade the processor with outright cancelling the order, and that price is no longer available. I guess they use a dynamic pricing algorithm, and their customer service is apparently shit?

So, that's why I've been discussing with the folks here how I might adjust the settings to get a performance envelope that is more suited to my use and preferences (quiet). I'm optimistic that adjusting PL1 and PL2 will allow me to find a sweet spot, if I even need to do anything at all.

And, even thermally / power throttled, this little machine will still utterly obliterate any machine I've ever used (supplied by companies for whom I've worked). They're all cheap AF.
 
Last edited:

bug

Joined
May 22, 2015
Messages
13,717 (3.97/day)
Processor Intel i5-12600k
Motherboard Asus H670 TUF
Cooling Arctic Freezer 34
Memory 2x16GB DDR4 3600 G.Skill Ripjaws V
Video Card(s) EVGA GTX 1060 SC
Storage 500GB Samsung 970 EVO, 500GB Samsung 850 EVO, 1TB Crucial MX300 and 2TB Crucial MX500
Display(s) Dell U3219Q + HP ZR24w
Case Raijintek Thetis
Audio Device(s) Audioquest Dragonfly Red :D
Power Supply Seasonic 620W M12
Mouse Logitech G502 Proteus Core
Keyboard G.Skill KM780R
Software Arch Linux + Win10
New question: Why are there no 5k displays available that are made for PCs (and not just targeted to macs)?
That one is simple, really: video cards can't game at 5k. Thus, 5k panels are confined to professional usage for the time being and you won't find many of them in typical consumer products.
 

Mussels

Freshwater Moderator
Joined
Oct 6, 2004
Messages
58,413 (7.97/day)
Location
Oystralia
System Name Rainbow Sparkles (Power efficient, <350W gaming load)
Processor Ryzen R7 5800x3D (Undervolted, 4.45GHz all core)
Motherboard Asus x570-F (BIOS Modded)
Cooling Alphacool Apex UV - Alphacool Eisblock XPX Aurora + EK Quantum ARGB 3090 w/ active backplate
Memory 2x32GB DDR4 3600 Corsair Vengeance RGB @3866 C18-22-22-22-42 TRFC704 (1.4V Hynix MJR - SoC 1.15V)
Video Card(s) Galax RTX 3090 SG 24GB: Underclocked to 1700Mhz 0.750v (375W down to 250W))
Storage 2TB WD SN850 NVME + 1TB Sasmsung 970 Pro NVME + 1TB Intel 6000P NVME USB 3.2
Display(s) Phillips 32 32M1N5800A (4k144), LG 32" (4K60) | Gigabyte G32QC (2k165) | Phillips 328m6fjrmb (2K144)
Case Fractal Design R6
Audio Device(s) Logitech G560 | Corsair Void pro RGB |Blue Yeti mic
Power Supply Fractal Ion+ 2 860W (Platinum) (This thing is God-tier. Silent and TINY)
Mouse Logitech G Pro wireless + Steelseries Prisma XL
Keyboard Razer Huntsman TE ( Sexy white keycaps)
VR HMD Oculus Rift S + Quest 2
Software Windows 11 pro x64 (Yes, it's genuinely a good OS) OpenRGB - ditch the branded bloatware!
Benchmark Scores Nyooom.
Why buy an i9 in the first place, it sounds like a i5 would have suited your use case and profile considerably better than what you bought.

Disable the E core's, manually set 90 watts max usage and undervolt it, you don't need the performance so just throw that potential in the bin by configuring it as a lower spec chip.
^ This
You cant review and judge the hardware if the CPU is user changeable
Maxed out settings on an i5 is going to be quieter than a PL1 limited i9 (and likely faster)


Oh and PC's can totally game at 5K, easily with DLSS involved. I game at 4K 165 at present using DLSS.

The reason we dont have 5k or 8k high refresh displays is because displayport and HDMI tech havent caught up enough yet, they're using compression at the moment to get what we currently have.
 
Joined
Nov 8, 2022
Messages
34 (0.05/day)
Maxed out settings on an i5 is going to be quieter than a PL1 limited i9 (and likely faster)
Yes, well, unfortunately I can't get customer service from HP to match the pricing I got, or I would switch down in a hot second.

I mean, if the i9 is really just intolerably loud, I can always sell the machine and buy something different. I hope that isn't necessary, for a number of reasons, hence my discussions here. Reviews are useful, but you never really know until you have the thing sitting in front of you doing the work, and making the tweaks...
 
Joined
Nov 8, 2022
Messages
34 (0.05/day)
Interesting turn of events… I guess HP must have their dynamic pricing bottom out on the weekends; I can get the low pricing today that I had last week, if I'm willing to wait until mid January for the machine!

Dropping to an i7k shaves off another $100, and dropping to an i5k shaves off almost $200. I'm not that worked up over the cost — over the lifespan of this machine, one or two hundred bucks doesn't matter that much. (And, I guess it's built such that I could drop in a future chip that was faster / more efficient?)

Anyway, the i7k review shows a 15% synthetic performance boost over the i5k, while also running cooler under load. But I don't know how meaningful that is since those data are using all three chips in their 'stock' configuration.

In my preferred oem machine, all three chip configs are set with LP1=90; while LP2=241 for i9k, LP2=190 for i7k, and LP2=150 for the i5k. None of the reviews or tests really show what I'd like to know:

- how would the i9k compare to the i7k with the same LP2=190?
- how would both the i9k and i7k compare to the i5k with all three set to LP2=150?
- how would all three compare at LP2=125?

The only hard data I have from this review (9:00) where they actually put a stock i7k into this machine and it benched faster than the i9k for rendering stuff, and pretty close to even frame rates at CS:GO. That's super interesting, and clearly something more than just LP2 limiting is going on… Presumably due to the lower LP1=90 limit for both chips, and throttling encountered under load.

I've read multiple articles (including this one) and also here — that discuss power limiting the i9k. But neither of them discuss the tradeoffs of that relative to a (similarly limited) i7k or i5k. So I'm left to wonder (money notwithstanding) why someone would choose an i5k or i7k vs a power-limited i9k?

Obviously, I'd prefer to have the fastest machine possible while remaining quiet. But what does that mean — stock i5k? stock i7k? LP2 throttled i9k? The best option I can surmise in the absence of more information is to get the i7k (and still potentially drop the LP2 down the road, but probably not need to)...

Thoughts/advice?
 
Last edited:
Joined
Jun 22, 2012
Messages
301 (0.07/day)
Processor Intel i7-12700K
Motherboard MSI PRO Z690-A WIFI
Cooling Noctua NH-D15S
Memory Corsair Vengeance 4x16 GB (64GB) DDR4-3600 C18
Video Card(s) MSI GeForce RTX 3090 GAMING X TRIO 24G
Storage Samsung 980 Pro 1TB, SK hynix Platinum P41 2TB
Case Fractal Define C
Power Supply Corsair RM850x
Mouse Logitech G203
Software openSUSE Tumbleweed
The bigger chip (i.e. with more cores) should always get better MT performance for the same power limit. Intel i9 CPUs also tend to be better binned than lower SKUs, requiring less voltage on average for the same frequency.

LTT results at 9:00 in the video, in particular for rendering, make no sense—I suspect other problems there.
 
Joined
Nov 8, 2022
Messages
34 (0.05/day)
The bigger chip (i.e. with more cores) should always get better MT performance for the same power limit. Intel i9 CPUs also tend to be better binned than lower SKUs, requiring less voltage on average for the same frequency.

LTT results at 9:00 in the video, in particular for rendering, make no sense—I suspect other problems there.

You're saying the rendering results are not what you would expect? Maybe putting the retail i7k in there somehow reset the LP1 to 125? Or maybe the additional i9k cores generated enough extra heat that the clock had to go down to stay within the LP1=90 limit, diminishing overall MT performance compared to the i7k? I don't know its weird, which is why they put it in the video, I think?

So what would you do in my situation? Keep the i9k order and lower the LP2 (if necessary)? Or switch to an i7k order and save $100?

Maxed out settings on an i5 is going to be quieter than a PL1 limited i9 (and likely faster)

But that isn't my option — All three chip options have PL1=90. I mean, I guess I could change that too, but I wasn't planning to do so.

How would a i5k compare to a SAME SPEC (PL1=90, PL2=150) i7k, and i9k? Especially for less-heavily multi-core workloads. I would expect those loads to more or less scale with the core count? That's the type of apples-to-apples scenario that would actually help me choose!
 
Last edited:
Joined
Jun 22, 2012
Messages
301 (0.07/day)
Processor Intel i7-12700K
Motherboard MSI PRO Z690-A WIFI
Cooling Noctua NH-D15S
Memory Corsair Vengeance 4x16 GB (64GB) DDR4-3600 C18
Video Card(s) MSI GeForce RTX 3090 GAMING X TRIO 24G
Storage Samsung 980 Pro 1TB, SK hynix Platinum P41 2TB
Case Fractal Define C
Power Supply Corsair RM850x
Mouse Logitech G203
Software openSUSE Tumbleweed
You're saying the rendering results are not what you would expect? Maybe putting the retail i7k in there somehow reset the LP1 to 125? Or maybe the additional i9k cores generated enough extra heat that the clock had to go down to stay within the LP1=90 limit, diminishing overall MT performance compared to the i7k? I don't know its weird, which is why they put it in the video, I think?

So what would you do in my situation? Keep the i9k order and lower the LP2 (if necessary)? Or switch to an i7k order and save $100?

On shorter renders (BMW) the 12700K appeared to be almost twice as fast while on longer ones (Gooseberry CPU) the the 12900K was faster despite large short-term losses. I don't think changes in Tau and PL2 alone can justify such a difference.

It seems as if the cooler-cpu thermal contact with the specific i9 that was in that HP was poor, while the i7 was better in this regard. There have indeed been reports of some 12th gen Intel CPUs slightly "bending" once installed and suffering from higher temperatures than normal (up to 10-15 °C in some cases), but not all were affected.

Or possibly it might have been a thermal paste issue, and the i7 tested was installed with something better (or better applied) than what the i9 came with from the factory. Difficult to tell for sure without more information.

Personally, if money was not an issue I would get the i9 and limit PL2/Tau to the same level of the i7 or less. It should be always faster for the same power draw, assuming that there are no BIOS-related issues with the i9 on that system.

1668278421942.png
 
Joined
Nov 8, 2022
Messages
34 (0.05/day)
Or possibly it might have been a thermal paste issue, and the i7 tested was installed with something better (or better applied) than what the i9 came with from the factory. Difficult to tell for sure without more information.

You wouldn't put too much import on that one anecdotal sample? (And, also, the frame rate in the games test was pretty similar, FWIW.)

Personally, if money was not an issue I would get the i9 and limit PL2/Tau to the same level of the i7 or less. It should be always faster for the same power draw, assuming that there are no BIOS-related issues with the i9 on that system.

Faster because i9k has more cores and larger cache? Does the lower base clock (3.2ghz vs 3.6ghz) mean the i9k suffers compared to i7k for single core performance, holding power draw constant?

The $100 delta is not a deal-breaker. I feel equally great about the price, wither its $1400 for the i9k system, or $1300 for the i7k system. What kind of BIO-related issues might crop up?

Is it in any way meaningful that none of the 3rd party sales channels (eg B&H etc) sell an i9k config? They all top out at i7k...
 
Top