*Official Benchmark Thread* - Post it here or it didn't happen :D

[Papusan]

1 hour ago, Custom90gt said:

Here is a quick pic, not sure what good values look like though.

 

Neither fish or bird (nor bad or fantastic). If you care about the E-core’s it’s above average. And P-cores ain’t bad. Could still run decent  And none should use air cooling with this chips so… Follow bro @electrosoft good advice for testing. 

 

Edit

 

[Tenoroon]

Alright, everything works!

I planned on turning down the speed to 4.0ghz all cores to reduce power, which helps a lot, but damn this thing uses a lot of power. I need to get or modify my bracket and get the special heatsink soon, this sucker runs HOT.

 

 


I'll be selling the 6700k here in a few days, still thinking about price, but it wont be outrageous like on Ebay, its only worth about 100-120 dollars max in my opinion due to its age and competition.

[Custom90gt]

2 hours ago, electrosoft said:

 

Much better than m SP81 and SP82 I had and the SP82 was actually a decent CPU. Once you get in there and do some auto stock runs at 4.9 -> 5.x (wherever it craps out), it will let you know what you're working with pull and temps wise.

 

Could be a decent chip with a good IMC.

 

Well fingers crossed.  Sadly it'll take me some time to actually get around to messing with it since my daughter takes up most of my free time.  I'd love to hit 6600mhz memory with reasonable timings at least.

 

1 hour ago, Papusan said:

Neither fish or bird (nor bad or fantastic). If you care about the E-core’s it’s above average. And P-cores ain’t bad. Could still run decent  And none should use air cooling with this chips so… Follow bro @electrosoft good advice for testing. 

 

Edit

 

 

Hah I haven't had an air cooled desktop in quite some time. Hopefully my 360 and 280 rads will keep it cool enough to run decently. 

[Papusan]

47 minutes ago, Custom90gt said:

 

Well fingers crossed.  Sadly it'll take me some time to actually get around to messing with it since my daughter takes up most of my free time.  I'd love to hit 6600mhz memory with reasonable timings at least.

 

Hah I haven't had an air cooled desktop in quite some time. Hopefully my 360 and 280 rads will keep it cool enough to run decently. 

I guessed so 🙂 For the SP ranking. Higher SP rating is no guaranteee for lowest possible power consumption/lower temps. 

 

 

[johnksss]

Didn't pass brother @Papusan , but still a decent run though.

[electrosoft]

Uninstalled the 12900ks SP94 and boxed it back up. My frugality has kicked in hardcore.

 

Popped the 12900k SP91 back in under 1403.

Letting Asus AI Optimizations do its thing with the SP91 12900k (I giggled at the reported 75% overclock):

 

 

 

 

[Papusan]

1 hour ago, electrosoft said:

Uninstalled the 12900ks SP94 and boxed it back up. My frugality has kicked in hardcore.

 

Popped the 12900k SP91 back in under 1403.

Letting Asus AI Optimizations do its thing with the SP91 12900k (I giggled at the reported 75% overclock):

 

 

 

 

Could be the monitoring software but from this it seems you saved of around 25W vs. your KS chips 🙂

[Mr. Fox]

After tinkering with some of my BIOS settings, an "Unkown Device" suddenly appeared in Device Manager on Windows 7. This driver that I modded will fix that if the unknown device HWID is ACPI\INTC1056. It is not important in the grand scheme of things, but I hate having things in Device Manager remain unresolved. Now I am back to everything being as it should be in Device Manager with everything recognized and working normally as long as the silly WiFi 6/BT crap is disabled in the BIOS.

 

https://mega.nz/file/AZ91QSCT#GHXoZS9zeP4V1Uf8Ngp7Nsz6kP-DRITzMEpO3id7Sao

[Reciever]

Light net search seems to be an issue born from Win10 era systems since all the posts regarding it are 2016+

 

Intel(R) Serial IO GPIO Host Controller - INTC1056

 

Some recommend going to "Treexy", but I have never heard of that site so I wouldnt desire to go there as a download source. However, may prove useful as a last resort.

 

https://treexy.com/products/driver-fusion/database/system/intel/serial-io-gpio-host-controller-intc1056/?o=id-/acpi/

[Custom90gt]

7 hours ago, electrosoft said:

Uninstalled the 12900ks SP94 and boxed it back up. My frugality has kicked in hardcore.

 

Popped the 12900k SP91 back in under 1403.

Letting Asus AI Optimizations do its thing with the SP91 12900k (I giggled at the reported 75% overclock):

 

 

 

 

 

What Asus utility are you using? That's pretty impressive, I've never been much into the "autoclocking" but that may allow me to mess around with this CPU...

 

 

14 hours ago, ryan said:

new internets....yeaaaaaa...its fast what are you guys getting you better not best

 

Just for fun:

 

 

[Mr. Fox]

9 hours ago, Reciever said:

Light net search seems to be an issue born from Win10 era systems since all the posts regarding it are 2016+

 

Intel(R) Serial IO GPIO Host Controller - INTC1056

 

Some recommend going to "Treexy", but I have never heard of that site so I wouldnt desire to go there as a download source. However, may prove useful as a last resort.

 

https://treexy.com/products/driver-fusion/database/system/intel/serial-io-gpio-host-controller-intc1056/?o=id-/acpi/

Yeah, just extra garbage payload that comes along with newer tech adding things that add no value, and lying about them not being compatible with W7. I have Treexy and it is reputable. It works kind of like Snappy Driver Installer (also a good tool) but Treexy has other features. Both are useful for people that find driver hunting to be intimidating or boring. But, neither one will make the anti-7 NeoNazi 10+ drivers work on W7 that have to be modded. I'd pay a nice amount of money if someone would develop an automatic driver modding tool. It would probably have to be Russian or Chinese to escape malicious litigation from the creeps at Intel, NVIDIA, AMD and Micro$lop and their bogus DMCA violation suits.

 

I am pretty good at driver modding as long as it only involves the INF. Where it gets into modding the SYS files with a hex editor, it's above my pay grade. I wish I knew how and I'd teach others how to use that knowledge and give a double-fisted nasty-finger salute to the smelly Kontrol_Klown_Klan.

[electrosoft]

16 hours ago, Papusan said:

Could be the monitoring software but from this it seems you saved of around 25W vs. your KS chips 🙂

 

Clock for clock, the SP91 12900k draws much less than the 12900ks at 4.9 and 5.0. 5.1 they're close. 5.2 is where the 12900ks pulls ahead and stays WAY ahead at 5.3.

 

12900k can boost to 5.5 on a few cores.

12900ks boosts to 5.7 on three cores.

 

12900k can do 5.3 all core but 5.4 is a bust. I'm going to go back in and push LLC7 just to see.

 

12900ks can do 5.4 all core and prolly 5.5 but the thermals. A delid and a CL would have it singing.

 

 

 

 

 

[electrosoft]

2 hours ago, Custom90gt said:

 

What Asus utility are you using? That's pretty impressive, I've never been much into the "autoclocking" but that may allow me to mess around with this CPU...

 

 

 

Just for fun:

 

 

 

Asus BIOS.

Leave MCE enabled (Do not turn on AI OC BIOS control)

Turn on AI Optimizations for P cores (E cores will auto enable)

Go under TVB

 enable +1 (or +2 if your system can handle it)

 enable both other options

 

save and restart.

 

 

 

[Papusan]

1 hour ago, electrosoft said:

 

Clock for clock, the SP91 12900k draws much less than the 12900k at 4.9 and 5.0. 5.1 they're close. 5.2 is where the 12900ks pulls ahead and stays WAY ahead at 5.3.

 

12900k can boost to 5.5 on a few cores.

12900ks boosts to 5.7 on three cores.

 

12900k can do 5.3 all core but 5.4 is a bust. I'm going to go back in and push LLC7 just to see.

 

12900ks can do 5.4 all core and prolly 5.5 but the thermals. A delid and a CL would have it singing.

 

Yep, you normally have more headroom with an binned chips. But you have 25W headroom for you K chips to come closer to the KS chips in all core boost. If I read your results correct, you had 5.2GHz all core for KS and 5.1GHz for the K in your comparison. Putting single core/dual/three core boost as for the KS is probably more difficult due need for even higher voltage increase to stay within reasonable/targeted power consumption (232W as for the KS). Everything is up to what you do. Run the KS stock (at +30% premium) or and want run the K closer to the KS chips 🙂

[electrosoft]

2 hours ago, Papusan said:

Yep, you normally have more headroom with an binned chips. But you have 25W headroom for you K chips to come closer to the KS chips in all core boost. If I read your results correct, you had 5.2GHz all core for KS and 5.1GHz for the K in your comparison. Putting single core/dual/three core boost as for the KS is probably more difficult due need for even higher voltage increase to stay within reasonable/targeted power consumption (232W as for the KS). Everything is up to what you do. Run the KS stock (at +30% premium) or and want run the K closer to the KS chips 🙂

 

Exactly. It happens to be an exceptionally cooler running, less hungry 12900k that only >=5.2 does the 12900ks show why it is a good binned 12900ks.

 

Still, too rich for my blood since I'll never delid or WC it this time around. So back it goes to Best Buy.

 

If anyone is interested in it since it is a pretty decent SP94, my exact price paid was 831.66 ($779.99 + tax). So $855 shipped non-cc CONUS. I'll be returning it Monday if no one wants it. In other words, it's out of here Monday one way or the other.

 

 

[Rage Set]

3 hours ago, electrosoft said:

 

Exactly. It happens to be an exceptionally cooler running, less hungry 12900k that only >=5.2 does the 12900ks show why it is a good binned 12900ks.

 

Still, too rich for my blood since I'll never delid or WC it this time around. So back it goes to Best Buy.

 

If anyone is interested in it since it is a pretty decent SP94, my exact price paid was 831.66 ($779.99 + tax). So $855 shipped non-cc CONUS. I'll be returning it Monday if no one wants it. In other words, it's out of here Monday one way or the other.

 

 

 

I'd have jumped on that in a heartbeat but alas I promised my wife to put all into my business until it's fully operational. Unless you're up for some kind of trade. Likely not but no harm in asking. 

[Mr. Fox]

OK, well... I did that washer mod thing that has been talked about relating to the 12th Gen ILM crushing the CPU into a banana shape. I used some M4 1mm thick  nylon washers, and it did lower the temps a bit, even after a delid with the copper IHS. I started the stress test and left at 3 minutes in to run to the store, and got back after the test had run for a total of 40 minutes. I put this into a side-by-side for ease of viewing. It was certainly easier to latch with the washers in there. It was so tight before that it almost felt like something is going to get broken before you can latch it down.

 

 

 

[Mr. Fox]

22 hours ago, Papusan said:

I guessed so 🙂 For the SP ranking. Higher SP rating is no guaranteee for lowest possible power consumption/lower temps. 

Part of it could also be some variance in the quality of the solder job attaching the IHS to the die. You could have a CPU wih better SP rating that runs hotter and is more of a power pig than a lower SP rating silicon sample.

[electrosoft]

6 hours ago, Rage Set said:

 

I'd have jumped on that in a heartbeat but alas I promised my wife to put all into my business until it's fully operational. Unless you're up for some kind of trade. Likely not but no harm in asking. 

 

I'll gladly relieve you of your Asrock OC Formula 6900xt for it. 😁

 

 

4 hours ago, Mr. Fox said:

Part of it could also be some variance in the quality of the solder job attaching the IHS to the die. You could have a CPU wih better SP rating that runs hotter and is more of a power pig than a lower SP rating silicon sample.

 

This. Many 12900k/s chips that won't hit 5.4 or 5.5 suddenly hit them with a delid and/or a CL. Throw a chiller in there and temps and pull drop accordingly.

 

 

[Papusan]

Wprime ain't so popular nowadays. Hence you don't get much points with an ok score. But nice be in between the LN2 boys with an AIO🙃

 

https://hwbot.org/submission/4980953_papusan_wprime___32m_core_i9_12900k_1sec_384ms?recalculate=true

 

 

[Mr. Fox]

7 hours ago, Mr. Fox said:

Part of it could also be some variance in the quality of the solder job attaching the IHS to the die. You could have a CPU wih better SP rating that runs hotter and is more of a power pig than a lower SP rating silicon sample.

 

3 hours ago, electrosoft said:

This. Many 12900k/s chips that won't hit 5.4 or 5.5 suddenly hit them with a delid and/or a CL. Throw a chiller in there and temps and pull drop accordingly.

I think mine is a good example of that based on the fact that a delid didn't improve the temps as much as other examples we have seen, but those where the delid made a bigger difference seem to have run hotter than mine did before a delid. But, some of that may also have been related to my cooling setup as well. The examples seen on YouTube were using systems with more mainstream cooling setups with an AIO or a much smaller custom loop. Without the chiller running, my system normalizes temperatures within about 3 to 5 minutes and the only time it changes up or down is dependent on the ambient temperatures. Those stress test comparisons I showed were at the temperature shown within about 3 or 4 minutes and whether I let the stress test run for 4 minutes or 40 minutes was irrelevant. The temperatures reached a certain point and then remained flat, both before and after delid. The stress wasn't able to overcome the cooling system.

[Mr. Fox]

https://hwbot.org/submission/4981077_

https://hwbot.org/submission/4981094_

[Mr. Fox]

This could be useful info. Sacrificial Celeron idea actually worked.

 

This is a good find as well. MUCH NEEDED to avoid the OEM motherboard software filth.

 

[johnksss]

1 hour ago, Mr. Fox said:

This could be useful info. Sacrificial Celeron idea actually worked.

 

This is a good find as well. MUCH NEEDED to avoid the OEM motherboard software filth.

 

That fan controller looks real promising.

 

And already the replacement caps on a 7980XE, but those were from a cracked die 7980XE though.

[johnksss]

On 4/22/2022 at 10:35 AM, electrosoft said:

 

Exactly. It happens to be an exceptionally cooler running, less hungry 12900k that only >=5.2 does the 12900ks show why it is a good binned 12900ks.

 

Still, too rich for my blood since I'll never delid or WC it this time around. So back it goes to Best Buy.

 

If anyone is interested in it since it is a pretty decent SP94, my exact price paid was 831.66 ($779.99 + tax). So $855 shipped non-cc CONUS. I'll be returning it Monday if no one wants it. In other words, it's out of here Monday one way or the other.

 

 

Just did a 5.5Ghz test on the KS and it ran 385 watts.