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

[Papusan]

14 hours ago, johnksss said:

Also in the grand scheme of things....it's not temps killing Intel cpus, it's the voltage. Now temps might kill an AMD CPU though, but then that is speculations on my part as I would never let an AMD CPU over heat, but an intel cpu, sure, no problem. as they dial back when they reach the set max temp, not just get hotter and hotter.

Intel have reduces Tjunction temp for KS chips down to 90C (100C for the K chips) and added in Intel® Thermal Velocity Boost Temperature who kick in at 50C. Higher boost clocks = Higher voltage. But this was needed for the KS chips to try counter the new locked down v-cashe chips from AMD to stay pretty alone on top as the gaming king. Yep, Intel don't want you to run high voltage paired with high temps over prolonged time, hence they dropped the ball on the Tjunction for the KS chps who have higher boost clocks. 

 

From Intel's engineers at their OC lab...

 

As Ragland explained, the amount of time a processor stays in elevated temperature and voltage states has the biggest impact on lifespan. You can control the temperature of your chip with better cooling, which then increases lifespan (assuming the voltage is kept constant). Assuming voltage remains constant, each successive drop in temperature results in a non-linear increase in life expectancy, so the 'first drop' in temps from 90C to 80C yields a huge increase in chip longevity. In turn, colder chips run faster at lower voltages, so dropping the temperature significantly by using a beefier cooling solution also allows you to drop the voltage further, which then helps control the voltage axis. All is about the voltage. 

https://www.tomshardware.com/uk/features/inside-intels-secret-overclocking-lab/7

 

Yep, the OEM have added 115C as an option in the firmware, but I don't think that temp is good over time if paired with higher voltage due you oc' your chips. Probably not as dangerous if you run your chips at stock clocks with default voltege (Dell is the bellsheep for 100C but you don't get much overclock features from Dell. They don’t let you  increase the voltage out of the chips range). High temps = No problem if the voltage stay within the default target for the chips. Dell do a lot weird things but they are correct in this one.

[Tenoroon]

10 minutes ago, Papusan said:

As Ragland explained, the amount of time a processor stays in elevated temperature and voltage states has the biggest impact on lifespan. You can control the temperature of your chip with better cooling, which then increases lifespan (assuming the voltage is kept constant). Assuming voltage remains constant, each successive drop in temperature results in a non-linear increase in life expectancy, so the 'first drop' in temps from 90C to 80C yields a huge increase in chip longevity. In turn, colder chips run faster at lower voltages, so dropping the temperature significantly by using a beefier cooling solution also allows you to drop the voltage further, which then helps control the voltage axis. All is about the voltage. 

https://www.tomshardware.com/uk/features/inside-intels-secret-overclocking-lab/7

 

This would have been such a useful source when I wrote my paper about how bad laptops are nowadays, its a shame I didn't find this!

[johnksss]

6 hours ago, electrosoft said:

 

I did and I agree with you 100% but there is no logical reason to raise temp caps in my setup. It will just heat up higher and pull more power and either throttle at a different point or crash. If I can't keep it under 100c, I won't keep it under 115c with the resulting scaling in heat and pull. If the objective is to pull even more (heat:power curve) for demonstration sure but outside of that? My golden rule is 100c cap (or lower even depending on purpose) because at that point I need to address my cooling, delid my chip , adjust my expectations based on silicon or all of thee above.

 

As for my previous 12900ks vs 12900k, setting 100c as the default cap, conditions being equal, you can see both chips bouncing off of the threshold and still see the 12900ks is superior not just in Vcore (1.421 vs. 1.465) but as a result in score as the 12900k bounces off of it much more frequently than the 12900ks.

 

Based on early data of dying AMD processors and AMD curtailing their boost options from 500mhz to 200mhz along with PBO overall I agree with your speculation.

 

Whew, that is a beefy, impressive run on the 12900ks.

 

 

 

 

Short answer.

It was to answer your own question, considering you were already 10C over on the KS.....So....You would need to go back and retest my friend.🙂

My answer was only to see where it bottomed out and using 115C for that particular test would have answered that ceiling question you had.

1 hour ago, Papusan said:

Intel have reduces Tjunction temp for KS chips down to 90C (100C for the K chips) and added in Intel® Thermal Velocity Boost Temperature who kick in at 50C. Higher boost clocks = Higher voltage. But this was needed for the KS chips to try counter the new locked down v-cashe chips from AMD to stay pretty alone on top as the gaming king. Yep, Intel don't want you to run high voltage paired with high temps over prolonged time, hence they dropped the ball on the Tjunction for the KS chps who have higher boost clocks. 

 

From Intel's engineers at their OC lab...

 

As Ragland explained, the amount of time a processor stays in elevated temperature and voltage states has the biggest impact on lifespan. You can control the temperature of your chip with better cooling, which then increases lifespan (assuming the voltage is kept constant). Assuming voltage remains constant, each successive drop in temperature results in a non-linear increase in life expectancy, so the 'first drop' in temps from 90C to 80C yields a huge increase in chip longevity. In turn, colder chips run faster at lower voltages, so dropping the temperature significantly by using a beefier cooling solution also allows you to drop the voltage further, which then helps control the voltage axis. All is about the voltage. 

https://www.tomshardware.com/uk/features/inside-intels-secret-overclocking-lab/7

 

Yep, the OEM have added 115C as an option in the firmware, but I don't think that temp is good over time if paired with higher voltage due you oc' your chips. Probably not as dangerous if you run your chips at stock clocks with default voltege (Dell is the bellsheep for 100C but you don't get much overclock features from Dell. They don’t let you  increase the voltage out of the chips range). High temps = No problem if the voltage stay within the default target for the chips. Dell do a lot weird things but they are correct in this one.

Yes, this is true. KS=90C and K=100C.

 

OEM added after Intel opened their mouth when at that time it was 105C up from either 90C/95C or 100C. Then it went to 110C, 112C (Speculations on the 110/112 ranges) then 115C depending on manufacture and chip over the years.

The higher setting is not for you to run it at that, but to cover the spikes. Example. If you run Cinebench R23 it will spike really high temp, but as the benchmark runs the temps start to fall. (For a bencher that can make or break a run. That spike temp hits the wall and). Ergo we try to have proper cooling in place as to not even bring that into play.🙂

 

Also well known is that intel CPU's have an internal temp monitoring system that will not all the chip to run faster than that temp will allow it. I had tried explaining that along time ago but was met with resistance. And the ranges I was talking about started at 50C and below under load. And when you go subzero brings in another set of bins....

 

 

I have to also say, not really caring about chip degradation as I wont have this chip that long and it's under warranty. lol

 

[Papusan]

 

What a e-waste! Damn the customers that let them continue with this.

 

Just fine....

 

[Clamibot]

16 hours ago, johnksss said:

Yes. 385W at 14C water.

And you didn't get the Intel memo when they were the ones that came out and said it was fine to run up to 115C like since what, the 10900K days? Also in the grand scheme of things....it's not temps killing Intel cpus, it's the voltage. Now temps might kill an AMD CPU though, but then that is speculations on my part as I would never let an AMD CPU over heat, but an intel cpu, sure, no problem. as they dial back when they reach the set max temp, not just get hotter and hotter. This comes from past experiments and just plain user error when I found these things out. So no reviewer or surfing the net to get the answers. 🙂

 

edit:

Also that 385W run was with 5.5Ghz/43 E core/45 cache/LLC8/1.4V/Load 1.288V

 

Wait, you can do that with Z490 boards? How would I set the TjMax temperature to over 100°C?

 

I want to see how far I can push my 10900K in my Clevo X170SM-G with that extra thermal headroom. Yeah yeah I know, not a good idea for prolonged use, but I don't plan to subject my CPU to temperatures over 100°C for an extended period of time. Even if I did, the CPU would still outlive it's useful lifespan for me since electronics are really robust.

[johnksss]

24 minutes ago, Clamibot said:

 

Wait, you can do that with Z490 boards? How would I set the TjMax temperature to over 100°C?

 

I want to see how far I can push my 10900K in my Clevo X170SM-G with that extra thermal headroom. Yeah yeah I know, not a good idea for prolonged use, but I don't plan to subject my CPU to temperatures over 100°C for an extended period of time. Even if I did, the CPU would still outlive it's useful lifespan for me since electronics are really robust.

Depends on your bios. If it was unlocked by Prema then chances are fair he may have unlocked that option in the advanced CPU menu. Not sure what your bios looks like though....

[johnksss]

1 hour ago, Papusan said:

 

What a e-waste! Damn the customers that let them continue with this.

 

Just fine....

 

Now why he using one or Mr. Fox's disses. Don't he know that @Mr. Fox has these diss words and phrases trademarked!😂

[Mr. Fox]

2 hours ago, Papusan said:

 

What a e-waste! Damn the customers that let them continue with this.

 

Just fine....

 

This should come as no surprise to anyone, but it's good that Steve is calling the balls and strikes.

1 hour ago, johnksss said:

Now why he using one or Mr. Fox's disses. Don't he know that @Mr. Fox has these diss words and phrases trademarked!😂

That's OK. As they say, "imitation is the greatest form of flattery" and as long as proper credit and public recognition is given for Dell's moral, ethical and technical failures, that's what matters most.

[Reciever]

 

First boot.

 

Decided to grab my Taichi 5700xt and strap 3x 45mmx92mm fans. I think they do 6k rpm's.

 

Went the AMD route for the perceived hope of an ease of use for Linux but also I cant find my x570 stock motherboard mounts so that system is down for now which means AMD will be migrated first.

[electrosoft]

2 hours ago, Papusan said:

 

What a e-waste! Damn the customers that let them continue with this.

 

 

Hilarious.

 

I had commented on a Mac coder's YT channel I watch  a week or so back who compared compile times of his Mac Studio Ultra (M1 Ultra chip) vs the 12900kf and the M1 won handily but looking at his video closely I saw his 12900kf was throttling down to 3.43ghz under load and said it was Dell/Alienware that was to blame because of their shoddy design (had a few M1 users try to defend the results).

 

I just went and added a reply pointing to this video:

 

 

 

 

[electrosoft]

This is exactly why I was not happy with HWU not including DDR4-3800 for the 12900k in their 5800X3D but included DDR4 for the 5800X3D. Intel scales very well with DDR4 and DDR5 in performance while the 5800X3D barely scales at all in some games. WoW is one of those examples.

 

If every system is running low end, sloppy memory timings, the 5800X3D with that V-cache just powers through. Once you start tweaking and tightening up your memory, the 12900k climbs in leaps and bounds as well as the other CPUs (including the 5800x).

 

From worst config (3200) to absolutely pushed (3670, Bclk OC), the 5800X3D gains *1* fps.

On the other hand, the 12900k goes from 168fps to 217fps.

 

Still, the 5800X3D is a fantastic last hurrah for AM but shows that cache is hit or miss and compared to the 5800x which managed to gain 23fps (140->163fps). For the 5800X3D everything is just sitting in that cache, memory be damned.

 

We're also getting yet another glimpse into the future scaling of DDR5.

 

1080p slop memory all around:

 

 

 

Tighten up timings. push frequency:

 

 

Every chip tight and max frequency / config push:

 

 

 

[Tenoroon]

25 minutes ago, Reciever said:

 

First boot.

 

Decided to grab my Taichi 5700xt and strap 3x 45mmx92mm fans. I think they do 6k rpm's.

 

Went the AMD route for the perceived hope of an ease of use for Linux but also I cant find my x570 stock motherboard mounts so that system is down for now which means AMD will be migrated first.

I love the D15, glad to see someone else using one!

[Reciever]

1 hour ago, Tenoroon said:

I love the D15, glad to see someone else using one!

D15 w/ 2150rpm gentle typhoons holds 5.2Ghz @ mostly low 80's @ 255 watts with cheapo nt-h1 paste

 

Pretty happy with the setup so far @Mr. Fox

 

Ill do some tests via remote tomorrow while at work. As it stands it's lovely to be able to let a system flex a bit. 

 

Power saver let's plex do its a thing @ 5 watts 

[Mr. Fox]

1 hour ago, Reciever said:

D15 w/ 2150rpm gentle typhoons holds 5.2Ghz @ mostly low 80's @ 255 watts with cheapo nt-h1 paste

 

Pretty happy with the setup so far @Mr. Fox

 

Ill do some tests via remote tomorrow while at work. As it stands it's lovely to be able to let a system flex a bit. 

 

Power saver let's plex do its a thing @ 5 watts 

Awesome. It makes me super happy to hear that. Have you posted any photos yet (I haven't looked, but will).

 

Another thing I am happy about is I finally got rid of the blue 200mm front panel fans. I couldn't get any in white at the time I completed this build, but the ones I wanted came back in stock. Now it's all black with white lights, just how I like it.

 

[Mr. Fox]

3 hours ago, Reciever said:

 

First boot.

 

Decided to grab my Taichi 5700xt and strap 3x 45mmx92mm fans. I think they do 6k rpm's.

 

Went the AMD route for the perceived hope of an ease of use for Linux but also I cant find my x570 stock motherboard mounts so that system is down for now which means AMD will be migrated first.

Sweet. Looking forward to seeing more. 👍

 

Looks like you got one of those build-your-own square aluminum frame chassis, too.

[Reciever]

19 minutes ago, Mr. Fox said:

Sweet. Looking forward to seeing more. 👍

 

Looks like you got one of those build-your-own square aluminum frame chassis, too.

Yep! It has some shortcomings but for what I wanted it suits the bill quite nicely. I suppose I could bring Temps down with liquid metal, still have some left over. 

 

Next I need to alter something in the bios so that it retains the xmp profile and executes it. Right now it's defaulting to 2133, likely your old mem oc still in place. 

[Mr. Fox]

14 minutes ago, Reciever said:

Yep! It has some shortcomings but for what I wanted it suits the bill quite nicely. I suppose I could bring Temps down with liquid metal, still have some left over. 

 

Next I need to alter something in the bios so that it retains the xmp profile and executes it. Right now it's defaulting to 2133, likely your old mem oc still in place. 

Probably so. I normally do manual timings and memory clocks. If you go into the BIOS and change it to XMP it should override the manual memory overclock. You may have to set it to memory default, save and exit, then select XMP to get it to apply the XMP profile correctly.

[Rage Set]

If anyone else wants to give their higher end GPU's a workout, check out the UE5 Matrix City demo. You will have to actually download Unreal Engine 5 (I think there are some compiled unofficial standalone demos out there) and compile it yourself, but it a beautiful demo to say the least. 

[Mr. Fox]

Incoming... http://www.supercoolcomputer.com/en/product/39/direct-die-intel-gen12

7 hours ago, Rage Set said:

If anyone else wants to give their higher end GPU's a workout, check out the UE5 Matrix City demo. You will have to actually download Unreal Engine 5 (I think there are some compiled unofficial standalone demos out there) and compile it yourself, but it a beautiful demo to say the least. 

Thanks, bro. I will check it out.

[Mr. Fox]

12 hours ago, Rage Set said:

If anyone else wants to give their higher end GPU's a workout, check out the UE5 Matrix City demo. You will have to actually download Unreal Engine 5 (I think there are some compiled unofficial standalone demos out there) and compile it yourself, but it a beautiful demo to say the least. 

 

4 hours ago, Mr. Fox said:

Thanks, bro. I will check it out.

Looks like it is going to force me to install the latest DCH GeFarts driver filth in order to launch it. It's pitching a fit about me not using a DCH cancer driver.

 

Edit: Spend a fair amount of energy trying to figure it out and I can't get it to work right, so I will have to wait until it is in a distributable runtime format.

[Reciever]

What is a good 24/7 voltage to have on the 10850K?

[Papusan]

59 minutes ago, Reciever said:

What is a good 24/7 voltage to have on the 10850K?

Again, Intel's engineers have no problem with 1.4V if the cooling is adequate for their chips. See how far you can come with that target. If temps is too high then reduce voltage to match the cooling/clockspeed. See also Techpowerup's review talking about volt/clock speed for the overclock of 10850K.

https://www.techpowerup.com/review/intel-core-i9-10850k/20.html

[Reciever]

This is Fox's old bundle, so far small tests of 5.1Ghz @ 1.32v (according to Throttlestop) @ 195w and 5.2Ghz pulled 255w but I think that was auto settings.

 

It does 5.2Ghz at 1.38v pulling 211-215w. Tried lower but crashed the system lol

 

Early days

[electrosoft]

5 minutes ago, Reciever said:

This is Fox's old bundle, so far small tests of 5.1Ghz @ 1.32v (according to Throttlestop) @ 195w and 5.2Ghz pulled 255w but I think that was auto settings.

 

It does 5.2Ghz at 1.38v pulling 211-215w. Tried lower but crashed the system lol

 

Early days

 

 

That is a very above average 10850k. It either slipped past Intel QC or they needed more 10850k's than 10900k's and purposely dipped into the 10900k pool.

 

 

[Mr. Fox]

1 hour ago, Reciever said:

What is a good 24/7 voltage to have on the 10850K?

 

14 minutes ago, Reciever said:

This is Fox's old bundle, so far small tests of 5.1Ghz @ 1.32v (according to Throttlestop) @ 195w and 5.2Ghz pulled 255w but I think that was auto settings.

 

It does 5.2Ghz at 1.38v pulling 211-215w. Tried lower but crashed the system lol

 

Early days

 

8 minutes ago, electrosoft said:

That is a very above average 10850k. It either slipped past Intel QC or they needed more 10850k's than 10900k's and purposely dipped into the 10900k pool.

Yes, it is an extraordinary 10850K sample and that voltage 52x all core with 1.375-1.400V is right where it ought to be. You may be able to run a little bit lower voltage with lower core temps and higher LLC. If the LLC is too low the voltage will drop too much under load and crash. I would run the LLC at Level 6. The Level 4 (default OC for ASUS) allows too much vdroop.

 

Most 10850K won't even run at 52x, some use that much voltage at 50x all core.