(Anti) SFF fun house

[tabascosauz]

It was a perfect fit for B550 Phantom Gaming-ITX/ax in my Silverstone ML-09B that I used with my 5950x likely because the NVMe heatsink doesn't stick out like other boards heatsinks and bump into the fan.
View attachment 345049

Also I was able to use the unused offset parts (see below) on my other B550 Phantom Gaming-ITX/ax in my NR200P that was using NH-C14S with 5800x. Noctua doesn't list it as compatible with NH-C14S but while the north plate uses the adjusted 7mm offset I discovered I just had to flip the south plate upside down to make up the longer distance of the NH-C14S mount points. This puts slightly more NH-C14S heat pipe area over the CCD's although a bit hangs over the chip.
View attachment 345048

Did you do any temp comparisons on offset/no offset?

 

[Selaya]

direct-die maybe? think noctua's got an am5 direct die kit?

 

[tabascosauz]

direct-die maybe? think noctua's got an am5 direct die kit?

Would be cool but it's a little hard to justify on a 7800x3d

 

[dgianstefani]

Would be cool but it's a little hard to justify on a 7800x3d

Mate

Succumb to the final iteration.

Go straight to the finish line.

There is no "justification" this is SFF enthusiasm

Although, my temps are excellent, and can tune harder because of it.

No need for traditional direct die with low wattage. Removing the IHS and TIM bottleneck and using Conductonaut Extreme with TG upgrade heatspreader achieves same goal.

These are temps while gaming, with multiple chrome tabs open, skype (for work), signal, steam, OCCT, and all the usual background stuff, plus two high refresh 1440p monitors. NVCP driver lock of 237 FPS which is why GPU isn't pegged at 100%. There's a bit of core delta because I use Process Lasso to set affinities, and have SMT off for latency and security.

CPU boosts to 5.15 GHz from BCLK and RAM is tuned to 56.6 ns too, so it's not like good temps because tuned low.

It's almost a shame because I'll be getting rid of this for Arrow Lake, since I'm tired of no granular control over clocks with Zen X3D. I do not need an algorithm deciding what performance I should get. Also the platform kinda mid and the efficiency sucks at anything except full load.

Chipset/VRM temps are fine too without those stupid little fans.

As is DDR5

 

[A Computer Guy]

Did you do any temp comparisons on offset/no offset?

I don't have any hard numbers recorded but generally the L12S with the offset mount seemed to perform slightly better (lower idle temps, faster cooldown) than NH-C14S (with no offset) in comparison with the same 5950x CPU despite having less mass and less heat pipes. However the comparison is very apples and oranges because of other differences too such as case configuration (ML09B vs. NR200P), fan types, and fan airflow (intake vs. exhaust). Noctua says it's only going to be a 1c to 3c difference at most anyway so I wasn't expecting much. If I wasn't observing temps in Ryzen Master I probably wouldn't notice.

So far NH-C14S with the offset using 5800x in the NR200P seems a little better too (lower idle temps, faster cooldown). Since I'm working in that system I think I might run some additional testing and record some hard numbers with and without offset when I get a chance before I swap it over to this summers custom loop project.

 

[tabascosauz]

I don't have any hard numbers recorded but generally the L12S with the offset mount seemed to perform slightly better (lower idle temps, faster cooldown) than NH-C14S (with no offset) in comparison with the same 5950x CPU despite having less mass and less heat pipes. However the comparison is very apples and oranges because of other differences too such as case configuration (ML09B vs. NR200P), fan types, and fan airflow (intake vs. exhaust). Noctua says it's only going to be a 1c to 3c difference at most anyway so I wasn't expecting much. If I wasn't observing temps in Ryzen Master I probably wouldn't notice.

So far NH-C14S with the offset using 5800x in the NR200P seems a little better too (lower idle temps, faster cooldown). Since I'm working in that system I think I might run some additional testing and record some hard numbers with and without offset when I get a chance before I swap it over to this summers custom loop project.

Heatpipe orientation is different on the two downdraft coolers so idk if they would respond differently. C14S orientation is usually disadvantaged for 2CCD, but it's surprising to hear you say that L12S is better, since on APUs the C14S is very confidently ahead.

I wanna say the U12A has a bigger cold plate than most other Noctua coolers, but it might be my eyes playing tricks on me.