Log FormD T1 meets MinisForum BD790i! (Build Log & benchmarks)

[MarcParis]

-------------------------------------------------------------------------------------------------------------------------------
FormD T1 : SSD Heatsink trial (22th July 2025)
-------------------------------------------------------------------------------------------------------------------------------
Here is another trial inside the FormD T1, while using my Core i5 12600K setup.

It is time to try some heatsink on my verso SSD. By default, its idle temperature is not that terrible, i.e. 60°C while gaming (heat coming from the RTX 5090 FE). It is still safe, but as there is an offset with the riser, thanks to the 2-slots GPU, why not using this empty space to test some SSD heatsink.

First candidate will be the a Thermalright model, with bottom & top heatsinks :

Despite its size it is fitting nicely behind the riser. There is no clearance issue and I was able to put back the RTX 5090 FE…however I was unable to start the computer.

After some troubleshooting, I found the root cause : the bottom plate of the heatsink is making contact with the motherboard, creating some shortcuts. Hopefully the motherboard detected them and refused to power up.

Conclusion? It is suited to the recto SSD slot, not verso!

Second trial then.
I used the same heatsink as on my U-ITX setup : 3mm tall, held by rubbers.
It is also an easy fit!

PS : Note I found the best way to assemble my RTX 5090 FE inside the FormD T1. Indeed it can be tricky to put the entire screw through the GPU bracket. Finally the best way is simply to unscrew partially (you can see on above photo), and then to rotate the GPU bracket. Very simple & effective!

This time, no booting issue!

Time to test!
Best trial is a simple 3DMark Speedway torture test with a stock RTX 5090 FE (575W)!

The result is without surprise : -5°C versus without the heatsink…basically the same result as the U-ITX..

One interesting point is that the SSD temperature rose up to 58°C (and decreased quickly) after the benchmark finished…the GPU airflow (in/out) is finally well used by the heatsink..

PS : I’ve moved back my RTX 5090 FE and the SF1000 inside the Cerberus-X in order to define a truly stable undervolt of the RTX 5090 FE…but not only as I suppose that the remaining instabilities are coming from either my undervolted 12600K, XMP profile or the Nvidia drivers. The issue is that I need RTX 5090 FE to be fully utilized…and the R7 9800X3D is way better at this task. Based on my first testing, I measured a gap of 30%, even at 5120x1440 in Stellar Blade, especially in open areas.


More to come…

 

[MarcParis]

-------------------------------------------------------------------------------------------------------------------------------
FormD T1 : System Stability (at last) & 6mm SSD Heatsink & comparison vs my Cerberus-X (27th July 2025)
-------------------------------------------------------------------------------------------------------------------------------

Debugging the instability!

My Cerberus-X setup with its R7 9800X3D proved very helpful as it allowed me to find the proper undervolting parameters for my RTX 5090 FE! So far no more crashes due the GPU in Stellar Blade, either in the Cerberus-X or the FormD T1.

However, Stellar Blade was very stable on my R7 9800X3D, but it was not the case with my FormD T1. After thinking, I supposed I could use the error message shared by the Unreal Engine (I got them on either Expedition 33 : Clair Obscur or Stellar Blade). Finally games using the Unreal Engine could be very useful to debug an unstable PC.

Here is my understanding (so far) on the Unreal Engine Debugger :

• Error Message : “FATAL ERROR”
  • This is translating an unstable computer, mostly on the CPU.
  • I got this message on both my R7 9800X3D (Curve Shaper -30/-10) and my Core i5 12600K (undervolt -0.085v). By reducing or removing the undervolt, instability disappeared!
• Error Message : “Lost Display….long message”
  • This is translating an unstable computer due to the GPU.
  • Once again, it appeared on both of my systems. By increasing slightly the tension (from 900mV to 910mV) removed the message on both systems!
• Error Message : None
  • The Unreal Engine debugger is not catching the error. I suppose it could be linked to either a Game Bug (if repeatable) or RAM stability issues (random)
  • Keep in mind, it is only my guess as the system was unable to catch the error nature.

Here is a long gameplay capture on my FormD T1, perfect!


6mm-height SSD heatsink

Very quickly I also tested 6mm-height SSD heatsink instead of my initial test with 3mm-height on my verso SSD :

Does it help? Quick answer is Yes by a 2°C additional decrease on the SSD temperature (using 3DMark Speedway torture test).
Here is a summary of the temperatures on the back/verso SSD, while heavy GPU load (575W) :

• Without heatsink : 60°C
• With a 3mm-height SSD heatsink : 55°C
• With a 6mm-height SSD heatsink : 53°C

FormD T1 vs Cerberus-X, initial comparison

After one month of intense use, it is time to make a first comparison between my beloved Cerberus-X and the FormD T1.

• AIRFLOW / COOLING / NOISE :
  • FormD T1 : Dual Chambers (CPU & GPU) design, herited from the Dan A4 case, is very, very effective, especially with the RTX 5090 FE 2-Slots passthrough design. Leaving a gap of 20 mm between the GPU and the motherboard/PSU tray is great for cooling as it allows an excellent airflow inside the case, without overheating the motherboard/PSU!
    With a 3-slots GPU, the case airflow is too restricted (based on my trial of the RTX 4080 FE).
    The 2 top exhaust 120 mm fans (Corsair RS120 Max + Noctua Offset Grid) are very effective.
    Overall the FormD T1, with my setup, can handle a stock RTX 5090 FE (575W), even if an undervolt (510W, ie 0.895mv) is a better compromise performance/noise for me.
  • Cerberus-X : With my latest cooling setup (dual fans bracket just in front of the GPU), the Cerberus-X is keeping the crown of the GPU cooling, either on gaming load & on idle (keeping the GPU fans inactive).
    My only issue remains the relatively poor (below my expectations) performance of the Noctua C14S / D12L on a R7 9800X3D. I got very little improvements by increasing the fan speed from 1000 rpm to 1500 rpm (1-2°C improvement on the CPU core). There is clearly a bottleneck on the heat dissipation of the R7 9800X3D. Neither Noctua nor the Cerberus-X are to blame here.
• MOTHERBOARD
  • FormD T1 : Mini-ITX motherboards have been a long topic of frustration for me, i.e. poor longevity, i/o, aso. I must admit, since 2024, the average level of mini ITX motherboards is improving, but slowly. On the AM5 side, B850 models are pretty decent to good (like Asus B850i or MSI B850i). But so far X870 models are pretty dull (Asus or Gigabyte). I’m putting high hopes in the future MSI X870i with its 4 m.2 slots, but let’s see.
  • Cerberus-X : Simply the best as it is compatible with ATX motherboards.
• MAINTENANCE
  • FormD T1 : It is a very important point for me. Well the FormD T1 is very easy to maintain. Thanks to its design, it is possible to open it while still powered/plugged in. Its tool-less side panels are a must.
  • Cerberus-X : It is still my reference for maintenance. Every maintenance operation is easy & fast. It is simply the best!
• EVOLUTION
  • FormD T1 : Simply the best. The FormD T1 managed to evolve overtime, while keeping its initial overall dimensions (10L). I loved its modular (Lego) & smart design. The only drawback I find is the bad handling manual.
  • Cerberus-X : My biggest criticism of the Cerberus-X would be its almost complete lack of evolutions over the last 8 years. This is a common criticism with other great Sliger cases like the SM550/560/570/580 or the S610/620. They were ahead at their release date but now they are outdated versus competition.

As a conclusion, my FormD T1 has managed to divert me from my Cerberus-X. The U-ITX managed the same but fell short due to some points, one of them was the mini-ITX motherboards versus ATX ones, especially for m.2 slots. Depending on the MSI X870i mini-ITX MB, I might consider switching my main setup to the FormD T1 as it is working very well with a RTX 5090 FE, undervolted or not.
Meanwhile I’m also frustrated with my R7 9800X3D. It is very true that it is faster than my i5 12600K, but it is also way harder to cool, despite using less power (W).

So let’s see…

 

[MarcParis]

For information I've just stopped the trial of the FormD T1 with a 12600K & a RTX 5090 FE. Overall, it is a viable combo with some standard drawbacks (mini ITX Form Factor, hotter temperatures & louder overall) but nothing critical/prohibitive.

For now I'm back to my Cerberus-X as I have an interesting experimentation to do...

Stay Tuned !

 

[MarcParis]

Well, well, as I just finished Stellar Blade for the 3rd time (4th run ongoing) and I don't have just yet another game to platinum, I'm considering to solve those instabilities issues met on my 12600K mini ITX.
After some thoughts I guess the issue is linked to the DDR5 speed, ie 6000MTS. Alder lake is only guaranted up to 4800MTS with DDR5 while Raptor Lake is 5600MTS. Thus I suppose those random issues I faced on Stellar Blade are linked to my DDR5, especially as Stellar Blade is pretty gourmand at RAM (~22-23Go). Issue is not the RAM itself but the RAM controller I would say.

I will let you know!

 

[MarcParis]

What is happening to FormD T1 (Ncase) : Why are they changing the orientation of their gen 5 riser??! This making the T1 no longer compatible with RTX 5080/5090 FE models!