This is probably the smallest PC one can build with a graphics card. At just 4.5L, it’s insanely small and easily fits in a backpack (weighs 8lbs fully built).
I know technically Velka 3 case is smaller than this, but this looks better, is sturdier, is much better quality and can fit a Nvidia 5000 series card.
Of course, at this size the usual GPUs and SFX PSUs won’t fit in here. The 5060 is a low profile GPU and using a Flex 1U PSU of 500W.
Cooling isn’t bad at all (Noctua cooler lives up to its name) and benchmarks are decent too.
Bravo man that build is tight!
(pun intended)
I know you do this as a hobby and to challenge your skills. but, for me at least once you get down to that size I think I would just buy a MacMini and call it a day. I thought you used a 5050 but a low profile 5060 is smart and makes it a little more useful. Great Job!
Playing “Devil’s Advocate” here for a moment - why did you choose to go with the 9600X over something like a Minisforum ITX board? They even have an X3D version now. Wouldn’t that have made more sense for this size of a build for both performance and thermal reasons?
Agreed. I love mini PCs and Mac mini myself. This is just better for gaming while being small enough to be carried in a backpack for LAN parties (if that’s still a thing )
That is a compelling alternative. However, this part list costs $150 less and since 9600X is a desktop CPU and 7945HX3D is “laptop” CPU, the unrestricted TDP may even push 9600X ahead for lower cost.
An interesting take! The 7945HX3D’s TDP is 55W, against the 9600X with 65W. It’s doubtful the TDP restriction on the 7945HX3D would cause it to lose out in any gaming scenario - the 3D V-Cache would easily turn the tables. Additionally, the 16 physical cores would handily trounce the 9600X in productivity. You’re absolutely right about the cost being higher.
Games typically don’t benefit from a higher core count. A 3D V Cache certainly does. But also a higher base clock speed of 9600X.
I am guessing that’s the reason why 9600X did slightly better in gaming benchmarks I linked above. Most games are just not using more than 6 cores and each core is faster in 9600X.
We’re taking those Passmark results with the world’s most giant grain of salt for multiple reasons. We found some fatal flaws - some of the positioning makes no sense, such as:
13900KF positioned hundreds and hundreds of points behind a regular 13900K, when the 14900KF and 14900K are separated by only 38 points
We tend to use information we gather from reviews. Some of our favorite YouTube channels to gather this information from include Hardware Unboxed, Gamers Nexus, and JayzTwoCents, they tend to run many benchmarks and provide an accurate overall picture.
Thanks! During gaming, the GPU is typically 75-80C (100% load) and the CPU is at 65C (25-35% load). A little on the hotter side, but nowhere close to throttling.
This is incredibly cool! I have always built in mid-tower/full towers, but there is something super appealing about building high performance machines in tiny little cases. I recently ordered a MoraIV and I’ve seriously considered doing a custom loop in a tiny case then hook it up to the big rad. I’m always so impressed with these types of builds. This one looks so clean. Well done!!!
Connecting a mini ITX custom loop to an external radiator seems like an intriguing idea. I recently built my first custom water loop in a somewhat small ATX case (My first custom waterloop build) and have been thinking about doing a custom loop in a mini ITX case like Lian Li Dan A4H2O.
)
