Comments by "Mikko Rantalainen" (@MikkoRantalainen) on "Gamers Nexus" channel.

  1. I actually can compile Linux kernel, modify GPU driver, cook my food, change the oil of my car myself and I'd be still happy to buy a pre-built PC if it's upgradable in the future and worth the price. It seems pretty clear that this PC is not worth anything.
    6
  2.  @DarkAttack14  I totally agree that a correctly implemented heat pipe should depend on wicking only. My trust in average hardware manufacturer is low enough that I think the implementation quality should be validated and that's why I think at least one test done upside down would be warranted. If that doesn't show a difference, then you can be pretty sure that the wick implementation was actually correctly done.
    2
  3. The "80 Gold" rating does not mean that "1200 W" PSU can provide "0.8 * 1200 W" to the components. It will provide 1200 W to the components with 80+ % efficiency meaning it will eat max 1200 W / 0.8 = 1500 W from wall socket. The only thing you need to check is that the PSU is able to output enough amps to single rail + the manufacturer is good enough not to lie in specs. If your area suffers from brown outs, spend the extra money on UPS, not to more expensive PSU.
    1
  4. 21:30 I think the fans should be tested both with the methodology you explained here plus after attaching them to steel or aluminium case (round hole with the usual 4 mounting holes) using the vendor provided mounting material. That way the quality of the mounting hardware would also be tested. If the vendor didn't provide any mounting hardware it would be cool to have numbers for "let's use steel strews directly" vs "soft silicone fan mount pin" to have understanding about the range for practical installation (if you test the fan in round hole without a grill the results are still optimal compared to real world). As for how I would love to see fan performance reporting, I would like to have multiple graphs (one for 0.1 mmAq, 0.2 mmAq, 0.3 mmAq each) with noise (preferably dB(C) instead of dB(A)) on the horizontal axis and CFM on vertical axis. And if you only draw 3 graphs, you could probably put all in the same diagram.
    1
  5. Really interesting product but $1000 is too expensive for my taste. If they could manufacture something similar (with focus on functionality, I don't care about the looks) for around $250–300 would sound much better.
    1
  6. Nice test! I wish you would test all cases normalized to 25 or 30 dB under full CPU+GPU load among other tests. I think there's a market for case with no window, full padding on both sides, top and bottom and airflow focused front and back. That would optimize the airflow in combination with maximum high frequency sound isolation. Having filtered air input with high airflow is hard to do, though.
    1
  7. Great work! This is as close to scientific method as you can easily do with coolers. The only thing possibly missing is testing for the gravity effects (e.g. you could run the whole test bench upside down and see if it affects the performance of heat pipes – logically cooled liquid should fall downwards and if you have the system upside down, the liquid inside the heat pipes might never contact the CPU end of the heat pipe).
    1
  8. I would like you to test air cooler performance with different orientations for the cooler. As heat pipe designs (may?) depend on gravity, it should make a difference if the pipes are pointing upwards vs to the side. Most test benches have CPU socket pointing up and heat pipes pointing up which is the optimal orientation for heat pipes if I've understood correctly. However, in most cases the tower cooler will not end up in that orientation and it would be interesting to see how much a different that makes. If it turns out that heat pipe designs work equally well no matter the orientation of the pipes, that would be a nice finding, too.
    1
  9.  @davapod  I agree. I've since studied heat pipe technology more and it indeed doesn't mind gravity if executed correctly.
    1
  10. I'm hoping to create a sleeper PC and I would like to find something like this without any window. Is it really that rare to look for full mesh front with proper cooling but without a window?
    1
  11. Why not standardize on USB-C? If every fan and RGB light had USB-C connector, you could use practically unlimited amount simply by chaining enough USB-C hubs together. Plus running LCD displays over USB-C is well standardized. And a single USB-C connector can transfer up to 200 W if you go up to USB PD specs.
    1
  12. Am I the only one interested in worst case behavior? It seems that Gen4 is about 15-25 % faster than Gen3 in 0.1% LOW numbers with realtime ray tracing.
    1
  13. Scythe FUMA 2 with very low fan speed (like in less than 500 rpm instead of 1500+ rpm) seems like a better option. I'm still running Scythe Kabuto with 190 rpm (not a typo!) CPU fan in my current system (with system running light loads around 40 W power usage). The CPU fan will ramp up to 1300 rpm for higher loads, though.
    1
  14. I think passive cooling should be at least 65 W because that seems to be the target that AMD and Intel focus for mid-tier CPUs. If some not-too-expensive passive cooler can keep 65 W CPU parts cool enough, that would be very interesting product.
    1
  15. 23:40 Case this expensive should come with an extra metal panel with air holes so you could sacrifice the glass for more air.
    1
  16. Great summary of the AMD marketing. It seems like it would have been better for AMD to speak about automatic "turbo" or "boost" clocks instead of "Precision Boost", and "overboost" or "overclocking" instead of PBO.
    1
  17. Honestly, the marketing was so bad that I would avoid getting this product on principle. However, where I live, Zalman CNPS20X is 20 EUR cheaper than Noctua NH-D15 which makes it hard to pass considering pretty equal performance. And DeepCool ASSASSIN III is priced exactly half-way between those.
    1
  18. I found this video only now after all these years! Great content and you should consider wearing a shirt with "I 🖤 scientific method"! This video should be linked from the description of every heat pipe cooler you test!
    1
  19. I still cannot fathom why EVGA didn't change the connector when they decided to change the pin order. Molex connectors have square and non-square pin so it would make sense to have different set of shapes whenever the pin order changes. If I were to decide, there should be different shape for every voltage level and if the connector physically fits, the voltages are always correct. However, due historical reasons PSU manufacturers are using basically random pin order. The best solution would be to use identical connectors for both ends of modular cables so that SATA power cable would look the same for both ends etc and the end of cable you decide to use would be purely cosmetic matter.
    1
  20. 12:40 Transient spikes 2.5–3.0x the spec value are insane! How on Earth GPU manufacturers are not required to disclose the actual power requirements of their cards?
    1
  21. 17:00 I don't think the cross-over point for this graph means anything because both red and blue graphs in the diagram use different vertical axis so you can move cross-over point simply by changing scale of either axis. Looking at the maximum value of derivate of P-Q graph would make more sense. In the example here, that would be appear to be around 37 CFM and if I understand the graph correctly, this is the optimum point of the performance where even a slight reduction in the backpressure will increase the CFM rating the most.
    1