Jump to content
Search In
  • More options...
Find results that contain...
Find results in...

bowrilla

Member
  • Content Count

    1,521
  • Joined

  • Last visited

Reputation Activity

  1. Like
    bowrilla got a reaction from GhostRoadieBL in AIO vs Custom loop temp theory   
    It has little influence. As long as it flows and is enough to overcome some restrictions like jet plates (if your block has these) then it's basically fine. 5-7K more or less is not a big thing unless you want to push to the absolute limit.
     
    I did my own testing (because der8auer's data raises some questions; his flow rate is unexplainably low at all times for the kind of pump he has, the reading must be either wrong or has some unit mishap, it's not GPM though and GPH is unlikely either) and the results (for my loop) are pretty obvious: flowrate has minimal influence. Between 100% pump speed and 25% pump speed (range of ~4x the flow rate between 168.9l and 41.5l) the maximum corrected difference was ~6K at 100% fan speed. Dropping the fan speed by half to 50% reduced difference to 3K. 
     
    Some random Windows process popping up in the background and hogging some resources sometimes has a bigger difference. Keep in mind, my data was the result of 30mins Prime95 runs. A very very unrealistic load. Under every day loads the difference will most likely be smaller.
     
    See data and discussion in this thread:
     
  2. Agree
    bowrilla got a reaction from Opossum in Help with CSS Responsive layout   
    "Ctrl + Shift + I" and then "Ctrl + Shift + M" ... welcome to Chrome's (and Firefox's) dev tools to simulate other resolutions, devices and aspect ratios.
  3. Agree
    bowrilla reacted to For Science! in Transient Hot Wire Thermal Conductivity Testing of Fluids   
    To what extent is the thermal conductivity of the liquid important? I ask because my understanding is that the pump and flow is what is primarily the way in which energy is transferred from A to B (I.e. water block to radiator), and by far the rate limiting step is heat being removed from the radiator into ambient air. 
     
    Is there a big bottleneck in the heat transfer between nickel (waterb lock)and the coolant, and then coolant to brass/copper (radiator).
     
    While the numbers are appreciated, ultimately what it looks like is that they are all give or take similar to water, and so some more context for the general reader would be appreciated.
     
    What do these numbers mean and what is an impactful difference. what is the significance of a high delta - did it just get nuked more (higher current?) or is it a reflection of some property of the coolant.
  4. Agree
    bowrilla got a reaction from Mark Kaine in Ryzen 5950x achieving rated speed of 4.9 Ghz seems to be a longshot   
    Nothing of that is abnormal. This sounds like exactly the behaviour AMD intended for the Zen2 (and beyond) architecture. Fluctuating idle temps on the higher side are normal. Max boost on only a single core for only very short bursts is normal. Lower all core boosts are normal. Overall higher temps on a 16 (!!!) core CPU are normal.
  5. Informative
    bowrilla got a reaction from Stahlmann in standard push VS. slim push/pull on radiators   
    I linked you the long elaborate radiator test at xtremerigs (even though its old). Slim rads (especially the GTS series) do not benefit much from push/pull. However you will double the fans resulting in more sources of noise.
     
    Since you want low noise, you'll be running fans at lower fan speeds. At 750rpm the difference between push and push/pull for the 360mm GTS radiators is ~14W/10KΔT. So push/pull gives you ~7% more cooling but you have double the noise sources. I sincerely doubt 7% lower rpm (if it was all linear which it isn't) can make up for having double the fans. On the Alphacool Monsta it's ~22%.
     
    About the same results (checked on only a few examples) at mid rpm and even at full blast. What we can see however is that thicker rads scale better with revs. From 750rpm to 1850rpm you'd gain ~89% more performance, the Alphacool Monsta gains over 100% in performance.
     
    Again, you aim for low noise first and best performance second. That means the less fans the better.
     
     
  6. Like
    bowrilla got a reaction from Chrismike in Help with gpu liquid cooling build price   
    Not sure how the US handles this but in Europe duties and taxes may apply to orders from AliExpress and that might just push the price into the realms of components from EK, Watercool and Corsair.
     
    Pump res and block are not that much cheaper, fittings aren't either as long as you go for the EK classic line or similar (Alphacool is also interesting, so might be XSPC). The ball valve actually seems pretty expensive even but then the radiators are dirt cheap.
     
    Oh and to be honest: you're already paying that much that I'd just add teh CPU. CPU blocks even from EK or Watercool aren't that expensive and 2 fittings aren't either.
     
    I'd also be careful with those displays on terminals or inline sensors: to my knowledge burn in can often happen easily - and the information aren't useful on a display but on a fan and pump controller that regulates fan speed according to coolant temps.
  7. Agree
    bowrilla reacted to For Science! in PWM Pump control - not working?   
    Not so familiar with armory crate unfortunately. It's a bit weird with the pump header afaik defaults to 100% constantly and you have to go out of your way to activate a curve for it....so for it to default to a fan curve makes me think you have some software installed that is changing the settings in Windows. I guess for example if you were to boot into safe mode, do you see the same effecT?
  8. Agree
    bowrilla got a reaction from Blai5e in 'High' Coolant Temps - Rads/ Fans not enough or something else?   
    Those little spinners are worth nothing. All they show is that there is some flow. You cannot visually determine how much and even little flow will spin it fast enough for you not to be able to see anything. 
    30% is not much flow but it should not have that much if an impact on the overall temps, especially since it's a D5.
     
    PBO is pretty aggressive with the voltage. That will obviously increase heat output but it should not lead to those high coolant temps.
     
    That depends on flow. The lower the flow the higher the heat gradient. Ideally you'd have 2 sensors: one directly in front of the radiators and one directly after. Then you can see that once you drop flow far enough the heat gradient grows probably to values in the range of 5-8K. If flow is sufficiently high the heat gradient shrinks to about 1K. At that point more flow does not give you any significant advantage but it will increase noise. With the blocks it's similar: high flow means lower heat gradient between inlet and outlet ports, low flow means higher heat gradient.
     
    Considering you have 2 360 radiators, coolant temps well beyond 40°C (almost 50°C) however indicate something else is the culprit because flow alone should not play such a big of a role. This means that there's probably nothing majorly wrong with the heat transfer to the coolant. The issue is rooted with dumping the heat energy into the air.
     
    Of course, temp sensors can varry in precision and small production differences can lead to misreadings (like more or less epoxy, 1-2mm variance in probe position. The probe itself is also not calibrated. I have 2 sensors (one XSPC plug type and one in my aquacomputer MPS flowmeter) and they usually are 1-2K apart. 
     
    Taking all of that into account your reading might be 4-6K higher than the actual average coolant temps considering you're measuring directly at a heat source, potentially lower than ideal flow and variances in sensor accuracy.
     
    I'd probably do the same as @Stahlmannsuggested and go for the Phanteks Halos frames. Personally I prefer the look of dark blades with an led ring around over white blades with everything being lit.
     
    Whatever you do though, I'd use push pull configs on radiators >45mm thickness. Which is why I wouldn't use rads that thick. Their performance benefits can only be achieved with more fans and high revs. The best 30mm radiators (HWlabs GTS) match the performance of midrange 45mm rads even in push pull and match the performance of midrange 56mm in push or pull (iirc). Even the best >56mm radiators need push pull to offer significant advantages over the best 30mm radiators. I rather spend 20 bucks more on the best rad but save the expenses for 2-3 fans per radiator and have ultimately less noise sources with less overall noise.
    Well, first off I'd unplug the side fans that you think create too much positive pressure. I doubt it but check that first.
     
    Next is setting the lowest pump setting to something like 40-50% which should not change anything noticeable in regards to noise.
     
    If that doesn't help: tilting and shaking to make sure everything is properly bled. 
     
    Then go to the radiator fans. If you have more fans around, through them on for a push pull config. That should improve performance. You don't need to screw them in just laying them on top should be enough as long as there are no huge gaps. Not ideal but should work.
     
     
  9. Informative
    bowrilla got a reaction from Stahlmann in 'High' Coolant Temps - Rads/ Fans not enough or something else?   
    Those little spinners are worth nothing. All they show is that there is some flow. You cannot visually determine how much and even little flow will spin it fast enough for you not to be able to see anything. 
    30% is not much flow but it should not have that much if an impact on the overall temps, especially since it's a D5.
     
    PBO is pretty aggressive with the voltage. That will obviously increase heat output but it should not lead to those high coolant temps.
     
    That depends on flow. The lower the flow the higher the heat gradient. Ideally you'd have 2 sensors: one directly in front of the radiators and one directly after. Then you can see that once you drop flow far enough the heat gradient grows probably to values in the range of 5-8K. If flow is sufficiently high the heat gradient shrinks to about 1K. At that point more flow does not give you any significant advantage but it will increase noise. With the blocks it's similar: high flow means lower heat gradient between inlet and outlet ports, low flow means higher heat gradient.
     
    Considering you have 2 360 radiators, coolant temps well beyond 40°C (almost 50°C) however indicate something else is the culprit because flow alone should not play such a big of a role. This means that there's probably nothing majorly wrong with the heat transfer to the coolant. The issue is rooted with dumping the heat energy into the air.
     
    Of course, temp sensors can varry in precision and small production differences can lead to misreadings (like more or less epoxy, 1-2mm variance in probe position. The probe itself is also not calibrated. I have 2 sensors (one XSPC plug type and one in my aquacomputer MPS flowmeter) and they usually are 1-2K apart. 
     
    Taking all of that into account your reading might be 4-6K higher than the actual average coolant temps considering you're measuring directly at a heat source, potentially lower than ideal flow and variances in sensor accuracy.
     
    I'd probably do the same as @Stahlmannsuggested and go for the Phanteks Halos frames. Personally I prefer the look of dark blades with an led ring around over white blades with everything being lit.
     
    Whatever you do though, I'd use push pull configs on radiators >45mm thickness. Which is why I wouldn't use rads that thick. Their performance benefits can only be achieved with more fans and high revs. The best 30mm radiators (HWlabs GTS) match the performance of midrange 45mm rads even in push pull and match the performance of midrange 56mm in push or pull (iirc). Even the best >56mm radiators need push pull to offer significant advantages over the best 30mm radiators. I rather spend 20 bucks more on the best rad but save the expenses for 2-3 fans per radiator and have ultimately less noise sources with less overall noise.
    Well, first off I'd unplug the side fans that you think create too much positive pressure. I doubt it but check that first.
     
    Next is setting the lowest pump setting to something like 40-50% which should not change anything noticeable in regards to noise.
     
    If that doesn't help: tilting and shaking to make sure everything is properly bled. 
     
    Then go to the radiator fans. If you have more fans around, through them on for a push pull config. That should improve performance. You don't need to screw them in just laying them on top should be enough as long as there are no huge gaps. Not ideal but should work.
     
     
  10. Agree
    bowrilla got a reaction from SpiderMan in ASUS HYPER M.2 EXPANSION CARD   
    I'm not sure I can follow you. Ryzen has indeed 24 PCIe gen4 lanes: 16 for the main/first PCIe Slot (GPU), 4 for the first/primary M.2 slot and 4 for the chipset. The chipset can and does provide additional lanes but the uplink to the CPU is limited by the 4 lines the chipset is connected with. All chipset features (WiFi, ethernet) and ports (additional USB, SATA, additional M.2) go through that and share the bandwidth. That means that the 2nd M.2 slot will share bandwidth with those features and ports resulting in potentially non-ideal performance (depends on what features are used/active and how many ports are being used). The chipset also provides additional lanes for the additional PCIe slots on the board (this is why the GPU always goes into the main/primary/first slot which is (usually) connected to the CPU directly).
     
    Bottom line: 2 M.2 drives with Ryzen might (the chipset uplink is PCIe gen4 after all so those 4 lanes have almost the bandwidth of 8 gen3 lanes) result in somewhat impeded performance for the 2nd M.2 but it will work. 
     
    Since the Asus hyper M.2 expansion plugs into PCIe slots it is still limited by the chipset's uplink of 4 lanes. If you plug in 4 nvme drives into that expansion card, you will at best cut their performance in half since you'd need 16 gen3 lanes but the chipset is only connected through 4 gen4 lanes which roughly equates to the bandwidth of 8 gen3 lanes. USB, sound, ethernet on top an performance will most likely drop below those 50% theoretical peak.
     
    Since you only use a 1650 which definitely cannot saturate 16 gen4 lanes (probably not even 16 gen3 lanes) you could bifurcate that first PCIe port. I'm not sure about that specific mainboard and if this has an onboard option to split the lanes into 2x8 between first and second PCIe slots but you can use a special gen4 riser cable with a PCB and some PCIe switching capability to split each gen4 lane's bandwidth into 2 gebe lanes. Not sure if that exists as a mass market product.
     
    I'd say: 2x2TB or 1x4TB is a wiser choice.
  11. Like
    bowrilla got a reaction from chaz69 in VRM grease fried gpu   
    How do you know it's the grease that killed it? Could be many things. The oils in thermal pads should be non conductive but of course dust and dirt can change this. 
     
    Thermal pads as far as I know are being soaked with these kind of oils to actually make them work as thermal pads. Once they dry out (or oil out) they become less heat conductive to the point of useless eventually. Again: the oils should not be electrically conductive.
     
    What is more likely is that one of those mosfets was not of great quality and might just have popped. A 1080FTW is several years old. It's rare for chips these days to die after years of usage (it's more likely for caps to pop) but it can happen. It could also be a bad solder joint that aged badly increasing resistance resulting in more heat.
  12. Informative
    bowrilla got a reaction from Ali90 in optimum airflow be quiet pure base 500   
    tl;dr
    Both intake or exhaust work, the differences are not huge. The best orientation depends on the specific system and the components in it.
     
    Longer explanation:
     
    AiO as intake makes heat transfer from AiO to air more efficient (due to larger heat gradient) but it also keeps warmer air inside the case. That might increase the internal temps by a few K and it will increase component temps of all air cooled components. So the GPU might run a bit hotter. We're talking about a handful of K.
     
    However, the amount of heat inside of the case will be comparable to an all air cooled system. Exhaust fans become more important in this scenario and so does good airflow. If the AiO is set as exhaust then the case temps will be a bit lower resulting in slightly lower GPU temps (and everything else that is directly air cooled) but due to the fact that the air inside the case is hotter than ambient air, the heat gradient between the water in your AiO and the air going through it (now the internal air volume of the case) shrinks which will result in slightly higher water temps and therefore higher CPU temps.
     
    Cooling means moving around heat energy and the larger the difference between two energy states is the more efficient and therefore quickly the transfer happens.
  13. Informative
    bowrilla got a reaction from Newblesse Obblige in Is using a MDF board or a thin Plywood bad for SFF case modding a good idea?   
    Everything under 8mm will probably be too wobbly and not enough material to mount things together. As I've said, MDF or any form of wood material is probably not the best solution for an SFF case project.
  14. Informative
    bowrilla got a reaction from Newblesse Obblige in Is using a MDF board or a thin Plywood bad for SFF case modding a good idea?   
    It's fine, you can do it. However it might not be a good material for your application since you need more material thickness in order to be strong enough. That means fan cutouts will need to be entirely open or very restrictive. When talking about extreme SFF builds, every mm counts so by having >12mm of material all around adds up compared to 1mm. 
  15. Informative
    bowrilla got a reaction from LOST TALE in Safe voltage for ddr4   
    1.4V is perfectly fine and up to 1.45V is usually no problem. I have yet to encounter some DIMM modules that do not have temp sensors, however those sensor data is not always shown in every tool. HWInfo64 is to my knowledge the best you can use supporting almost all sensors available. 
  16. Agree
    bowrilla reacted to WereCat in Safe voltage for ddr4   
    1.36V is perfectly safe.
    1.4V to 1.45V is safe for most DDR4 DIMMs but some can go as high as 1.5V to 1.55V for daily use. At that point you need an active airflow over the RAM sticks though.
  17. Like
    bowrilla got a reaction from rockyroller in Extra cooling additives   
    The bottleneck is not the water, it's the small area of the die that's producing heat, then going to the heatspreader (which is inefficient) and then going to the waterblock (which is inefficient) and then going to the water (also inefficient). The more media the heat energy needs to go through the less efficient it is.
     
    The waterblocks with their micro fin structure are a solution to increase the surface at least between block and water. But there's still too little surface area between die and IHS and between IHS and waterblock. 
     
    Most custom loops run water temps ~30°C +- a handful. That's maybe a ΔT of ~10K over ambient. 10% difference would be 1K. The result on your CPU or GPU temps would most likely be around the margin of error.
     
    Car or motorcycle engines run their coolants A LOT hotter and they have A LOT more surface area with A LOT LESS media changes.
  18. Agree
    bowrilla got a reaction from Jeppes in IMPORTANT PSA FOR WRAITH PRISM USERS   
    To believe is not to prove. Do you have any data (i.e. from HWInfo64 or similar programs) that demonstrate a severe raise in temperatures on the DIMM modules? 
     
    Whatever it is you're doing, I'd expec tthe 3900X to overheat before the cooler could actually transfer enough heat over the air to the DIMM modules for those to overheat and take damage. JEDEC defines regular operating temps of memory modules to be <=85°C. Up to 95°C operation at half refresh rates is safely possible. Beyond that there is significant risk of damage. However, in order to heat up the chips to beyond 95°C you'd definitely need air that is hotter than that hitting the modules and that would mean CPU temps WELL BEYOND 100°C.
     
    So without any solid data to make your case I very much doubt that the Wraith cooler has anything to do with your failed memory and the fact that the module closest to the cooler died is most likely pure coincidence.
     
    P.S.: the JEDEC specs obviously do not say anything about stability of faster than spec memory. Instability and errors however do not mean damage.
  19. Agree
    bowrilla got a reaction from SavageNeo in IMPORTANT PSA FOR WRAITH PRISM USERS   
    To believe is not to prove. Do you have any data (i.e. from HWInfo64 or similar programs) that demonstrate a severe raise in temperatures on the DIMM modules? 
     
    Whatever it is you're doing, I'd expec tthe 3900X to overheat before the cooler could actually transfer enough heat over the air to the DIMM modules for those to overheat and take damage. JEDEC defines regular operating temps of memory modules to be <=85°C. Up to 95°C operation at half refresh rates is safely possible. Beyond that there is significant risk of damage. However, in order to heat up the chips to beyond 95°C you'd definitely need air that is hotter than that hitting the modules and that would mean CPU temps WELL BEYOND 100°C.
     
    So without any solid data to make your case I very much doubt that the Wraith cooler has anything to do with your failed memory and the fact that the module closest to the cooler died is most likely pure coincidence.
     
    P.S.: the JEDEC specs obviously do not say anything about stability of faster than spec memory. Instability and errors however do not mean damage.
  20. Informative
    bowrilla got a reaction from Jason 57 in Maximum number of applications a CPU can support?   
    If you really want to understand how computers work: https://www.youtube.com/c/BenEater/videos
     
    But be warned. That's a deep dive.
  21. Agree
    bowrilla got a reaction from -rascal- in What fittings do I need from EKWB's Quantum lineup to add a fill port and a drain port?   
    Both Alphacool and HWLabs have those. They're coold x-flow. Alphacool also has slimmer radiators with multi-port designs (makes filling, draining and bleeding A LOT easier when you have top and/or bottom mounted radiators). I think the Alphacool ST30 are the slimmest multi-port rads out there (note: only v2). The ST30 X-Flow comes also with 4 G1/4" ports so. However, the HWLabs radiators are better in terms of performance.
     
    Since the P360 is not multi-port and not x-flow, front mounted radiator it is then, since filling and bleeding a top mounted rad would be a lot more tricky without a multi-port radiator. Question would be now: ports up or down. Ports at the top means bleeding is super easy, but draining isn't (requires tilting obviously on its back or top). Ports at the bottom makes bleeding trickier (you will have some air bubble at the end tank probably but not in the fin area). Filling and draining is fairly straight forwards though. If the top of you reservoir is below the end tank of the radiator, filling through the reservoir might be tricky.
     
    I agree with what @-rascal- suggested. Ports down with a T-splitter and a ball valve. Pump res either at the rear fan mount or mounted higher up at the front (it might block a bit of airflow but not by that much). 
     
    With an x-flow radiator you'd have more freedom in terms of pump-res placement, since you'd add the same T-splitter and a ball valve at the top end tank of the radiator. 
  22. Like
    bowrilla got a reaction from 1982 Original in Anyone know the difference between these fans?   
    Well, I love the P12 and P14 fans, but the Noctua fans are better. They also have different bearings for example. Price/Performance-wise the Arctic are unbeatable though.
  23. Like
    bowrilla got a reaction from 1982 Original in NZXT Aer 120mm vs Noctua NF-A12x25   
    PWM PST CO version. Yep, saw the 15mm as well, interesting for small builds.

    I agree that the Silent Wings 3 (regular and High Speed both 19€+) are objectively a bit better than the P12 and P14, the Pure Wings though have a sleeve bearing. That's enough to disqualify them in my book. NF-A12x25 PWM 12V is ~28€+, NF-F12 PWM is ~18€+.
     
    But we're getting OT here imho. Let's agree to disagree. 
  24. Like
    bowrilla got a reaction from phoon in PC crashing in Prime95   
    Perfectly fine.
  25. Like
    bowrilla got a reaction from 1982 Original in NZXT Aer 120mm vs Noctua NF-A12x25   
    Same. I have all of them (and more) here, used them, tested them. Let's agree to disagree then.
×