for every day use you must not setup more than 80% fan speed all factorys have a standar use about 80% of the maximum performance of gpu , if you can keep it cool the gpu at 80% fan spped then the moding works
With the card you have, it probably won't matter, but I wouldn't plug and unplug the power cable more often than necessary. The contacts are only rated for a limited amount of plug in events. Also one should check the temperature of a self-made plug after a while of intensive use to see whether there is any unnecessary resistance due to a suboptimal crimp. Again, this does not matter for graphics cards that aren't very power hungry, but cards like the 4090 may draw a lot of amps over their 12 pin connectors.
You should absolutely test other of dynatron CPU coolers as well considering they seem to be ultra compact and way better than any commercial version for consumers.
I did this to my 6900 formula oc z yesterday, 30 degree drop after doing so.(binned xtxh chip, bigger cooler) The paste from factory was NOT a good job. The right side didn't have much paste, seems all AMD GPU's silicon are minorly warped on top, so don't do a super thin layer of paste as you would a CPU, do a slightly thicker layer(think preapplied paste thickness).
2:25 If the card gets this hot at idle, turning the max power limit down does absolute nothing, since it is not even close to being power limited at idle ...
Right you are, I didn't even think of that!! Man if this thing was getting THAT hot pulling like 15w at idle there was a major issue. Thanks for pointing that out 🤦♂️
Please don't be so negative on yourself! Anyone who watches this video will get a clear understanding of cable modding and your step by step method was really clear, the tools you used and why ! You've gained me as a subscriber because I initially watched your vid on GPU cooling and the shroud you sell via Etsy....brilliant idea!
Running your card at the boiling point of water is not good for it's longevity lol and some people here are claiming that it's fine lol makes me glad I got the sapphire toxic aio 360mm at load it only goes to about 55'c
I noticed that your vram clocks don’t drop at idle. That is partially why your card is baking at idle. It happens when you have multiple monitors at various and mismatched resolutions and frame rates. I had this problem on a gtx 1060 6gb and a 6700xt. There is a program called CRU that can change monitor timings and you can fix this idle bug. Custom resolution utility
*Turns out the RTX A4000 wasn't a good daily driver GPU* >Guys its toasty as hell I regret buying this crap >I got it a little cleaning >Turns out the RTX A4000 is a great card I will keep it EDIT: ok I got a little angry cause at the end I felt like watching it was totally wasted time as there was no message or reliable info other than "Check your vents". Besides this particular example, you got that something that kept me watching. Just don't make it 100% filler and you will do just fine ;3
No one, apart from nvidia ever said workstation cards are any good at all, nvidia ones specifically have mostly always been absolute garbage (i am talking about the design here, not absolute performance) with tiny heatsinks and terrible cooling. I mean you could have done the fag packet maths yourself - a 3070 but with a heatsink and fan 5 times smaller.....it's never gonna be any good when you think about it like that, is it?
it was the paste being trash. When your paste starts going bad again in a couple months replace it with the thermalright Heilos AMD paste (its AMD IHS shaped, read square, so itll fit your gpu die)
I think one of the main issues too with this particular card is the opening in the heatsink for air to flow through is small and very prone to getting blocked by dust.
There is no indication that the KPX won't be squished out after an arbitrary number of thermal cycles. KPX is a LN2 paste which does not necessarily make it good for ambient use. Liquid metal is best for scenarios where there is little mass to soak up the heat but the heatsink is not Ni coated so you will have faster amalgamation. So thermal paste will be "less better" than LM on a 3090 FE for example than on an A4000. All modern thermal pastes are subject to the dreadful pump-out effect it seems and I found the stock pastes to be far superior in this regard than something like MX4 or MX6. I didn't mention MX5 because it simply disappears before it has any chance to squeeze out. I recently repasted my 3090 FE with MX6 and it pumped out in less than 2 months and started performing worse than the dried up stock paste of 4 years. Then I tried PTM7950 and it has been perfect so far and should stay perfect for far, far longer than the service life of the card. I used LM on my previous GPU, a 1080ti Gaming X and I used it like that for more than a year, sold it to a guy and he has been rendering stuff on it even today with no performance degradation whatsoever. I'd use the PTM if I had it at the time though. Does LM perform far better than PTM7950 thermally? Hell yeah. Would that have mattered for the 1080 ti Gaming X? No. Would it have mattered for the A4000? Very much yes. Would I prefer it to PTM7950? No, but only because the oxide layer on the Cu heatsink is more permeable than the oxide on the Ni coating. The moral of the story is that any particular thermal paste would perform differently on different surfaces. Just because it's good on the CPU IHS it wouldn't necessarily be good on bare silicon. Also, I kind of find it funny that LM is not really more useful on an IHS, I mean it works and the performance is still superior than a with a thermal paste but it's not worth the damage on the IHS at all.
I recently found out first hand that kpx doesn’t last long term due to pump out, as I’ve been using it for a couple years now. I hear Thermal Grizzly is launching 2 new thermal pastes that specifically focus on longevity instead of maximum performance, so I’d be interested to try those out. We should also mention Thermal Grizzly Carbonaut pads, as they offer great temps and longevity, but they are expensive per application.
That is false. How much heat the heatsink can soak has nothing to do with the TIM. The TIM only cares about the Watts per area per temperature difference. This is a big chip and low power. Meaning the difference between LM and regular paste is small. Think of it like the heat transfer between the heatsink and air is the bottleneck, not the TIM.
@@Violet-ui you are missing one point: having less surface area on the interface limits the amount of heat that it could give away at any unit of time. This therefore makes the efficiency of the thermal interface material that much more important. This sounds like it doesn't make sense but you have more time to take the heat away with a larger heatsink with more surface area. I had a 7700K that I had to delid, I delidded two of them actually, and I tried both mx-4 and TG conductonaut between the chip and the IHS, then mx4 between the IHS and the heatsink. The difference was more than 15 degrees C. The results didn't change across multiple applications of mx4 and conductwhatever. 7700K had a very small die. Keep this in your mind. Then I did the same with a 1080 ti gaming X. GP102 was a huge chip with like 5 times the surface area of 7700K. The difference between mx-4 and conductwhatever was around 5 degrees C. The main problem was the pump-out effect so I had to use LM regardless. Visualise this: There is a small die, then there is an IHS on top of it which is for all intents and purposes a heatsink then there is a larger heatsink on top of the IHS. 7700K produces X joules of heat, then this heat has to be transferred from the surface of the silicon which is Y mm^2 to the IHS with around 6Y mm^2 surface area only to be transferred to another surface with a bit larger area. The bottleneck is between the silicon and the IHS. It is always more difficult to extract heat from a smaller surface. This is exactly why 100 and 102 dies have considerably larger surface areas than their brethren with far less heat output. Engineers leave empty spaces of the silicon only to make it possible to take the heat away.
@@ultraprecisiontechnologies that thing is the most electrically conductive material in the world (in room temps). It's far more dangerous than LM because the shit chips and crumbles to produce little circuit shorters that could travel anywhere with some airflow. PTM7950 seems to be the better choice.
@@ahmetrefikeryilmaz4432 Read my comment again. You are missing how the surface area is big for the amount of power and the power density is low. Your 7700K will have far higher power density, which is why the LM made a big difference. The 1080 Ti has a lower power density and the A4000 has an even lower one than that. So by your numbers it would probably be 3 to 4 C better than MX4.
If anyone wants a permanent fix to this, you need to use thick thermal paste. What's occurring is called pump out. Heat and pressure steadily causes thermal paste to press out from the center of the die. Thick pastes help with this.
I know those hotspots high temp aren’t safe for gpu at all. After almost 2 yrs my rx 6600xt gigabyte eagle is now reaching 95 to 98c of hotspot temp with core temp being around 76 to 78c.
Dumb question but would you make custom fan covers slot covers as well? My case has vents I'd like to close to have better air channelling (the whole bottom section of my Torrent Compact).
I just did a build in the meshify c, and I happen to have a 3d printer. I was actually thinking of designing a new cover for the right portion of the psu shroud. I took the metal one out so I could put a 3rd fan for front intake, but all the cables are showing. So I could make a new one that slants up for airflow and still has cutouts for the gpu power cable. Thoughts on this?
Термопаста с фазовым переходом, например, Honeywell хорошо показывает себя на чипах AMD . Дельта температуры между gpu и hotspot будет ещё меньше и со временем не увеличивается.
Is anyone having issues with audio not working on the hive? Realtek refuses to install and nothing for hive shows up in the sound settings. ASUS Reddit, rog forum, and ASUS support are all useless.
The "onboard" audio hardware has been moved off-board into the "strix hive". (Strix hive is a triangular shaped object with a dial on it with some buttons and ports). The strix hive has a dedicated USB type C port it needs to be plugged into in the motherboards I/O. You should also make sure the hive is enabled in BIOS. Hope this can be of some help or point you in some helpful direction.
Something you could try since it does not have glass from top to bottom is cut it down into an itx case and have something unique, maybe make it into a sandwich style and Yes, it still needs to be painted it's yellowed
fyi for the sapphire nitro+ RX 6900 XT the fan curve is set to hit 80C edge temp when i last checked the bios/softpowerplay tables with the morepowertool It is hot and quiet by design.
