Hi Der8auer I work for a company that uses Alphacool blocks on our high end open loop systems we sell. As far as im aware we are advised that 60-100% pump speed are recommended settings for the 40 series generation of GPU blocks. They are designed for high flow configurations opposed to EK's more restrictive small channeling designs. We have found that when vertical mounted the top left corner does get some air trapped, but after tilting the system anyway to help get air out of the radiators, this also fixes the air trapped in the GPU.
This was exactly my first thought... Why so much talking about that air trapped? Rush the pump to 90%, tilt the GPU a bit and the air is gone... Also, very well said! Pumps are designed to run 60% speed in standard conditions. Some also are not recommended running below 50%. And anything above +75% will drastically reduce life expectation of the pump. (I'm talking about normies pumps, like d5 and some distribution blocks too, and I'm not sure and well educated on EK pumps but should not be far off). Now, in his defense, he's also right. I was working on RMA services, and he's right. Ppl want great experience out of the box. Normies does not have a functional brain and they might send the GPU back to manufacturer straight away because "it's defective, the water doesn't flow through there". Like, no joking... I'm 100% sure there will be buyers doing so, because they don't understand physics and I've had these cases on my desk, so Roman has a solid point too. :/
@@ShankayLoveLadyL I wouldn't insult people over a design flaw. You should be able to turn your pump down if the cooling is adequate and still have the entire block fill with water. Unless they are going to specify a pump and minimum speed, they should make a block that will fill with water completely with the most common pump people would be using. The whole point of WC is a reduction in sound for many people, so it only makes sense that they would drop pump speed if temps are fine at lower speeds than would push the air out of that block.
presumably in a company that pc doesn't stand next to you and in a noise sensitive (living) room. Having a pump (or anything really) running loud is a VERY legitimate criticism. So if they "recommend" something annoying, that doesn't make it better. And there must be other pump solutions.
Just use higher pump speed and move/shake the build a bit to break the surface tension. I guess it's not really possible in an open air bench. Also temp wouldn't be an issue as long as it is cooled by the block
@@TheDeeGeeNL He doesn't have to run the higher pump speed constantly. You only need to do 100% pump speed and change the angle of the card a few times to bleed the air out of the loop, then you can go back to a lower pump speed. Its water cooling basics.
@@eliadbu true true, but it's next to impossible to figure out if a component may whine without small cursorary checks by independent youtubers like this. so i appreciate it, even if i do get one & it whines.
Remove the air pockets using 100% pump speed/rotating the pc a little. After it is removed, return pump speed to 30% and put some "aura" additive to the coolant, this would show if there is flow through the VRM section. ( i guess you could also do it with bubbles into the intake of the pump from a fill bottle too) I would bet there will be flow, albeit slow. So long as that air pocket is removed and water is present there it will be removing more heat, very sceptical if you would notice the difference in temperature though as your own tests showed. The VRM doesn't really produce a lot of heat, and it's spread out over many chips, with that size of copper block it's just going to average out. Would seem to just work better if they machined off that extra baffle/ridge, it's not like the VRM really needs all that guided water flow.
Love this kind of content Der8auer. I do wish you would have included more about the watercooling setup such as temperature delta under load. I have the original variant of this block for the Gaming OC and was wondering if the "Core" version of the block was noticeably better. I will say with my block, there is about a 3C drop in temp going from 50% to 100% on the D5 by the way
usually with any water cooling system I always try to bleed the system from air, as much as possible - so it will move to the reservoir and then fill the system again. Did you try to move the system around? I know open bench is bit trickier but since you use flex tubing you could in theory remove the card, and tilt in several direction allowing the air to bleed.
With the flow close to zero, getting the air bubble out doesn't matter much. Stagnant water would move heat better than air, but copper is so much better than either that stagnant water adds nothing.
This was common knowledge 20 years ago when I watercooled my first PC and I'm pretty sure it still is in the watercooling community. I can't believe Roman doesn't know this. It's so silly. Alphacool has nothing to fix here.
You've entirely missed the point. A well-designed block would be able to move water through those areas. He even explicitly said he moved it a bit to disturb the surface tension. So, no: you haven't discovered a piece of knowledge he doesn't have. He was just demonstrating to you that you will get very little water flow through that area. Even if you move the the card around and get the air out, it will just be relatively stagnant water.
@@rustler08 I had cards and blocks with similar issue Of flow impeded by air, every single time I bled it, the issue resolved. This is no new knowledge I never claimed it is. It is common practice, and I could not have found mention that he did it. I'm not talking about bit disturbing the tension I'm talking about removing the air completely.
Just pick up the board! It is a bubble. Air will always rise so turn the pump on and rotate the board as needed until all the bubbles are in the radiator. Common Sense is apparently a super power.
High flow or low flow... water is a fluid, it will follow the path of least resistance. It doesn't really matter in this card, but ideally in a center inlet, you want each side to have the same resistance so the fluid flows equally.
@@MikeTimbers I totally agree. I was going by what Der8auer was saying, even with that poor design, the mass of the block was still effective at cooling the VRAM.
Water doesn't only flow through the path of least resistance. It flows along all available paths in proportion to their resistances. Same with electricity. It flows along all paths, splitting according to the resistance of each path.
It is very odd that someone as Derbauer complaint so long about this and never fixed it in the video. Very odd, the whole comment section knows the fix but the master himself didn't mention it...
The air pocket is blocked by liquid on both sides, of course it won't move. Such pockets are kind of inevitable at the first fill of the loop and it is not the water blocks fault. Tilt it a couple of times to get the fluid going, it'll be fine then....
I'm disappointed with this review. You didn't show hotspot and memory temperatures. You didn't compare vrm, hotspot, and memory temperatures between the horizontal and vertical positions of the card.
I have an Alphacool block mounted vertically, i can see some micro bubbles going inside the water channel over the vrms and some going down, liquid is flowing there.
3:36 . I disagree that having a cover makes the card look better. I have an "asus ekwb 3090", and the black plate on the bottom looks out of place and like it was just tacked on as an afterthought, just to cover the tiny led strip. I wish it wasn't there, but removing it isn't an option. to me, this card looks much cleaner and better without that cover plate
Hi i got a rx 7900 xtx reference from sapphire mounted on a water block from Alphacool witch is reaching 110 degree Celsius on hotspot. If u want to further investigate the problem.
hey roman first of great video as always. Im curious since the pcb of 40 series are short and gpu waterblocks follows the design maybe you can make a hybrid by machining aluminum heatsink at the end of the waterblock cooling plate that can be bolted in like extra 2-3 inches more? since this gpus are mostly inside the case with fresh air flowing thru
If they had made the outlet on the bottom slightly smaller than the one on the top then the water would have filled the side channel since it would be the path of least resistance. Having a slightly smaller outlet on the bottom would also more than likely create a bit of a venturi effect and help to pull the water from the channel on the side therefore increasing the flow in that channel. While in a case I'm sure the pump noise would be greatly diminished but still they should provide documentation if the block needed faster pump speeds to completely fill the channels. Seems like it works great though so I guess its just nitpicking on my part. As always, great video!
Can I suggest a topic for a future video? Seeing how most manufacturers of modern gaming laptops are starting to use thermal putty instead of thermal pads for v-ram etc. I would love to see a in-depth video about putty vs pads and how this affects maintenance. Should you replace the factory thermal putty with pads or not. Which has better thermal conductivity, pads or putty.
I swapped out the Alphacool Eisblock on my Inno3d Frostbite 3080 for an EK equivalent. The Eisblock was fine on the core but seemed not great at cooling the RAM whereas the EK cools both well. Always used with a D5 on full power.
Hi Der8auer! Do you plan to make vram temperature test on 4060ti 16Gb? Half of memory chips are placed on the back side of pcb because of that clamshell configuration and cooled only by backplate. It would be very interesting to see is there any problems with temperatures. In msi 4070 ventus 2x/3x engineers made a huge mistake not covering one of the VRM phases with the radiator. Because of that mosfet and coil can heat up to 100°C, so I don't believe in infallibility of engineers.
I'm planning/getting parts for a new watercooled PC at the moment (Thanks for answering my emails!!! looking forward to taking the mycro for a spin ) and I got the Alphacool core Master, for my Gigabyte 4090. This block on the Inno3D ultra is also from the Alphacool core series, which was already available for some time, for reference boards like the Inno3D. I just compared it to my core block, and mine doesn't have these secondary outer channels 🤔My loop is not up yet, so I can't tell you if it works any better, but it's very peculiar to me, to see exactly this part different on the core Master block. I wonder if is a newer revision, or if the second channel is not there because the VRMs are just more compact ? Anyway. Interesting video. Thanks!
Does the power usage on normal include the waterpump? If not then the card is using more because the other models are using fans that also consume power. 10 watts or more for 3 fans is easy to obtain?
Getting the air out is relatively simple, you gently tilt the case and coax the air into the reservoir. Though generally speaking over the course of a couple of days and a few power cycles, the air will all just go away by itself. Don't know about a 7900XT version, but they also do a 4070Ti and a 4080 with the pre-installed waterblock. Their 4070 iChill aircooled model is compatible with that same waterblock from Alphacool that also fits the 4070ti, but you'll have to install it yourself.
I think surface tension isn't the issue, the air bubble is held in place by gravity. Imagine if the bubble suddenly broke up into smaller bubbles. Then those small bubbles would need to travel downward before they have a chance to escape the block. But they can't do that because if they did, the weight of the water would push the bubbles back to where they started. Running the pump at higher speed might fix it, but maybe not, because the water will just take the path of least resistance. Rotating/tilting the block 90 degrees should fix it. But like you said, it's fine even with no water directly flowing over it, copper is a really good conductor.
I cannot see the problem ? This happens in every liquid installation with air pockets. It is called: air lock. And can only be resolved by flushing them out...
The low flow rate performance issues are unfortunately a problem with alphacool 40 series cards. I got the Eiswolf aio for my zotac 4090, and the flow rate was abysmal and the card was hitting 70+C in OCCT at 600w even with a perfect mount. I swapped the terminal and did a regular loop with D5 at 100% and my max temp is now 54C at 600w and I can see flow visually around the sides.
Please, would it be worth changing my frostbite normal version? for this? the 2 doing oc would reach the same values? Could you put the bios of this ultra to the normal version??? would that be a problem? or just overclock the 2950 normal version and they will be almost similar?
Has this been released yet? I can't find any vendor with it Also could we just flash the bios for this one onto the normal frostbite to get the higher power draw?
He was complaining about the sun being hot in this video... Some times YTbers need to complain about something that doesn't exist to have click bait video's...
I flashed my Inno3D 4090 ichill frostbite with Galax 666w bios and have seen it going over 600w in some games. Maybe I should flash it with this cards bios instead, don't know how well this reference board vrm takes that 600w+ abuse.
I have alphacool block for 4090 strix looks similar and have total no air insde - 0 Zero. Works great and its rly good quality. Also so quiet coil whine on
Yes all of my hyper cooled and regular flow and not using a jacuzzi water pump lol / had burn in and paste and pad setting aka getting hot changing then making it structural form for its life span
With all that RGB in the water distribution, the card is absolutely designed for vertical mounting. It should be easier to get working than this, regardless of expected flow rate from the manufacturer.
6:30 this is 100% completely normal. Just let it run for a few hours and the block will fill up completely. Faster pump speeds will speed up the process significantly, but even at 10% speed the water will always fill the block eventually.
@@nik12937It will, I have this water block and I had that "problem". Setting my pump speed to 100% for some time and just using my PC regularly fixed the "problem". Also my pump speed is pwm controlled, from 800 to 2600rpm max, once went away the air pocket never returned.
Here in the UK this card costs £2,000 most likely equivalent to 2200 euros and it's only available on one single store. It is quite disappointing we can't find it for the same price as in Europe.
Get a train from King's Cross to Luton, than airplane to Frankfurt for a City-break and also get the GPU. I'm in the same boat and this looks like an option for me. :)
I have VPP755 which can be heared above 80%, but that might bee due to the bay reservoir and a VPP Apex which is on an Aurora Reservoir and I cannot hear it outside the case and it's running only full speed Indeed, the trapped air is a water path design flaw, but it can be mittigated
I can confirm the air goes out after some time and does not really affect the heat. I have the older 4080 iCHILL and my D5 pump chills at 40-50% all the time
Really, it would make more sense to direct water only over the GPU/VRAM and then bring the MosFET heat in with heatpipes, rather than have this asymmetrical flow impedance. Probably cheaper, too -- much less copper to CNC.
bought original last year, temp was 60deg...weird, hotspot was 100-115deg ! dismantled, bad thermal paste a part of the gou die was clean, reoasted with kryonaut my hotspot is now 58max and gpu vram are usually at 40ish...so dont buy for a warranty when thermal interface is random quality you probably will need to repaste, it now runs 3090mhz 24ghz vram btw
@@N0N0111 for some power system mod. I found some good ass DrMOS SPS from Alpha & Omega with tons of protection, pcb unload features and so on with 110A rate and QFN5x6 package. Same goes to Monolithic and their 90A SPS with LGA-41 profile that's used for MSI SUPRIM DrMOS (although circuit board of iCHILL doesn't look reliable like MSI's big fatty crunch of textolite). I could not find it anywhere and you know, I have a better alternative with MSI suprim as I said and 26 power stages. Don't criticize, I know that the 4090 power system is already overkill, but just in hunt for the lowest and the most stable temps possible
Alphacool is 50/50.. I do use them on my RX 5700, eis wolf.. but I had to clean the whole AIO to get the superb temps. (Lots of junk in the AIO system)
Just turn your pump on and let it run for a week at 100% speed. Make sure to keep topping off the reservoir as the water level goes down. I guarantee with absolute 10000% certainty, that block will be completely full. This is how GPU waterblocks have been for 20 years. EK's blocks do the EXACT same thing.
I must have gotten lucky with my two d5 pumps. I have them set at minimum 50% all the time. When loading up the system they will slowly go up to 75% then up to 100%. I NEVER hear them till 100% and even then barely. And no I don't have a case. Obt just sitting on a table with two pump/res
Oh man, I water-cooled my Strix 4090. The temps I am getting I was never expecting. I sit at 28 degrees C on idle. I have not seen the temp go higher than 35 degrees C yet but that is because I am not playing very demanding games yet. I am sure it will spike a bit higher once I start to play cyberpunk 2077.
At 100% load I get temps in the 68-70 region with fans at 100% rpm (Alphacool Eiswolf). I would be curious to know what you get because it's unfortunate having to hear the jet engine take off plus I can also hear coil whine. :(
can't you just rotate the card around to force that air bubble out? once you do that then it will flow through the whole block no problem... I would think.... It's just air trapped with no way to get out, even if you ramp up the pump to 100% it would most likely still be trapped there..... If it were me, and I had it al in a case and no the open air setup like you, I would pick up the case and rotate it around so the air can get push over to the outlet... then you could bring the pump back down to 30% and you wouldn't see or hear that trapped air anymore..... - - then again YOU are the pro, I've personally never had a water cooled system... but this just make sense to me :/
oh look..I was right LOL (i made this first comment before you did any testing) so you moved it around and SOME of the air got removed.... and you didn't even try to lol - yea its a real pain in the butt.... but I would think you would have to do this in ANY water cooled system... ESPECIALLY if you want the water to be SILENT
I have an Eiswolf2 GPU AIO from Alphacool for my AMD-made 7900XTX. It has an air bubble in the same spot and changing the pump RPM does nothing. Their fix is for you to take the whole thing - rad and card - and shake them around to try to get the air to the top of the rad and then add more coolant via the fill port there. Tried that several times now and I still have the same air pocket reform at the top left of the water block after a couple of days use. Not ideal. Alphacool will not be getting any more business from me sadly.
I just bought the for my 7900xtx as well. I just got rid of my bubbles but be honest I'm not impressed by the watercooler's performance. I have a pulse 7900xtx and I was running 60c gpu and 80c hotspot @ 2906 with the watercooler I get 57c gpu 77c hotspot @ 2966. What utility are you using to turn up the pump speed? I'm current using the Bios.
So easy to fix! You just need to separate the two plates and from the inside narrow the path leading to the center of the GPU (it is too wide), then the water will go to the left side of the map on the VRM. I suggest narrowing the path with epoxy glue! It's invisible and transparent! With love from Russia! I Come in Peace!
CORRECT, I see all the RTX 4090 prices are going up! Mainly because RTX 4090 is the best cheap fit for AI tasks (powerful GPU and 24GB VRAM) This AI craze is bumping up the price by a lot!!! Nvidia is a company that is worth $1.9T already and it has just begun ramping up AI tech!
You saw Gigabyte is secretly making a new RTX 4090 revision after not complying to consumers guarantee? I don't think this GPU will have cracks, it is super compact and less overhang beyond the PCIe slot.
@@N0N0111 Gigabyte has some shady business in the past unfortunately also. It's sad because it was one of my first motherboards manufacturer. I'm sad to see them doing this to their customers.
I was considering this exact alphacool core block for a gigabyte 4090 gaming OC but went for the phanteks looks like that was the right decision as the bad water flow path was something my alphacool 3090 block had as well
Well duh, if you don't get the air out of the system there's not going to be water flow. When you're filling a loop with water you're supposed to tilt things around until there is no visible air trapped in the loop. The problem isn't in the water block, it's in the way you've decided not to do all the necessary steps required when building a water cooled system. "I didn't bother installing it properly and now it doesn't function properly. This can only mean that the product is not designed well!"
I really cannot understand how there are people who are able to buy such an expensive graphics card (or any other hardware component), thinking only about overclocking it, and forgetting to use it for its main purpose - playing games or using it for productivity, such as for making animated films. Do these people remember the times when they played computer games? Did they ever played computer games? Or am I the freak, who only thinks about using them to play computer games?
I definitely agree with your conclusion that practically speaking the poor flow in the left channel has no meaningful impact on temperatures, but it still shows a lack of attention to detail in the block design. If the channel isn't doing anything, they shouldn't have bothered milling the slot in the first place and saved themselves a bit of machine time.
Use horizontally and air won't accumulate. Or at least fill in that position rather than talking about problems. Some container can be used to collect the air, from which the air is evacuated by means of a valve.
Hi Der8auer I work for a company that uses Alphacool blocks on our high end open loop systems we sell. As far as im aware we are advised that 60-100% pump speed are recommended settings for the 40 series generation of GPU blocks. They are designed for high flow configurations opposed to EK's more restrictive small channeling designs. We have found that when vertical mounted the top left corner does get some air trapped, but after tilting the system anyway to help get air out of the radiators, this also fixes the air trapped in the GPU.
This was exactly my first thought... Why so much talking about that air trapped? Rush the pump to 90%, tilt the GPU a bit and the air is gone... Also, very well said! Pumps are designed to run 60% speed in standard conditions. Some also are not recommended running below 50%. And anything above +75% will drastically reduce life expectation of the pump. (I'm talking about normies pumps, like d5 and some distribution blocks too, and I'm not sure and well educated on EK pumps but should not be far off).
Now, in his defense, he's also right. I was working on RMA services, and he's right. Ppl want great experience out of the box. Normies does not have a functional brain and they might send the GPU back to manufacturer straight away because "it's defective, the water doesn't flow through there". Like, no joking... I'm 100% sure there will be buyers doing so, because they don't understand physics and I've had these cases on my desk, so Roman has a solid point too. :/
@@ShankayLoveLadyL I wouldn't insult people over a design flaw. You should be able to turn your pump down if the cooling is adequate and still have the entire block fill with water. Unless they are going to specify a pump and minimum speed, they should make a block that will fill with water completely with the most common pump people would be using. The whole point of WC is a reduction in sound for many people, so it only makes sense that they would drop pump speed if temps are fine at lower speeds than would push the air out of that block.
presumably in a company that pc doesn't stand next to you and in a noise sensitive (living) room. Having a pump (or anything really) running loud is a VERY legitimate criticism.
So if they "recommend" something annoying, that doesn't make it better. And there must be other pump solutions.
@@--_DJ_--Maybe the pumps are like internal combustion engines which need a minimal rpm to not stall?
EK blocks are also designed to work better with higher pump speed.... it's a matter of fins.
Just use higher pump speed and move/shake the build a bit to break the surface tension. I guess it's not really possible in an open air bench. Also temp wouldn't be an issue as long as it is cooled by the block
@@TheDeeGeeNL what I mean is to bleed the air with higher pump speed, then lower the speed for normal use.
Water-cooling 101
@@TheDeeGeeNL a D5 at 100% it's quiter than any fan at 50%.
@@TheDeeGeeNL He doesn't have to run the higher pump speed constantly. You only need to do 100% pump speed and change the angle of the card a few times to bleed the air out of the loop, then you can go back to a lower pump speed. Its water cooling basics.
@@Mildly_Amusedthat's kind of odd that he didn't bleed the air out
@@TheDeeGeeNL Before starting, you MUST turn on the pump to 100% to pump the system.
Thank you for mentioning coil whine, or the lack of it. Legit very important for me.
Yea me as well. I can't stand coil whine.
While card electrical design might change things - this really varies from one unit to another and other things like PSU can change things.
@@eliadbu true true, but it's next to impossible to figure out if a component may whine without small cursorary checks by independent youtubers like this. so i appreciate it, even if i do get one & it whines.
Remove the air pockets using 100% pump speed/rotating the pc a little. After it is removed, return pump speed to 30% and put some "aura" additive to the coolant, this would show if there is flow through the VRM section. ( i guess you could also do it with bubbles into the intake of the pump from a fill bottle too)
I would bet there will be flow, albeit slow. So long as that air pocket is removed and water is present there it will be removing more heat, very sceptical if you would notice the difference in temperature though as your own tests showed. The VRM doesn't really produce a lot of heat, and it's spread out over many chips, with that size of copper block it's just going to average out.
Would seem to just work better if they machined off that extra baffle/ridge, it's not like the VRM really needs all that guided water flow.
Love this kind of content Der8auer. I do wish you would have included more about the watercooling setup such as temperature delta under load. I have the original variant of this block for the Gaming OC and was wondering if the "Core" version of the block was noticeably better. I will say with my block, there is about a 3C drop in temp going from 50% to 100% on the D5 by the way
usually with any water cooling system I always try to bleed the system from air, as much as possible - so it will move to the reservoir and then fill the system again. Did you try to move the system around? I know open bench is bit trickier but since you use flex tubing you could in theory remove the card, and tilt in several direction allowing the air to bleed.
With the flow close to zero, getting the air bubble out doesn't matter much. Stagnant water would move heat better than air, but copper is so much better than either that stagnant water adds nothing.
This was common knowledge 20 years ago when I watercooled my first PC and I'm pretty sure it still is in the watercooling community. I can't believe Roman doesn't know this. It's so silly. Alphacool has nothing to fix here.
You've entirely missed the point. A well-designed block would be able to move water through those areas. He even explicitly said he moved it a bit to disturb the surface tension.
So, no: you haven't discovered a piece of knowledge he doesn't have. He was just demonstrating to you that you will get very little water flow through that area. Even if you move the the card around and get the air out, it will just be relatively stagnant water.
@@yup8388 The only thing silly is you not getting the point he was making.
@@rustler08 I had cards and blocks with similar issue Of flow impeded by air, every single time I bled it, the issue resolved. This is no new knowledge I never claimed it is. It is common practice, and I could not have found mention that he did it. I'm not talking about bit disturbing the tension I'm talking about removing the air completely.
Just pick up the board! It is a bubble. Air will always rise so turn the pump on and rotate the board as needed until all the bubbles are in the radiator. Common Sense is apparently a super power.
High flow or low flow... water is a fluid, it will follow the path of least resistance. It doesn't really matter in this card, but ideally in a center inlet, you want each side to have the same resistance so the fluid flows equally.
it does matter though if the water flow doesn't push water around the edges to effectively cool the VRAM. That's why I swapped out my Alphacool block.
@@MikeTimbers I totally agree. I was going by what Der8auer was saying, even with that poor design, the mass of the block was still effective at cooling the VRAM.
Water doesn't only flow through the path of least resistance. It flows along all available paths in proportion to their resistances.
Same with electricity. It flows along all paths, splitting according to the resistance of each path.
@@Falcrist according to the resistance, yes. so if the resistance i.e. an airblock allied to a restrictive channel can prevent any flow.
12:44 can’t we tilt the board to take the air gap out?
It is very odd that someone as Derbauer complaint so long about this and never fixed it in the video.
Very odd, the whole comment section knows the fix but the master himself didn't mention it...
The air pocket is blocked by liquid on both sides, of course it won't move. Such pockets are kind of inevitable at the first fill of the loop and it is not the water blocks fault. Tilt it a couple of times to get the fluid going, it'll be fine then....
I'm disappointed with this review. You didn't show hotspot and memory temperatures. You didn't compare vrm, hotspot, and memory temperatures between the horizontal and vertical positions of the card.
Same.
Good Point. Will make sure to include it next time
@@der8auer-en Thanks.
Does anyone know where to buy this card in the US? Can’t find any listing anywhere to purchase/ship it to me
This is why i love DerBauer, brands i see in EU stores that i never hear about on more North American channels.
I have an Alphacool block mounted vertically, i can see some micro bubbles going inside the water channel over the vrms and some going down, liquid is flowing there.
3:36 . I disagree that having a cover makes the card look better. I have an "asus ekwb 3090", and the black plate on the bottom looks out of place and like it was just tacked on as an afterthought, just to cover the tiny led strip. I wish it wasn't there, but removing it isn't an option. to me, this card looks much cleaner and better without that cover plate
Hi i got a rx 7900 xtx reference from sapphire mounted on a water block from Alphacool witch is reaching 110 degree Celsius on hotspot. If u want to further investigate the problem.
Hey Der8auer would you do a video on your w3175x pc? Curious how it's holding up nowadays
The fitting connectors at the top need goggly eyes - :)
I wonder if you moved the card into a horizontal and/or inverted it, if the position would help with the air pockets.
hey roman first of great video as always. Im curious since the pcb of 40 series are short and gpu waterblocks follows the design maybe you can make a hybrid by machining aluminum heatsink at the end of the waterblock cooling plate that can be bolted in like extra 2-3 inches more? since this gpus are mostly inside the case with fresh air flowing thru
Time Spy extreme my score was 155 fps 12900k RTX 4090 Suprim x air cooled version no over clocking. Great RUclips channel always great work
If they had made the outlet on the bottom slightly smaller than the one on the top then the water would have filled the side channel since it would be the path of least resistance. Having a slightly smaller outlet on the bottom would also more than likely create a bit of a venturi effect and help to pull the water from the channel on the side therefore increasing the flow in that channel. While in a case I'm sure the pump noise would be greatly diminished but still they should provide documentation if the block needed faster pump speeds to completely fill the channels. Seems like it works great though so I guess its just nitpicking on my part. As always, great video!
Where can I get the version you show. All I find is the older version.
Watercool Heatkiller V Pro Strix or FE test next please! I have these blocks and the performance and build quality is phenomenal.
Maybe there is a problem with milling the water block. They may need to widen the channel. It seems like only a small tweak should resolve the issue.
5:10 cut meow ❤😹
i mean . . you could just take the card out with the water connected and move the air bubbles by moving the card around right?
Can I suggest a topic for a future video? Seeing how most manufacturers of modern gaming laptops are starting to use thermal putty instead of thermal pads for v-ram etc. I would love to see a in-depth video about putty vs pads and how this affects maintenance. Should you replace the factory thermal putty with pads or not. Which has better thermal conductivity, pads or putty.
I swapped out the Alphacool Eisblock on my Inno3d Frostbite 3080 for an EK equivalent. The Eisblock was fine on the core but seemed not great at cooling the RAM whereas the EK cools both well. Always used with a D5 on full power.
Great Video, just wandering if you considered bleeding the air out of the block and seeing wether or not it re-formed in those spots ?
Hi Der8auer! Do you plan to make vram temperature test on 4060ti 16Gb? Half of memory chips are placed on the back side of pcb because of that clamshell configuration and cooled only by backplate. It would be very interesting to see is there any problems with temperatures.
In msi 4070 ventus 2x/3x engineers made a huge mistake not covering one of the VRM phases with the radiator. Because of that mosfet and coil can heat up to 100°C, so I don't believe in infallibility of engineers.
Ya did good , mods that support and also improve in design reality based on facts of time of use
I'm planning/getting parts for a new watercooled PC at the moment (Thanks for answering my emails!!! looking forward to taking the mycro for a spin ) and I got the Alphacool core Master, for my Gigabyte 4090. This block on the Inno3D ultra is also from the Alphacool core series, which was already available for some time, for reference boards like the Inno3D. I just compared it to my core block, and mine doesn't have these secondary outer channels 🤔My loop is not up yet, so I can't tell you if it works any better, but it's very peculiar to me, to see exactly this part different on the core Master block. I wonder if is a newer revision, or if the second channel is not there because the VRMs are just more compact ? Anyway. Interesting video. Thanks!
Does the power usage on normal include the waterpump? If not then the card is using more because the other models are using fans that also consume power. 10 watts or more for 3 fans is easy to obtain?
*Is there an ichill for the 7900 XT that would be perfect for an upgrade for my GPU, also how do you guys get all the air out of a water cooled build*
Getting the air out is relatively simple, you gently tilt the case and coax the air into the reservoir. Though generally speaking over the course of a couple of days and a few power cycles, the air will all just go away by itself.
Don't know about a 7900XT version, but they also do a 4070Ti and a 4080 with the pre-installed waterblock. Their 4070 iChill aircooled model is compatible with that same waterblock from Alphacool that also fits the 4070ti, but you'll have to install it yourself.
Did you modify \ change the voltage while OC it or was that option left on stock settings (1.05v)
Stock voltage. More/less didn't help
I wish they made blocks for the RX7900XT. Having a hard time finding one for my vanilla ASUS model.
Should've bought 4080
Shouldn’t have bought either, both terrible value.@@GewelReal
You guys are so helpful...
I just wonder which paste did he use ?
I use Thermal Grizzly Kryonaut on my 6700XT GPU - wondering if thats the best choice 🤔
I think surface tension isn't the issue, the air bubble is held in place by gravity. Imagine if the bubble suddenly broke up into smaller bubbles. Then those small bubbles would need to travel downward before they have a chance to escape the block. But they can't do that because if they did, the weight of the water would push the bubbles back to where they started.
Running the pump at higher speed might fix it, but maybe not, because the water will just take the path of least resistance.
Rotating/tilting the block 90 degrees should fix it.
But like you said, it's fine even with no water directly flowing over it, copper is a really good conductor.
I have the same water block, I had an air bubble in the top left corner, it went away with time.
I cannot see the problem ? This happens in every liquid installation with air pockets. It is called: air lock. And can only be resolved by flushing them out...
Because YTber needs clickbait video...
The low flow rate performance issues are unfortunately a problem with alphacool 40 series cards. I got the Eiswolf aio for my zotac 4090, and the flow rate was abysmal and the card was hitting 70+C in OCCT at 600w even with a perfect mount. I swapped the terminal and did a regular loop with D5 at 100% and my max temp is now 54C at 600w and I can see flow visually around the sides.
I don't know about the aio version, but I have a Core water block for my 4090 suprim and it's performing great.
Just out of curiosity, what core did that card have, AD102-300 or AD102-301?
Please, would it be worth changing my frostbite normal version? for this? the 2 doing oc would reach the same values?
Could you put the bios of this ultra to the normal version??? would that be a problem? or just overclock the 2950 normal version and they will be almost similar?
Has this been released yet? I can't find any vendor with it
Also could we just flash the bios for this one onto the normal frostbite to get the higher power draw?
Flash the Galax 666W vBIOS to it instead, it'll give you 1.1V Vcore too. My 4090 FBU did ~3 GHz on water with said vBIOS.
Does the flow problem only occur on vertical mount?
wouldn't placing the rad higher solve this problem? isn't this the same issue with aios
i read somewhere that there's a 600w bios mod for this. would you ever be interested to revisit this GPU with a more reasonable power target?
There's not really a BIOS mod, what you're thinking of is the Galax 666W vBIOS.
Worms Armagedon! Man of culture....
I just got one, you should do test horizontal not vertical. 🙂
He was complaining about the sun being hot in this video...
Some times YTbers need to complain about something that doesn't exist to have click bait video's...
Where can I buy it?
I flashed my Inno3D 4090 ichill frostbite with Galax 666w bios and have seen it going over 600w in some games. Maybe I should flash it with this cards bios instead, don't know how well this reference board vrm takes that 600w+ abuse.
Place it in the horizontal position for some time and you are good.
I have alphacool block for 4090 strix looks similar and have total no air insde - 0 Zero. Works great and its rly good quality. Also so quiet coil whine on
Yes all of my hyper cooled and regular flow and not using a jacuzzi water pump lol / had burn in and paste and pad setting aka getting hot changing then making it structural form for its life span
With all that RGB in the water distribution, the card is absolutely designed for vertical mounting. It should be easier to get working than this, regardless of expected flow rate from the manufacturer.
hi, i wonder if you could try and upgrade a intel a750 to 16GB vram and compare it to a a770 with 8GB abd 16GB
We get air pockets in car cooling systems, we pull a vacuum on the system and let the fluid in slowly until its full of fluid, no air.
You need a thermostat on the inlet.
You got a mill!!!
Chop off the divider!
We must know :)
6:30 this is 100% completely normal. Just let it run for a few hours and the block will fill up completely. Faster pump speeds will speed up the process significantly, but even at 10% speed the water will always fill the block eventually.
It literally won't. You can shake the card to fill the chamber and it will proceed to empty itself at any pump speed.
@@nik12937It will, I have this water block and I had that "problem".
Setting my pump speed to 100% for some time and just using my PC regularly fixed the "problem". Also my pump speed is pwm controlled, from 800 to 2600rpm max, once went away the air pocket never returned.
Hmm better bios in the future maybe?
Then the Voltage regulator controller VRM I know module / also has hot warm and cold ops
its a non issue the air pocket will resolve itself with some time. i have the waterblock not sure how u didint get this.
and where thid you see thad price in germany
why 30% pump speed?
noise
@@MaxIronsThird D5 pump goes BRRRRRRRRRR
Here in the UK this card costs £2,000 most likely equivalent to 2200 euros and it's only available on one single store. It is quite disappointing we can't find it for the same price as in Europe.
Get a train from King's Cross to Luton, than airplane to Frankfurt for a City-break and also get the GPU.
I'm in the same boat and this looks like an option for me. :)
Pressure as well as vacuum mapping tower wars
I have VPP755 which can be heared above 80%, but that might bee due to the bay reservoir and a VPP Apex which is on an Aurora Reservoir and I cannot hear it outside the case and it's running only full speed
Indeed, the trapped air is a water path design flaw, but it can be mittigated
I can confirm the air goes out after some time and does not really affect the heat. I have the older 4080 iCHILL and my D5 pump chills at 40-50% all the time
Why not bleed it?
Because YTber needs clickbait video...
Really, it would make more sense to direct water only over the GPU/VRAM and then bring the MosFET heat in with heatpipes, rather than have this asymmetrical flow impedance. Probably cheaper, too -- much less copper to CNC.
bought original last year, temp was 60deg...weird, hotspot was 100-115deg ! dismantled, bad thermal paste a part of the gou die was clean, reoasted with kryonaut my hotspot is now 58max and gpu vram are usually at 40ish...so dont buy for a warranty when thermal interface is random quality you probably will need to repaste, it now runs 3090mhz 24ghz vram btw
please, circuit board hi-res pictures, I beg 🙏
Why you need that so bad?
@@N0N0111 for some power system mod. I found some good ass DrMOS SPS from Alpha & Omega with tons of protection, pcb unload features and so on with 110A rate and QFN5x6 package. Same goes to Monolithic and their 90A SPS with LGA-41 profile that's used for MSI SUPRIM DrMOS (although circuit board of iCHILL doesn't look reliable like MSI's big fatty crunch of textolite).
I could not find it anywhere and you know, I have a better alternative with MSI suprim as I said and 26 power stages. Don't criticize, I know that the 4090 power system is already overkill, but just in hunt for the lowest and the most stable temps possible
Alphacool is 50/50.. I do use them on my RX 5700, eis wolf.. but I had to clean the whole AIO to get the superb temps. (Lots of junk in the AIO system)
Those quick connects are beaut
Fucking LOVE my Koolance QD3s!
Mem chips like to be same as gpu but run at 6 to 16ghz so a 2.7ghz gpu does get hot or fast enough
Also mem is a form of cpu design just with cache cells
Just turn your pump on and let it run for a week at 100% speed. Make sure to keep topping off the reservoir as the water level goes down. I guarantee with absolute 10000% certainty, that block will be completely full. This is how GPU waterblocks have been for 20 years. EK's blocks do the EXACT same thing.
i promise you he already knows this
@@cjbtlrI'm sure he knows, it just strange he didn't mention it in the video, (if he did my bad)
I must have gotten lucky with my two d5 pumps. I have them set at minimum 50% all the time. When loading up the system they will slowly go up to 75% then up to 100%. I NEVER hear them till 100% and even then barely. And no I don't have a case. Obt just sitting on a table with two pump/res
looks nice
if the video card was in a pc case and the card was horizontal then there would be no water flow problem
Great video
Oh man, I water-cooled my Strix 4090. The temps I am getting I was never expecting. I sit at 28 degrees C on idle. I have not seen the temp go higher than 35 degrees C yet but that is because I am not playing very demanding games yet. I am sure it will spike a bit higher once I start to play cyberpunk 2077.
At 100% load I get temps in the 68-70 region with fans at 100% rpm (Alphacool Eiswolf). I would be curious to know what you get because it's unfortunate having to hear the jet engine take off plus I can also hear coil whine. :(
Well, you look younger in this video. It is the haircut or eyebrow trim? Just can't tell. Nice video as always. Weight loss.
can't you just rotate the card around to force that air bubble out? once you do that then it will flow through the whole block no problem... I would think.... It's just air trapped with no way to get out, even if you ramp up the pump to 100% it would most likely still be trapped there..... If it were me, and I had it al in a case and no the open air setup like you, I would pick up the case and rotate it around so the air can get push over to the outlet... then you could bring the pump back down to 30% and you wouldn't see or hear that trapped air anymore.....
- - then again YOU are the pro, I've personally never had a water cooled system... but this just make sense to me :/
oh look..I was right LOL (i made this first comment before you did any testing) so you moved it around and SOME of the air got removed.... and you didn't even try to lol - yea its a real pain in the butt.... but I would think you would have to do this in ANY water cooled system... ESPECIALLY if you want the water to be SILENT
I have an Eiswolf2 GPU AIO from Alphacool for my AMD-made 7900XTX. It has an air bubble in the same spot and changing the pump RPM does nothing. Their fix is for you to take the whole thing - rad and card - and shake them around to try to get the air to the top of the rad and then add more coolant via the fill port there. Tried that several times now and I still have the same air pocket reform at the top left of the water block after a couple of days use. Not ideal. Alphacool will not be getting any more business from me sadly.
I just bought the for my 7900xtx as well. I just got rid of my bubbles but be honest I'm not impressed by the watercooler's performance. I have a pulse 7900xtx and I was running 60c gpu and 80c hotspot @ 2906 with the watercooler I get 57c gpu 77c hotspot @ 2966. What utility are you using to turn up the pump speed? I'm current using the Bios.
So easy to fix! You just need to separate the two plates and from the inside narrow the path leading to the center of the GPU (it is too wide), then the water will go to the left side of the map on the VRM. I suggest narrowing the path with epoxy glue! It's invisible and transparent! With love from Russia! I Come in Peace!
put core count on your packs Hetzner
Its meh, i think other companies have those sections of the waterblock, curved off, like heatkiller blocks for example
3 months later.. 2.149,00 €
CORRECT, I see all the RTX 4090 prices are going up! Mainly because RTX 4090 is the best cheap fit for AI tasks (powerful GPU and 24GB VRAM)
This AI craze is bumping up the price by a lot!!! Nvidia is a company that is worth $1.9T already and it has just begun ramping up AI tech!
the cat at @5:08 xD
Another contender for cracking pcb xD
You saw Gigabyte is secretly making a new RTX 4090 revision after not complying to consumers guarantee?
I don't think this GPU will have cracks, it is super compact and less overhang beyond the PCIe slot.
@@N0N0111 Gigabyte has some shady business in the past unfortunately also. It's sad because it was one of my first motherboards manufacturer. I'm sad to see them doing this to their customers.
I was considering this exact alphacool core block for a gigabyte 4090 gaming OC but went for the phanteks
looks like that was the right decision as the bad water flow path was something my alphacool 3090 block had as well
Linus says it should work fine on a 3090...
HAHA, this joke is perfect for this video.
When you know you know it!
Makes me think I am running my D5 too fast. Has been at 80-100% for the last like 8 years, but its no louder than the 800ish RPM fans.
so remove the air and fill the loop properly..... !
Big aluminum back plate without any thermal pads going to it? Seems like a missed opportunity.
I have an alphaccol blcok for my 4090 and there are thermal pads
He actually shows there are pads you might missed that he shows on a side view.
Well duh, if you don't get the air out of the system there's not going to be water flow. When you're filling a loop with water you're supposed to tilt things around until there is no visible air trapped in the loop. The problem isn't in the water block, it's in the way you've decided not to do all the necessary steps required when building a water cooled system.
"I didn't bother installing it properly and now it doesn't function properly. This can only mean that the product is not designed well!"
I really cannot understand how there are people who are able to buy such an expensive graphics card (or any other hardware component), thinking only about overclocking it, and forgetting to use it for its main purpose - playing games or using it for productivity, such as for making animated films. Do these people remember the times when they played computer games? Did they ever played computer games? Or am I the freak, who only thinks about using them to play computer games?
You overclock to have better fps when playing games of faster render time in productivity
I definitely agree with your conclusion that practically speaking the poor flow in the left channel has no meaningful impact on temperatures, but it still shows a lack of attention to detail in the block design. If the channel isn't doing anything, they shouldn't have bothered milling the slot in the first place and saved themselves a bit of machine time.
This card is probably made for vertical mounting.
Use horizontally and air won't accumulate. Or at least fill in that position rather than talking about problems.
Some container can be used to collect the air, from which the air is evacuated by means of a valve.