CORSAIR Hydro X Series XG7 RGB 10-SERIES GPU Water Block

[VSG]

Continuing our coverage of CORSAIR's Hydro X custom water cooling products, we take a look at the XG7 RGB GPU water block. It features integrated dRGB lighting, a flow indicator, backplate, and pre-applied thermal pads and paste. Pricing is solid, offering good value for your money.

Show full review

 

[Zareek]

Wow nothing like greasing the wheels with that kit! I'm not trying to imply there is any bias, just saying when something like that shows up I'm sure it motivates you to open it up and mess around with it. I know I would.

That being said, I wish I had the dough to do a custom loop! Maybe I will spend next years upgrade fund on it. That is going to depend a lot on what the video card market looks like then...

Nice job as always...

 

[Gan77]

Who is the OEM manufacturer of this Block?

 

[VSG]

Wow nothing like greasing the wheels with that kit! I'm not trying to imply there is any bias, just saying when something like that shows up I'm sure it motivates you to open it up and mess around with it. I know I would.

That being said, I wish I had the dough to do a custom loop! Maybe I will spend next years upgrade fund on it. That is going to depend a lot on what the video card market looks like then...

Nice job as always...

Hah actually what it does is make me think about where I have space to put it.

Who is the OEM manufacturer of this Block?

Blocks are their (CORSAIR) own thing, that's where the ex-EK guys come in for the engineering and design.

 

[Zareek]

Hah actually what it does is make me think about where I have space to put it.

I'd be figuring out where to display it!

 

[RH92]

In my book that's a decent start for Corsair , although as mentioned in the review for a '' value '' oriented product well ..... value ( perf/dollars ) is average at best . Quality review as always !

 

[VSG]

I'd be figuring out where to display it!

After a while things just pile up, so it's comes down to most efficient packing and storage

In my book that's a decent start for Corsair , although as mentioned in the review for a '' value '' oriented product well ..... value ( perf/dollars ) is average at best . Quality review as always !

That's not the best chart to be honest, as explained in the final page. Prices for water blocks have increased a lot in the last 2-3 years, so comparing this new block for an older GPU with older blocks kinda puts it in a negative light. I tried to expand further on this in the conclusion section to help make more sense.

 

[AnarchoPrimitiv]

I cannot stand how PC brands just can't use the addressable LED industry's 3 pin JST connector which is industry standard. Instead, like this corsair product, they have to use proprietary connectors just for the nefarious reason on locking you into an ecosystem where, for example, a 30 inch, 30 led per meter corsair branded strip costs $40 while I could go on aliexpress and buy a ws2812b or an even newer ap102 addressable strip with 60 or 90 leds per meter that is 1 meter long for $15. Or how for some reason, PC branded controllers cost $50 and can only control 150 pixels while I could buy a generic controller, again off aliexpress l, than can control 2048 pixels for $10.

It's just like how Asus made the addressable header on their motherboards that ridiculous 4 minus 1 (=3) pin header instead of the ubiquitous 3 pin JST connector and somehow its become something of an PC brand standard that's now been copied by other mobo and accessory manufacturers. That's why I have my PC lighted with generic addressable LED strips controlled by a digital, generic controller that is programmed with open source software and is powered by my PSU with a sata power connector I soldered myself. I can control and power 4096 pixels with it, rewired my fans to be controlled by it, and now I have my own fully integrated ecosystem that cost a tenth of the price of something like NZXTs hue+....

... Sorry for the rant, but it's true

 

[VSG]

I cannot stand how PC brands just can't use the addressable LED industry's 3 pin JST connector which is industry standard. Instead, like this corsair product, they have to use proprietary connectors just for the nefarious reason on locking you into an ecosystem where, for example, a 30 inch, 30 led per meter corsair branded strip costs $40 while I could go on aliexpress and buy a ws2812b or an even newer ap102 addressable strip with 60 or 90 leds per meter that is 1 meter long for $15. Or how for some reason, PC branded controllers cost $50 and can only control 150 pixels while I could buy a generic controller, again off aliexpress l, than can control 2048 pixels for $10.

It's just like how Asus made the addressable header on their motherboards that ridiculous 4 minus 1 (=3) pin header instead of the ubiquitous 3 pin JST connector and somehow its become something of an PC brand standard that's now been copied by other mobo and accessory manufacturers. That's why I have my PC lighted with generic addressable LED strips controlled by a digital, generic controller that is programmed with open source software and is powered by my PSU with a sata power connector I soldered myself. I can control and power 4096 pixels with it, rewired my fans to be controlled by it, and now I have my own fully integrated ecosystem that cost a tenth of the price of something like NZXTs hue+....

... Sorry for the rant, but it's true

Nothing to be sorry for, feedback is always something that companies should appreciate and be aware of.

 

[bogami]

To this product can not be added the name of a block! Consisting of inconsistent materials, broken slots worth about $ 30 max. Do not buy these garbage and I resent this kind of support for such a poor product with an unworthy rating !! Copper blocks are much more efficient and I personally use them. GPU temperatures, however, stick around 15 to 17 degrees celsium above liquid temperature max in SLI conf. The impact of advertising is very noticeable in this review. deliberate support for a bad product. Be wise and don't buy this garbage for the money . You get the right block for that money !

 

[VSG]

To this product can not be added the name of a block! Consisting of inconsistent materials, broken slots worth about $ 30 max. Do not buy these garbage and I resent this kind of support for such a poor product with an unworthy rating !! Copper blocks are much more efficient and I personally use them. GPU temperatures, however, stick around 15 to 17 degrees celsium above liquid temperature max in SLI conf. The impact of advertising is very noticeable in this review. deliberate support for a bad product. Be wise and don't buy this garbage for the money . You get the right block for that money !

I am sure the language barrier is playing a role here, but this is a copper block (nickel-plated copper) so I am not sure what this particular rant is on about.

 

[Tsukiyomi91]

another unwanted ranting where he/she has never tested the product but judging them from mere pictures. Get a life, slime. Also, if Corsair's proprietary connectors is bad for you, go for those generic RGB LED standards & compare how "great" they are when the quality of the bulbs, controllers etc are still inferior compare to Corsair's.

 

[N3utro]

Thanks for the review, but it's missing key points like the radiator and fan used, their speed, if the loop was only on the GPU side, what was the water temperature.

If you used a cooling system that was not enough to remove all the W from the loop this could lead to improper results on the waterblock performance.

Also using Gelid thermal paste, even if it has good performance, can not compare to using liquid metal, thus restricting the cooling power of the waterblock.

My 1080Ti has a delta temperature of around 10°C with water staying really close to ambiant temperature. Your results are way above that, which is strange.

 

[VSG]

Thanks for the review, but it's missing key points like the radiator and fan used, their speed, if the loop was only on the GPU side, what was the water temperature.

If you used a cooling system that was not enough to remove all the W from the loop this could lead to improper results on the waterblock performance.

Also using Gelid thermal paste, even if it has good performance, can not compare to using liquid metal, thus restricting the cooling power of the waterblock.

My 1080Ti has a delta temperature of around 10°C with water staying really close to ambiant temperature. Your results are way above that, which is strange.

The thermal performance metrics were measured as a difference of the component temperature over the coolant temperature as mentioned in the review, so the radiator/fans etc do not come into play as long as it is sufficient and prevents throttling of any kind. I made sure that was the case (Black Ice Nemesis 480 GTX with NB-eLoop fans at 1600 RPM, since you were curious), and I definitely do not advocate liquid metal on a bare PCB for the average end user. Finally, you may have missed the part where I shunt-modded the GTX 1080 and pushed more current and volts through it to compensate for the relatively smaller die, which in turn makes for a hot GPU. That way I can see better any differences between the various blocks in how well they cool.

 

[N3utro]

Thanks for your answer.

Ah yes i didn't see that it was Dt with the coolant temperature, my bad. But in some way that's even worse because the coolant temperature does not stay stable when the GPU gets hotter if the cooling is not enough so the ambiant air would actually be a better point of reference in that case.

And of course the rad/fans come into play even without throttling: if the coolant can't evacuate enough heat into the air then it becomes warmer and so the GPU gets warmer, which give false results on its real cooling capabilities.

I'm not an advocate of LM for the average end user either, but since you're an editor at TPU you're definitively not an average user and it would make your test more accurate

Also my Dt was with 1.1v. With 1.2 it's pretty close to that and since it's a 1080ti its also consuming 70W more than a regular 1080. Just saying that 45°C or more for a 1080 under custom watercooling is just wrong. That's the kind of temperature you can get with an AIO system. You shouldn't be above 35°C with a proper setup.

 

[erocker]

I can pick up an AIO for my 1080 for 89 bucks where GPU OC limitation will kick in well before heat does. These blocks are not a very good value at all.

 

[VSG]

Thanks for your answer.

Ah yes i didn't see that it was Dt with the coolant temperature, my bad. But in some way that's even worse because the coolant temperature does not stay stable when the GPU gets hotter if the cooling is not enough so the ambiant air would actually be a better point of reference in that case.

And of course the rad/fans come into play even without throttling: if the coolant can't evacuate enough heat into the air then it becomes warmer and so the GPU gets warmer, which give false results on its real cooling capabilities.

I'm not an advocate of LM for the average end user either, but since you're an editor at TPU you're definitively not an average user and it would make your test more accurate

Also my Dt was with 1.1v. With 1.2 it's pretty close to that and since it's a 1080ti its also consuming 70W more than a regular 1080. Just saying that 45°C or more for a 1080 under custom watercooling is just wrong. That's the kind of temperature you can get with an AIO system. You shouldn't be above 35°C with a proper setup.

Of course, and I use an environmental chamber with set point and steady state monitoring to keep variables as uniform as possible. There is no perfect solution, even fan speeds and clock bins vary over time but averaged out pretty well.

It's more than 1.2V, that's just the set value on Precision X before the shunt mod. I wasn't sure if I should be advertising it in the graphs at the time of the first review but now it doesn't matter anyway. My sample is also one of the most inefficient ones I've seen, but that's kinda what I wanted for thermal testing.

I've tested some blocks on a reference 1080 Ti as-is and the dT values are pretty much in range of expectations, including your own values.

 

[N3utro]

I can pick up an AIO for my 1080 for 89 bucks where GPU OC limitation will kick in well before heat does. These blocks are not a very good value at all.

You're wrong. The OC limitation on pascal cards (10xx series) is mainly due to their temperature. So with a 1080 card running at around 50°C you'll never get near the frequency that you could reach with one of these blocks and a proper custom watercooling.

My oced 1080ti starts freezing when temperature gets above 37°C, which you can never get with an AIO.

There is no perfect solution

Of course not, but if you can i suggest you include more rads in your loop. I have a phobya 1260 rad with 9x140mm fans with high static pressure and even with that i think if i added a second one it would probably yield even better results.

 

[EarthDog]

You're wrong. The OC limitation on pascal cards (10xx series) is mainly due to their temperature. So with a 1080 card running at around 50°C you'll never get near the frequency that you could reach with one of these blocks and a proper custom watercooling.

My oced 1080ti starts freezing when temperature gets above 37°C, which you can never get with an AIO.



Of course not, but if you can i suggest you include more rads in your loop. I have a phobya 1260 rad with 9x140mm fans with high static pressure and even with that i think if i added a second one it would probably yield even better results.

There is a point of diminishing returns... 9x140mm has squarely crossed that threshold for quite a bit.

37C??? Your card freezes where an air cooled card sits in idle??

 

[gamefoo21]

I have two issues with that block... the flow wheel you'll never see. Just a needless cost adder, due to parts, machining, and design time. Imho.

As well as the separate cold plate. It just adds another leak path. It also means the massive chunk of copper on the card, can't help cool/stabilize the GPU temps.

Also, as someone who has Koolance blocks with surprise Aluminium in them... I really hope there's no aluminium anywhere in that thing.

 

[erocker]

You're wrong. The OC limitation on pascal cards (10xx series) is mainly due to their temperature. So with a 1080 card running at around 50°C you'll never get near the frequency that you could reach with one of these blocks and a proper custom watercooling.

My oced 1080ti starts freezing when temperature gets above 37°C, which you can never get with an AIO.



Of course not, but if you can i suggest you include more rads in your loop. I have a phobya 1260 rad with 9x140mm fans with high static pressure and even with that i think if i added a second one it would probably yield even better results.

I'm not concerned with 50Mhz. I could add the card to my loop, but there's no point and it's a bad value.