The AIO cooled GPU relies on a single 120 radiator to manage power consumption between 250 and 500 watts.
The AIO cooled GPU relies on a single 120 radiator to manage power consumption between 250 and 500 watts.
Hello,
This situation seems quite unusual. We understand that at least 240 radiators are required to overclock a CPU with an OC TDP around 140 watts. How can a 120 radiator support dual GPU cards with a TDP of 500 watts and still cool them effectively? And what about cards like the GTX 1080 Ti or Titans, which also have a TDP of 250 watts and are overclocked?
For example:
1. Radeon R9 295X2 with 500W TDP and only 120 radiators – how can it handle two GPUs?
2. GTX 1080 Ti, Hydro models, Titan Hydro, etc., with 250W TDP and 120 radiators – will they work?
Comparing this to an AIO water cooling solution like the I7 K at 140 watts OC, we’d need at least two times the radiator count... This raises questions about its practicality.
I would say 68C at load isn't particularly impressive. The most puzzling part is the production of numerous AIO coolers with 240 or 280 radiator sizes for CPUs rated between 80-130w, yet for GPUs with a TDP that's 2 to 3 times higher, they use just one 120mm unit. This highlights the limited mounting options in a case, making it hard to fit two separate 2x120 radiators, and points to a focus on CPU cooling over GPU needs even when thermal demands are greater at standard speeds.
Well, a 120 mm radiator is sufficient for cooling a GPU or CPU. The main issue with this configuration is the noise level. Fans need to spin very quickly to move enough air and maintain effective cooling.
I installed my overclocked i7 + overclocked GTX 1070 on a single 240 mm radiator. Although it was mostly copper, it had a low FPI (on most AIOS, the radiator is made of aluminum with a high fin density). It performed well in thermal tests, though not ideal. It was somewhat noisy, but definitely noticeable.
I decided to add another 240 mm radiator. Manufacturers often use 120 mm radiators because they are more compatible with cases. These "hybrid" cards also rely on liquid cooling for the GPU only, which means less heat needs to be dissipated compared to full cover blocks that actively cool the GPU, VRAM, and VRM (the latter generates a lot of heat). It's also worth noting that AIO radiators are typically used as exhausts, quickly removing heat from the case. With most air cooling setups, case fans handle the removal of hot air.
This approach is clearly superior to a reference or blower cooler, enabling lower GPU temperatures—leading to higher and more stable boost clocks.
Regarding your question, it makes sense for manufacturers to do this. However, there are very few practical scenarios where this solution is beneficial for end users, especially considering the price. Good custom coolers from brands like EVGA, Asus, Zotac, GB, and MSI provide adequate cooling at a much lower cost. Full cover blocks are not significantly more expensive (if any) and offer a better option if you already have a loop to connect your GPU. In the long run, a custom loop can even be cheaper than an AIO cooler.
Well, a 120 mm radiator is sufficient for cooling a GPU or CPU. The main issue with such a configuration is the noise. Fans need to spin very quickly to push enough air and maintain effective cooling. I installed my overclocked i7 + overclocked GTX 1070 on a single 240mm radiator. Although it was full copper, it had a low FPI (on most AIOs, the radiator is made of aluminum with a high fin density). It performed well in thermal tests, though not ideal. It was somewhat noisy, but definitely noticeable. I added another 240 mm radiator for better performance.
Manufacturers prefer 120mm radiators mainly because they are more compatible with cases. These "hybrid" cards often use liquid cooling for the GPU only, which means less heat needs to be dissipated compared to full cover blocks that actively cool the GPU, VRAM, and VRM (the latter generates a lot of heat). It's also worth noting that AIO radiators are typically used as exhausts, quickly removing heat from the case. With most air cooling setups, case fans handle the removal of hot air.
This approach is clearly superior to "reference" or blower coolers, enabling lower GPU temperatures—this in turn supports higher and more stable boost clocks.
So, your question makes sense: it’s logical for them to adopt this method.
However, there are very few practical scenarios where such a solution is worthwhile (especially from a cost perspective) for the average user. Good custom coolers from major brands like EVGA, Asus, Zotac, GB, and MSI provide adequate cooling at much lower prices. Full cover blocks are not significantly more expensive (if any) and are a better option if you already have a compatible loop to connect your GPU. In the long run, a custom loop can be even more economical than an AIO cooler.
Man, look at the AMD 295x2—it has a 500-watt TDP. There’s absolutely no way a 120 mm radiator can handle that much heat... it simply lacks the capacity. Your i7+1070 is only 300 watts total... and you’re using a 240 mm radiator, right? That’s a lot of pressure...
Your i7+1070 are 300 watts... and you’re putting 240 RAD in there. I really need an answer to this, lol.
The concern centers on delta values, and they acknowledge that using a higher delta on a GPU remains quite reasonable compared to a CPU, given GPUs are usually designed for 80-100°C temperatures. In that sense, 50 or 60°C load temperatures seem acceptable. However, a 500W TDP device with a 120mm radiator still seems excessive, which has been my point of view for a long time. The main advantage is that they offer a 'liquid cooled' high-end graphics card, giving the impression of a premium GPU experience and also being water-cooled from the start. I often emphasize this – simply because liquid cooling isn’t always the best doesn’t mean it’s ineffective.
The concern centers on delta values and the fact that running higher delta on a GPU is still considered reasonable compared to CPUs, given GPUs are usually designed for 80-100°C temperatures. In this context, 50 or 60°C temperatures seem acceptable. However, using a 500W TDP part with a 120mm radiator has been a point I've raised before. The main advantage is that it's marketed as a 'liquid cooled' high-end graphics card, promising a premium feel and cooling performance. I emphasize that liquid cooling doesn't automatically guarantee good cooling quality. Performance remains strong, with temperatures around 68°C during load and 33°C when idle. You can find more details here: http://www.guru3d.com/articles_pages/amd...ew,13.html And on TOM'S site: How is this possible with only a 120 radiator and a 500W power draw?
I believe 68C at load isn't particularly impressive. The most puzzling aspect is the production of numerous AIO coolers with 240 or 280 radiator sizes for CPUs rated between 80-130 watts, yet for GPUs with significantly higher TDP ratings they use a single 120mm unit. This raises questions about space constraints in cases and whether CPU cooling solutions are favored over GPU cooling despite greater thermal demands at standard speeds.
It's quite amusing how the R9 295x2 manages with both liquid and air cooling. The radiator is a bit more substantial than usual for AiOs, which helps maintain a stable temperature. It doesn't reach 500 watts; it stays around 400 at most. Given that older components are bigger (like a 28nm node), they naturally dissipate heat more efficiently due to increased surface area and better thermal conductivity.
P.S.
My i7+1070 handles at least 400 watts under heavy synthetic load. The CPU draws about 150-160 watts, the GPU around 250 watts (exact value uncertain because it's power modded). Overall system draw from the wall exceeds 500 watts (around 520-540). For safety, consider a 70-watt PSU and a 50-watt RAM setup. Even at 1300 RPM with low FPI 240 and high fan speeds, it can manage over 400 watts at 20°C difference between coolant and air. That would push GPU temps above 60°C, which is unsuitable for liquid cooling. My system rarely hits 50°C, usually staying between 38-45°C based on room conditions, fan settings, and workload.
n0ns3ns3 :
you are so funny.
R9 295x2 is cooled by both liquid and air. the rad is a bit thicker than "normal" AiOs. together, they can keep the card cool enough. this rad is not cooling 500wats, it gets to cool 400 at most. since older parts are relatively large (older node like 28nm), they are much easier to cool - larger die, more heat transfer at lower delta.
P.S.
my i7+1070 is at least 400watt combined under synthetic load. CPU is ~150-160w, GPU is ~250w (Can't tell exactly since it's power modded). Total system draw from the wall is well over 500w (520-540). so take 70 for PSU, and 50 for RAM and everything else beside CPU and GPU.
The point, is that even at ~1300 RPM, low fpi 240rad was enough to keep it well below any air cooler.
So thick 120 rad with high FPI + fans at 2000 or more RPM can dissipate over 400watts at delta 20C (difference between coolant and air).
that will result in GPU temperature in over 60C. this is too high for liquid whores. mine never reaches even 50C. mostly in 38-45C range depending on room temperature, fan profile and load.
First, no need for such a language okay ? lets keep this civil.
second , there is no way your system is 400watts at all , to say at least !!! even at Max it is no way .
i7+1070 will never reach 400watts .. it will be in the 300-350 range max ...