Companies Need To Stop Releasing Top-end Components With Poor Cooling Capabilities

Our friends over at Gamers Nexus have been playing around with their new NVIDIA TITAN V graphics card a lot, to the point where if the card had feelings, it’d undoubtedly feel violated. As per usual with a top-end card, Steve’s decided to take the stock cooler off, and make his own makeshift liquid cooler to see how the card performs afterwards.

In his exploration of the TITAN V, Steve noted that the use of the Founders Edition cooler restricts the card’s full performance. It’s a great-looking cooler, but not a great-performing one – a surprise to no one at this point (I’d hope). It was noted that the TITAN V could easily lose 100MHz simply because the GPU gets hotter than the cooler can actually handle, and thus, performance is lost.

This ties into a complaint that’s been building inside of me for a while, that I haven’t felt like writing about to a great degree up to this point. This year, we’ve seen numerous examples of high-end components being released with abysmal cooling. Even AMD’s Radeon RX Vega cards have generic blower coolers that prevent the cards from keeping cool, but maybe it’s easier to accept that from a $400 or $500 GPU more than a $1,000 or $3,000 one.

When AMD released its Radeon Vega Frontier Edition, Gamers Nexus testing revealed the same issue we’ve seen with the TITAN V: the GPU runs too hot, and the cooler can’t do enough to help. It’d be one thing if better air solutions didn’t exist, but they do. People run out and buy vendor cards more than Founders Editions because they know that temperatures are important.

NVIDIA’s biggest, baddest GPU ever – held back by a sub par cooler

High temperatures kill performance – it’s that simple. The TITAN V does not operate at its peak performance because its cooler throttles it, and for a $3,000 offering, I am not sure how a legitimate excuse for it could be made. Again, it’s not just NVIDIA here, but AMD, and even Intel. Intel’s biggest chips – including the mammoth 18-core i9-7980XE, use sub par thermal interface material that prevents it from reaching its full potential without the most effective cooling.

These issues really began to speak to me when I took at look at the 10-core i9-7900X this past summer. It was the first CPU I’ve tested to make me realize that Corsair’s H100i V2 is no longer sufficient for the biggest chips going. I then upgraded to the Kraken X62, which handles the heat a bit better, but even it leaves me wanting more.

But here’s the problem: it doesn’t have to be like this. If Intel used decent TIM, like AMD does with its Ryzen and Threadripper chips, maybe these $150 coolers wouldn’t feel so ineffective. Maybe if NVIDIA used a decent cooler on the TITAN V, more people would be interested in spending $3,000 on it. It doesn’t matter if you’re an enthusiast or someone exploring deep-learning – everyone wants the best possible performance from their components. If you pay $3,000 for a GPU, why would you be happy to get $2,600 worth of performance? Or $900 worth of performance from a $1,000 GPU?

Intel’s biggest, baddest CPUs ever – held back by sub par thermal paste

At least on the desktop side, if you don’t like the Founders Edition cooler, you have a plethora of other options to ogle over. With TITAN V, there are no other options. The card is exclusively sold through NVIDIA, so the only way companies like EVGA would be able to release a special version is to buy the cards and modify them (and it’s extremely unlikely that’d happen given the limited market potential).

Call me naive, but I feel like if AMD can use good thermal paste on its lowest-end Ryzens, Intel should be able to use the same TIM on its highest-end processors. The same thought applies to the fastest GPUs out there. If cost is the issue, these companies should just tack on that extra dollar to the SRP. I’d pay that for significantly improved cooling.