The stuff Hacker News actually recommends.

GPUs and CPUs · NVIDIA GeForce RTX

NVIDIA GeForce RTX 4070

11owners with an opinion
83%of verdicts positive
5/1positive/negative since 2024
0happy after 3+ years

Praised for performance and reliability.

What owners wrote

Neutral
This was with a nvidia 4070 attached to a 500 hz monitor. Even so P95 difference between worst and best was closes to 5ms. Meanwhile 60 hz monitors are incredibly common and every time a 16.7 ms frame deadline it causes an additional 16 ms of latency worse this is inconsistent and spiky and will be noticeable. The length of a blink is irrelevant. People notice latency especially inconsistent latency. The perception of xwayland being laggy in latency sensitive context like gaming is accurate.
Neutral
9b with 4bits runs with around 60 tok/s on my RTX 4070 with 12GB VRAM and 35b-A3B runs with around 14 tok/s and partial offloading. For roleplaying I prefer the faster 9b Version but for coding tasks both aren't really usable and Claude is still way better especially if you manage to persuade your employer to give you unlimited access.
Neutral
I've been on the same PSU for I think 13 years now, its currently running my Ryzen 7 3700x and RTX4070 desktop. I suppose if its not a great quality PSU or its already suspected of causing issues then replacing is a good idea.
Recommends it
Yeah I was exactly the same as you for years, holding out against what I considered to be unecessary exrtravagence. That was until I got a 4k monitor at work and experienced 4k HDR gaming. I immediately went out and bought an RTX 4070 and a 4k monitor and I will never be going back. The experience is glorious and I was a fool for not jumping sooner. 4K HDR gaming is not the future, is has been the standard for many years now for good reason.
Likes it
Depends what fast means. I’ve run llama and gemma3 on a base MacMini and it’s pretty decent for text processing. It has 16GB ram though which is mostly used by the GPU with inference. You need more juice for image stuff. My son’s gaming box has a 4070 and it’s about 25% faster the last time I compared. The mini is so cheap it’s worth trying out - you always find another use for it. Also the M4 sips power and is silent.
Likes it
I agree but it's a little deceptive. For example I have a 4070 right now and I paid $600 for it. That's good money but it is more than likely far more than most people actually need. If you watch or read reviews, you'd think only people in poverty use 4070s. I play competitive games and everything else I want to do with this card without issues and even with gas left in the tank. They crank up settings in reviews to ultra settings and then try to make it look like if you don't have a $1200 GPU that you have trash. Reality is that these GPUs are overkill and in many games medium settings look nearly identical to ultra. I swear GPU manufacturers pay to have ultra settings available, with their nearly imperceptible improvements. The option is mostly there as far as I can tell, simply to upsell GPUs.
Neutral
Thanks for the reply. I am currently testing the limits and got llama 3 70B in a 2bit-quantized form to run on my laptop with very low specs RTX3080 8GB VRAM (laptop version) and 16GB system RAM. It runs with 1,2 tokens/s which is a bit slow. The biggest issue however is the time it takes for the first token to be printed which fluctuates and takes between 1.8s to 45s. I tested the same model on a 4070 with 16GB VRAM (desktop pc version) and 32GB system RAM and it runs at about 3-4 tokens per second. The 4070 also has the issue with quite long time for the first token to be displayed i think it was around 12s in my limited testinh. I still try to find out how to speed the time to initial token up. 4 tokens a second is usable for many cases because that's about reading speed. There are also 1bit-quantized 70B models appearing so there might be ways to make it even a bit faster on consumer GPUs. I think we are at the bare edge of usability here and I keep testing. I can not tell exactly how this strong quantization affects output quality information about that is mixed and seems to depand on the form of quantization as well.