NVIDIA GeForce RTX 5060 8GB review: 8GB just isn’t enough for a modern $300+ graphics card

The RTX 5060 runs well enough when it doesn't run out of VRAM, basically providing a horizontal shift from the RTX 4060 Ti 8GB for $100 less (two years later).

GPU Testbed
Asus RTX 5060 Dual OC
AMD Ryzen 7 9800X3D CPU
Asus ROG Crosshair 870E Hero
G.Skill 2x16GB DDR5
Crucial T705 4TB SSD
Corsair HX1500i PSU
Cooler Master 280mm AIO

The NVIDIA GeForce RTX 5060, with 8GB of VRAM, lands at the $299 price point, with factory overclocked cards pushing prices slightly higher. It replaces the RTX 4060 8GB in the product stack, with largely similar specifications that we'll get to in a moment. Built on the Blackwell architecture, the card promises higher performance than its predecessor, with GDDR7 memory boosting bandwidth by an impressive 65%. Still, 8GB of memory on a card that costs over $300 is a concern, much bigger now than when the 4060 launched — and it was a problem then as well!

We’ve been seeing Nvidia GPUs with 8GB of VRAM since the GTX 1070 in 2016. That was a $379 card at launch, which would be about $500 in today's money... but we do expect a certain element of computer components getting faster for the same price. Regardless, the VRAM is a potential sticking point. 1080p medium to high settings should work well enough, but maxed out settings, or even high settings with upscaling and frame generation, could push beyond the GPU's VRAM capacity. 8GB of VRAM simply doesn’t cut it anymore for a “new” GPU launching in 2025.

We’ve benchmarked the ASUS Dual RTX 5060, comparing it to a variety of current and previous generation GPUs, across our full suite of modern titles at 1080p, 1440p, and 4K. The short version? It’s generally a solid performer for 1080p gaming, but like its predecessors, the limited memory capacity becomes a serious concern the moment you crank up the settings. Indiana Jones and the Great Circle for instance flat-out refused to run on 8GB NVIDIA cards, even though AMD and Intel GPUs with 8GB managed to give it a go. In short, it's NVIDIA doing its best to marginalize the lower tier GPUs.

NVIDIA GeForce RTX xx60-class Specifications
Graphics Card	GeForce RTX 5060	GeForce RTX 4060	GeForce RTX 3060 12GB	GeForce RTX 2060
Architecture	Blackwell	Ada Lovelace	Ampere	Turing
GPU Codename	GB206	AD107	GA106	TU106
Process Technology	TSMC 4N	TSMC 4N	Samsung 8N	TSMC 12FFN
Billions of Transistors	21.9	18.9	12	10.8
Die Size in Square Millimeters	181	158.7	276	445
GPU Core Clusters	30	24	28	30
GPU Shader ALUs	3840	3072	3584	1920
Tensor Cores	120	96	112	240
Ray Tracing Cores	30	24	28	30
Render Output Units (ROPS)	48	48	48	48
Texture Mapping Units (TMUs)	120	96	112	120
Boost Clock in MHz	2497	2460	1777	1680
TFLOPS FP32	19.2	15.1	12.7	6.5
TFLOPS FP16 (FP4/FP8 TFLOPS)	153 (614)	121 (242)	102	52
Gigabytes of VRAM	8	8	12	6
VRAM Speed in GT/s	28	17	15	14
VRAM Bus Width	128	128	192	192
Memory Bandwidth in GB/s	448	272	360	336
L2 / Infinity Cache	32	24	3	3
Total Graphics Power in Watts	145	115	170	160
PCIe Lanes and Version	x8 5.0	x8 4.0	x16 4.0	x16 3.0
Launch Date	May 16, 2025	Jul 21, 2023	Feb 25, 2021	Jan 15, 2019
Official Launch MSRP	$299	$299	$329	$349

Compared to the previous generation 4060, the 5060 delivers a decent step up in compute, primarily due to using a larger die with more Streaming Multiprocessors (NVIDIA's name for GPU core clusters). It has the same MSRP, with the newer Blackwell features like GDDR7, PCIe 5.0 support, and FP4 support for the tensor cores.

Gen-on-gen, the 5060 offers a 27% improvement in base FP32 compute, 65% more memory bandwidth, and for AI workloads that actually utilize the FP4 format, there's a potential 154% boost in performance. Don't get too excited about that last bit, however, as so far the number of AI workloads that utilize FP4 seems very limited. Even NVIDIA's own DLSS 4 Transformers still uses FP8, or FP16 on earlier RTX architectures.

Note that there's also a 30W increase in power use compared to the 4060. Which, if we're keeping score — and we always keep score — gets pretty close to the RTX 4060 Ti 8GB card's 160W TGP. That's basically what you're going to end up with: a sidestep form the 4060 Ti 8GB, with a bit less raw compute, a lot more raw memory bandwidth, and some architectural enhancements.

ASUS RTX 5060 Dual Design

The ASUS Dual model used for testing follows the company’s tried-and-true budget to midrange design language. It comes with dual axial-tech fans, a full-length heatsink with a flow-through rear section, and a minor factory overclock. The card draws power from a single 8-pin connector and fits comfortably within a 145W envelope. Despite the seemingly higher TGP rating, it was still one of the most efficient GPUs we’ve tested in this price class.

Build quality is typical of ASUS’s Dual line: solid but not luxurious. It has a metal backplate, there's a Performance/Quiet VBIOS switch — frankly a luxury on this level of GPU — and of course no RGB lighting if you're into that sort of thing. Cooling was more than sufficient, with temperatures averaging 63°C under sustained gaming load, and fan noise peaking around 34dBA. Cool, quiet, and efficient — no complaints here. Let's hit the benchmarks.

Performance

For... reasons... our charts here aren't presented in what might be the best way possible, but it should work okay. The overall results are visible as a large image, and if you click on that, you'll get a lightbox where you can click through the rest of the individual game results. (Note that this will "overflow" into the other galleries, so like I said... not ideal.) The full set of charts can also be viewed in the video at the top. (I'm manually coding a table with images for these "galleries," so if someone has a better way to do a good, FREE gallery on Blogger, let me know in the comments! I've tried hosting images offsite, and that has other issues that I'd rather not deal with any more than necessary.)

Let’s start with the good news. At 1080p medium settings, the RTX 5060 largely matches the 4060 Ti while consuming less power. It’s also 23% faster than the RTX 4060, and 26% faster than the RTX 5050. Stepping back another generation, the 5060 delivers a 45% improvement over the 3060 12GB, and if we go all the way back to the 2060 6GB, it's just over double the performance.

Nearly every game in our test suite manages more than 60 FPS, with the exceptions being Cyberpunk 2077 (medium RT) and Final Fantasy XVI, and many of the games break the 100 FPS mark. Using quality mode DLSS upscaling would easily get every game above 60, and there's always frame generation, though the benefits are a bit harder to nail down in that case.

Things take a turn for the mediocre at 1080p ultra, though for the most part the 5060 remains viable. Overall, the relative performance between the 5060 and the 4060 and 4060 Ti 8GB stays the same as at 1080p medium. However, the lack of VRAM already starts to show up, with Indiana Jones and the Great Circle simply crashing to desktop on all 8GB NVIDIA GPUs — even though AMD and Intel cards with the same memory size continue to function. So the 5060 only comes in 30% faster than the 3060 12GB, and it's 11% slower than the 4060 Ti 16GB; it's also still 20% faster than the RX 7600 XT, down from 30% at 1080p medium.

If you're gunning for 60 FPS, of course, the story becomes even less positive. The overall average is exactly 60, but that means quite a few games are falling below that mark. In fact, just half of the games tested (seven) break 60 FPS, though other than the big donut on Indiana Jones, all remain playable at 40 FPS or higher. Also note that the 3060, which can run Indiana Jones thanks to its 12GB, manages 75 FPS in the game at 1080p ultra.

1440p ultra settings clearly pushes beyond the comfort zone of the RTX 5060. Many of the games are technically playable at over 30 FPS, with an overall average of 40 FPS. Again, older GPUs with more VRAM narrow the gap, by about 5% relative to 1080p ultra — so now the 5060 beats the 7600 XT by 16%, and the 3060 12GB by 25%). Only three games outright fail to come anywhere near 30 FPS (Indiana Jones and the Great Goose Egg, joined by Control Ultimate and Cyberpunk 2077 — not coincidentally, all three use ray tracing in our tests), though two more (Black Myth and Final Fantasy) are right on the borderline.

And it's worth pointing out that ultra settings are often more of a placebo than a massive leap in image fidelity. If you're playing a game at 1440p high, with DLSS upscaling enabled, performance ends up looking a lot like 1080p medium at native resolution. By that metric, you can totally handle 1440p on the 5060... just not at native resolution and maxed out settings. Still, the 8GB VRAM is clearly a limiting factor in many games at higher settings.

Obviously, our 4K ultra results aren't particularly useful for a card like the RTX 5060. Only three of the 15 games break the 30 FPS mark (God of War, Space Marine 2, and Witchfire), and none manage to reach anywhere near 60 FPS. Overall average frame rates dip into the teens, with some games landing in the single digits. Power draw also drops by about 25W relative to 1080p, not because the GPU is efficient but because it’s idling half the time waiting for data to move in and out of VRAM. Even the RTX 3060 12GB comes out ahead of the 5060 in our 4K testing, as does the RX 7600 XT.

DLSS upscaling and framegen — multi-frame generation, even — can help, especially if you drop to high settings. DLSS support continues to be wider and more mature than AMD’s FSR or Intel’s XeSS as well. Still, no amount of upscaling or multi-framegen can make up for inadequate VRAM. Borderlands 4 at high settings needs more than 8GB even at 1080p, for example, and we anticipate more games will behave in a similar fashion in the coming years. 8GB GPUs have been common for nearly a decade now, so anything claiming to be a midrange or higher tier card needs to move to at least 12GB going forward.

Power, Clocks, and Thermals

Our primary image in these galleries will be the 1080p ultra results here, as they're perhaps the most pertinent, though you can click through to see the other results. Average power use across our test suite comes in at 136W, still 9W below the rated 145W TGP. It's about a 15W increase from the prior generation 4060, pretty much matching our expectations. NVIDIA didn't move to a more advanced process node, so nearly all the gains end up being a matter of spending more power to achieve higher performance levels.

This becomes more obvious when looking at the overall efficiency charts. The 5060 ends up tied with the 4060 Ti, only slightly ahead of the 4060, and well behind cards like the 5060 Ti 16GB that don't faceplant into VRAM limitations. It is interesting to note that AMD's 9060 XT 16GB card only matches the 5060 as well.

There's not too much to say about clockspeeds, other than pointing out the huge step function gain when NVIDIA shifted from the Turing and Ampere architectures to the Ada and Blackwell architectures. The newer chips are designed to hit higher clocks, and they definitely do so. (AMD did the same thing a generation earlier with the RX 6000-series.)

Finally, we have temperatures, and they're a non-issue. The ASUS cooler easily keeps the GPU below 65°C under load, even in long-duration gaming sessions. Acoustic performance is similarly solid — not silent, but generally as quiet as our CPU cooler and case fans.

Conclusion: A Decent Card That Misses the Moment

The RTX 5060 is a capable, efficient 1080p gaming card. It easily outperforms the RTX 4060, runs cool and quiet, and benefits from NVIDIA’s strong software ecosystem (DLSS, frame generation, Reflex). For esports and lighter AAA workloads at 1080p, it’s a fine option. It can also handle 1440p at medium to high settings, especially if you enable DLSS and framegen — but then we can largely say the same for the prior generation RTX 4060.

But for a 2025 product, 8GB of VRAM feels painfully outdated. The 3060 from two generations ago shipped with 12GB on a 192-bit bus, and we really need mainstream GPUs to have at least 12GB. And NVIDIA could have done exactly that if it hadn't decided to stick with 2GB GDDR7 chips. Just imagine how much better than upcoming 'Super' refreshes with 3GB GDDR7 chips will look when they arrive next year... and then ask yourselves why we even needed these stopgap solutions. When a $300 GPU can’t run certain high-profile games at ultra settings due to VRAM limits, it’s hard to recommend the RTX 5060 without caveats.

NVIDIA’s focus has clearly shifted toward AI and data center markets, and the gaming division feels more like an afterthought these days. Instead of pushing boundaries, we’re treading water. The RTX 5060 doesn’t move the needle forward—it just holds the line while NVIDIA’s attention (and profits) are elsewhere. And we get that. It's hard to argue with over $100 billion in revenue from the data center side of things, compared to about one tenth as much for gaming GPUs. Hopefully the AI bubble bursts at some point and we can get back to playing games rather than generating mediocre text and images.

If you can stretch your budget, the RTX 5060 Ti 16GB or AMD’s Radeon RX 9060 XT 16GB offer more breathing room for the future. AMD's card is only a $50 upsell, the price of a single game. For that, you get about 20–30 percent more performance at 1080p/1440p ultra, and 75% higher performance at 4K ultra — with actually viable framerates in a lot of games. But you also lose out on DLSS support, if that matters to you.

Bottom line: For anyone on a strict budget who primarily games at medium to high settings with DLSS enabled, the RTX 5060 will get the job done. Just don’t expect it to age gracefully.