The RX 7600 and 7600 XT only improved over the RX 6650 XT by a few percentage points, but the 9060 XT should be a bigger upgrade.
You can already see that in the lower total board power (TBP) figures for both cards: The 16GB 9060 XT starts 30 W lower than the RX 7600 XT, and the 8GB 9060 XT starts 15 W lower than the vanilla RX 7600.
At 1440p, the 9060 XT usually runs between 22 and 62 percent faster than the RX 7600 XT in games without ray-tracing effects enabled.
Compared to Nvidia’s offerings, the 9060 XT sits between the $299 8GB RTX 5060 and the supposed-to-be-$429 16GB 5060 Ti.
Cyberpunk and even the somewhat less-demanding Returnal struggle more on the 8GB RX 7600 than they do on the 16GB RX 7600 XT.
With the release of the majority of Nvidia’s GeForce RTX 50-series graphics cards, it is evident that AMD and Intel have the best chance this decade to regain some market share and even out the playing field in the dedicated graphics card industry.
The 50-series GPUs haven’t been bad cards, but their emphasis on occasionally useful, occasionally non-AI-generated frames and their lack of significant manufacturing improvements over the 40-series have reduced Nvidia’s typically dominant performance and power efficiency.
The new Radeon RX 9060 XT graphics cards, which retail for $299 and $349 for 8GB and 16GB, respectively, are priced and configured to completely undercut Nvidia’s RTX 5060 and 5060 Ti. AMD is attempting to capitalize on this advantage. Similar to the RX 9070 series earlier this year, the RDNA 4 architecture significantly improves the mediocre power efficiency and poor ray-tracing performance of the RX 6000 and RX 7000 series. Additionally, a reasonably priced 16GB version will help shield consumers from the RAM constraints that are gradually becoming more of an issue for 8GB cards.
Let’s dig into one of the few truly exciting midrange GPUs we’ve seen in a long time, with the standard 2025 GPU disclaimer that the difference between a “good” and a “bad” GPU is whether you can actually find one and purchase it for anything near the list price.
Features and speeds of the RX 9060 XT.
Compute units (stream processors) RX 9070 XT RX 9070 RX 9060 XT RX 7600 XT RX 7600 64 RDNA4 (4,096) 56 RDNA4 (3,584) 32 RDNA4 (2,048) 32 RDNA3 (2,048) Boost Clock 2,970 MHz 2,520 MHz 3,130 MHz 2,755 MHz 2,655 MHz Memory Bus Width 256-bit 256-bit 128-bit 128-bit 128-bit 128-bit 128-bit Memory Bandwidth 650GB/s 320GB/s 288GB/s 288GB/s 288GB/s 288GB/s 288GB/s 288GB/s 288GB/s 288GB/s 288GB GDDR6 8 or 16GB GDDR6 16GB GDDR6 8 or 16GB GDDR6 16GB GDDR6 8GB GDDR6 16GB GDDR6 16GB GDDR6 16GB GDDR6 GDDR6 Total board power (TBP) 304 W 220 W 150 (8GB) or 160 W (16GB).
The RX 7600 and RX 7600 XT from the previous generation appear to be very similar to the 9060 XT series on paper. The 32 AMD compute units found in each of those GPUs are linked to 8GB or 16GB of GDDR6 through a 128-bit memory interface.
When the 9070 series was first released, AMD claimed that each RDNA4 CU could render about twice as quickly as an RDNA2 CU or 2–5 times faster for ray-tracing workloads. However, AMD claims that each RDNA4 CU is far more capable than in earlier architectures. Although the RX 7600 and 7600 XT only made a slight improvement over the RX 6650 XT, the 9060 XT ought to be a more significant improvement.
The Navi 44 silicon in the 9060 XT is also produced using a 4 nm TSMC process instead of the 7600 series’ 6 nm process. This is evident from the lower total board power (TBP) numbers for both cards: the 8GB 9060 XT starts 15 W lower than the vanilla RX 7600, and the 16GB 9060 XT starts 30 W lower than the RX 7600 XT.
The power consumption of the cards will differ slightly depending on the settings you choose in the Adrenalin software’s Tuning section and the settings used by the manufacturer of your GPU. Instead of releasing a single default TBP figure for the 9060 series, AMD formally releases a range of figures. That range is 150 W to 182 W for the 8GB card and 160 W to 182 W for the 16GB card.
According to AMD, performance-per-watt is at its peak at the 150 and 160 W baselines, which should be the optimal location on the efficiency curve. However, OEMs can select a higher default in an attempt to extract a little more performance, and users can typically use the “favor efficiency” or “favor performance” buttons in the Adrenalin app to configure their own GPUs to run at the bottom or top of that TBP range.
Performance can be improved by a card with a higher TBP because it should be able to sustain higher clock speeds for a little longer as long as it is properly cooled. The cards’ overclocking headroom should be somewhat increased by using a higher TBP setting, but you’ll also be at the mercy of the silicon lottery.
All of our tests show that the ASRock review unit’s default TBP was 170 W, which is marginally higher than the baseline. It only features one 8-pin power connector, and we’re willing to wager that the majority of GPUs from AMD’s partners come in the same configuration.
Users of slightly older systems will appreciate that the 9060 XT has a full 16-lane PCI Express 5.0 interface. The GPU doesn’t require 16 lanes of PCIe 5.0 bandwidth, but older PCs with only PCIe 4.0 or even 3.0 support should still have enough bandwidth for it. This has occasionally been a problem for midrange cards that switched to eight-lane PCIe interfaces.
Notes from the testbed.
The gaming testbed’s AMD Ryzen 7 9800X3D CPU, Asus ROG Crosshair X670E Hero motherboard, and 32GB (2x16GB) of RAM. DDR5-6000 SSD 1TB Samsung 990 Pro/1TB Crucial P3 Plus Power supply Thermaltake Toughpower GF A3 1050 W CPU cooler 360 mm MSI MAG CoreLiquid I360 Case Montech XR ATX Mid-tower with three 120 mm cooling fans installed and side panel removed OS Windows 11 24H2 with Core Isolation on and Memory Integrity off Drivers AMD RX 9060 XT: Beta driver 25.10. 09 points.
24.30 is the beta driver for the AMD RX 9070 series. Point 31.
Game Ready Driver 576.52 is for the Nvidia RTX 5060 series.
Beta driver 572.50 for the Nvidia RTX 5070.
32.0 is the driver for the Intel Arc B580. 101 Point 6874.
Nvidia cards other than the Game Ready Driver 566.36.
Additional AMD cards: 24.12 Adrenalin. 1. .
We’ve been using the same setup for our gaming testbed since the RTX 5090 review earlier this year. An eight-core Ryzen 7 9800X3D processor powers it, which is significantly more CPU than any of these midrange GPUs require. Generally speaking, they will function equally well with any newly released AMD Ryzen 5 or Intel Core i5 processor. But when we observe sluggish or otherwise strange results in our benchmarks, it guarantees that the CPU is never restricting the GPU, removing one crucial factor.
The 9060 XT was primarily tested at 1080p and 1440p resolutions, and its main competitors were the Radeon RX 7600, 7600 XT, 7700 XT, and 9070 from the previous generation; Nvidia’s GeForce RTX 4060, 5060, and 5060 Ti; and Intel’s Arc B580. For those who haven’t upgraded their GPU in a few years, we’ve also included some results for older cards, such as the AMD Radeon RX 6650 XT and the GeForce RTX 3060 and 3070, where available.
performance and power usage.
The RX 7600 and RX 7600 XT from the previous generation, which, despite the “XT” in the name, were essentially the same GPU with 16GB of RAM rather than 8GB, are essentially dropped-in replacements for the 9060 XT’s 8GB and 16GB models. Although we were only able to test the 16GB version of the card, we can draw some general conclusions by contrasting the 8GB and 16GB 7600 models and determining which games appear to suffer disproportionately from lower RAM.
When playing games without ray-tracing effects enabled, the 9060 XT typically outperforms the RX 7600 XT by 22–62 percent at 1440p. The 9060 XT is a compelling argument for an entry-level 1440p GPU because of that significant generational leap. These $300–$400 GPUs have long been marketed mainly as 1080p cards, and the 9060 XT will work well with a high-refresh-rate 1080p monitor. However, if they can find this card at or close to its list price, gamers who want to play games at 1440p for less than $400 have a good option. A last-generation RX 7700 XT (originally $449) or RX 6800 (originally $580, in 2020 dollars) are faster than the 9060 XT, but it’s not quite there.
When ray-tracing effects are enabled, the 9060 XT easily outperforms the 7700 XT and 6800. It also runs some games, such as Cyberpunk 2077, about twice as quickly as the previous 7600 XT. In this case, Nvidia still has a slight advantage because the 9060 XT is slower than the 5060 Ti but marginally faster than the RTX 5060. However, AMD has made significant progress, to the point where ray tracing is no longer nearly as problematic for AMD as it formerly was. Ray-tracing support is already starting to be required for some games, and it will probably become more widespread over the coming years, regardless of how much you care about these glitzy lighting effects.
The 9060 XT falls between the $299 8GB RTX 5060 and the rumored $429 16GB 5060 Ti in terms of Nvidia’s product line. In general, it performs slightly better than the 5060, but consistently, with the exception of Black Myth Wukong, where Nvidia cards continue to outperform AMD’s. In contrast, it typically lags behind the 5060 Ti by 10 to 15 percent. In fact, 16GB 5060 Ti cards are currently closer to $500, so if you can find one, the 9060 XT, which costs $349, is still a more cost-effective option.
The fact that the 16GB 9060 XT is typically 40–60% faster than the 2021 RTX 3060 is a significant victory for AMD. You’re probably still getting good performance out of one of those cards if you have one. However, unlike the RTX 4060 and 5060, AMD offers a reasonably priced upgrade path that increases your RAM from 12GB to 16GB instead of forcing you to reduce it to 8GB. AMD’s new card might help you play more games at 1440p or at higher settings if you have a 3060 and are looking for a reasonably priced solution.
Although it uses about the same amount of power as the 8GB RX 7600 or Nvidia’s RTX 3060, the 9060 XT performs significantly faster than both, making it a significant improvement in terms of power efficiency. The 5060 Ti still has a slight advantage in this case—power consumption is comparable, and performance is marginally better—but, similar to ray-tracing performance, the difference has narrowed enough that it is no longer a decisive factor in favor of Nvidia.
Can 8GB suffice?
8GB cards are useless garbage that you would throw in a hole before damaging your gaming PC if you frequently watch any of the tech YouTube channels that specialize in GPU reviews. Much of the complaining—not all of it, but a lot of it—involves setting the resolution of a game to 4K and all of the textures to Ultra, then complaining that it doesn’t work. This is similar to purchasing a Zippo lighter and complaining that it isn’t a flamethrower.
In these situations, 8GB of RAM is undoubtedly insufficient, but you’re also discussing settings and resolutions that the GPUs would find difficult to handle even with infinite video RAM.
All things considered, we are observing a few instances where our 8GB cards occasionally struggled with Ultra settings at 1440p or even 1080p resolution. These problems occasionally occurred intermittently; for example, cards such as the 8GB RTX 4060 would function normally in tests like Black Myth Wukong on some runs but suddenly transform into a flipbook on others. At high settings in 1440p, the 8GB version of the 7060 would not run the Black Myth Wukong benchmark at all, and when it was run at 1080p, it would eventually slow down or crash halfway through. On the 8GB RX 7600, Cyberpunk and even the somewhat less demanding Returnal perform worse than on the 16GB RX 7600 XT.
These are edge cases in specific games, and the problems are fixed by reducing the settings or resolution. However, you should try to stay away from 8GB cards if you’re spending $350 to $400 or more on a GPU in 2025. The 5060 Ti and 9060 XT are both fast enough to be respectable 1440p graphics cards when they have the proper amount of RAM installed. Previously, these low-end-to-midrange GPUs weren’t good for much above 1080p anyway. Are the 8GB models completely useless? No. However, they fail to consider performance, and the issue will worsen over time.
The GPU market as it stands right now.
The main problem with the 2025 graphics card market is not that popular cards like the 5060, 5070, or 9070 aren’t available, but rather that they aren’t often available for the listed retail price. The board partners of AMD, Nvidia, and Intel have always struggled with this, as they attempt to justify price increases by adding features like upscale cooling, lighting, or slight overclocks. MSRP cards are currently extremely uncommon, so it’s difficult to predict where the 9060 XT’s price will end up.
The best price for Intel’s “$249” Arc B580, according to Newegg listings for cards that are in stock, is $299. Gigabyte offers one RTX 5060 for $299, but the others cost at least $320. In reality, the $349 8GB model of the RTX 5060 Ti starts at $390, while the $429 model starts closer to $490. More like $600 or $610 is the $549 RTX 5070. In contrast, the $549 RX 9070 starts at $650, and the $599 RX 9070 XT costs at least $720.
As you can see, even though a lot of graphics cards are available, their true costs are typically between $50 and $100 higher than they should be.
If you’re in the market for a GPU and the 16GB RX 9060 XT is available for $349, you should buy it because it’s obviously better and more future-proof than anything else at that price, and the benefits of the 5060 Ti are completely negligible at a price that’s nearly $150 higher. If, however, you are unable to find anything at the advertised list price, be mindful of your other options and their true costs.
excellent under condition.
In my ideal world, graphic card design and manufacturing companies would announce new products and their prices, and customers could then visit their physical or virtual stores to purchase the items at the aforementioned price. Wouldn’t that be pleasant?
Due to a number of factors, including the demise of Moore’s Law, inflation, tariffs, and overall economic uncertainty, as well as the fact that businesses can profit significantly more from the sale of GPUs as AI accelerators than from the sale of similar silicon to video game players, that is not the reality of our world.
Every GPU review I’ve done in the past six months has included some discussion of the pricing and availability situation, which makes me feel like a broken record, but at this point, it’s simply inevitable. I can compare this GPU’s likely fictitious retail price to those of several other GPUs, but I’m not sure how much it will actually cost or how readily available it will be.
Assuming all current-generation graphics cards were sold at their stated MSRPs, I would argue that the RX 9060 XT 16GB model is currently the best midrange graphics card available and that most people should purchase it rather than an RTX 5060 or RTX 5060 Ti. In order to future-proof it, it is worth the extra money to leave the 8GB cards in the pre-built PCs at this time, even though the $299 version with 8GB of memory is probably still fine.
What will the RX 9060 XT really cost after everything is said and done? However, compared to AMD’s previous two GPU generations, the RX 90-series is more appealing and the RTX 50-series is unimpressive at baseline. Simply choosing an AMD card no longer means you have to put up with a significant drop in power efficiency or ray-tracing performance. Although Nvidia can still claim advantages like frame generation and DLSS upscaling, they aren’t as valuable as their contemporary FSR counterparts to warrant a high price tag.
the positive.
At least with the 16GB version, gaming performance in 1080p and 1440p is excellent.
reduced trade-offs compared to previous Radeon GPU generations, including significantly better ray-tracing and power efficiency.
The negative.
Even at 1440p, the RAM of the 8GB version will be a hindrance, and these issues will worsen over time rather than improve.
The DLSS from Nvidia is still more widely supported than FSR 3.1/FSR 4 and FSR frame generation.
The ugly.
the 8GB version’s RAM limitations.
The GPU market’s current situation.