AMD Needs to Just Shut Up: AMD Radeon RX 9060 XT 16GB GPU Review

Unfortunately, we’re back to that because AMD can’t just launch a product and shut the f*ck up. It has to shove its foot so deep into its own mouth that it comes out the other side. 

Today, we’re reviewing AMD’s RX 9060 XT. The 9060 XT comes in 8GB and 16GB options, but beyond this, it comes with the usual side of AMD executive tweets.

AMD’s Frank Azor and guy who once bet Twitter $10 that he’d have better supply than NVIDIA during COVID said this:

“Majority of gamers are still playing at 1080p and have no use for more than 8GB of memory. Most played games WW are mostly esports games. We wouldn’t build it if there wasn’t a market for it. If 8GB isn’t right for you then there’s 16GB. Same GPU, no compromise, just memory options.”

First of all, AMD is the king of building things that there’s no market for. For example, these include:

• The 2004 Personal Internet Communicator it abandoned in under 2 years
• Bulldozer
• Anti-Lag+ the first time, before pulling it because it got people VAC banned
• The FX-9370
• The Radeon Pro Duo with almost no actual market
• Athlon 64 X2 3800+ EE SFF
• Radeon VII
• Vega Frontier Edition
• Vega 64
• AMD RX 6300
• The Radeon 6500 XT with PCIe Gen 4 x 4
• The HD 2900 XT
• AMD Fusion Render Cloud in 2009
• AMD Live!
• And finally, the AMD mountain bike, which is still the proudest we’ve ever been of a product that GN killed entirely on its own.

Secondly, Azor wrote, “same GPU, no compromise” while having half the VRAM is literally a compromise. That is the definition of a compromise in the same way Corsair’s i500 was such a compromise as an SFF PC that it removed all the “no compromise” language after our review. Rebranding “compromise” as “options” is the type of thing you put on your job application to become a Chief Architect of Marketing. 

Oh… OK, bad example.

In June 2020, AMD posted a blog post saying that 4GB of VRAM “is evidently not enough for today’s games.” In January 2022, AMD hides the blog post just ahead of the launch of the RX 5500 XT 4GB GPU, then panicked when it was discovered and unhid the blog post.

In 2020 after NVIDIA’s paid 8K marketing campaign, Azor tweeted “I guess we’ll scratch our 16K gaming message” in a jab at NVIDIA. 

In November 2020, AMD partners with Sony to slap a giant “8K” sticker on the PlayStation 5 boxes that it later removed the label from.

Also in November, 2020, Frank Azor tweeted about successfully buying his own product in a market with a massive GPU shortage presumably to save $10 from a bet.

In 2017, AMD posted a video that said “Poor Volta” before having its ass handed to it by NVIDIA for the next several generations.

Also in years past, the Fury is called the “overclocker’s dream.”

In 2016, AMD former Corporate Vice President Roy Taylor sat next to us at the RX 480 (watch our review) launch event and looked at us when we were looking up the Fury X (read our review) specs. He asked why we needed them. We said because we were trying to remember if it was 4GB of memory, and he said, “we’ll never make that mistake again.”

That’s the same launch event that had two RX 480s in CrossFire to show them beating a single GTX 1080 in Ashes of the Singularity, even though AMD’s marketing team ran the CPU benchmark for one brand and the GPU benchmark for the other in a completely incomparable benchmark.

AMD is single-handedly keeping Podithodontists employed by requiring so many emergency foot-from-mouth extraction surgeries.

The point is that an 8GB card should not exist for $300 today. Not for NVIDIA, not for AMD. 8GB cards do have a place — he’s right about that. But it’s not in $300 solutions. We had 8GB cards for $200 years ago when the RX 580s (read our review) were on fire sale. It’s as if VRAM has stopped advancing, and it is actively hurting gaming today. You’ll definitely notice this within a few years, but it’s also likely that it’ll affect the gaming experience for many people today.

If AMD wants to sell an 8GB “esports card” or whatever the company wants to call it, it can do that — but do it for the true low-end. AMD is just right back to playing NVIDIA minus $50. AMD has to invest in its brand, like Intel is doing with Arc, where the company is willing to take less money to establish market share to try and eventually grow longterm. AMD could also shut the f*ck up on Twitter, specifically getting the executives off the social media platform. It’s not helping. The investments are supposed to go into products and getting market share, not into one-upping NVIDIA’s f*ckups. Giving out review samples doesn’t undo weaponization of incompetence from AMD marketing.

GPU Test Bench

RX 9060 XT Thermals

We’ll run a thermal test to evaluate how the Sapphire Pulse model does. These lower-end cards sometimes get shafted with poorly applied thermal solutions or just mounting issues.

In a 100% rendering workload, the 9060 XT GPU temperature sensor climbed to about 56 degrees Celsius steady state. That’s completely acceptable. The hot spot temperature was 79 degrees, meaning that there’s about a 23-degree delta between the two. The highest we’ve seen for GPUs has been in the 40s for a poorly installed GPU cooler, which is why we pay attention to the delta to begin with. In this instance, 23 is higher than we’d like to see, but both numbers are well within tolerances. We can thank the relatively low power limit of this particular GPU for that more than the cooler.

Memory temperatures ran warm at 85 degrees Celsius in an ambient of 21C. We’d like to see this cooler as it leaves little room for aging and ramping internal case ambient temperatures, but there’s still enough room here that it’s OK overall. We evaluate this closer in our tear-down video.

One thing we noticed was the bump right at the start: The fans took a little bit to kick in and didn’t hit until the GPU bounced off of what we assume is the VBIOS temperature target of 56 degrees for the GPU. After this, they aggressively cooled down the card, then slowly ramped back to that temperature target.

Frequency Target

Next, we checked the frequency log to ensure the card is hitting the advertised clock speed. In the same test where the GPU leveled-out at 56 degrees Celsius, the core frequency held at about 3130 MHz or so. The advertised boost frequency is “up to” 3130 MHz according to AMD, so it’s hitting that marker.

RX 9060 XT Acoustics

We also ran a quick acoustic test in our hemi-anechoic chamber that we built. The RX 9060 XT Pulse ended up almost being too quiet to get a good read on, as it’s near our noise floor. The device measured at about 16 dBA under full load and tested at 1 meter distance in the chamber, shown by the green line in this plot with some 9070 (read our review) models for reference.

The 9060 XT Pulse ended up with a slight spike at 500 Hz, which we’ve seen in the other cards on this chart, and then otherwise a relatively gradual frequency falloff later on in the plot. Nothing jumped out at us here, which is a compliment for the Pulse. The low power draw of the card allows it to spin relatively slowly. It is likely that the other fans in your system would be louder than this GPU.

RX 9060 XT Game Benchmarks

Dragon’s Dogma 2 – 4K

Dragon’s Dogma 2 is up first. Tested at 4K, the RX 9060 XT outperforms the RTX 5060 GPU. We’re in unplayable territory, but this is still a useful tool to understand scaling across resolutions.

The RX 9060 XT leads the 5060 by 7% here, with both behind the former flagship 2080 Ti GPU. The jump from the 9060 XT to the 9070 is huge, at 79% improved to the next card up. AMD has a large performance gap between these models. In between, we can find the 7900 XT (read our revisit) and 7800 XT (read our review) also from AMD, with the 6950 XT around the levels of the 7900 GRE (read our review). NVIDIA competition additionally includes the 5060 Ti (read our review), which ran at 40 FPS AVG and outperformed the 9060 XT by 11%.

Dragon’s Dogma 2 – 1440p

At 1440p, the RX 9060 XT ran at a playable 65 FPS AVG with lows proportionately spaced. The lows for all the latest NVIDIA and AMD cards are indicative of consistent frametime pacing for this game.

The 65 FPS result positions the 9060 XT near the level of the 7700 XT and 3070 Ti, just below the 5060 Ti. The 5060 Ti has a 7% lead, with the 9060 XT leading the 5060 non-Ti by 14%. The 9070 is significantly better, at 106 FPS AVG to the 9060 XT’s 65 FPS AVG, so there is a large gap between these. Currently, that gap is filled by the 7800 XT, 7900 GRE, and 6950 XT of prior generations from AMD, or the 3080, 4070, and 5070 (read our review) from NVIDIA.

Dragon’s Dogma 2 – 1080p

At 1080p, the 9060 XT again sits around the 3070 Ti and 7700 XT levels of performance. That positions it ahead of the RTX 5060 by 13%, with the 5060 Ti ahead of the 9060 XT by 8%. The 9070 improves to 134 FPS AVG from the 86 FPS AVG of the 9060 XT, or 56%, with the 9070 XT (read our review) improving by 68.5% to 145 FPS AVG, just below the 5070 Ti and prior XTX flagship.

Generationally and by name only (which, again, has shifted in meaning for both AMD and NVIDIA), the 9060 XT improves on the 7600 (watch our review) by 42% and the 6600 XT by 48%. The 6600 XT launched at around $380, but was available later in its life commonly for $220 to $260.

FFXIV 4K

Final Fantasy 14: Dawntrail is up now, first at 4K.

By framerate alone, the RX 9060 XT is basically the exact same product as the RX 6700 XT here. The two are nearly identical in performance, though there is a spec difference, most notably with VRAM. The original MSRP for the 6700 XT was around $480, and later in its life it was often available for around $300 to $350. In other words, against only the launch price, AMD has come down about $130 to equal the performance of a card from 4 years ago. Listings for a used 6700 XT range from about $250 (sold) to about $400, with common options in the $280-$320 range.

The 4060 Ti is also about tied with the 9060 XT, with the 7700 XT slightly ahead. The 5060 leads the 9060 XT in this game and benchmark, part of our established fact that AMD just generally struggles in this benchmark right now as compared to NVIDIA. In fact, even Intel’s B580 (read our review) is outperforming the 9060 XT. In this test, the 9060 XT just doesn’t do that well overall. The same was true of the 9070 and 9070 XT in this testing.

FFXIV 1440p

At 1440p, the 9060 XT ran at 85 FPS AVG, which has it equivalent to the 6700 XT from 2021, slightly below the B580 (although with better 0.1% lows), and just barely ahead of the RTX 4060 Ti. This game is one of Intel’s more promising for performance gains against the incumbents and remains one of AMD’s weaker games.

The 5060 leads the 9060 XT by 2% for AVG FPS, with lows functionally the same. The RTX 5060 Ti at 104 FPS AVG leads the 9060 XT by 22%. As we’ve seen in the past, AMD’s competitive performance in FFXIV just isn’t as strong as in other titles.

Generationally, the 9060 XT leads the RX 6600 XT’s 63 FPS AVG by 36%, the 6600 by 62%, and NVIDIA’s once most common GPU, the GTX 1060 6GB, by 158%.

Starfield – 4K

Starfield is up next. Sometimes, people ask why we still test this game. Well, we figured it made sense to benchmark a game nobody plays with cards nobody wants to buy.

AMD’s RX 9060 XT ran at 39 FPS AVG, with lows proportional to the average. None of the NVIDIA or AMD cards on this chart have particularly strong or weak 1% and 0.1% lows. The RTX 5060 Ti 16GB at 41 FPS AVG outperforms the 9060 XT 16GB by 5%. 

Against the new RTX 5060 that we bought for $330, the RX 9060 XT leads by 18.8% in average framerate. Lows scale along with the average. We need to drop resolution for more cards to compare.

Starfield – 1440p

Dropping to 1440p, the RX 9060 XT ran at 62 FPS AVG and ended up between the 5060 Ti and 3070 Ti. The RTX 5060 Ti’s 65 FPS AVG is indistinguishably different from the 9060 XT from an experiential standpoint, but technically ahead by 6%.

Buying a higher-end card instead, like the 9070, would improve the performance in this title by 56%, from 62 FPS AVG to 96 FPS AVG. One of the cheapest RX 9070s in stock at the time of writing was $650, about $100 over MSRP, which means that a $350 9060 XT 16GB (if you could find it at that price) would achieve 64% of the performance for 54% of the price. We expect it’ll be more than $350.

Down the stack and against older generations, the 9060 XT distances itself from the 6700 XT this time (despite similar performance in other games), and improves upon the 6600 XT by 63%, the 6600 by 94%, and the 2060 by about the same. Intel’s B580 is similar to the RX 7600 in performance, but suffers in frametime consistency in this game. Its experience would be noticeably worse as a result.

Starfield – 1080p

Tested at 1080p, the 9060 XT ran at 77 FPS AVG and maintained consistent lows, just like the other AMD and NVIDIA cards at its flanks. The 5060 Ti runs just 4.7 FPS AVG higher framerate, or a 6% advantage. Against the 5060 non-Ti, the 9060 XT leads by 17%. 

AMD’s prior 7900 GRE, 6950 XT, and NVIDIA’s new 5070 all end up ahead of the 9060 XT in a similar clustered ranking. Beyond that, the 9070’s 118 FPS AVG is 52% ahead. By MSRP, which the 9070 still regularly isn’t, that’d be a 52% performance gain for a 57% price increase by MSRP, but we can’t be sure where the 9060 XT will land until launch.

Resident Evil 4 – 4K

Resident Evil 4 is up now. At 4K, the 9060 XT’s 56 FPS AVG has it about equal to the 5060 Ti. Neither has a frametime consistency advantage over the other and all of these metrics are within variance. They’re the same result. The 3070 Ti is just behind the 9060 XT, with the 7700 XT just ahead of the 5060 Ti and 9060 XT. AMD improves upon the RTX 5060’s 46 FPS AVG by 20%, which is the same improvement from the Intel A770 (read our revisit). In this game, the Arc GPUs do better for frametime consistency than they have elsewhere. The 9060 XT is also 22% improved over the Intel B580, although the B580 is commonly cheaper.

Generationally, noteworthy markers in this chart include the RTX 2060, RX 6600, RX 6600 XT, and RX 7600, all of which would be significantly improved upon with a 9060 XT or 5060 Ti-class GPU.

Resident Evil 4 – 1440p

At 1440p, the RX 9060 XT held a 106 FPS AVG, putting it between the 5060 Ti and 3070 Ti. The 5060 Ti is 6% ahead in average framerate. The 7700 XT and 4070 outperform both of these, with the 5070 notably ahead both in price and performance.

The 9070 again leaves a large gap between it and the 9060 XT, with a 62% improvement over the new card. Versus the 5060, the 9060 XT runs 17% ahead, about the same as its uplift over the 3070, 4060 Ti, and similar to the 6700 XT. Intel’s B580 does OK to hang on here, but isn’t keeping up with the newer 5060 and 9060 XT. It’s getting closer though, so we’ll see what Intel achieves in the next generation.

Versus older cards, users of the 2060 would see an improvement of 127% to the 9060 XT, 6600 owners would see the jump from 58 FPS to 106 FPS AVG, nearly doubling, and 6600 XT users would also see meaningful uplift.

Black Myth: Wukong – 4K

Black Myth: Wukong is up now, tested first at 4K just for scaling purposes. 

The 9060 XT ran at 25.6 FPS AVG here. That’s obviously not playable, but let’s look at scaling: Against the 5060, performance is about the same. The 5060 Ti is 22% ahead, so similar to what we’ve seen even in higher framerate scenarios. The 9070 improves on the 9060 XT by 61% here. Let’s move to a lower resolution.

Black Myth: Wukong – 1440p

At 1440p, the 9060 XT climbs to 49 FPS AVG. That has it at about the same level as the 7700 XT and 3070 Ti. The 5060 Ti is about 17% ahead of the 9060 XT in this test, with the 7800 XT similarly ahead. As for the new 5060, that provides functionally the same experience when not running into VRAM issues. The 3070 is around the same spot as the 5060.

The 9060 XT at least improves over prior generation parts in a meaningful way: Against the RX 6600 XT’s 29 FPS AVG, we saw a 66% improvement. Versus NVIDIA’s old and popular RTX 2060, the uplift is 94% from 25 FPS AVG.

Black Myth: Wukong – 1080p

1080p puts the 9060 XT into playable territory without any form of upscaling and with these quality settings. The 9060 XT ran at 71 FPS AVG with lows where we’d expect them. The 5060 Ti’s 81 FPS AVG has it 14% ahead of the 9060 XT here, with the 5060’s 66 FPS AVG giving up an 8% lead to the 9060 XT. This is a stronger title for NVIDIA even when rasterized.

Intel’s B580 does OK for frametime pacing this time, but ends up far down the stack and adjacent to the old RTX 3060 and prior generation RX 7600. It’s just not competitive here.

Dying Light 2 – 4K

Dying Light 2 is up now. 

4K isn’t worth much time: First of all, as we all know, the human eye can only see 30 FPS without running two of them in SLI, which enables 60 FPS visibility. One of our viewers with glasses in the last review pointed-out that his spectacles enable MFG 4X, multiplying the visible framerate by a further 2X on top of SLI eyeballs.

All this to say that, at 32 FPS AVG for the 9060 XT, clearly we’re past perfection anyway.

The 5060 Ti outperforms the 9060 XT by 20% here, similar to some of the other tests even in spite of the heavy workload. That’s why we include these tests, though. The 5060 is about the same. Let’s move on.

Dying Light 2 – 1440p

At 1440p, the 9060 XT ran at 63 FPS AVG with good lows — but not any better than any of its competitors’. The 3070 is slightly ahead and, critically, the B580 is actually about the same. Intel does well in this one, outperforming the 5060 in average and lows, while being functionally equal to the 9060 XT’s metrics. That’s more exciting than NVIDIA or AMD here.

But as for NVIDIA: The 5060 and 9060 XT would feel about the same in play in this game, with the 5060 Ti improving on the 9060 XT by 17%. Users of the now venerable RTX 3080 10GB can feel pretty good about their purchase, because the card is still hanging in there and keeping pace with modern GPUs, not too distant from the RX 9070.

Speaking of, the 9070’s uplift over the 9060 XT is 68%, putting them in totally different price and performance classes.

Cyberpunk 2077: Phantom Liberty – 4K

Cyberpunk 2077: Phantom Liberty is up next.

Tested at 4K, the RX 9060 XT ran at about 30 FPS AVG, so although there’s only a 6% lead for the 5060 Ti, we’re too constrained in performance to have a full picture yet. 

The 5060 fell notably below the 9060 XT here, with the AMD device holding a 21% advantage. Let’s see if that sticks at 1440p.

Cyberpunk 2077: Phantom Liberty – 1440p

At 1440p, the 9060 XT ran at 64 FPS AVG and landed between the 3070 Ti and 5060 Ti. The lead over the 5060 non-Ti drops to 16.7% here, down from about 21% at the less playable 4K. As for the 5060 Ti, its lead actually didn’t change much: It’s at 7.5% ahead now, from around 6% previously.

The 9060 XT is playable with these settings. Moving to something like a 5070 or 9070 would obviously be a huge jump up and would give some more room for higher settings or just future games, or even mods in this game, but that’s also a big price jump.

Cyberpunk 2077: Phantom Liberty – 1080p

At 1080p, the 9060 XT pushed up to a 100 FPS AVG. The 5060 Ti’s lead is now 8.8%, so we’ve seen it slowly increase its advantage over the 9060 XT as resolution has decreased. That’s why we run the 4K tests even when performance is too limited to play. As for the 5060, the 9060 XT leads it by 12% now, so it has consistently dropped from the 4K lead of about 21%. 

RX 9060 XT Ray Tracing Benchmarks

Ray tracing benchmarks are next.

Ray Tracing – Black Myth: Wukong 4K

We’ll start with the worst case scenario for everyone except NVIDIA, which is Black Myth: Wukong.

First at 4K upscaled, the 9060 XT ends up at the bottom of the stack. AMD has at least improved its ray tracing performance over the 7000 series, shown with the 7900 XT equaling the 9060 XT, but it’s not good enough to contend with NVIDIA (even if ignoring the low framerate at this resolution). The 5060 runs a 52% higher average framerate than the 9060 XT here. Let’s move to something that might actually run.

Ray Tracing – Black Myth: Wukong 1440p

At 1440p upscaled and ray traced, the 9060 XT ran at 31 FPS AVG and landed just ahead of the 2080 Super. That gives the 5060 non-Ti and its 45 FPS AVG a lead of 44%, down from the 52% previously. We’ll see if that continues to the more playable next resolution. The 5060 Ti ran at 54 FPS AVG, leading the 9060 XT by 72%. Even the 9070 XT was technically below the 5060 Ti, showing just how this particular game remains undefeated for NVIDIA.

Ray Tracing – Black Myth: Wukong 1080p

At 1080p upscaled, the 9060 XT ran at 47 FPS AVG. As always, for us, it’s not about the absolute framerate but instead about the relative framerate. In a relative sense, the 5060 ends up 37% ahead of the 9060 XT. In an absolute sense, it’s also just at a more playable FPS. The interesting part is the shrinking of the gap at lower resolutions here: Possibly because of how overrun it was, AMD’s GPU went from yielding a 52% advantage to the 5060, to 44% at 1440p, to 37% at 1080p. At this rate, maybe they’ll be at parity at 144p or something.

The 5060 Ti ends up 60% ahead here, with the 9070 XT around the same level. AMD gets left behind in this particular game with ray tracing on.

Ray Tracing – Dragon’s Dogma 2 4K

Dragon’s Dogma 2 with RT is up next. At 4K and not upscaled, just at normal 4K, the 9060 XT ran at 32 FPS AVG, basically tying the 2080 Ti and just behind the 3070 Ti. The 5060 Ti leads by 11% here, with the 9060 XT leading the 5060 non-Ti by 6.4%. Better performers include the 5070 at 49 FPS AVG, or 53% improved, and the 9070, at 76% improved. That breaks rank from what we saw in most rasterization tests. Of course, you could keep scaling up and we have results for that as well.

Ray Tracing – Dragon’s Dogma 2 1440p

At 1440p with RT and no upscaling, the 9060 XT’s 56 FPS AVG puts it well into playable territory with these settings. That has the 7700 XT just barely ahead, but the same from an experiential standpoint, and the 3070 is behind. The lead over the 5060’s 50 FPS AVG is 12% for a somewhat meaningful advantage. The 5060 Ti has a similar lead of 10% over the 9060 XT.

Compared to prior generations, AMD has improved on its architecture significantly and boosted its RT performance meaningfully. This is something we already saw in the 9070 XT review, but it continues at the lower end: The 9060 XT is a big jump over cards like the 6600 XT and 6600 previously.

Ray Tracing – Dragon’s Dogma 2 1080p

At 1080p, the 9060 XT’s 78 FPS AVG put it between the 7700 XT and 5060 Ti. The 5060 Ti’s lead drops from the 10% we saw at 1440p to just 5.5% at 1080p, with the 9060 XT increasing the gap against the 5060, now at 17% from 12% before. AMD is scaling better in this test as resolution decreases, even with ray tracing enabled.

Ray Tracing – Dying Light 2 1440p

We’re moving on to Dying Light 2 with ray tracing at 1440p upscaled. 

The 9060 XT was at 51 FPS AVG here and didn’t have any issues to speak of for frametime consistency. That has it ahead of the 5060 by less than 2 FPS, which wouldn’t be noticeable in real play. The 5060 Ti’s lead is more meaningful, at 17.8% over the 9060 XT. Overall, AMD is mostly improving on its own past performances here: Nearly hitting 7800 XT levels with the 9060 XT so at least that’s an improvement in RT.

Ray Tracing – Dying Light 2 1080p

At 1080p, the 9060 XT’s 77 FPS AVG has it again at about the level of the 7800 XT, just ahead of the 7700 XT. The 5060 Ti is about 13% better in average framerate, with the 9070 48% better than the 9060 XT and similar to the 5070.

AMD’s 9060 XT is 7.8% ahead of the RTX 5060 non-Ti here. The B580 is down below that, though not far below, and has OK frametime pacing this time. The 3060 Ti remains a weirdly relevant comparison to the 5060, which sort of feels like a 3060 Ti Super or something.

Ray Tracing – Resident Evil 4 4K

Resident Evil 4 ray traced is next, first at 4K upscaled. The RX 9060 XT 16GB ran at 63 FPS AVG here, putting it relatively close to the 5060 Ti’s 67 FPS AVG. The 5060 Ti has a lead of just 7% here, which is good for the 9060 XT. The 5060 gives the 9060 XT a 14% advantage, so it’s positioned better in this test than some of the other RT benchmarks. Lows are also consistent and proportional to the average, similar to the flanking devices from NVIDIA and AMD. Intel actually also had an OK overall performance here, with a 50 FPS AVG that begins to threaten the RTX 5060, though not the 9060 XT. Intel is mostly contending with the fact that it’s not consistently competitive, despite having improved this significantly with Battlemage.

Ray Tracing – Resident Evil 4 1440p

At 1440p upscaled and with RT, the RX 9060 XT’s 102 FPS AVG puts it just below the 7700 XT. The 5060 Ti is now 10% ahead, up from its 7% lead at 4K. The 9060 XT’s 10 FPS lead over the RTX 5060 non-Ti also reduces its advantage to 11% from 14% previously.

Ray Tracing – C77 VRAM Issues (1080p, RT Medium)

Finally, we’ll show the VRAM issues on the 8GB RTX 5060. These will also apply to the RX 9060 XT 8GB model.

In this ray tracing benchmark of Cyberpunk 2077: Phantom Liberty with 1080p/RT Medium settings, the RTX 5060 ran at 51 FPS AVG — which sounds OK — but had awful lows at 21 FPS for 0.1%. What’s important is looking at its neighbors: The RX 7800 XT had nearly the same average, but its lows indicated far more consistent and less choppy frametime pacing with the 43 FPS 0.1% result. The 9060 XT held a 54 FPS AVG and had well-timed, consistent frames, resulting in good 1% and 0.1% lows that represent an overall consistent experience. 8GB cards struggle here today, already, as shown with the 5060 and RTX 3070 Ti.

Ray Tracing – C77 VRAM Issues (4K, RT Medium)

To illustrate the point, here’s 4K with RT Medium. It’s ridiculous, but that’s the point: Despite struggling hard and overall being stuttery with its 15 FPS AVG, the 9060 XT had, to its credit, excellently paced frametimes for its average at 13.7 FPS and 13.6 FPS. There’s not a ton of data to work with here at this framerate, but the point is that even this sh*t experience is infinitely more playable than what we observe on cards like the RTX 3070 Ti, 3070, and RTX 5060, all of which are so variable from exceeding VRAM that the numbers are meaningless. We really can’t tell them apart: They’re all just unplayable in more ways than the 9060 XT is here.

So this is an illustration of VRAM limitations in real-time.

RX 9060 XT Efficiency Benchmarks

Up next, we’ll look at efficiency benchmarks when capturing GPU power consumption via the PCIe cables and the PCIe slot combined. This is isolated power draw to just the GPU. We’re representing efficiency in the form of FPS/W, or frames per joule, with higher being more efficient. The product name also contains the power drawn during the specific test.

Efficiency: Starfield 1440p

In Starfield at 1440p, the RX 9060 XT ended up slightly more efficient than the 9070 XT, though obviously with a lower framerate. The 9060 XT pulled 169W in this test, resulting in a 0.36 FPS/W result. The 4060 has a much lower framerate that’d push it into lower graphics settings, but is more efficient and has lower overall power draw. The RTX 5060 was also behind the 9060 XT in framerate, but also ahead in efficiency as a result of reduced total power draw of 126W for its framerate. The 5060 Ti pulled 145W and had a higher framerate than the 9060 XT, allowing it to approach the top of the charts with the 4060 Ti. Both of these are the 16GB models.

Efficiency: F1 24 RT 4K

F1 24 with RT is next and at 4K. The RTX 5060 couldn’t run this test without major stuttering, so it’s not on the charts.

The 9060 XT had a 0.17 FPS/W result, with the 5070 tying it also at 0.17 FPS/W, but with a 246W power draw. The 5060 Ti pulled 176W here. The 9060 XT is about the same power consumption as the 5060 Ti and only slightly behind in framerate, as this is one of AMD’s stronger RT titles. As a result, the efficiency is almost the same.

Efficiency: F1 24 RT 1080p

At 1080p with RT, the 9060 XT continued to pull about 178W. Its efficiency is better than that of the 9070 XT here, which is beyond the peak point for efficiency for the V-F curve for the 9070 XT. The 9060 XT is lower in framerate, but its lower power draw allows it this efficiency lead over its larger alternative.

The 5060 Ti ends up slightly more efficient than the 9060 XT, with both pulling around the same power and pushing a framerate within single digits of each other. This is a good showing for AMD’s new card.

Efficiency: FFXIV 1440p

We’ll look at a worse one for AMD as well. Final Fantasy XIV has poor overall gaming performance for the 9060 XT, which means the 9060 XT ends up lower down in the chart. It’s at 0.51 FPS/W, which at least has it improved on the 9070 XT and 7800 XT. The card is also tied with Intel’s B580, but related to NVIDIA, it’s behind the RTX 4060 and significantly behind the 5060 Ti.

RX 9060 XT Conclusion

8GB cards can exist, but we don’t think they should at $300. If AMD wants to sell an OEM piece of sh*t edition, the company is welcome to do that, but we think those should be relegated to true low-end devices. 

Selling the lower VRAM alternative opens up the door to less informed users, who are likely not in our audience, and getting them tricked into buying something worse than they’re led to believe. It also artificially deflates the MSRP when people talk about pricing.

As far as performance versus the RTX 5060, it’s at least better in nearly all instances. It’s often around levels of the 3070 Ti. It tends to be between the 3070 and 3070 Ti in a lot of tests. 

The 16GB 5060 Ti is currently $490 without a combo and without open box. The 5060 Ti is up to 14% better in our 1080p tests, up to 17% better at 1440p, and up to 22% better in our 4K tests. In ray-tracing games like Black Myth, it really pulls away, but that depends on the game.

As for price commentary, we’re not going to delve into that yet until we can see where the prices land. We’ll then do a GPU pricing comparison once things settle down as usual.

---