Intel's Arc B580 has finally settled at its $250 MSRP, and for the first time in nearly a year, you can reliably find it at that price. That's significant news in a GPU market where alternatives like the GeForce RTX 5050 are complete garbage. The RTX 5060 is a better option, but it also costs more at $300, leaving the Radeon 9060 XT 8GB which is often available for ~$270 as the closest competitor, a modest 8% premium over Intel's card.
The problem is that all the budget Radeon and GeForce alternatives only come with 8GB of VRAM. This is what makes the Arc B580 so appealing. Intel equips the B580 with 12GB, enough to sidestep performance bottlenecks in modern titles, even at 1440p.
For budget gamers in 2025, limited VRAM has become the defining constraint, while Arc's main issue has been poor availability and inflated pricing. With pricing now apparently resolved, a tougher challenge remains in the form of CPU overhead, an inefficiency Intel publicly acknowledged five months ago, promising ongoing optimizations.
It's also been five months since we last examined that issue. So before updating our benchmarks against the RTX 5050, RTX 5060, and other relevant GPUs, we wanted to revisit how the B580 handles CPU overhead today.
To do that, we'll be testing the Arc B580 alongside the Ryzen 7 9800X3D, Ryzen 5 5600, and Ryzen 5 2600.
We will also include the 8GB version of the Radeon 9060 XT as a measuring stick – not necessarily to compare these two GPUs directly in performance, but rather to use the 9060 XT as a guide to determine whether the results are CPU- or GPU-limited. With that framework in mind, let's dive into the data.
Test System
| CPU | AMD Ryzen 7 9800X3D AMD Ryzen 5 5600 AMD Ryzen 5 2600 |
| Motherboard | Gigabyte X670E Master (BIOS F33f - ReBAR enabled) MSI MPG X570S Carbon MAX WiFi (BIOS 7D52v1B1 - ReBAR enabled) |
| Memory | G.Skill Trident Z5 RGB DDR5-6000 [CL30-38-38-96] G.Skill Ripjaws V Series 32GB DDR4-3600 [CL14-15-15-35] |
| Graphics Card | Radeon RX 9060 XT 8GB Intel Arc B580 12GB |
| ATX Case | Antec Flux Pro |
| Power Supply | Kolink Regulator Gold ATX 3.0 1200W |
| Storage | TeamGroup T-Force GE Pro M.2 PCIe Gen5 NVMe SSD 4TB |
| Operating System | Windows 11 24H2 |
| Display Driver | AMD Radeon Adrenalin 25.9.2 Intel Arc 32.0.101.8132 WHQL Certified |
Benchmarks
Rainbow Six Siege X
First up, we have Rainbow Six Siege. Although the 9060 XT is 30% faster in this example, that's not the main point of this test. What matters is that the Arc B580 delivered the same performance relative to the 9060 XT when paired with both the 9800X3D and the 5600, proving that the overhead issue isn't a problem here – at least for CPUs equal to or faster than the Ryzen 5 5600.
Where things get messy for the Arc GPU is with the much slower Ryzen 5 2600. At first glance, the data might look favorable, as it's only 18% slower than the 9060 XT, compared to 28% slower with the 5600.
However, this result is misleading. With the 2600, performance should be entirely CPU-limited, meaning the B580 should match the 9060 XT. Due to the overhead issue, that's not possible with this weaker CPU. Still, the 2600 is a very old and slow processor, so the fact that we see no overhead problem with the 5600 is encouraging.
Horizon Zero Dawn Remastered
Next, we have Horizon Zero Dawn Remastered, where the B580 shows some issues when paired with slower CPUs. With the 9800X3D, performance is strong, coming in just 14% slower than the 9060 XT, though the margin for 1% lows is a larger 22%.
When paired with the Ryzen 5 5600, the 9060 XT appears to be CPU-limited at 125 fps, though this is an assumption without broader testing. Even so, the B580 clearly struggles here, dropping to 99 fps – a 21% hit compared to what we saw with the 9800X3D. This is an issue because we know the B580 can render 126 fps on average when not CPU-limited. Assuming no overhead issue, this level of performance should also be possible with the 5600, as demonstrated by the 9060 XT.
With the Ryzen 5 2600, both GPUs are heavily CPU-limited, but not to the same frame rate. Typically, all GPUs should perform equally under these conditions – say, around 80 fps. Instead, the B580 is almost 20% slower, showing roughly a 20% CPU overhead.
Kingdom Come: Deliverance II
Testing with Kingdom Come: Deliverance II, we see no real issues for the B580, at least at the start of the game where we tested. The B580 and 9060 XT delivered similar results with either the 9800X3D or the 5600, though the Radeon GPU was about 25% faster.
We also see similar margins with the Ryzen 5 2600, so CPU performance isn't a concern in this case.
Warhammer 40,000: Space Marine 2
Next up we have Space Marine 2 . This game has previously highlighted the B580's CPU overhead issues, and unfortunately, that remains true today. The B580 manages only around 60 fps on medium settings, and this limit shows up with both the 9800X3D and the 5600, suggesting that current optimizations cap its performance.
When paired with the Ryzen 5 2600, the B580 falls further behind. While it ends up closer to the 9060 XT in raw performance, it should actually be capped at 52 fps, just like the Radeon GPU. Instead, we see a 21% drop, again entirely due to the CPU overhead problem.
Borderlands 4
Moving on to Borderlands 4, we have another example where the B580 is able to deliver similar performance in a CPU-demanding title when paired with both the 9800X3D and 5600, which is encouraging.
However, while it averaged 89 fps with the 9800X3D and 87 fps with the 5600, performance dropped to just 54 fps on the 2600. This is a real problem, considering the same CPU delivered 81 fps with the 9060 XT.
That translates to roughly a 33% overhead penalty for the B580 in this test, which is highly problematic for those with older systems.
Spider-Man 2
The B580's overhead issue was very apparent in Spider-Man Remastered, so we decided to test Spider-Man 2. Unfortunately, the results aren't great – just 64 fps on average with the 9800X3D and 60 fps with the 5600, along with a noticeable dip in 1% lows.
Using the Ryzen 5 2600, the B580 was nearly 40% slower than the 9060 XT, rendering only 44 fps. Without the overhead problem, it should realistically be pushing at least 60 fps here.
Marvel Rivals
Performance in Marvel Rivals looks much better in terms of CPU overhead. The B580 produced similar results with both the 9800X3D and 5600. With the 2600, some overhead is still evident, as it's 9% slower than the 9060 XT where performance should be equal. Still, this is far better than the 20 – 30% gaps we've observed in other titles.
Dying Light: The Beast
The good news for B580 owners interested in Dying Light: The Beast is that the CPU overhead issue doesn't appear to be a factor here. We observed very similar performance across all three CPUs. There was a slightly bigger drop-off with the 2600, but overall the difference was minor.
Cyberpunk 2077: Phantom Liberty
The results in Cyberpunk 2077: Phantom Liberty shocked me. Previously, when using ultra settings with high crowd density, the B580 struggled with CPUs like the 5600, and performance on the 2600 would have been disastrous. So what changed? Before explaining, let's look at one more unexpected case.
Spider-Man Remastered (Old side-by-side testing)
About five months ago, we revisited the Arc overhead issue after a large Russian YouTube channel publicly claimed the problem wasn't real. Ironically, after they doubled down, Intel confirmed the very next day that the overhead issue existed and reassured Arc owners that they were working on optimizations (Intel's official statement).
As a quick recap: with the high preset on the 9800X3D, the B580 was 13% faster than the RTX 4060. But when we switched to the 5600, the B580 dropped to 27% slower – a dramatic and concerning shift.
Spider-Man Remastered (New test)
Given those past results, we were puzzled to find that the B580 now matches the 9060 XT when paired with the 5600. Note that this data set uses the "very high" preset rather than "high," since it wasn't originally intended as a direct comparison to results from six months ago. Still, the B580 is clearly performing better, even with higher in-game quality settings.
That said, the slower 2600 still shows a sharp performance drop – around a 30% overhead hit. The problem hasn't disappeared, but Intel appears to have reduced the overhead seen in earlier versions. This led us to dig deeper into where the change occurred, and we found the answer.
Drivers Improvement: Comparison in Spider-Man Remastered
Here's how the B580 performs using a range of display drivers released in mid-2025. With the 9800X3D, all seven driver versions tested delivered the same or very similar results – no surprises there.
Switching to the Ryzen 5 5600 and checking four driver versions released between July and August, we see how slow the B580 was: just 87 fps on average, when it should have been CPU-limited to the same 117 fps as the 9060 XT. That's a typical 26% overhead hit.
What shocked us was the result with the latest driver. The B580 now reaches 118 fps, matching the 9060 XT. For systems with CPUs like the 5600, that's a 36% performance boost thanks solely to a display driver update. We narrowed down this major leap in performance to driver version 7028, released just four days after version 7026 in late August.
Confidence Boost: Arc Gets the Fix It Needed
Over the weekend, we reached out to our Intel contact on the Arc team to ask whether recent driver updates had finally addressed the long-standing CPU overhead issue. The response was clear: yes. Beginning with driver version 7028, Intel rolled out game-specific optimizations aimed at boosting performance on older processors. The team also confirmed that these fixes targeted titles we had previously flagged – explaining the improved results in Marvel's Spider-Man Remastered and apparent gains in Cyberpunk 2077.
It's a breakthrough, and the first tangible evidence that Intel can meaningfully mitigate the overhead problem – enough to make it a non-issue on a 3-year-old CPU like the Ryzen 5 5600.
Nine months ago, Intel privately acknowledged to us that our findings were accurate: Arc GPUs suffered sharp performance drops when paired with slower CPUs. At the time, the company promised it was investigating but offered no timetable. Months passed without visible progress. Even five months ago, when we re-tested the issue, nothing had changed. Just last week on the Hardware Unboxed podcast, we discussed how the Arc B580 and B570 had finally reached their MSRPs, and noted that re-examining the overhead issue in modern games would be worthwhile.
That curiosity led us to fresh testing – and to this discovery. As far as we know, no one outside Intel has reported on these optimizations. For nine months, the assumption was that this was an unfixable hardware flaw. That may still be partly true, but the new drivers show Intel can at least soften the problem in meaningful ways.
To be clear, the issue isn't gone. Even in optimized titles, overhead hasn't been eliminated. But the improvements make it far less of a concern. For most players, it may not matter anymore.
This speaks to how aggressive Intel's driver team has been. While it took nearly a year for this fix to surface, the company has released a steady drumbeat of driver updates in that time, patching issues and expanding support. That track record inspires more confidence in Arc than we've had before – and makes us more comfortable recommending hardware like the B580.
Arc B580 vs 9060 XT 8GB Performance
So how does the B580 actually stack up? Across the 10 games we tested, the Radeon 9060 XT 8GB was on average 34% faster with the Ryzen 7 9800X3D and 30% faster with the Ryzen 5 5600. On older CPUs, the gap widens further: with the Ryzen 5 2600, the 9060 XT was 27% faster, entirely due to the B580's overhead.
As things stand today, the 9060 XT 8GB costs ~10% more, yet delivers about 30% greater performance. The only drawback is its 8GB of VRAM. VRAM matters, and 12GB is certainly preferable, but raw performance still carries more weight. If the performance gap were only 10%, the VRAM advantage would make the B580 a stronger case. As it stands, the Radeon card remains in a different performance class.
Had we tested at 1440p with upscaling, the gap might have narrowed slightly, but we would still expect the 9060 XT 8GB to be at least 20% faster on average. That performance-to-price ratio is also why the 9060 XT remains available below MSRP. At $300, it's 20% more expensive than the B580 – a point where the Arc card starts looking more appealing.
We're keen to hear which budget GPU makes the most sense to you. We'll continue expanding this testing with more games and more budget GPUs, so stay tuned.