Intel Arc B570 'Battlemage' GPU Review & Benchmarks, Low-End CPU Tests, & Efficiency

Anyway, let’s get into the B570 review.

Intel Arc B570 Overview

Check out our B580 review if you want more detail on frametime pacing and on the basics of the architecture. For a quick recap of the specs:

The B580 GPU is a 20 Xe-core configuration on 5 render slices, a 190W TBP, and 12GB of GDDR6. Memory bandwidth is relatively high for a card of this class, at 456 GB/s. The B570 card drops 2 of the Xe cores to land at 18 and is still set up on 5 render slices. RT units get cut down as well, TBP is at 150W, and memory is set up at 10GB of GDDR6 with a 380GB/s bandwidth. The memory interface is a smaller 160 bits.

Here’s a quick pricing recap:

At the time of writing this ahead of launch, we checked Newegg and Amazon for several GPUs.

The Intel B580 GPU is mostly out of stock at the actual MSRP and being scalped at the time we’re writing this. On Newegg, it’s $340 and up right now, which is far too much to pay. It’s likely that when this story is published, the retailers will restock with B580s that are closer to normal, but we can’t review the future.

The RTX 4060 is currently around $295 to $320, with units regularly available at $300.

The AMD RX 7600 is $250 as of writing this, making it the closest price competition on a relatively modern architecture.

Enough of that. We have a lot of charts to get through today, so let’s start with the alternative CPU testing.

We are going to present charts that have 3 CPUs per card on them. The way to use these is not to compare the GPUs to each other, but rather to compare the same GPU against itself across 3 CPUs. This is to test whether Arc is disproportionately impacted by lower-end CPUs, meaning to test whether relative percent scaling against its competition worsens on a more likely pairing of a lower-end CPU.

We locked the 5600X to its single-core frequency except across all cores for one set of tests, then we used a 12400 for the other set. 

The current topic, at least as we’ve seen it mostly discussed in comments, is that Intel may be disproportionately hurt by driver overhead. Our testing is done with the intent to create a GPU bottleneck, as this isn’t a CPU review, and our content is done with maximum controls with the intent that you’d check both the CPU and GPU review. In this instance, we are still enacting controls, but potentially using a lower-end CPU. We kept our game settings the same, though, which means that even 1080p will often be GPU bottlenecked.

Intel Arc B570 Bounded Benchmarks

FFXIV - 1080p

This chart is for Final Fantasy 14 at 1080p.

We’ll look at all of the B580 rows first:

• With a 9800X3D, the B580 ran at 124 FPS AVG
• With a 5600X and a fixed all-core frequency, it was at 122 FPS AVG
• With the 12400, it was at 120 FPS AVG

There is no substantial change to the B580’s performance across these three CPUs with this test configuration.

Moving to the RTX 4060: The 4060 on the 9800X3D bench, 5600X bench, and 12400 bench performed identically. It was about 112-114 FPS AVG on all of them. There is no substantial change to performance.

The RX 7600 saw a performance drop on the i5-12400, but on the other two systems, was about the same.

With the B570, we observed approximately equivalent performance from the 12400, 5600X, and the 9800X3D (read our review). As such, the percent difference between all of these devices is overall comparable in this test, despite which CPU is used.

Across all tested devices, we did observe a reduction in 0.1% low performance as the CPU was reduced to lower performance levels. As an example, the RX 7600 with the 9800X3D, at 84 FPS 0.1%, versus with the 5600X at 62.1. This is expected.

FFXIV - 1440p

Here’s the test at 1440p. In this benchmark, we did not observe a meaningful difference in B580 performance across the 3 CPUs. The same is true for the RTX 4060 and the RX 7600. We are not able to detect any differences beyond slight variance across platforms. The B570 also does not scale differently and scores almost identically across the three platforms. The data consistency for average FPS is actually remarkable. This means that the unbound data later will be sufficient for establishing differences between averages.

As for lows, we are seeing consistently reduced lows on the lower-end CPUs in this game. The 5600X and 12400 both had worse 0.1% lows than the 9800X3D with the B570. With the RTX 4060, we’re seeing stable 0.1% lows across all 3 devices. With the RX 7700 XT (watch our review), we saw differing lows between the 9800X3D and 12400. With the RX 7600, we saw stable lows across all 3.

It would appear that, based on the B580, B570, and A770 (read our revisit) data, there may be an argument to worse lows on lower-end CPUs for Arc. We’ll need more games.

FFXIV - 4K

At 4K, everything is remarkably consistent. This mostly includes lows as well, although exceptions like the 9800X3D and 5600X for the A770 apply.

Broadly speaking, the GPUs are about the same here across all configurations here.

Cyberpunk - 1080p

This is Cyberpunk at 1080p/Ultra. We’ll go through it fast: The results do not change CPU-to-CPU. The only differences are on the RTX 4060, where we see worse lows on the lower-end CPUs, the A770, where we see a very slightly lower average, and the 7700 XT, where we see a change in 0.1% lows. The B580 and B570 do not exhibit a change. In fact, realistically, the CPU change does not affect any of these results with regard to how we do a standard review. They are the same. This is GPU-bound, so that’s what you’d expect. Some of the internet has lost the plot and think “Arc bad at 1080p,” but it’s not that simple. It depends on the GPU load and the game.

Starfield - 1080p

Starfield is known to be problematic for Arc. We have that discussion in the standalone chart, but we’ll focus only on scaling within a given card here. We did not observe a difference in the relative ranking of the B580 or B570 versus their competition in this test. Relative percent scaling did not get worse with lower-end CPUs.

At 1080p, there is a slight difference at the top-end with the 7700 XT and the 12400 as a potential CPU bind comes into play.

The RTX 4060 is identical in AVG FPS performance across all 3 CPUs; however, we observed worse frametime pacing on the 5600X and 12400 for the RTX 4060. Ultimately, they are the same level of reduction from the 9800X3D, so the two lower-end CPUs are comparable to each other.

The RX 7600 is identical in AVG FPS performance and within variance on all 3. Again, the 9800X3D has a slight advantage in frametime pacing.

The B580 is identical on all 3 for AVG FPS. There is also no meaningful difference in frametime pacing or 0.1% low performance. Note though that the B580 has some troublesome lows in this title already, but they don’t get worse on lower-end CPUs.

The B570 is the same and the results are within variance for all 3 metrics.

Resident Evil 4 - 1080p

This is Resident Evil 4 at 1080p.

The 7700 XT exhibits a slight ceiling with the 12400, but is mostly comparable. The B580 ran at 128 FPS AVG with the 9800X3D and 122 with the 5600X. We didn’t run this test on the 12400 configuration. The B570 had all 3, though: The 9800X3D has it at 110 FPS AVG, or 5% ahead of the 5600X configuration and 5.7% ahead of the 12400 configuration.

The RTX 4060 ran at 107.9 FPS AVG on the 9800X3D and 106.5 FPS AVG on the 12400. The RTX 4060 shows a top-to-bottom range of 1.3%, smaller than the observed difference on the B570. This slightly reconstructs the stack such that the B570 ends up tied or slightly under the RTX 4060 rather than ahead of it by 2.1%; however, the reduction puts the RTX 4060 on like-for-like hardware maximally 2% ahead. Between the two, they’re realistically about the same.

As for 0.1% low performance, the Arc cards underperform compared to the RTX 4060 on all CPUs. This does not get disproportionately worse on the 5600X or 12400 and instead scales with the average. The disadvantage in lows exists on the 9800X3D standard CPU and on the lower-end ones.

Baldur’s Gate - 1440p

Next up is Baldur’s Gate 3. We only run this test at 1440p and 4K due to the low load. 

In this test, the RX 7700 XT found a hard bottleneck with the i5-12400. This capped out at 79 FPS AVG, as opposed to the unbound test at 133 FPS AVG. This would be appropriate for a CPU review, but does not show GPU performance. The RTX 4060 on the 12400 further reinforces these limits (although bounces off of them way harder, with much worse 0.1% lows as a result of the CPU, not the GPU). The RTX 4060 exhibits identical AVG FPS with all 3 configurations. The lows are better on the 9800X3D as it allows the GPU to hit frame pacing without external binds, which is expected behavior.

The RX 7600 is restricted by the CPUs with the 5600X and i5-12400, with both lower than the 9800X3D’s result. Again, this is normal and is the expected result from a CPU bind. Lows are hurt as a result of limiting the ceiling.

The B580 also experienced a performance reduction with the 12400, but a much larger one with the R5 5600X. To do some quick math:

The RX 7600 at 77.1 FPS AVG with the 12400 gains 11.1% by moving to the 9800X3D.

The B580 with the 12400 at 63.7 FPS AVG gains 16.2% by moving to the 9800X3D. That’s not as a big of a difference as we might expect, but it is larger than we saw on the 7600.

The 5600X is a far different story, with a maximum uplift of 32% from the 5600X unit to the 9800X3D at 74 FPS AVG. This may be the driver overhead.

The B570 shows similar behavior, with scaling worse on the 5600X than on the 12400. Lows are not disproportionately worse on any of them; in other words, the lows follow the behavior relative to other devices also scaling with the CPU.

This would possibly change our recommendation for a 5600X user, but not a 12400 user where there is only a 3 FPS difference.

Baldur’s Gate 3 - 4K

At 4K, the 7700 XT is the same on both CPUs for AVG FPS, but worse in frametime pacing on the 12400. This is expected behavior when using a worse CPU.

The A770 has identical performance between the 9800X3D and 5600X.

The RTX 4060 has identical performance between all 3. There is a 1.3 FPS AVG difference as a result of not only run-to-run variance between the 3, but also the frametime pacing is also much better on the 9800X3D for the RTX 4060, which will influence the AVG FPS results.

The RX 7600 has the same AVG FPS on all 3 CPUs, with better frametime pacing on the 9800X3D.

The B580 has identical performance on all 3 CPUs for AVG FPS, with a slight 1% low advantage on the 9800X3D. 

The B570 is identical on all 3 CPUs, with better lows on the 9800X3D.

All of this behavior is expected for a GPU bind. 

Black Myth - 1080p

Black Myth will be brief as well: At 1080p/High, which is our standardized test platform, there is no meaningful, broad difference in scaling.

The numbers do not change in any meaningful way. The 7700 XT is the same on both, the 4060 is the same on all 3, the 7600 is the same on all 3 (with the exception of reduced lows on the 12400), the B580 is the same on all 3 and lows are within variance run-to-run. The B570 is the same on all 3, with lows within the wider variance at the low-end. The A770 is the same on both.

Dragon’s Dogma 2 - 1080p

We’ll look at one more. 

In Dragon’s Dogma 2, they are all the same from CPU-to-CPU, with a few notes.

The 7700 XT is roughly identical. The RTX 4060 is mostly identical, with the 12400 potentially showing marginally worse performance. The RX 7600 is the same across all 3, with slightly worse 0.1% lows. 

The B580 is the same across all 3, with worse 0.1% lows on the 5600X specifically. The B570 is the same across all 3 in all 3 metrics.

The A770 is the same in all 3 metrics.

Discussion

Let’s quickly talk results: First, in most scenarios, there is no meaningful impact. The biggest situation was with the 5600X in Baldur’s Gate 3. The 12400 did not react as violently. The secondary impact is in some 0.1% low differences, but these often also manifested on other devices.

This does not affect how we review GPUs. We have to make that clear here. There is no data to support that our conclusions would be different if we ran lower-end CPUs with these cards. It would just cap the ceiling of how much change you can see. 

That’s not to say there isn’t driver overhead or that there can’t be. We are aware that multiple outlets have now shown this. However, we tested with High and Ultra settings and simply were not able to replicate the impact. Our belief is that in scenarios of extremely high framerate, such as 200+, or in situations where the CPU is heavily bound from low settings and lower resolutions, that’d be where you’d see the difference. But we don’t test GPUs for review with low settings and we currently don’t test esports titles, so these would not show up in our data.

Others do. We’d encourage you to check them, as everyone in this space provides a different set of tests for unique value. This story is long enough as is, but what we can say is that our entire usual suite has now been validated on two low-end CPUs and we have found the performance does not meaningfully change the conclusions except perhaps in one main scenario tested.

Intel Arc B570 Unbound Testing

FFXIV - 4K

Final Fantasy 14 at 4K is up first for the unbound benchmarks on the 9800X3D.

In this one, the B570 ran at about 41 FPS AVG, establishing a 16% advantage for the B580 at 47 FPS AVG against a 14% price bump by MSRP. This is pretty close to linear. The B570 roughly ties the A770 in AVG FPS and 1% low, with 0.1% lows close enough. It also outperforms the RTX 4060 by 11%, the 3060 by 23%, and the A580 (read our review) by the same. The RX 7600 card landed at around 32 FPS AVG, giving the B570 a notable lead, especially at the price. There are also substantial improvements on the RX 6600, which was the previous leader in this price class (though it eventually sold for cheaper).

For some references to older cards, we also have the GTX 1070 (watch our review), GTX 1060 watch our review), and GTX 1650 (watch our review) present. 

FFXIV - 1440p

At 1440p, the B570 ran at 75 FPS AVG against the B580’s 86 FPS AVG, which puts the B580 about 15% ahead of the B570. It had a marginally higher lead at 4K. The B570 now sits between the A770 and A750 (watch our review), so the A770 has slightly re-positioned itself as resolution came down. The RTX 4060 also gets closer to the B570, now at 72 FPS AVG, nearing equivalence despite a larger gap previously.

This is the same behavior we saw in the B580 review: The higher the resolution goes, generally speaking, the more of an advantage Battlemage has over its peers in the class. The lead over the RX 7600 is 14% here, with the RX 6600 XT and 3060 (watch our review) below it. Some older reference anchors include the RTX 2060 (watch our review), GTX 1070, 1060, and 1650.

Frametime pacing on the B570 is good overall, performing similarly to the RTX 4060 for frame-to-frame interval. 

FFXIV - 1080p

At 1080p, the B570 loses its advantage against the RX 7600 and RTX 4060. The B580 maintains a lead over these two cards with its 124 FPS AVG, although its frametime pacing, indicated by the 0.1% lows, is worse than what we’re seeing on the RTX 4060 and RX 7600. We dove into this in the B580 review.

Overall, Battlemage and the B570 remain a large improvement over Alchemist, especially at the price; however, we still see that reduction in relative performance as resolution decreases, even with an unbound configuration. At 1440p and higher, Battlemage when unbound seems to be more competitive than it is at 1080p.

Starfield - 1440p

Starfield is next. Intel had a lot of problems with Starfield when the game launched -- namely, the problem was that it didn’t work.

But it does now. Despite working, NVIDIA and AMD retain an advantage with the RTX 4060 and RX 7600 cards. The B580’s 41 FPS AVG leads the B570’s 36 FPS by 12%. The RX 7600 leads by 17%, with the 4060 leading by a staggering 31%. This is an example of one of the games where Battlemage is disadvantaged even when unbound, even at 1440p.

Starfield - 1080p

At 1080p, the B570 GPU ran at 47 FPS AVG, planting it between the A770 of last-gen and the RX 6600 XT. Against Alchemist, it’s clear that Battlemage is a huge improvement when considering the configuration size of the A770 GPU against the B570 GPU. Against itself, Intel has definitely improved in huge ways. Against competition, the RTX 4060 and RX 7600 both hold advantages at both the resolution and with the game.

The B580 ends up 10% ahead of the B570 in this test, so its value isn’t always linear.

Resident Evil 4 - 4K

Resident Evil 4 at 4K was one of the B580’s strongest titles. In this test, the B580 marginally outperformed the RTX 4060 Ti (watch our review) in averages, with the 0.1% lows close enough to be comparable given wider margins. The B580 at 46 FPS AVG leads the B570’s 40 FPS AVG by about 16%, aligning with some of the other 4K results. The B570 ends up outperforming the RX 7600 by 7% and the RTX 4060 by 14%, which is an incredible lead for the B570 when fully GPU-bound. Considering the RTX 4060’s price, this is a big lead for Intel.

The improvement on the A580 is also noteworthy, with similar gains.

Lows on the B570 are not meaningfully different from those on the RX 7600, 6600 XT, or RTX 4060. The frame pacing is comparable.

Resident Evil 4 - 1440p

At 1440p, the B580 is still spaced about 16% over the B570. The B570’s lead over the RX 7600 has reduced and they are now mostly equivalent, with the B570 experiencing less consistent lows. This is something we showed in our B580 review, if you’d like to learn more about why it hits this ceiling.

The RTX 4060 also climbs relative to the B570, with its 70 FPS AVG result reducing the B570’s advantage to just 6%. This is while the RTX 4060 maintains slightly better 0.1% low results for a better average frame-to-frame interval. The B570 is still overall better here, but as we continue to reduce resolution, that’ll change.

Resident Evil 4 - 1080p

This is 1080p. With these settings, the B570 is now roughly tied with the RX 7600. The RTX 4060 is encroaching on the B570, which has had its lead when GPU-bound cut to 2%. That’s still impressive for the price and for the age of Intel’s new architecture. If we imagine the B570 line item were some hypothetical new NVIDIA card at $220, this would be an amazing move in the right direction for NVIDIA. The fact that Intel is still proving itself and working through some teething pains is the only hesitation that remains, though they have reduced that generationally so far.

Even still, the B570’s frametime pacing is inconsistent. It bounces around in this title, with run-to-run variance higher than we see on any AMD or NVIDIA device. This remains a challenge in some specific games. It’s overall acceptable, but not as consistent as its peers.

As for the B580, its lead over the B570 is about 18 FPS, or 16.4%.

Baldur’s Gate 3 - 4K

Next is Baldur’s Gate 3, which remains one of the best titles launched in the last couple years.

At 4K, the B570 ran at 40 FPS AVG with our Ultra settings. This has it just below the RTX 3060 and above the A580, RX 6600 (watch our review), and RTX 2060. The B580’s 44.8 FPS AVG result leads the B570 by 13%, with both exhibiting limited scaling in the 0.1% lows, indicating frame-to-frame consistency issues as compared to NVIDIA neighbors.

Baldur’s Gate 3 - 1440p

1440p has the B570 at 67 FPS AVG, between the A750 and A580. The generational uplift isn’t nearly as impressive here as in some of the other games. The B580 is about 11% better than the B570 here, with the RTX 4060’s 80 FPS AVG putting it 20% improved in average FPS, plus a significant uplift in 0.1% lows to represent smoother frametime pacing. The RX 7600 outperforms both the RTX 4060 and B570, including in lows.

Cyberpunk: Phantom Liberty - 1440p

Let’s move on to Cyberpunk: Phantom Liberty at 1440p.

In this test, the B570 ran at 46 FPS AVG, with the B580 at 54 FPS AVG. The B580’s lead is about 16%, which is consistent with other high load scenarios; in fact, the B580 bordered on the RTX 4070 in this test, which we didn’t see happen elsewhere but which has consistently happened in this game. Its lows aren’t nearly as good as what the RTX 4070 offers (at 50 FPS 0.1% to 31 FPS), but this was still a good showing for Battlemage.

The B570 reinforces that, functionally tying in average with the RX 6700 XT (watch our review). This is an impressive feat. For both, it’s the memory bandwidth benefitting the cards here. Unfortunately, the low pacing just isn’t as good on the B570 as its neighboring devices, including the larger config A750. The B570 leads the 4060 by 18% here in a staggering flip favoring Battlemage for averages, again with the caveat of the lows being better on NVIDIA.

Cyberpunk: Phantom Liberty - 1080p

At 1080p, the B570 ran at 71 FPS AVG, with the B580 sustaining a 13% uplift in average framerate. Battlemage still has worse lows than neighbors. At this resolution, the RTX 4060 begins to encroach on B570 and functionally ties it in average framerate while offering substantially improved frame pacing behavior. The RX 7600 is the same, roughly tied with the RTX 4060 and overall a better experience than the B570.

Black Myth: Wukong - 1080p

Black Myth: Wukong is up next, tested at 1080p/high first. With these settings, the B570 maintains about 40 FPS AVG with lows around or below 30 FPS averaged. This has the RX 6600 XT (watch our review) and RTX 3060 both above the B570, with the B570 above the RX 6600 non-XT (watch our review) and RTX 2060. The B580’s 46 FPS AVG gives it a lead of 15%, with the RX 7600 16% ahead and RTX 4060 (at 54 FPS AVG) ahead of the B570 by 36%. That’s a huge lead for the 4060.

Black Myth: Wukong - 1440p

At 1440p, the B570 is reduced to 30 FPS AVG. This gives the RTX 4060 a 22% lead, with the B580’s 35 FPS about 15% ahead.

Dragon’s Dogma 2 - 1440p

Dragon’s Dogma 2 is up next, first at 1440p/Max without RT. In this test, the B580 sticks to a 15% lead over the B570, so that’s consistent. The RTX 4060 leads the B580 and the B570 alike, with the RX 7600 about tied with the RTX 4060 and within run-to-run variance. The B570 is not particularly strong here. It’s about the same as the 3060 and outperforms the RX 6600. The price is noteworthy, although the B580 is overall more competitive in raw performance. In the least, lows are comparable in this test to non-Intel parts.

Dragon’s Dogma 2 - 1080p

At 1080p, performance falls overall as compared to competition. This is consistent with other tests. The gap from the B570 to the B580 remains 15%, with the 4060 now leading the B570 by 30% and the B580 by 7%. AMD’s RX 7600 also holds a similar lead.

Dying Light 2 - 1440p

Dying Light 2 is up now. At 1440p, the B570 ran at 53 FPS AVG in what’s one of the strongest relative showings unbound for the card. This has it ahead of the RTX 4060 by 8%. Lows are also better, which is important for Arc. The lead over the RX 7600 is 13%. As for the B580, Intel’s $250 baseline model is about 18.6% ahead of the B570 here.

Dying Light 2 - 1080p

At 1080p, the B580 maintained an 18.7% advantage at 87 FPS to 73 FPS AVG over the B570. The RTX 4060 lands between the two and is closer to the B570. Intel has marginally better 0.1% low averages here. The RX 7600 is roughly tied with the B570 in this test.

Intel Arc B570 Ray Tracing Benchmarks

We’re moving on to ray tracing now, tested with the unbound configuration. We’ll keep this as focused as possible. These tests are not comparable to the rasterization performance, especially as some settings (such as upscaling) may change.

Ray Tracing: Resident Evil 4 - 4K

In Resident Evil 4 at 4K with RT and FSR, the B570 ran at 43 FPS AVG and with lows overall acceptable for its average. That has it ahead of the RTX 4060 in a strong showing for Intel. The B580 leads the B570 by 16%, which puts it functionally identical in performance to the RTX 4060 Ti.

Ray Tracing: Resident Evil 4 - 1440p

At 1440p, the B570 ran at 66 FPS AVG and tied with the RX 7600 in every metric. Considering the B570’s price advantage at time of writing, that’s a competitive spot to be. The RTX 4060 leads the B570 by 8%, flipping from what we saw at the higher resolution and consistent with knowledge that Arc disproportionately scales better at higher resolution. The B580 is 15% ahead of the B570 here.

Ray Tracing: Resident Evil 4 - 1080p

At 1080p, the B570 ran at 81 FPS AVG. The RX 7600 now slightly outperforms it. Realistically, these are the same perceived performance, but lower resolution performance scaling remains stronger on NVIDIA and AMD. The RTX 4060 is now up at 95 FPS AVG, a lead over the B570 of 17%. The B580 is about tied with the 4060.

Ray Tracing: Dying Light 2 - 1080p

Dying Light 2 with RT is up now. At 1080p, the B570 ran at 55 FPS AVG and held proportional lows, which is what we want. These results have it close enough to the RTX 4060 that they’re experientially the same. The RTX 4060 has a lead of 5% over the B570, with the B580 leading both at 65 FPS AVG. The B580’s average framerate is 18% higher than the B570 in a notable distinction. The RX 7600 lands below all of these, closer to the RTX 3060 and below the A580. Intel’s Arc has overall strong RT performance in this title.

Ray Tracing: Dragon’s Dogma 2 - 1080p

In Dragon’s Dogma 2 with RT and at 1080p, the B570 ran at 39 FPS AVG. Unfortunately for Intel, this is one of its worst relative results. Ray tracing in this game is hard for the Intel cards. They don’t maintain the scaling from elsewhere when tested against the 7600 or RTX 4060 here.

Ray Tracing: Dragon’s Dogma 2 - 1440p

At 1440p, the B570 holds a 31 FPS AVG, the B580 a 36 FPS AVG, and the 7600 and 4060 are both just under 40 FPS AVG. Again, Intel is disproportionately disfavored in this test set.

Ray Tracing: Cyberpunk - 1080p Medium

Cyberpunk with RT Medium is next. Battlemage remains super competitive here, with the B570 performing better than the RTX 4060. The average framerate is about the same, but the RTX 4060 suffers from wildly inconsistent frametimes in this test. We showed that in the B580 review with a frametime plot, where the 4060 was spiky as a result of its VRAM limitations.

The B580 keeps its 16% lead over the B570 in this test, with both outdoing the RTX 4060. The RTX 4060 Ti even struggles in this, also suffering from inconsistent run-to-run variance and low performance for 0.1%. AMD has problems of its own, mostly just that it’s unable to compete in this test. The RX 7600 is down under 30 FPS AVG. At this price point, Battlemage is actually the best in this specific test and setup.

Ray Tracing: Black Myth Wukong - 1080p (Experimental)

Black Myth: Wukong is up now. This is one of our experimental charts. This designation is reserved for charts where we’re still evaluating the methodology and the accuracy, and so you shouldn’t assign as much weight to these since we are still leaving open the possibility of some sort of test limitation or issue.

As it stands now, the B570 lands between the A750 and A580 with RT. The B580 is about the same as the RX 7800 XT (watch our review), with the B570 outperforming the RX 7600, so both are competitive with AMD. NVIDIA is up in orbit though with this game, with the RTX 4060 up at 52 FPS AVG in this test with these settings.

Power Consumption & Efficiency (Experimental)

Efficiency testing is up next. This is also a part of our suite of experimental tests, meaning we are still developing methods and that there is more room than typical for possible error or future revisions. This testing is done with an Elmor Labs PMD2 that has been calibrated personally by Elmor on our request. Testing includes both PCIe slot and PCIe cable power consumption, but excludes all other power, including the CPU.

Power Consumption: Idle

In idle testing, Intel Arc remains power hungry as compared to its competition. Using our test platform as configured for benchmarking, the B570 pulled about 30W idle. The B580 was around 35W. This is overall high.

Efficiency: Baldur’s Gate 3 (1440p)

Baldur’s Gate 3 at 1440p shows FPS/W, meaning that higher framerate and lower power will create the most efficient result. Higher is better.

The most efficient device here so far is the 4060 Ti of those tested. The B580 was in the middle of the pack, at 0.52 FPS/W. That had it between the 6600 series. The B570 is more efficient than the B580, at 0.55 FPS/W and below the RX 7800 XT.

Efficiency: FFXIV (1440p)

Final Fantasy 14 at 1440p is an OK showing for the B570. At 0.54 FPS/W, it’s near the RTX 4060 and better than the B580. This is an OK spot to be in for Intel.

Efficiency: FFXIV (1080p)

At 1080p, the B570 produces 0.81 FPS/W and sits again below the RTX 4060 and above the B580 and 7800 XT. Against AMD, Intel is looking good, but NVIDIA maintains its advantage here.

Efficiency: Black Myth Wukong (1080p)

In Black Myth at 1080p, the B570 sits between the RTX 4060 and B580 again. Performance is acceptable overall and far improved from the A-series, like the A580 or A770.

Cyberpunk: Phantom Liberty (RT, 1080p)

Cyberpunk is the last one. This is a pretty good showing for Intel: The B570 isn’t tied with the RTX 4060, but it’s getting a lot closer. The 0.28 FPS/W result is far improved over the 7800 XT and puts it in 4th of the devices tested so far. Note that the 4090 (watch our review) looks inefficient here as it is becoming bound by other components at this resolution.

Intel Arc B570 Conclusion

Providing some highlights, our B570 conclusion generally feels very similar to our conclusion with the B580 in that Intel has definitely improved over Alchemist. The B570 and B580 are often outperforming the A750 and A770, which are larger configs, especially the A770. That’s a good place for Intel to be against itself. Against the competition, there’s a lot of trading back and forth. There are some games where the B570 is similar to or slightly exceeds the RTX 4060, but there are situations where the opposite is true. The trend that persists is that the higher the resolution is, regardless of the discussion around driver overhead, the Battlemage series cards will typically either gain on the competition or pull away from them, depending on where they started. You could also look at it the other way and say that as you approach lower resolutions like 1080p, results get closer together, and many times they’ll trade blows. 

Like with our B580 review, we didn’t experience any spectacular failures and crashes this time. Compared to Alchemist, the card is far improved. While NVIDIA’s drivers are known to be the most stable, Intel is certainly improving here.

In terms of value, the B580 remains a very competitive price, and the B570 is similar.  On average in our results, the B580 tends to be about 15-16% better than the B570. At the upper end of that, we’ll see about a 18.7-19% advantage for the B580 over the B570. At the lower-end, we’ll see about a 10% gain or so. That performance difference largely aligns with the price difference between the 2 cards with the B580 going for roughly $250 vs the B570’s roughly $220 price.