It's fair to say that PC gamers have faced a challenging year with very little in the way of positive news for anyone planning to upgrade or build a new gaming PC. So, from a hardware perspective, it's fair to say it's been nothing but bad news. However, we have received some pretty great games this year, and that trend continues with this latest release, The Witcher 3: Wild Hunt Remastered. But before we get into that, today's sponsor spot is brought to you by MSI and their GeForce RTX 50 series, built with Nvidia's Blackwell architecture and enhanced by MSI's premium board design, power delivery, and thermal solutions. RTX 50 series cards deliver major gains in games, AI workloads, streaming, and productivity. Enjoy many of the latest games such as Control Resonant with the full suite of RTX features such as DLSS 4.5 ray reconstruction, path tracing, and dynamic multiframe generation. Ray reconstruction improves image quality in supported ray traced and path traced games, offering better lighting accuracy, motion clarity, and stability, while dynamic frame generation helps to better utilize high refresh rate monitors. MSI offers a number of higher quality models from the Shadow range to the Supreme. So, for more information on MSI's entire range, please click the link in the video description.
Okay, so incredibly this massive update is free for existing owners of the game, allowing players to relive the experience like never before. And for those yet to play the game, the update brings countless improvements. The update includes overhauled combat, a rebalanced skill tree, new brutal finishes, and refined monster hitboxes. The Red Engine 3 game engine has also been heavily upgraded, allowing for modern rendering techniques such as ray tracing and path tracing. Along with the realistic lighting and shadows with ray tracing, there's also improved textures, major enhancements to foliage, and an expanded photo mode. All of which help to make the 11-year-old game feel like a modern release. The remaster is launching on PlayStation 5, Xbox Series X, and PC via Steam, GOG, Epic Games Store, and Battle.net. It will also be released natively on the Nintendo Switch 2. There's a lot to cover with this update, but for the purposes of this video, we want to focus on PC performance. Namely, how does the updated remastered version compare to the original? And how does AMD and Nvidia's latest generation of GPUs compare using a range of quality settings? Normally, we'd aim to test a lot more, but with relatively short notice and Steve being absolutely smashed by a flu virus over the past week, we've had to settle on a smaller sample of tests for this one. But you can expect the remastered version to appear in future benchmarks as we'll be adding it to the test suite. Now for this testing, we're going back to Novigrad as the city areas of the game are by far the most demanding. And as we recall, Novigrad was the most demanding of them all, heavily utilizing not just the GPU, but also the CPU. Of course, for our testing, we will be using the Ryzen 7 9800X3D armed with 32 GB of DDR5 6000 memory. So, this should help to minimize any CPU bottlenecks. Just lastly, please note all testing was conducted using a pre-release AMD driver developed for this remastered version of The Witcher 3. The latest publicly available Adrenaline 26.9.1 version is completely broken, resulting in lots of rendering glitches. The early release driver we use for testing is version 26.20.15.02. Finally, the current GeForce driver appeared to work just fine. The latest version is of course game ready for this specific title, though that isn't to rule out a future version which could improve performance further. Okay, let's get into the data.
Actually, before we jump head first into the blue bar graphs, take a look at some quick image quality comparisons. Firstly, here's a look at the RT Ultra presets for each version of the game, comparing the original to the remastered. Opinions are going to vary quite a bit on this one, as it's very subjective as to which version looks better. In this comparison, the remastered version is certainly a lot brighter, and as a result, there's more visible detail in the early hours of the morning. There's also better lighting detail in the distance, though, it's not obvious to me that textures have improved. As we get to this fork in the road, I'll slow the footage right down so you guys can get a good look at how they compare. Again, it's subjective. There's more lighting detail in the distant trees, for example, in the remastered version. But given it's very early in the morning, I feel the original version looks more realistic. Here's a look at the full screen version. And please note, we are testing at the same time of day for both versions. I do believe the light bleeding through the foliage looks better for the remastered version, but is there meant to be that much light at 3:44 a.m.? It's really difficult to say which version looks best. I feel like neither is technically better, just different. Now, here's how the ultra presets compare. Again, I'd say they're pretty similar overall. The lighting is certainly more powerful for the remastered version, but it's difficult to say if that makes the game look better or not. Personally, I feel Geralt looks better lit, but again, that's going to be a subjective thing. Now, here's the medium presets compared between the versions. It's a similar story really. The remastered version is better lit. Whether or not that makes sense at 3:44 a.m. is hard to say, but if it's meant to be better lit, then the remastered version certainly looks better.
Now, let's focus on the remastered version. Looking at path tracing versus ray tracing. For some reason, path tracing generally looks worse in this game, either overexposing characters or adding a strange blue filter. There are some instances where it does look better, but in my opinion, path tracing more often than not looks worse than ray tracing, and in some examples much worse. For example, in my opinion, Yennefer's face looks much worse with path tracing enabled here, flatter and more blown out. In game, it's a mixed bag, but like I said, for the most part, I think path tracing looks worse than ray tracing, but that's just my opinion. And like I said, this is subjective. Finally, here's a look at RT Ultra versus standard ultra. And in my opinion, ray tracing does look much better, particularly the shadow quality. A lot of the ultra shadows are very fuzzy, featuring a lot of noise. The rasterized shadows and lighting in this remastered version are lacking. They're not nearly as clean as you'll find in a good modern title. So, that's a bit disappointing to see. Anyway, let's now jump into the benchmark data.
Okay, so let's compare FPS performance between the original version, which has been labeled as classic, to the remastered version. And for those comparisons, we're using the 5070 Ti and 9070 XT. At 1080p, using the ultra preset, the GeForce GPU was 8% faster using the original version of the game. And interestingly with the remastered version, it remained 8% faster when looking at the average frame rate. But we're seeing some strange behavior for the 1% lows. And this is because there is a strange bug, probably a driver issue that is limiting the performance of Radeon GPUs. And we'll explore this more shortly with the medium preset. Looking at the 5070 Ti data though, we see that the remastered version runs just 7% slower, which is amazing given the visual upgrade. At 1440p, we see that previously the 9070 XT and 5070 Ti were neck and neck, but with the updated version, the Radeon GPU is now 16% faster. Though this could change with future driver updates from Nvidia, but leading up to the release, this is how things stand. We're also seeing a 23% reduction in performance for the 5070 Ti with this new version of the game with just a 10% hit for the 9070 XT. Then at 4K, we again see similar performance from these two GPUs when running the original version of the game. Interestingly, the tables have turned at 4K with the remastered version. Here, the 5070 Ti is 7% faster, making it around 10% slower when compared to the original performance.
For this batch of tests, I've enabled ray tracing using the RT Ultra preset along with what we deem to be an appropriate level of upscaling for each resolution. Again, we find some really interesting and even odd results. For example, at 1080p, the 5070 Ti was previously 16% faster than the 9070 XT, but testing with the remastered version eliminated this margin entirely. In fact, RT performance for the 9070 XT went unchanged, still averaging 93 FPS, while the 5070 Ti dropped 14% of its original performance. This data looks wrong, but after multiple retests, we assure you it's correct. In this example, the Nvidia GPUs run into some kind of bottleneck. Again, this could just be a driver related issue, but even with the RTX 5090 installed, we're seeing the same limitation of just over 90 fps. Switching to 1440p, we see that previously the 5070 Ti was a massive 24% faster than the 9070 XT. And while it still remains ahead with the remastered version, the margin has been reduced to 14%. This also means that ray tracing performance for the 9070 XT between the two versions is much the same, whereas the 5070 Ti took a 12% hit. Then at 4K, we see that previously running the original version of the game that the 5070 Ti was 20% faster than the 9070 XT. But interestingly, with the remastered version, that margin comes down to 7%, which is a lot less than what we saw at 1440p. So, I'm not sure why the Radeon GPU is doing so well here relative to the 5070 Ti.
Now, out of interest, we also tested with the medium preset. And this data is particularly interesting as the remastered version of the game looks significantly better and despite that runs much better, or at least it can. At 1080p, we see some oddities, and this is a result of the Radeon GPUs running into some kind of bottleneck. Again, this is likely a driver related issue. Running the original version of the game, the 9070 XT and 5070 Ti were basically neck and neck. The GeForce GPU was just 3% faster. But when we tested with the remastered version, the 5070 Ti wound up 31% faster. Again, we believe this is due to a Radeon driver issue. Still, the 9070 XT was good for 149 FPS, and you'll be able to achieve this level of performance with an 8 GB 9060 XT, which we'll look at shortly. What's interesting to note here is the fact that the 5070 Ti was 5% faster running the much better looking remastered version of the game. Now, at 1440p, the 9070 XT and 5070 Ti were again neck and neck when running the original version of the game. But when we switch over to the remastered version, the 5070 Ti ends up 26% faster and more crucially was 23% faster running the remastered version when compared to the 11-year-old original. Now, at 4K, we're below the artificial Radeon performance limitation of 148 to 149 FPS. So, like the 5070 Ti, the 9070 XT is also much faster running the remastered version and in fact is now faster than the 5070 Ti by a 12% margin. So, some very odd scaling data here, but what it does indicate is that the remastered version of the game not only looks significantly better when comparing the medium settings, but it also runs significantly better, 45 to 58% better in this example.
Okay, so now it's time to compare the current generation GeForce and Radeon GPUs. And we'll start with the ultra preset at 1080p. The RTX 5090 was good for 183 FPS on average, though it did only make it 6% faster than the 5080 and 20% faster than the 5070 Ti. The 5070 Ti was 8% faster than the 9070 XT, but that's likely due to a driver issue. And we can see just how compressed the 1% low and average frame rate performance is for the Radeon GPU. The 9070 XT and 9070 also achieved the same average frame rate despite the 1% lows of the XT being 12% greater. The 9070 GRE also matched the RTX 5070 while the 5060 Ti and 9060 XT models were neck and neck. We're also looking at very similar performance between the 8 and 16 GB versions using these quality settings. Finally, the Arc B580 and RTX 5060 delivered similar performance at around 80 fps.
Increasing the resolution to 1440p changes the results quite a bit. The RTX 5090 is now able to stretch its legs, increasing the margin over the 5080 out to 42%. The 9070 XT also looks much better, as here the 5080 was just 3% faster, allowing the flagship Radeon GPU to beat the 5070 Ti by a 16% margin. In fact, the 5070 Ti was only able to match the RX 9070 here, though the 9070 GRE and RTX 5070 were still very evenly matched. The 9060 XT 8 and 16 GB models both average 63 FPS, but the reduced PCIe bandwidth of the 5060 Ti hurt the 8 GB model here as it was almost 10% slower for the average frame rate while being a little over 20% slower for the 1% lows. As a result, the 5060 Ti 8 GB wasn't much faster than the vanilla 5060 and therefore also only just edged out Intel's Arc B580.
Finally, at 4K, you will want to enable upscaling for most of these GPUs. That said, the RTX 5090 needs no such help, averaging 107 FPS to make it 62% faster than the 5080 and 75% faster than the 5070 Ti, which rendered 61 FPS on average. Beyond that, we are dropping below 60 fps with the 9070 XT. So, you'll probably want to enable upscaling here. Now, for the ray tracing ultra preset data, and we'll of course start with the 1080p results. Here we see that higher GeForce GPUs slam into a 94 FPS performance barrier along with the 9070 XT. So, this appears to be some kind of game or possibly system limitation. The first GPU to fall below this limit was the RTX 5070 with 87 FPS, making it 7% faster than the RX 9070. This meant that the 9070 GRE was a lot slower than the 5070 with 70 FPS, making it 20% slower. The 5060 Ti also smashed the 9060 XT, offering 31% greater performance, meaning the Radeon GPU could only slightly outpace the RTX 5060. The Arc B580 also struggled with ray tracing enabled, delivering just 38 FPS on average while using quality FSR upscaling at 1080p.
When we moved to 1440p, the strange performance ceiling persisted for the top tier Nvidia GPUs. The RTX 5090 and 5080 both slammed into a 93 to 94 FPS wall, while the 5070 Ti followed closely at 90 FPS. The RX 9070 XT delivered 79 FPS, which trailed the 5070 Ti. Further down the stack, the RTX 5070 maintained a slight lead over the RX 9070 as it averaged 74 FPS compared to 70 FPS. VRAM capacity severely impacted the low end of the chart. The RTX 5060 Ti 16 GB achieved 56 FPS, but the 8 GB version plummeted to just 32 FPS, which placed it at the very bottom below the Intel Arc B580. Similarly, the 16 GB RX 9060 XT comfortably outperformed its 8 GB counterpart.
Scaling up to 4K with performance upscaling enabled, the upper end bottleneck finally broke, allowing the RTX 5090 to stretch its lead with 93 FPS on average. The RTX 5080 followed at 72 FPS, while the RTX 5070 Ti managed 62 FPS, edging out the 9070 XT by 7%. Lower down the chart, the RX 9070 narrowly surpassed the RTX 5070 with 52 FPS versus 50 FPS. However, memory capacity limitations became catastrophic for lower tier GPUs under these demanding conditions. While the RTX 5060 Ti 16 GB delivered a somewhat playable 37 FPS, its 8 GB sibling was a mess with just 13 FPS, falling behind the RTX 5060 and the Arc B580. Similarly, the 8 GB RX 9060 XT took a massive hit, dropping to just 19 FPS compared to the 31 FPS rendered by the 16 GB model.
Now, testing at 1080p with path tracing and quality upscaling enabled revealed a very odd performance ceiling for the top Nvidia GPUs. The RTX 5090 and 5080 slammed into a hard limit of just 62 FPS, while the RTX 5070 Ti followed right behind at 61 FPS. AMD's flagship RX 9070 XT struggled significantly, averaging just 43 FPS, which allowed the mid-range RTX 5060 Ti models to close the gap at 41 FPS. The RX 9070 managed only 37 FPS, barely edging out the RTX 5060 at 36 FPS. At the bottom of the stack, the RX 9060 XT variants produced unplayable averages of 24 FPS, while the Intel Arc B580 completely choked, rendering a mere 14 FPS.
Stepping up to 1440p with balanced upscaling, the bizarre artificial ceiling persisted for the high-end GeForce hardware. The RTX 5090 and 5080 capped out at 59 FPS with the RTX 5070 Ti sitting closely at 57 FPS. The Radeon GPUs continued to fall behind heavily under path tracing as the flagship RX 9070 XT achieved only 34 FPS, essentially tying the RTX 5060 Ti 16 GB, which rendered 33 FPS. Meanwhile, the 9070 dropped to 30 FPS, leaving it trailing the RTX 5070's 44 FPS by a significant margin. VRAM capacity did not save the RX 9060 XT cards here with both the 8 and 16 GB models plummeting to 18 FPS while the Intel Arc B580 again occupied last place with just 11 FPS.
At 4K, using performance upscaling, the engine bottleneck finally lifted allowing the RTX 5090 to pull away with a 59 FPS average. The RTX 5080 trailed behind at 43 FPS and the RTX 5070 Ti managed 40 FPS. Path tracing at this resolution crushed the Radeon lineup with the 9070 XT barely managing 23 FPS, effectively matching the RTX 5060 Ti 16 GB's 22 FPS. Severe VRAM limitations surfaced for the lower-end cards at 4K. The 5060 Ti 8 GB collapsed to just 10 FPS, falling significantly behind its 16 GB counterpart. The RX 9060 XT 8 GB also struggled at 10 FPS, slightly behind the 16 GB version's 12 FPS. While Intel's Arc B580 rendered a slideshow-like 7 FPS on average.
When testing at 1080p using the medium preset, we observed a bizarre performance ceiling. The high-end Nvidia GeForce cards, including the 5090, 5080, 5070 Ti, and 5070, all slammed into a hard limit between 194 and 198 FPS. Astonishingly, the entire upper stack of Radeon GPUs experienced a similar but much lower artificial limitation. The 9070 XT, 9070, 9070 GRE, and both 9060 XT models were completely flattened, delivering almost identical average frame rates, ranging from 147 to 149 FPS. Because of this severe limitation, the mid-range RTX 5060 Ti 16 GB easily outpaced AMD's flagship at 177 FPS. At the bottom of the stack, the Intel Arc B580 managed 136 FPS, which kept it well ahead of the baseline RTX 5060 at 113 FPS. For those of you wondering, testing outside of Novigrad did increase performance, but we were still seeing bizarre scaling. For example, in an open area, the GeForce GPUs would cap out at around 270 FPS, while the Radeon GPUs found their limits much sooner at 180 FPS. So, something odd is going on here.
Moving to 1440p on medium settings, the Nvidia performance cap began to lift slightly with the 5090 and 5080 reaching 198 and 193 FPS respectively. The RTX 5070 Ti followed closely behind at 186 FPS. However, the artificial limitation plaguing the high-end Radeon cards persisted. The flagship RX 9070 XT and the RX 9070 were still completely bottlenecked, rendering 148 and 147 FPS, respectively. This bizarre scaling allowed the RTX 5070 to easily surpass them with a 158 FPS average. Further down the chart, the 16 and 8 GB versions of the RX 9060 XT both achieved 121 FPS, placing them right between the 16 and 8 GB versions of the 5060 Ti. Finally, the Intel Arc B580 maintained a comfortable lead over the standard RTX 5060.
Testing at 4K native finally eliminated the strange performance ceilings, allowing the hardware to scale more naturally. The RTX 5090 stretched its legs significantly, dominating the chart with a massive 182 FPS average. The RTX 5080 trailed far behind at 126 FPS, though it managed to hold off the RX 9070 XT, which delivered 120 FPS. The 5070 Ti and RX 9070 were virtually neck and neck, rendering 107 and 106 FPS. Memory capacity once again proved critical at this resolution. While the RTX 5060 Ti 16 GB maintained a playable 67 fps, the 8 GB version plummeted to just 47 FPS, finishing near the very bottom of the stack. Interestingly, thanks to the increased PCIe bandwidth, the 8 and 16 GB RX 9060 XT models both produced 64 FPS, keeping them slightly ahead of Intel's Arc B580 at 61 FPS.
So, there you have it. Visually, The Witcher 3 is certainly improved, though not always massively, and I'm sure some will even argue that the original looked better. But like I've said, this is all highly subjective. Though, having said that, I think most will agree that path tracing is very underwhelming in this title, and in some cases just looks bad. Maybe they can fix that. Not sure. For me, the most impressive aspect of this update, at least in terms of performance, is the fact that at 4K we saw massive FPS gains using the medium quality settings. 36% for the 5070 Ti and 45% for the 9070 XT. That means for those of you running lower-end GPUs, you can expect much better performance. And I think overall the medium quality visuals do look better as an added bonus. There were of course some performance oddities such as frame rates not really going above 60 fps in Novigrad with path tracing enabled even with the RTX 5090, or the fact that Radeon GPUs hit a performance cap well before GeForce GPUs when using lower quality visual settings. I should also note when testing in Novigrad with path tracing enabled at 1080p using the RTX 5090, we saw CPU utilization max out at 35% with none of the 9800X3D's cores getting anywhere near 100% usage. Outside of Novigrad, away from the cities, the frame rate cap with path tracing enabled jumped from 60 fps to about 80 fps. But again, this meant that the RTX 5090 was no faster than the RTX 5080, for example.
The good news from a performance perspective is that the game is very easy to run using the medium quality settings at native 1440p. Budget GPUs like the Arc B580, which crucially you can still buy for a reasonable price in most regions, was able to average 102 FPS. The 9060 XT 8 GB was also very good, delivering 121 FPS under the same conditions. Those of you looking at an RT experience will probably have to upgrade to at least the RX 9070 GRE, which averaged 58 FPS at 1440p with the help of balanced upscaling. The RX 9070 was quite a bit better with 70 fps, and you can currently buy those for $650 US, which is, I guess, pretty good in the current market. If you don't care for ray tracing, then the remastered version is very easy to run using the ultra preset at 1440p. The 9060 XT 8 GB, for example, was good for just over 60 fps, while a 9070 will push you over 100 fps.
Anyway, that's going to do it for this video. This is actually Steve's video, and Steve's script. I've just recorded it for him because he's been a bit sick. I think his voice is not working at the moment. So, I thought I'd give him a chop out and I'd film this video. But these are all his thoughts on this particular game. That said, I also took a look — I personally, as in Tim here, I personally took a look at The Witcher 3 remastered earlier, and I do agree that path tracing in this game does look a little weird in a lot of the scenes that I looked at. So, I'm not 100% sure what's going on there. This is supposed to be one of the big visual improvements to this specific title. I didn't like the way it looked in a lot of scenes, especially it removes some of that warmer look that The Witcher 3 has and made it look a bit colder. Some of the indoor scenes looked a bit strange to me. So, I wasn't overly impressed with that. But overall, I do think this is an improvement to the game's visuals with this remastered version. Plus, there's all those supposed gameplay improvements as well. So, yeah, I'm sure you'll see all the thoughts on that from actual game reviewers. But anyway, that'll do it for this one. Thanks for watching to the end. As always, we do have Patreon if you want to support our testing here at Hardware Unboxed. If you sign up, you get access to things like our Discord community, monthly live streams, and BTS content. So, thanks for watching and I'll catch you in the next one.