CEOInterviews.AI
Start App
Get an email when Steven Walton gives a new interview
One email per new interview. Unsubscribe in one click.
Steven Walton
Co-founder & Host, Hardware Unboxed, Hardware Unboxed

Using an Old CPU? An RTX 5070 Might Be Pointless (CPU Scaling Benchmark)

📅 Nov 03, 2026 Hardware Unboxed 23 MIN 308422 VIEWS 43 SEGMENTS · 1 SPEAKERS
Support us on Patreon:   / hardwareunboxed   YT Membership:    / @hardwareunboxed   Buy relevant products from Amazon, Newegg and others below: GeForce RTX 5090 - https://geni.us/JyDFBvM GeForce RTX 5080 - https://geni.us/dRy7y07 GeForce RTX 5070 Ti - https://geni.us/AuxCu GeForce RTX 5070 - https://geni.us/Wg4vPDv GeForce RTX 5060 Ti 16GB - https://geni.us/5Kgs2I GeForce RTX 5060 - https://geni.us/xmaeB Radeon RX 9070 XT - https://geni.us/CH0CxfR Radeon RX 9070 - https://geni.us/3neQ6nF Radeon RX 9060 XT 16GB - https://geni.us/I28LWJ Video Index: 00:00 - Welcome to Hardware Unboxed 01:40 -...

What Steven Walton said

Written from the verified transcript and checked against it. Every figure links to the moment it was said.

Steven Walton tested the RTX 5060, RTX 5070, and RX 9060 XT with a range of CPUs from the Ryzen 7 9800X3D down to the Ryzen 5 5500 to examine Nvidia's CPU overhead issue. He found that with high-end CPUs, the RTX 5070 was 36-39% faster than the 9060 XT, but as CPU performance dropped, the Radeon GPU gained ground. With the Ryzen 5 5500, the 9060 XT was 24% faster than the RTX 5060 on average and 36% faster on 1% lows, and 12-14% faster than the RTX 5070. He attributed this to Nvidia's architectural CPU overhead, not AMD's Smart Access Memory, and noted the RTX 5060's x8 PCIe interface caused severe stuttering in Horizon Zero Dawn Remastered with PCIe 3.0. He concluded that budget gamers with older CPUs should choose a Radeon GPU for better performance.

Key takeaways

  1. With the Ryzen 5 5500, the RX 9060 XT was 24% faster than the RTX 5060 on average and 36% faster on 1% lows.
  2. The RTX 5060's x8 PCIe interface caused it to be more than twice as slow as the 9060 XT in Horizon Zero Dawn Remastered with PCIe 3.0.
  3. Nvidia's CPU overhead issue affects Intel CPUs too, not just AMD, and is not related to Smart Access Memory.
  4. Budget gamers with older CPUs should pair them with a Radeon GPU like the 9060 XT to maximize FPS.
  5. The Ryzen 5 5500 is roughly equivalent to the Ryzen 5 3600 in gaming performance.

Numbers and commitments

FigureWhat it refers toTypeAt
24% Average frame rate advantage of 9060 XT over RTX 5060 with Ryzen 5 5500 metric 16:31
36% 1% lows advantage of 9060 XT over RTX 5060 with Ryzen 5 5500 metric 16:31
12% Average frame rate advantage of 9060 XT over RTX 5070 with Ryzen 5 5500 metric 16:31
14% 1% lows advantage of 9060 XT over RTX 5070 with Ryzen 5 5500 metric 16:31
36-39% Performance advantage of RTX 5070 over 9060 XT with 9800X3D metric 15:10
20-30% Performance improvement range for Radeon GPUs over GeForce in some games with older CPUs metric 21:09
360 Total benchmark passes across 10 games metric 20:28
8 GB VRAM capacity of RTX 5060 and 9060 XT other 8:30

Chapters

  1. 0:00Introduction to CPU scaling test
  2. 2:34War Thunder results
  3. 3:42Marvel Rivals results
  4. 5:02Space Marine 2 results
  5. 6:16Hitman results
  6. 7:15Battlefield 6 results
  7. 8:30Horizon Zero Dawn VRAM issue
  8. 10:02Gears of War results
  9. 11:21Hogwarts Legacy results
  10. 12:38Call of Duty results
Steven Walton 0:00 ↗
I recently took an updated look at Nvidia's CPU overhead issue, and I found that it could still be a problem for GeForce 50 series GPUs, and this was seen when comparing the RTX 5070 Ti and 9070 XT using the Ryzen 5 5600, which admittedly is quite a slow CPU by today's standards, especially when pairing with such expensive modern graphics cards. But as I found in the past with this testing, this is an architectural issue. So, it will be seen with all GeForce GPUs, not just the RTX 5070 Ti as we showed in that video. And therefore, logically, you could apply the data that was shown in that video to something like an RTX 5060, for example.
But you know what they say, seeing is believing, which is possibly why so many people felt that it was a complete non-issue that didn't need to be investigated. So, for this video, we are going to be looking at more modest hardware configurations, testing with the GeForce RTX 5060, and that will be compared to the Radeon RX 9060 XT. It's highly conceivable that those of you shopping for an RTX 5060 or 9060 XT have a lower-end PC, and right now isn't a particularly great time to be upgrading to the latest platforms that use DDR5 memory. One of the most popular CPUs of the last few years has been the Ryzen 5 5500, which does bring me much pain as it's a particularly poor choice. But at $90, it seems people just can't resist.
And I think this is because most people see a part like the Ryzen 5 5600, which costs around $150 US right now, and think that the 5500 is basically the same thing. So, the 40% saving seems like a great deal. Sadly, however, despite the naming similarities, they're wildly different products. Anyway, whatever the reason, the 5500 has been selling like hot cakes for a long time now. So, there are plenty of them out there, and for that reason, I've included it. So, we have a good range of CPUs that should provide us with some very interesting scaling data. And the focus for this content is more on the CPU side of things. So, it's not a complete CPU/GPU scaling benchmark as I'll be focusing on low-resolution testing. That said, you could argue, and some certainly do, that 8 GB GPUs like the RTX 5060 and 8 GB 9060 XT are 1080p products.
Anyway, I don't want to get too bogged down in the basics of CPU testing, so let's just jump into the blue bar graph, shall we?
Starting with the War Thunder data, it's pretty clear that the RTX 5070 is a lot faster than the 9060 XT. As you'd expect, when paired with the 9800X3D, it was an average 35% faster, and even the standard RTX 5060 led the budget Radeon GPU by a 10% margin. Still, using the high quality preset at 1080p, all GPUs tested could push over 400 FPS in this title. But if we go and reduce CPU performance by using the Ryzen 5 7600, we see that frame rates become much more CPU limited. And now all three GPUs are seen delivering very similar results.
The 9060 XT was slightly faster than the 5060, but Nvidia's additional CPU overhead isn't hurting the GeForce GPU too much in this example. And similar results were also seen when using the Ryzen 5 5600. But once we drop down to the 5500, the 9060 XT does start to enjoy a small performance advantage, beating the 5060 by an 8% margin and the 5070 by a 5% margin. So when heavily CPU limited, the Radeon GPU does work better, though only slightly better in this example.
Now, the Marvel Rivals data is far more interesting. Firstly, using the 9800X3D, we see that the 5070 is massively faster than the 9060 XT, offering 42% more performance. But the Radeon GPU was able to edge out the 5060 by a slim 6% margin. Now, when retesting using the Ryzen 5 7600, not much changes between the 9060 XT and 5060. In fact, the margin is slightly reduced, aiding the GeForce GPU. However, the much faster RTX 5070 is now much more CPU limited, shrinking its margin over the Radeon GPU to just 23%.
Then, as we move to AM4 with the Ryzen 5 5600, the 9060 XT extends its lead over the RTX 5060 out to 7% and now the heavily CPU limited RTX 5070 is just 7% faster.
Up until this point though, the data has looked like a fairly typical set of results that are CPU bottlenecked. But that all starts to change when we use the Ryzen 5 5500. Using this low-end processor, the 9060 XT is now 20% faster than the RTX 5060 and 17% faster than the RTX 5070. And these strange results are due to Nvidia's additional CPU overhead.
Nvidia's CPU overhead issue is even more pronounced when testing with Space Marine 2. For example, when testing with the 9800X3D, everything looks pretty normal. The RTX 5070 is by far the fastest GPU tested here, rendering 127 FPS, making it 21% faster than the RTX 5060 and 31% faster than the 9060 XT. But that all changes with the Ryzen 5 7600. The results instantly become CPU bound. But in doing so, the 9060 XT is now 6% faster than the RTX 5060 when comparing the average frame rate and 10% faster for the 1% lows. It even beats the RTX 5070.
Then, as we continue to lower CPU performance, those margins continue to grow. Using the Ryzen 5 5600, the 9060 XT is now 15% faster than the RTX 5060, or nearly 20% faster when comparing the 1% lows. But we're not done. The margin balloons out to 32% with the lowly Ryzen 5 5500, with a 37% increase to 1% lows. The RTX 5070 did fare slightly better than the 5060, but even so, the 9060 XT was still 25% faster.
Hitman: World of Assassination proved highly CPU sensitive, and we saw when using the Ryzen 7 9800X3D that the RTX 5070 was able to average 245 FPS with 128 FPS for the 1% lows. This represented a 33% performance advantage over the 9060 XT, which averaged 184 FPS, while the RTX 5060 managed 181 FPS.
Moving to the Ryzen 5 7600 saw the RTX 5070's average frame rate fall by 44% to just 136 FPS with 1% lows dropping to 74 FPS. Interestingly, the 9060 XT pulls slightly ahead at 138 FPS while delivering stronger 1% lows of 77 FPS. And this trend continued with the Ryzen 5 5600 and 5500 where the Radeon GPU averaged 124 and 106 FPS respectively, outperforming the RTX 5070 by approximately 7 to 8%.
Now Battlefield 6 demonstrated substantial CPU bottlenecks with the Ryzen 5 processors. At first, the RTX 5070 managed to average an impressive 251 FPS using the Ryzen 7 9800X3D, making it 44% faster than the RTX 5060 and 48% faster than the RX 9060 XT. However, moving to the Ryzen 5 7600 saw that advantage disappear almost entirely with all three GPUs averaging between 167 and 169 FPS, though the 1% lows were noticeably better for the Radeon GPU.
Similar behavior occurred with the Ryzen 5 5600 where the 9060 XT edged ahead at 156 FPS compared to 151 FPS for the RTX 5070. Interestingly, the Ryzen 5 5500 saw both GeForce GPUs deliver identical results, averaging 126 FPS with 97 FPS for the 1% lows. So, they're CPU bound to this performance. But the 9060 XT was noticeably faster, averaging 143 FPS with 111 FPS for the 1% lows, making it 13% faster when comparing the average frame rate and 14% faster for the 1% lows.
Horizon Zero Dawn Remastered has some interesting results for us that don't really highlight the Nvidia CPU overhead issue, but they do in a way highlight a VRAM issue, but only for the RTX 5060. And that's despite the RTX 5060 and 9060 XT both having just 8 GB of VRAM. The problem here for the RTX 5060 is that it features just an x8 interface whereas the 9060 XT gets the full x16 PCIe bandwidth, doubling its read speed to system memory. As a result, it was 27% faster using the 9800X3D or a massive 47% faster when comparing the 1% lows. And that's despite the RTX 5070 beating the Radeon GPU by a 40% margin.
Dropping down to the Ryzen 5 7600 limits the 9060 XT more than the RTX 5060 as we're still using the same PCI Express 5.0 bus and DDR5 memory. But when we switch to the Ryzen 5 5600, we're now using PCIe 4.0 and DDR4 memory. And this cripples the RTX 5060, resulting in poor 1% lows, which saw the 9060 XT up to 61% faster.
But things get even worse for the budget GeForce GPU with the Ryzen 5 5500, which limits us to PCIe 3.0, and now the 9060 XT is more than twice as fast. The RTX 5060 was completely unusable, suffering constant stuttering with the occasional pause as data was loading in and out of system memory at a very low bandwidth.
Gears of War: E-Day delivered some surprisingly consistent results, particularly with the Radeon RX 9060 XT. Regardless of whether it was paired with the Ryzen 7 9800X3D, the Ryzen 5 7600, or the Ryzen 5 5600, the Radeon GPU averaged exactly 72 FPS, dropping only slightly to 70 FPS with the Ryzen 5 5500.
The RTX 5070 was considerably faster when using higher-end processors, averaging 97 FPS with the 9800X3D and 96 FPS with the Ryzen 5 7600. However, its performance dropped to 79 FPS with the Ryzen 5 5600 before falling to just 67 FPS with the Ryzen 5 5500. This allowed the RX 9060 XT to pull ahead while also delivering stronger 1% lows of 53 FPS compared to 45 FPS. That's an 18% increase along with a 36% boost over the RTX 5060.
The RTX 5060 generally trailed both GPUs, though its results remained relatively consistent. Overall though, the 9060 XT showed very little sensitivity to processor performance, while the RTX 5070 benefited noticeably from the faster CPUs, as you might expect at 1080p.
Hogwarts Legacy showed some of the biggest performance differences between the processors, particularly with the RTX 5070. Paired with the Ryzen 7 9800X3D, the 12 GB GPU averaged an impressive 133 FPS, comfortably ahead of the RX 9060 XT at 103 FPS and the RTX 5060 at 99 FPS. However, things changed considerably with the Ryzen 5 7600 where the RTX 5070 dropped to just 76 FPS.
Interestingly, the 9060 XT managed 91 FPS, giving it a comfortable 20% performance advantage despite being the slower GPU when paired with the 9800X3D. The Radeon GPU maintained its lead with the Ryzen 5 5600, averaging 76 FPS compared to 68 FPS for the RTX 5070. That is a 12% performance advantage over the higher-end RTX 5070 and 17% over the RTX 5060. Then with the Ryzen 5 5500, both GeForce GPUs managed just 60 FPS, while the Radeon GPU delivered 73 FPS along with stronger 1% lows of 44 FPS, making it up to 26% faster. Again, this strange data is entirely a result of Nvidia's CPU overhead issue.
Call of Duty: Modern Warfare 4 produced some interesting results, particularly for the Radeon RX 9060 XT. With the Ryzen 7 9800X3D, the RTX 5070 comfortably led the pack, averaging 237 FPS, followed by the 9060 XT at 195 FPS and then the RTX 5060 at 173 FPS. The RTX 5070 maintained its lead with the Ryzen 5 7600, although the gap did narrow to just 16 FPS as the data became much more CPU limited.
And then things became much more interesting with the Ryzen 5 5600. Here the 9060 XT actually pulls ahead of everything, averaging 187 FPS compared to 175 FPS for the RTX 5070. It also delivered noticeably stronger 1% lows at 133 FPS, making it up to 16% faster.
And this trend continues with the Ryzen 5 5500 where the 9060 XT averaged 167 FPS, comfortably placing it ahead of both GeForce GPUs. But it was the 1% lows that really stood out, offering at least 31% greater performance. Overall, the 9060 XT provided remarkably consistent performance across the different processors tested here, while the GeForce GPU suffered much larger performance losses as the CPU performance decreased.
Finally, we have Stalker 2. And once again, we have some interesting results here. The RTX 5070 comfortably led when paired with the Ryzen 7 9800X3D. It averaged 120 FPS compared with 86 FPS of the RTX 5060 and then 82 FPS for the 9060 XT. We're also looking at a massive 46% increase over the Radeon GPU. Then moving down to the Ryzen 5 7600 here, the RTX 5070 still maintained a reasonable advantage at 102 FPS, though we do see the gap narrow considerably between the three GPUs when we downgrade to the Ryzen 5 5600. Here they deliver between 75 and 78 FPS. And this data does look very much like your typical CPU bottleneck.
However, if we lower CPU performance further with the Ryzen 5 5500, the 9060 XT actually pulls further ahead, averaging 73 FPS compared to just 65 FPS for the RTX 5070. It also delivered noticeably stronger 1% lows at 61 FPS compared with 52 FPS. That is a 20% increase to the 1% lows and a 14% increase for the average frame rate.
Okay, so here's the 10 game average data and it paints a pretty clear picture. Firstly, as you'd expect when not CPU limited, the RTX 5070 is much faster than the RTX 5060 and 9060 XT, offering 36 to 39% more performance on average when paired with the 9800X3D. That margin's eroded to 16% over the RTX 5060 when paired with the Ryzen 5 7600 and just 7% when compared to the 9060 XT, but this is what you'd expect to see.
And this data is relevant for those of you trying to maximize FPS performance. For games like Counter-Strike, Apex Legends, Valorant, and Fortnite, just to name a few examples, your CPU is far more important than your GPU, at least within reason, of course. Something we did notice with the Ryzen 5 7600 data when compared to the 9800X3D is the fact that the 9060 XT went from just 2% faster than the RTX 5060 when using the higher-end CPU to 8% faster with the Ryzen 5 processor. And we saw an even larger increase for the 1% lows.
And this trend continued as the 9060 XT was able to match the RTX 5070 when both GPUs were paired with the Ryzen 5 5600 and that made it up to 17% faster than the RTX 5060.
Then with the Ryzen 5 5500, we ran head first into Nvidia's CPU overhead issue. And as I've said many times in the past now, this really is a problem that budget gamers need to be aware of. Using this low-end processor, the 9060 XT was an average 36% faster than the RTX 5060 when comparing 1% lows and 24% faster for the average frame rate. That said, some of this data was skewed by the 5060's limited PCIe bandwidth. So, if we compare with the RTX 5070, we still see that the 9060 XT was an average 12% faster or 14% faster when comparing 1% lows.
Okay, so the main takeaway from all of that testing being that if you have a lower-end CPU that you're not quite ready to upgrade yet, but you want to get yourself a new GPU, you really are best pairing it with a Radeon GPU such as the 9060 XT. It didn't take too much before those margins really did start to increase in the 9060 XT's favor.
For a typical benchmark review, so a product review where we include benchmarks, we'd normally aim to remove as many system limitations as we possibly can, which means testing with a flagship processor such as the 9800X3D. And for the most part, this really is the best way to go about that particular testing, especially when comparing it with higher-end products like the RTX 5070, for example.
Still, the 9800X3D configuration found that the 9060 XT was 2% faster than the RTX 5060, but that margin did grow to 8% with the Ryzen 5 7600, which admittedly doesn't change a huge amount, but it is more favorable for the Radeon GPU. Then with the Ryzen 5 5600, it grew to 12%, a statistically more significant margin. And then with the 5500, we were faced with a complete blowout that favored the 9060 XT by a massive 24% margin. But again, some of that was due to the 5500 only offering PCI Express 3.0 support and the fact that the RTX 5060 is limited to just eight lanes.
It's probably worth noting that the Ryzen 5 5500 is roughly equivalent to the Ryzen 5 3600 in terms of gaming performance. So, it's not that much slower than say a Ryzen 7 3700X.
And that means AM4 owners with a CPU equal to or slower than any of these CPUs will likely see better gaming performance when using a Radeon GPU, especially in examples where you're trying to maximize FPS performance. And I should note that this has nothing to do with AMD's claimed Smart Access Memory support, which is really just Resizable BAR. This PCI Express feature has nothing to do with what we're seeing here. These same margins will also be seen when testing without Resizable BAR enabled, and they'll also be seen when testing with Intel CPUs of roughly equivalent performance.
I've seen some people claiming that the Nvidia CPU overhead issue doesn't affect Intel CPUs, and that is completely untrue, and we've already proven this many years ago. Now, if you have an Intel CPU that delivers similar gaming performance to that of the Ryzen 5 5500, you're going to find similar margins between the RTX 5060 and 9060 XT.
Now, the reasons for why I didn't test any Intel CPUs in this video are pretty simple. Firstly, AMD is and has been for years now completely crushing Intel in the retail market. Ryzen's heavily outsold anything Intel and therefore the vast majority of you watching this content are using a Ryzen processor. So, we're providing data that caters to the majority of our viewers. Makes sense, right? Of course, ideally, I would like to include Intel CPUs as well. And if the market was more split, I'd probably make two versions of this video, but as it stands, adding more CPU models to the graphs, it does really start to clutter things up, making it more difficult to digest the data. And when all is said and done, it really adds no benefit.
For example, if I were to include the Core i5-10400F, it would provide us with the same or very similar results to the Ryzen 5 5500. So, we just have two data points telling us the same thing. Time is unfortunately also a real factor. Each game in this video required about 36 benchmark passes to gather the data. And with 10 games in total, it's about 360 benchmark passes. So adding in more CPUs just increases those figures. And as I said, if it doesn't really provide any real benefit, then there's not much point doing it. It's just a waste of my time and also a waste of your time as I have to present all that additional data that doesn't tell us anything that you haven't already learned from the data that was provided.
And really the takeaway is if you have an older system, whether that be an older AMD or Intel system, as in a system with a CPU that's old, and you want to upgrade the graphics card in that particular system, you are best off investing in a Radeon GPU, especially if you wish to maximize FPS performance. In some games, you will be enjoying anywhere from 20 to 30% greater performance. So, it is quite significant and it is what led us to investigate this series to begin with because some of you with older systems were going from a Radeon GPU to what should be a faster GeForce GPU and you're noticing performance either not improve or go backwards which is very strange and now we know why.
Anyway, that is going to do it for this video. I'll probably do some more in-depth GPU/CPU scaling content in the near future where we look at multiple resolutions, quality presets. So, more in-depth, more balanced look at it, whereas this one was a bit more focused on the CPU side of things and that overhead problem. So, make sure you subscribe for all of that upcoming content.
And if you do want to support the channel more directly, we have the join button or Patreon. Either one of those, the links for those or well, you can click the join button or the link to Patreon will be in the video description. So check that out if you're interested. You get some pretty cool perks. Access to our exclusive Discord server. Great community over there. Good place to chat about basically anything really, but in particular PC hardware. We do live streams, behind the scenes content, and Q&A stuff. So check that out if you're interested. If not, that is perfectly fine. And I would like to thank you for watching this video. I'm your host, Steve. See you again next time.

Cite this transcript

APA, MLA, BibTeX
APA

Walton, S. (2026, November 3). Using an Old CPU? An RTX 5070 Might Be Pointless (CPU Scaling Benchmark) [Interview transcript]. Hardware Unboxed. CEOInterviews.AI. https://ceointerviews.ai/interview/3229039/

MLA

Steven Walton. "Using an Old CPU? An RTX 5070 Might Be Pointless (CPU Scaling Benchmark)." Hardware Unboxed, 3 Nov. 2026. Transcript, CEOInterviews.AI, https://ceointerviews.ai/interview/3229039/.

BibTeX
@misc{walton2026_3229039,
  author       = {Steven Walton},
  title        = {Using an Old CPU? An RTX 5070 Might Be Pointless (CPU Scaling Benchmark)},
  howpublished = {Interview transcript, Hardware Unboxed. CEOInterviews.AI},
  year         = {2026},
  month        = {nov},
  url          = {https://ceointerviews.ai/interview/3229039/},
  note         = {Speaker-attributed transcript with timestamps}
}