On retailer cabinets, AMD’s Radeon RX 9060 XT seems to be like a compelling choice for avid gamers. It is reasonably priced, backed by a trusted model, and seems well-equipped for contemporary gaming. However there is a catch: the 8GB model is severely constrained by its restricted video reminiscence – a technical bottleneck that will not present up in commonplace benchmarks. That lacking reminiscence turns the 8GB mannequin right into a ticking time bomb for avid gamers hoping to future-proof their rigs.
The 8GB RX 9060 XT is marketed beneath the identical title because the 16GB mannequin, but performance-wise, they inform very totally different tales. In our overview, we highlighted a number of fashionable video games with rising VRAM calls for the place the 8GB model did not simply battle – it broke.
Body charges tanked. Texture streaming stuttered. What ought to have been easy gameplay at affordable settings turned a irritating expertise. That is problematic for a number of causes, however most significantly, a brand-new graphics card shouldn’t be bottlenecked by restricted VRAM in at present’s titles – and but, right here we’re.
This turns into a significant concern when you think about that the majority patrons count on to maintain their GPU for not less than three to 4 years. And if 8GB is already falling brief in 2025, the street forward seems to be even rougher.
The Way forward for VRAM Utilization
In these 3 to 4 years, video games will virtually definitely turn into extra demanding, requiring considerably extra VRAM than they do at present. Builders are already sounding the alarm. Even now, optimizing video games for 8GB GPUs is changing into a time-consuming hurdle – an issue many within the gaming group flagged properly earlier than it confirmed up in overview metrics. Some builders have spent weeks, even months, constructing workarounds simply to make video games practical on 8GB playing cards, solely to see gamers count on next-gen experiences on last-gen reminiscence footprints.
Testing VRAM limitations might be troublesome, and in lots of circumstances, a fast 30- to 60-second benchmark run will not reveal the efficiency points gamers encounter throughout prolonged gaming periods.
For instance, to see VRAM-related efficiency points in Halo Infinite on an 8GB GPU utilizing extremely settings at 1080p or 1440p, you usually must play the sport for a protracted interval. It takes time to overwhelm the VRAM buffer. Texture points, nonetheless, have a tendency to look extra rapidly, however you must be on the lookout for them – and so they will not present up in benchmark graphs.
One other issue reviewers typically overlook is {hardware} configuration. When testing GPUs, it is commonplace follow to pair them with the newest and quickest CPUs to showcase their full potential, and that is precisely what we do.
However in terms of VRAM limitations, testing on a high-end CPU with quick reminiscence and a contemporary platform represents a best-case state of affairs for pushing the VRAM buffer. We briefly demonstrated this in our RX 9060 XT 8GB overview by working checks on a Core i7-8700K, a processor restricted to PCI Categorical 3.0 and slower DDR4 reminiscence – and the outcomes have been surprising.
Testing Methodology
For this take a look at, we’re utilizing our high-end 9800X3D take a look at system to guage PCIe efficiency. This implies we’re isolating the PCIe bus and testing it particularly, giving us a transparent understanding of how PCIe bandwidth influences efficiency – significantly when dipping into system reminiscence.
We had thought-about utilizing older CPUs to characterize PCIe 3.0 and 4.0 configurations, however that will introduce too many variables. As an alternative, we’re sticking with the 9800X3D and manually adjusting the PCIe mode within the BIOS.
Additionally take a look at: The Inside Workings of PCI Categorical, How {Hardware} Works
This enables us to measure the efficiency of the 8GB RX 9060 XT utilizing PCIe 3.0, 4.0, and 5.0, and evaluate that information to the 16GB RX 9060 XT utilizing PCIe 3.0. These outcomes are particularly helpful for these with out PCIe 5.0-capable methods, as many platforms are nonetheless restricted to PCIe 4.0 and even 3.0.
Let’s dive into the testing.
Benchmarks
F1 25 – Very Excessive
Beginning with F1 25 utilizing the very excessive preset – the second highest high quality setting within the sport – we additionally checked efficiency utilizing the excessive preset, which we’ll cowl shortly. The 16GB RX 9060 XT averaged 73 fps on this take a look at, with 1% lows at 51 fps. That is stable, very playable efficiency.
With PCIe 5.0, the 8GB model was 18% slower, dropping to 60 fps, with 1% lows down by 22%. Keep in mind, that is the 8GB mannequin on PCIe 5.0 versus the 16GB mannequin on PCIe 3.0.
When downgrading the 8GB card to PCIe 4.0, efficiency dropped by 30% to simply 42 fps.
On PCIe 3.0 – the identical interface utilized by the 16GB mannequin on this take a look at – efficiency fell additional to simply 35 fps.
That is a 17% drop in comparison with PCIe 4.0, and 42% slower than PCIe 5.0. Worse nonetheless, on PCIe 3.0, the 8GB RX 9060 XT was 53% slower than the 16GB model on the identical interface – or to place it one other method, the 16GB mannequin was 111% quicker.
F1 25 – Excessive
This is F1 25 once more, this time utilizing the excessive preset at 1440p with high quality upscaling. The 16GB RX 9060 XT delivered 190 fps on common with 1% lows at 156 fps, all on PCIe 3.0. By comparability, the 8GB mannequin was 11% slower when utilizing PCIe 5.0.
Switching the 8GB card to PCIe 4.0 led to an extra 12% drop to 149 fps, and one other 13% drop utilizing PCIe 3.0. With each GPUs on PCIe 3.0, the 16GB mannequin offered practically 50% increased common body charges and 63% increased 1% lows. Once more, these are very affordable settings – excessive preset at 1440p with high quality upscaling – so in the event you’re utilizing an older PC, it is important to go for the 16GB mannequin.
Monster Hunter Wilds
In Monster Hunter Wilds, the sport merely would not run properly on 8GB GPUs utilizing the extremely preset at 1440p with high quality upscaling. The 16GB mannequin averaged over 60 fps on PCIe 3.0 – twice the efficiency of the 8GB card beneath the identical circumstances.
Even with PCIe 5.0, the 8GB model struggled, delivering horrible body time efficiency and 1% lows of simply 2 fps.
Curiously, elevated PCIe bandwidth did enhance common body charges for the 8GB mannequin, but it surely really worsened the 1% lows, creating extra stuttering. This appears to be a results of rendering extra frames total whereas nonetheless working out of VRAM. The takeaway right here is that PCIe bandwidth alone can’t resolve the VRAM bottleneck of the 8GB RX 9060 XT on this title – particularly with solely 8 lanes obtainable.
Marvel’s Spider-Man 2
Spider-Man 2 runs exceptionally properly at 1440p with upscaling on the 16GB RX 9060 XT, averaging 112 fps with 1% lows of 66 fps. That is 26% quicker than the 8GB mannequin utilizing PCIe 5.0, and 38% quicker for 1% lows.
Switching the 8GB card from PCIe 5.0 to 4.0 induced a 20% efficiency drop. Transferring to PCIe 3.0 resulted in a good bigger 27% decline, dropping the typical body price to simply 52 fps.
This implies in a PCIe 3.0 system, utilizing the settings proven right here, the 16GB model of the RX 9060 XT delivers not less than 115% increased common body charges and 175% increased 1% lows. Even in PCIe 4.0 methods, the 16GB mannequin continues to be 58% quicker on common and 74% quicker for 1% lows.
Ratchet & Clank: Rift Aside
Transferring on to Ratchet & Clank: Rift Aside at 1440p with high quality upscaling utilizing the very excessive preset, the 16GB RX 9060 XT on PCIe 3.0 averaged 128 fps, with 1% lows of 97 fps – wonderful efficiency total.
Underneath the identical circumstances with PCIe 3.0, the 8GB mannequin was 103% slower on common and 155% slower in 1% lows. Upgrading to PCIe 4.0 improved efficiency for the 8GB mannequin by 30%, however the 16GB card was nonetheless 56% quicker. Transferring from PCIe 4.0 to five.0 gave the 8GB mannequin one other 32% increase, making it usable, though the 16GB mannequin nonetheless maintained a 20% efficiency benefit.
Indiana Jones and the Nice Circle
Subsequent up is Indiana Jones and the Nice Circle. Right here, the 16GB RX 9060 XT utilizing PCI Categorical 3.0 rendered a mean of 135 fps, with 1% lows at 113 fps – very stable efficiency total. Underneath the identical circumstances, the 8GB mannequin utilizing PCIe 3.0 managed simply 45 fps, making the 16GB model 200% quicker, and 223% quicker when evaluating 1% lows.
Rising the PCIe bandwidth to PCIe 4.0 improved the 8GB mannequin’s common body price by a considerable 44%, reaching 65 fps. Nonetheless, this was nonetheless lower than half the efficiency of the 16GB mannequin. Switching to PCIe 5.0 delivered an extra 42% efficiency increase, and total efficiency turned fairly good. Even so, the 16GB card remained 47% quicker – whereas nonetheless working on PCIe 3.0.
Dragon Age: The Veilguard
Dragon Age: The Veilguard doesn’t run properly on 8GB graphics playing cards utilizing the extremely preset at 1440p with upscaling, whatever the PCI Categorical mode. On PCIe 3.0, the 16GB mannequin was 174% quicker.
Even after switching the 8GB card to PCIe 4.0, the 16GB model remained 70% quicker, and nonetheless held a 31% lead when the 8GB card used PCIe 5.0. Regardless of these positive factors, 1% lows on the 8GB card have been poor throughout all PCIe configurations.
Bonus: 5800X3D Testing
F1 25 (5800X3D)
To wrap up testing, we put in each GPUs into our Ryzen 7 5800X3D system, which is restricted to PCIe 4.0 – the best interface supported by the AM4 platform.
Curiously, the 16GB RX 9060 XT delivered practically equivalent efficiency on each the 5800X3D and 9800X3D methods, as GPU compute efficiency was the limiting issue. So, whatever the CPU used, the total potential of the 16GB mannequin was unlocked with these high quality settings.
The 8GB mannequin, nonetheless, continued to point out inconsistencies. Utilizing PCIe 4.0 with the 9800X3D, the 16GB model was already 27% quicker than the 8GB mannequin. However on the 5800X3D, the efficiency hole widened to 44%, attributable to a further 11% efficiency drop – seemingly attributable to the slower DDR4-3600 reminiscence used on AM4.
The takeaway: at finest, you may count on efficiency just like the 9800X3D when utilizing an older or slower CPU, however most of the time, the hole between the 8GB and 16GB fashions will likely be even bigger.
Ratchet & Clank (5800X3D)
It is vital to notice that this efficiency hole is not constant throughout all titles.
Take Ratchet & Clank: Rift Aside, for instance. Right here, the 8GB RX 9060 XT utilizing PCIe 4.0 delivered practically equivalent common body charges when paired with both the 5800X3D or the 9800X3D. Nonetheless, 1% lows – which have been already poor on the 9800X3D – have been even worse on the 5800X3D.
Making Sense of the Numbers
So there you’ve got it. As proven earlier with the Core i7-8700K, you completely don’t wish to run out of VRAM on a PCI Categorical 3.0 system, even when all 16 lanes can be found. These utilizing PCIe 4.0 methods will fare significantly better, however as we additionally noticed with PCIe 5.0, a major efficiency hit is not at all times avoidable.
In practically all the video games examined, the 8GB RX 9060 XT was successfully unusable when restricted to PCIe 3.0 – F1 25 with the excessive preset being the one actual exception, although even then, efficiency lagged properly behind the 16GB mannequin.
This highlights that even with extra modest settings, efficiency can drop sharply on older platforms. You solely must exceed the native video reminiscence restrict barely to come across severe efficiency points… issues that will typically go unnoticed on fashionable methods with ample VRAM and bandwidth.
PCIe bandwidth ought to by no means be a performance-limiting issue, and the one dependable method to make sure that is by having sufficient VRAM. At this level, 16GB is ample, 8GB will not be.
One other vital issue to think about is motherboard configuration. Many motherboards, particularly budget-oriented fashions, will scale back the PCIe lane rely to the first GPU slot when the secondary PCIe slot is populated. So, if something is put in within the secondary PCIe x16 slot, each the first and secondary slots could function in an x8/x8 configuration, one thing that’s extraordinarily frequent.
This successfully halves the obtainable PCIe bandwidth for the graphics card. For instance, on a PCIe 4.0 system with all 16 lanes energetic, there’s 64 GB/s of bandwidth. However with solely 8 lanes, that drops to 32 GB/s, or the identical bandwidth you’d get from PCIe 3.0 x16.
In the end, PCIe bandwidth ought to by no means be a performance-limiting issue, and the one dependable method to make sure that is by having sufficient VRAM. At this level, 16GB is ample, 8GB will not be. That stated, having the total 16 lanes obtainable nonetheless advantages the 8GB RX 9060 XT considerably. We’ll revisit this and a few new testing in a follow-up benchmark session with totally different GPUs and PCIe configurations.
