Palantir has helped the Internal Revenue Service’s Criminal Investigations office probe a variety of financial crimes in the U.S. for much of the last decade, The Intercept reported.
The IRS has paid the firm $130 million since 2018 to use its data analysis software to pore over financial records for investigative purposes, the outlet reported, citing public records detailing Palantir’s IRS contract that were obtained by the nonprofit watchdog group American Oversight.
It was previously known the IRS was using Palantir’s products, and that the agency sees the software as a way to automate and modernize audits. Last summer, it was also reported that Palantir was assisting DOGE, the “government efficiency” initiative launched by President Trump’s executive order with a project designed to access IRS records. However, the extent of the agency’s use of the company’s tools had not been previously reported.
The software, Palantir’s Lead and Case Analytics platform, is being used to aggregate and analyze data across a variety of federal agencies. The software can find “connections from millions of records with thousands of links” between various databases, and the tool is particularly good at mapping human relationships and communications, according to the outlet.
Earlier this week, American Oversight sued the Trump administration for public records related to numerous federal agencies’ use of Palantir tools, including the IRS. TechCrunch has reached out to Palantir for more information and will update the article if the company responds.
SimplePlanes is back! Build planes, cars, or anything else using powerful procedural parts! Explore detailed environments with friends in multiplayer, cause chaos, or download 1,000,000 community crafts for free.
BUILD PLANES Build planes using enhanced wing and fuselage parts. It’s easier than ever to build the plane of your dreams! Reshape the airfoil of your wings using real NACA profiles with overhauled flight physics, or use improved fuselage editing to create an endless variety of shapes. Then grab your input device of choice (mouse/keyboard, controller, and HOTAS are all supported) and fly!
BUILD ANYTHING Build anything else you can imagine using a host of new parts. With the new suite of procedural powertrain parts, you can build cars, propeller aircraft, forklifts, and so much more, with detailed physics. If you’re overwhelmed by the idea of building something from scratch, there are tons of stock crafts available from the start that you can load, tune, and reverse engineer, with an extra 40+ hidden around the map waiting to be discovered! Pratfall
CUSTOMIZE IT Make your crafts your own with paint texturing and engine tuning. Once you’ve put your craft together, you can use the improved paint tool to give it colour and texture. Then when that’s done, you can get into the nitty gritty with engine customization, tweaking the size, cylinder count, transmission, and more to get your craft running exactly the way you want it.
PLAY WITH FRIENDS (OR ENEMIES!) Cause chaos in a variety of activities in multiplayer, or just hang out. With multiplayer, you can play with up to nine other players (or even more in a private lobby) and finally test your crafts on something other than bots. Race each other, kill each other, co-operate with each other, and more! And with our new playable mascot, Major Chad, you can get out and stretch your legs with other players as well!
From Reactive to Ready: A 90-Day Supply Chain AI Roadmap for Mid-Market Ops Leaders
Most supply chain AI conversations stall in the same place. The ops leader knows the problem. The case for doing something is clear. The question that does not have a clean answer is: what does the first 90 days actually look like?
This is the roadmap USM Business Systems uses with mid-market manufacturing and logistics clients who are moving from interest to implementation. It is designed for organizations that do not have 18 months or a seven-figure platform budget. It is designed for teams that want to start, measure, and expand.
Before You Start: The Three Inputs That Determine Your Roadmap
A 90-day AI roadmap for supply chain is only as good as the three inputs that shape it. Get these clear before any build decision is made.
Input 1: The Problem With the Clearest Cost
Every mid-market supply chain operation has multiple AI opportunities. The teams that move fastest pick one. The one with the most direct and measurable cost attached.
Supplier lead time visibility. Inventory coverage calculation speed. Demand signal latency. Pick the one where someone can tell you what a miss costs in dollars, hours, or margin. That is where you start.
Input 2: Your Current Data Access Points
The roadmap is shaped by what you can connect the agent to. ERP API access. WMS data exports. Supplier EDI feeds. Order management integrations. You do not need all of these to start. You need the ones relevant to the problem you are solving.
A two-week scoping engagement with USM maps your data access reality and builds the agent architecture around what exists, not what would be ideal.
Input 3: The Success Metric
Before build begins, define what success looks like at 90 days. A number. Coverage calculation time reduced from 6 hours to 45 minutes. Near-misses surfaced with 72 hours of lead time instead of 24. Report generation recovered from Thursday manual build to automated Monday delivery.
That metric drives scope. It also drives the conversation about whether to expand.
Days 1-14: Scoping and Architecture
This is not a sales process. It is a working session.
Data environment mapping: what systems exist, what APIs are accessible, what exports are available
Problem prioritization: identify the one or two problems with the clearest ROI and the fastest measurement cycle
Agent architecture design: what the agent will connect to, what it will monitor, what it will surface
Success metric definition: specific, measurable, and agreed upon before build begins
At the end of day 14, you have an architecture document, a build scope, a timeline, and a defined metric.
Days 15-60: Build and Integration
The build phase runs in two tracks simultaneously.
Track one is data integration. The agent connects to your existing systems and begins ingesting live data. This phase surfaces the data quality issues that need to be addressed before the agent can produce reliable outputs. Those issues are resolved here, not discovered after go-live.
Track two is agent logic development. The monitoring rules, the exception thresholds, the scenario modeling logic, and the reporting templates are built and tested against real data from your operation.
By day 45, a test version of the agent is running against your data. The supply chain team begins evaluating outputs. Feedback shapes the final configuration before go-live.
Days 61-90: Go-Live and Measurement
Go-live is not a launch event. It is a transition. The agent moves from test to production. The team begins using it as the primary source for the problem it was built to solve.
The measurement cycle starts at day one of production. The success metric defined in scoping is tracked weekly. By the end of day 90, you have six weeks of live data showing the impact on decision time, report generation, near-miss visibility, or whatever metric was set.
That six weeks of measurement data is what drives the conversation about what to build next.
The Expansion Path
The teams that get the most out of supply chain AI do not deploy a platform across the entire operation on day one. They solve one problem, measure it, and expand.
After a successful first deployment, the common expansion paths are:
Adding supplier performance monitoring to an inventory visibility agent
Expanding from lead time tracking to landed cost scenario modeling
Connecting demand signal inputs from a second channel or geography
Integrating logistics lane performance data into coverage calculations
Each expansion is scoped and built with the same 8-12 week discipline. The architecture from the first deployment is designed to support expansion from the start.
The supply chain leaders who move fastest on AI do not have bigger budgets or cleaner data than their peers. They pick one problem, run a contained build, and measure it. That is the entire edge.
USM’s POC Commitment
For qualified supply chain and logistics engagements, USM fronts the proof-of-concept cost. You identify the problem. We scope and build the initial deployment. You measure the output before making a larger commitment.
The engagement starts with a scoping conversation. If the architecture is sound and the ROI case is clear, we move to build within two weeks.
Ready to scope your first supply chain AI deployment? Start with a 30-minute conversation at usmsystems.com. No pitch deck. Just the architecture conversation.
To paraphrase some lyrics from my mispent youth pining for a Hot Topic, ‘this ain’t a scene, it’s a goddamn AI arms race’. The preview of DeepSeek V4 is now live, and given that the AI giant was reportedly not so keen to grant Nvidia and AMD early access to the new model, the American AI industry has been equally keen to outpace China-based businesses any way it can. Case in point, Nvidia is really pushing OpenAI’s GPT-5.5.
DeepSeek wouldn’t give Nvidia the sneak peek, so 10,000 of its employees got an early look at OpenAI’s latest frontier model instead. The company went with a widespread rollout of Codex, specifically, which is OpenAI’s agentic coding application powered by GPT-5.5. This has apparently resulted in big efficiency wins. “Debugging cycles that once stretched across days are closing in hours,” says Nvidia.
The Nvidia blog continues in a similarly effusive tone: “Experimentation that previously required weeks is turning into overnight progress in complex, multi-file codebases. Teams are shipping end-to-end features from natural-language prompts, with stronger reliability and fewer wasted cycles than earlier models.”
As with anything agentic, security was top of mind for the rollout. So, to “ensure maximum security and auditability” (and presumably to stop Codex from getting into anything it shouldn’t), all participating Nvidia employees were given a cloud-based virtual machine to run the AI agent from. From there, Codex can read company data but can’t directly edit it (or indeed, delete). To oversimplify, they locked the bot in a virtual plexiglass box.
The blog post also surfaces similarly gushing praise from employees, with choice quotes calling Codex’s results “mind-blowing” and “life-changing”. Codex was rolled out across plenty of company departments besides engineering and development, with folks working in “product, legal, marketing, finance, sales, HR, and operations” also getting to grips with the application (personally, I would love to hear the honest Codex thoughts of a grumpy office admin).
We tried a new thing with NVIDIA to roll out Codex across a whole company and it was awesome to see it work.Let us know if you’d like to do it at your company! pic.twitter.com/Xjn6ShrRuqApril 23, 2026
Second, GPT-5.5 runs on Nvidia GB200 NVL72 rack-scale systems. This set-up is apparently “capable of delivering 35x lower cost per million tokens and 50x higher token output per second per megawatt compared with prior-generation systems,” making it an economically appealing enterprise model. Company head Jensen Huang quipped to OpenAI CEO Sam Altman over email, “Fire up those Blackwells. We need more tokens!”
Keep up to date with the most important stories and the best deals, as picked by the PC Gamer team.
A few years back now I started shooting videos for my wife who wanted to start her own YouTube channel.
Up until that point I was mainly involved in editing videos and as far as shooting them went, it was strictly “point and shoot” family and friends stuff for me!
So the style of video that I was going to be learning and shooting for YouTube was a single person in front of the camera in maybe medium shot or medium close up.
To light something like that all the advice I came across was to just slap up a three point arrangement and we are all good to go.
The reality proved to be something entirely different!
Available lights, how much space is there, what are you shooting with, what is the focal length of the camera lens, how can you position the person in that available space to light them… I mean the variables seemed to just go on and on!
Each one of those variables meant some kind of compromise in some other aspect of the overall activity.
So in light of that I came across this video from Gabriel VIP that covers this exact subject in great detail taking all of those variables into consideration.
3 Steps to Turn Any Idea into a Story
When it comes to anyone achieving any sort of success in the world of online video I tend to see one set of factors pushed while one simple one falls away.
What gets pushed are the technical aspects of optimizing videos, recording them and uploading a certain way, understanding search algorithms and the list goes on… and on!
The one that gets missed is one of the most basic points of video creation and that point is story.
We humans just loves us some story and if you give us story, we will wait around till the end to find out what happens.
I’m sure I am not alone in saying that I have watched some pretty awful movies in my day and while watching them have said to myself and others, man, this is a really crappy movie!
Yet despite that… I watched until the end because once I am following the story, I have to see it through to find out what happens!
And even if the ending was just as crappy, I still feel correct in having persevered through to resolution.
So in light of that here’s what I think is a very good walkthrough on working out what the story is for any video.
How to Crop a 16×9 Video and Save the Output in a 9×16 Format
Converting a 16:9 (landscape) video to a 9:16 (portrait) video is a relatively simple process these days but there are a few things you need to keep in mind.
First up is that often when you add 16:9 footage to the timeline, your editing software assumes you want to operate at 16:9 from start to finish.
So you have to let it know that’s not what you are doing!
The other thing to keep in mind is that often as you go from 16:9 to 9:16, when you crop the video you may also have to zoom in on it to fill the new frame size.
That zoom-in may be minimal with no real effect on your footage or it may be extreme resulting in loss of quality (resolution).
It always helps to record your video if you can at the highest possible resolution available to you in order to preserve quality.
As an example I shoot my wife’s videos at 4K widescreen all the time on the understanding that at some point I may re-use that footage in a short form 9:16 video..
Often she will want portions of those long form videos cut into pieces for YouTube shorts, Instagram Stories etc.
In that case I set up projects that are at 1080p resolution in 9:16 and using the 4k footage I can zoom in and crop easily with no real loss of quality.
How to Use Speed Ramp Like a Pro (Filmora Editing Tips)
Last week I added a tutorial from the guys at Movavi showing a few tips on how to speed ramp effectively.
This week the folks at Wondershare posted their own speed ramping tutorial and have added what I think are some great points to the conversation.
The key point they are pushing is that if your footage has not been shot with speed changes in mind, you may not get the results you imagined!
For example if you shoot at 30fps and want to slow the footage down, the motion is going to get very blurry in the process.
On the other hand if you are shooting with the intention of speeding the motion up, then your camera movement and even the movement in the frame has to be really smooth.
Any sudden movements or jerkiness within the shot will only be accentuated in fast motion.
Check out the video for some well considered tips related to speed ramping.
How to Cut a Video Simply
For anyone just starting out in the world of video editing, even the simplest action can seem like a bit of a nightmare!
Let’s face it, all video editing software user interfaces are unbelievably confusing in the beginning because of one key reason.
Video editing software can not mimic anything familiar to the new user from the real world.
Microsoft Word looks like a piece of paper in a typewriter, the drawing tools in Canva look like a sheet of paper with some pens and stuff.
There is nothing like that for video editing so… it’s all a bit confusing!
In the video below from the folks at Movavi you can see how to at least make those first basic cuts without chopping everything to pieces, losing it all (and your mind!) and starting again!
The (Real) Secret to Getting Beautiful iPhone Cinematography
Whether you want to go with Theodore Roosevelt saying, “Comparison is the thief of joy” or Mark Twain’s, “Comparison is the death of joy” the result is always the same.
And trust me when I say no-one understands that concept better than your average marketing guy in just about any company anywhere!
The reason I say this is because of the video below created by Zac Ramelan who has been producing videos at a pro level for years now.
In that video he points out that yes, you can create a fully cinematic, Hollywood level video with an iPhone.
All you really need on top of that is a million dollar budget and you are good to go!
When you compare those results with an iPhone to your own results I think we can safely say that any joy has made its way down the toilet!
So in the video he strips away that million dollar budget to make his own short movie but in the process delivers the real lesson.
The phone he uses to record the video is not the make/break point of the process.
The real driver of that process is the technical skill and creative ideas the person making the video brings to the table.
DaVinci Resolve 21 Quick Tip: Fix Your Audio Levels Fast
Over the past year or so I have been using various A.I driven tools to correct or enhance the audio on some of the projects I have been working on.
One thing that I have noticed very clearly is that as far as most of those solutions go, the better the quality of what you input, the better the output will be.
For example I have one client who sometimes finds himself recording video well away from any studio scenario.
Usually that means audio recorded though the onboard microphone of his smartphone and on top of that he is often either outside or in a noisy environment somewhere.
If I feed that audio into an A.I. tool like Adobe Podcast, I will get pretty decent results.
However, if I clean that audio up just a little in DaVinci Resolve or similar by taking out some background noise, adjusting levels, maybe doing a little Eq work before I send it off, the results are way, way better!
Here’s a little video from Jason Yadlovski showing how to clean up and level an audio track fast.
DaVinci Resolve 21 Can Now Animate Anything to Audio (No Plugins)
This is a preview of one of the features inside the new version of DaVinci Resolve to be released… soon?
However the official “latest version” is still Resolve 20.
Black Magic have generally followed this release path where they create the new version, test it, release as a beta to the public for a few months, go back to the drawing board for bug fixes then finally release the full version.
Anyhoo, a few brave souls (not me!) have installed the beta and are posting their experiences online.
Here’s one from Jason Yadlovski showing one of the new features.
Amazing Effects in Classic Films – How Did They Pull It Off?
This episode 12 in a regular series (irregularly posted!) by the folks at Film Riot.
These days it is becoming more and more a matter of being able to write a precise SFX prompt into an A.I. video model to create astounding SFX.
So I like to go back in time to when dinosaurs roamed the earth and the special effects department did not have things like A.I or even CGI!
A time when some really smart thinkers used a comprehensive knowledge of their craft to think up effects techniques all done with their human brains!
Key Takeaways
This week’s Friday Roundup covers various aspects of creating and editing videos for YouTube, starting with a lighting setup masterclass.
It emphasizes the importance of storytelling in video creation, suggesting that a compelling story keeps viewers engaged.
Readers can learn how to convert 16:9 videos to 9:16 format while maintaining quality by using high-resolution footage.
Additionally, the article discusses effective speed ramping techniques and basic cutting methods for beginners.
Finally, it highlights new features in DaVinci Resolve 21 for audio animation and quick audio level adjustments.
Welcome to the 1.118 release of Visual Studio Code.
Happy Coding!
April 23, 2026
Adopt the Copilot CLI SDK session-title APIs as the source of truth for session names. #311535
Add keybindings such as Ctrl+1 and Ctrl+2 to quickly switch between sessions in the Agents app. #310992
Add auto model support to the Copilot CLI agent. #311299
April 22, 2026
Add descriptions to the chat customization creator menu, helping users understand which skill location to use for each type of customization. #295151
Let users opt in to testing the TypeScript 7.0 nightlies. #311966
Copilot CLI responses in the chat panel show a model badge indicating which model handled the request. #303960
Add support for auto model selection in Copilot CLI. #311779
April 21, 2026
The Copilot CLI SDK resolves node-pty from VS Code via hostRequire, eliminating the need to copy node-pty binaries into the SDK’s prebuilds folder at build time or runtime. #307746
We really appreciate people trying our new features as soon as they are ready, so check back here often and learn what’s new.
Cross the local bureaucratic comedy of Parks and Recreation with the small-town strangeness of Twin Peaks, and you’ll get a sense of the singular tone of Apple TV‘s new genre gem Widow’s Bay.
Created by Katie Dippold, herself a writer for Parks and Recreation, the seriesartfully blends horror and comedy to create an enthralling portrait of a town you’ll want to get lost in… even if some of its locals would advise you to run the hell away.
What’s Widow’s Bay about?
Matthew Rhys in “Widow’s Bay.” Credit: Apple TV
A charming island 40 miles off the coast of New England, Widow’s Bay has a lot going for it. It boasts picturesque coastal views, invigorating ocean breezes, and atmospheric fog banks that have absolutely nothing hiding in them, do you hear me?
At least, that’s what Mayor Tom Loftis (Matthew Rhys) has to say. He’s trying to turn Widow’s Bay into the next Martha’s Vineyard, a quest that locals like former fisherman Wyck (Stephen Root) threaten to derail with their claims that Widow’s Bay is cursed.
Tom initially brushes Wyck and his allies off in favor of bringing in tourists. It’s a page straight from the playbook of Mayor Vaughn from Jaws, which Widow’s Bay pays loving homage to throughout. However, when haunts start knocking at his door, Tom has to face the truth: There’s something truly sinister at work on the island, and it’s only getting worse.
Widow’s Bay is a wonderfully frightening watch.
Matthew Rhys and Stephen Root in “Widow’s Bay.” Credit: Apple TV
To pull off its many scares, Widow’s Bay draws inspiration from a wide range of horror iconography. Stephen King‘s influence hangs over the series like the town’s unshakeable mist, present in everything from the New England setting to the show’s title font, a clear nod to the style of his earlier covers.
The aforementioned Jaws plays a huge role, too, and not just because of Tom’s role as a skeptical, tourism-hungry mayor. (To Tom’s credit, he’s far less willing to put people in danger for the sake of money than Mayor Vaughn.) Root’s salty veteran sailor Wyck bears shades of Jaws‘ Quint, and an episode devoted to a cursed ocean outing recalls Jaws’ third act. But the biggest lesson Widow’s Bay takes from Stephen Spielberg’s horror classic is the fear of the unknown. Just as Jaws generates suspense by holding off on showing us its biggest threat until later in the runtime, so too does Widow’s Bay keep its viewers in the dark about its many frights.
And what darkness it is! Visually, Widow’s Bay thrives in rich, inky black tones where you can crucially still make out every little detail, a rarity in TV nowadays and a testament to the show’s production team and crew. Series directors Hiro Murai (Atlanta), Sam Donovan (Severance), Andrew DeYoung (Friendship), and Ti West (the X trilogy) weaponize this darkness — along with fog and ocean depths — to its fullest obscuring extent. That expert build-up of tension made the subsequent reveals hit all the harder, to the point that I couldn’t get through an episode without screaming or cackling. (Or, most often, some mix of both.)
Widow’s Bay nails the balance between scares and laughs.
Kate O’Flynn in “Widow’s Bay.” Credit: Apple TV
Widow’s Bay‘s scares pull from sailor superstitions, classic slashers, and more, but they never feel cheap. That’s because the series roots them firmly in its characters’ anxieties, like Tom’s worries about tourism, or town hall staffer Patricia’s (Kate O’Flynn) desperate need to be liked. (She’s so desperate, in fact, that she may have falsely claimed to be the sole survivor of a serial killer who murdered some of her high school classmates.)
The latter produces a series highlight, an episode where a frantic Patricia attempts to host the perfect party. The social isolation she faces from her disdainful former classmates is just as wince-worthy as the uncanny events surrounding the event, resulting in a one-two punch of cringe comedy and horror.
Widow’s Bay oftenoperates in that sweet spot between horror and comedy, which tends more toward the dry and bizarre than Dippold’s previous work on sitcoms like Parks and Recreation or comedy films like Ghostbusters (2016). Instead of cracking jokes at a mile per minute, Widow’s Bay finds the funny in the eerie. Think historical wax figures, a cursed party game called “Teeth,” and sight gags about cannibalism.
Widow’s Bay‘s central trio of Rhys, Root, and O’Flynn play these occurrences as straight as they can, adding further to the show’s brand of offbeat strangeness. Rhys and Root are tremendous foils as a skeptic and believer forced to work together. And O’Flynn proves an absolute scene stealer as Patricia, marrying her hilarious scorn for others with the genuine pain of being scorned right back.
The rest of Widow’s Bay‘s ensemble, which includes Somebody Somewhere‘s Emmy-winning Jeff Hiller, Dale Dickey, and several surprising guest stars, further builds out the town of Widow’s Bay. Their efforts, as well as the weatherbeaten production design by Steve Arnold (Midnight Mass, another superb cursed island series), turn Widow’s Bay into a real town. You can almost taste the salt air (or hear the screams of cursed souls in the distance) just by watching.
In the end, the town of Widow’s Bay is as deliciously odd as the show itself. What begins with a bit of a haunt-of-the-week structure soon evolves into a deeper unraveling of Widow’s Bay’s darkest secrets, as well as the choices that kept them in place. While I’d love for the show to return to its haunt-of-the-week mode occasionally, I also adore what it became. That it can pull off this transformation is proof of its sheer uniqueness. Forget being the next Martha’s Vineyard: Widow’s Bay sets a new, wonderfully weird course all by itself.
I’ve spent the past few weeks testing the pants off of several graphics cards. Such is the life of a hardware writer. Alongside my 8 GB vs 16 GB VRAM testing, I’ve also reviewed the RTX 5060 Ti 8 GB and the RX 9060 XT 8 GB, and found them to deliver reasonable gaming performance on average—although the Nvidia card definitely wins the day. And yes, the 16 GB versions are faster overall.
One question has been nagging at the back of my mind, though. While I’ve proven that the 8 GB variants of these semi-budget GPUs are often slower on average compared to their 16 GB equivalents, can overclocking even the odds?
To find out, I’ve tweaked the twangers out of my 8 GB graphics card samples, and then run them across our GPU benchmarking suite. I’ve then compared that data with both the stock performance and the overclocked performance of the RTX 5060 Ti 16 GB and RX 9060 XT 16 GB.
It’s been a big old showdown, and I’ve come up with some interesting results. Yes, overclocking an 8 GB graphics card can often even the odds with its more-expensive 16 GB variant in the benchmarks. But, in some games, there’s simply no replacement for VRAM—and the tests don’t always tell the whole story.
Overclocking methodology
(Image credit: Future)
I’ve kept things real simple in regards to overclocking, as you can easily get into the weeds chasing tiny percentage gains and fighting stability issues as a result.
Essentially, it’s been a process of upping the core clock and memory speeds, then running demanding benchmarks over and over again until the GPUs crash, before backing things down to a state where each can complete all of our tests. This differs for each GPU, so your mileage may vary here.
Article continues below
Some of my tested cards have hard power limits, so for the most part the power levels have been kept stock for comparison purposes. The RTX 5060 Ti 8 GB, for example, couldn’t get past +350 MHz core speeds and a +500 MHz memory clock without a power boost, something its stock BIOS doesn’t allow—so that’s where I’ve stopped.
The exception to this has been the RX 9060 XT 8 GB. I had no luck undervolting my 16 GB sample to allow for higher clockspeeds, as it crashed almost immediately as soon as I dropped the voltage by a small amount, even with a higher power limit overall. Still, it accepted a reasonable +300 MHz core, 300 MHz memory overclock at stock power levels nonetheless.
My 8 GB Asus sample, however, needs an undervolt to accept any clockspeed tweaks without crashing, so I went for a stable -100 Mv. This will differ from card to card, but it did allow me to push the core clock to +300 MHz, and the memory to +200 MHz.
Our Jacob provided the figures for the Palit Infinity 3 RTX 5060 Ti 16 GB. As you’ll see from the results, a +400 MHz core overclock alone still gains some considerable performance from the Nvidia GPU. It’s also worth mentioning that all my tested cards kept within temperature limits at these speeds, which is something to keep an eye on if you do it for yourself.
Keep up to date with the most important stories and the best deals, as picked by the PC Gamer team.
Take these results as they’re intended—a demonstration of what basic overclocking techniques can potentially do to 8 GB graphics card performance, not any kind of guarantee over what you will gain from your own cards. If you’d like to buy one and break out the liquid nitrogen, be my guest. Me? I value my fingers too much, and I’d only end up spilling it everywhere anyway.
Overclocked 8 GB graphics cards
(Image credit: Future)
Let’s kick things off by looking at the performance gains you can get out of an overclocked 8 GB graphics card. At 1080p, we can see that all of our 8 GB samples show average frame rate improvements over the stock GPUs.
No great surprises there. However, while the gains are often small, there is the odd exception. The Talos Principle 2, is one such example, with the otherwise-unimpressive RX 9060 XT 8 GB gaining a whole 10 frames over the stock version. That allows it to soar from the bottom of my results to the very top.
Overclocked 8 GB GPUs – 1080p gaming benchmarks
RX 9060 XT 8 GB Asus Prime
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
RTX 5060 Ti 8 GB Palit Dual
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
Black Myth Wukong (1080p High) Data
Product
Value
RX 9060 XT 8 GB Asus Prime
71 Avg FPS, 58 1% Low FPS
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
75 Avg FPS, 67 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual
74 Avg FPS, 62 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
80 Avg FPS, 69 1% Low FPS
RX 9060 XT 8 GB Asus Prime
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
RTX 5060 Ti 8 GB Palit Dual
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
Cyberpunk 2077 (1080p RT Ultra) Data
Product
Value
RX 9060 XT 8 GB Asus Prime
41 Avg FPS, 26 1% Low FPS
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
47 Avg FPS, 30 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual
52 Avg FPS, 35 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
56 Avg FPS, 36 1% Low FPS
RX 9060 XT 8 GB Asus Prime
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
RTX 5060 Ti 8 GB Palit Dual
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
F1 24 (1080p Ultra high) Data
Product
Value
RX 9060 XT 8 GB Asus Prime
82 Avg FPS, 66 1% Low FPS
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
94 Avg FPS, 74 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual
90 Avg FPS, 73 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
92 Avg FPS, 75 1% Low FPS
RX 9060 XT 8 GB Asus Prime
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
RTX 5060 Ti 8 GB Palit Dual
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
Homeworld 3 (1080p Epic) Data
Product
Value
RX 9060 XT 8 GB Asus Prime
111 Avg FPS, 63 1% Low FPS
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
115 Avg FPS, 60 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual
108 Avg FPS, 32 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
121 Avg FPS, 55 1% Low FPS
RX 9060 XT 8 GB Asus Prime
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
RTX 5060 Ti 8 GB Palit Dual
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
Metro Exodus Enhanced Edition (1080p Ultra) Data
Product
Value
RX 9060 XT 8 GB Asus Prime
93 Avg FPS, 64 1% Low FPS
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
99 Avg FPS, 67 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual
101 Avg FPS, 65 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
108 Avg FPS, 71 1% Low FPS
RX 9060 XT 8 GB Asus Prime
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
RTX 5060 Ti 8 GB Palit Dual
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
The Talos Principle 2 (1080p Ultra) Data
Product
Value
RX 9060 XT 8 GB Asus Prime
66 Avg FPS, 35 1% Low FPS
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
76 Avg FPS, 49 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual
69 Avg FPS, 37 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
70 Avg FPS, 38 1% Low FPS
RX 9060 XT 8 GB Asus Prime
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
RTX 5060 Ti 8 GB Palit Dual
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
Total War: Warhammer 3 (1080p Ultra) Data
Product
Value
RX 9060 XT 8 GB Asus Prime
90 Avg FPS, 42 1% Low FPS
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
96 Avg FPS, 58 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual
81 Avg FPS, 59 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
85 Avg FPS, 62 1% Low FPS
In F1 24, the same 8 GB card miraculously gains 12 frames on average. This is a game that already seems to favour AMD GPUs, but the improvement is impressive nonetheless.
Overall, though, the RTX 5060 Ti 8 GB beats out its AMD competition when comparing stock to stock, and overclocked to overclocked. Adding a dose of enhanced speed to the little AMD GPU does allow it to jump ahead of the stock Nvidia card, though, which suggests it’s worth doing if you’ve decided to save some cash with the former.
Overclocked 8 GB GPUs – 1440p gaming benchmarks
RX 9060 XT 8 GB Asus Prime
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
RTX 5060 Ti 8 GB Palit Dual
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
Black Myth Wukong (1440p High) Data
Product
Value
RX 9060 XT 8 GB Asus Prime
48 Avg FPS, 43 1% Low FPS
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
51 Avg FPS, 46 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual
50 Avg FPS, 43 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
54 Avg FPS, 47 1% Low FPS
RX 9060 XT 8 GB Asus Prime
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
RTX 5060 Ti 8 GB Palit Dual
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
Cyberpunk 2077 (1440p RT Ultra) Data
Product
Value
RX 9060 XT 8 GB Asus Prime
26 Avg FPS, 17 1% Low FPS
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
26 Avg FPS, 19 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual
33 Avg FPS, 25 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
34 Avg FPS, 26 1% Low FPS
RX 9060 XT 8 GB Asus Prime
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
RTX 5060 Ti 8 GB Palit Dual
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
F1 24 (1440p Ultra High) Data
Product
Value
RX 9060 XT 8 GB Asus Prime
51 Avg FPS, 41 1% Low FPS
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
57 Avg FPS, 47 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual
59 Avg FPS, 47 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
60 Avg FPS, 47 1% Low FPS
RX 9060 XT 8 GB Asus Prime
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
RTX 5060 Ti 8 GB Palit Dual
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
Homeworld 3 (1440p Epic) Data
Product
Value
RX 9060 XT 8 GB Asus Prime
75 Avg FPS, 58 1% Low FPS
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
81 Avg FPS, 49 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual
87 Avg FPS, 42 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
94 Avg FPS, 37 1% Low FPS
RX 9060 XT 8 GB Asus Prime
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
RTX 5060 Ti 8 GB Palit Dual
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
Metro Exodus Enhanced Edition (1440p Ultra) Data
Product
Value
RX 9060 XT 8 GB Asus Prime
68 Avg FPS, 49 1% Low FPS
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
69 Avg FPS, 48 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual
75 Avg FPS, 53 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
80 Avg FPS, 56 1% Low FPS
RX 9060 XT 8 GB Asus Prime
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
RTX 5060 Ti 8 GB Palit Dual
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
The Talos Principle 2 (1440p Ultra) Data
Product
Value
RX 9060 XT 8 GB Asus Prime
47 Avg FPS, 27 1% Low FPS
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
51 Avg FPS, 36 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual
44 Avg FPS, 30 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
46 Avg FPS, 32 1% Low FPS
RX 9060 XT 8 GB Asus Prime
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
RTX 5060 Ti 8 GB Palit Dual
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
Total War: Warhammer 3 (1440p Ultra) Data
Product
Value
RX 9060 XT 8 GB Asus Prime
60 Avg FPS, 44 1% Low FPS
RX 9060 XT 8 GB Asus Prime | +300 MHz | +200 Mem | -100 mV
64 Avg FPS, 46 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual
57 Avg FPS, 40 1% Low FPS
RTX 5060 Ti 8 GB Palit Dual | +350 MHz | +500 Mem
60 Avg FPS, 44 1% Low FPS
Switch things up to 1440p, and the gains become smaller. Again, no great surprise. Still, the little AMD card does benefit considerably from the extra speed, although Cyberpunk 2077 once again proves to be its downfall.
The average frame rate remains exactly the same for the RX 9060 XT 8 GB, whether overclocked or stock, although the 1% lows do improve by two frames. This is a pretty crunchy result overall, although I’m testing all of our games here without upscaling or frame generation, in order to represent the raw rendered fps difference underneath the hood.
The RTX 5060 Ti 8 GB, meanwhile, offers a solid set of improvements once overclocked, although it’s nothing mind-blowing. Still, extra frames are extra frames, and while its advantage is eaten into slightly by the RX 9060 XT’s overclocked gains, it’s still the top performer at 1440p on average.
Overclocked 16 GB graphics cards
(Image credit: Future)
So, what do the 16 GB cards have to offer by comparison? Well, higher frame rates overall, for a start. However, the gains from overclocking aren’t quite so impressive, with the odd exception.
Check out the eight extra frames gained by the RTX 5060 Ti 16 GB in Cyberpunk 2077 at 1080p. That’s with purely a core overclock, too. Given just how demanding Cyberpunk can be with the ray tracing turned up, I’d happily overclock this card for the extra smoothness here alone.
Overclocked 16 GB GPUs – 1080p gaming benchmarks
RX 9060 XT 16 GB XFX | +300 MHz | + 300 Mem
RTX 5060 Ti 16 GB Palit Infinity 3
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
Black Myth Wukong (1080p High) Data
Product
Value
RX 9060 XT 16 GB XFX
73 Avg FPS, 63 1% Low FPS
RX 9060 XT 16 GB XFX | +300 MHz | + 300 Mem
75 Avg FPS, 66 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3
75 Avg FPS, 65 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
81 Avg FPS, 31 1% Low FPS
RX 9060 XT 16 GB XFX | +300 MHz | + 300 Mem
RTX 5060 Ti 16 GB Palit Infinity 3
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
Cyberpunk 2077 (1080p RT Ultra) Data
Product
Value
RX 9060 XT 16 GB XFX
49 Avg FPS, 39 1% Low FPS
RX 9060 XT 16 GB XFX | +300 MHz | + 300 Mem
49 Avg FPS, 40 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3
51 Avg FPS, 41 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
59 Avg FPS, 42 1% Low FPS
RX 9060 XT 16 GB XFX | +300 MHz | + 300 Mem
RTX 5060 Ti 16 GB Palit Infinity 3
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
F1 24 (1080p Ultra high) Data
Product
Value
RX 9060 XT 16 GB XFX
95 Avg FPS, 61 1% Low FPS
RX 9060 XT 16 GB XFX | +300 MHz | + 300 Mem
97 Avg FPS, 75 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3
90 Avg FPS, 71 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
94 Avg FPS, 77 1% Low FPS
RX 9060 XT 16 GB XFX | +300 MHz | + 300 Mem
RTX 5060 Ti 16 GB Palit Infinity 3
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
Homeworld 3 (1080p Epic) Data
Product
Value
RX 9060 XT 16 GB XFX
106 Avg FPS, 65 1% Low FPS
RX 9060 XT 16 GB XFX | +300 MHz | + 300 Mem
123 Avg FPS, 74 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3
117 Avg FPS, 65 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
123 Avg FPS, 68 1% Low FPS
RX 9060 XT 16 GB XFX | +300 MHz | + 300 Mem
RTX 5060 Ti 16 GB Palit Infinity 3
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
Metro Exodus Enhanced Edition (1080p Ultra) Data
Product
Value
RX 9060 XT 16 GB XFX
96 Avg FPS, 42 1% Low FPS
RX 9060 XT 16 GB XFX | +300 MHz | + 300 Mem
100 Avg FPS, 69 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3
98 Avg FPS, 42 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
102 Avg FPS, 42 1% Low FPS
RX 9060 XT 16 GB XFX | +300 MHz | + 300 Mem
RTX 5060 Ti 16 GB Palit Infinity 3
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
The Talos Principle 2 (1080p Ultra) Data
Product
Value
RX 9060 XT 16 GB XFX
73 Avg FPS, 37 1% Low FPS
RX 9060 XT 16 GB XFX | +300 MHz | + 300 Mem
77 Avg FPS, 53 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3
75 Avg FPS, 41 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
80 Avg FPS, 47 1% Low FPS
RX 9060 XT 16 GB XFX | +300 MHz | + 300 Mem
RTX 5060 Ti 16 GB Palit Infinity 3
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
Total War: Warhammer 3 (1080p Ultra) Data
Product
Value
RX 9060 XT 16 GB XFX
94 Avg FPS, 72 1% Low FPS
RX 9060 XT 16 GB XFX | +300 MHz | + 300 Mem
97 Avg FPS, 72 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3
83 Avg FPS, 54 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
89 Avg FPS, 56 1% Low FPS
And yes, the RX 9060 XT 16 GB doesn’t seem to benefit as much from overclocking as the rest of my cards. I’d wager that’s down to lower sustained clock speeds without an undervolt, but again, this XFX sample refuses to play ball once you start fiddling with the power levels.
Still, an extra 4 fps in Metro Exodus Enhanced edition is good to see, and with a massive 27 fps improvement in the 1% lows, too.
In fact, improved 1% lows might be reason enough to overclock this card, even if the averages don’t budge by significant amounts. Homeworld 3, F1 24, The Talos Principle 2—a couple of simple slider adjustments in AMD Adrenaline appears to be all it takes.
Overclocked 16 GB GPUs – 1440p gaming benchmarks
RX 9060 XT 16 XFX | +300 MHz | + 300 Mem
RTX 5060 Ti 16 GB Palit Infinity 3
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
Black Myth Wukong (1440p High) Data
Product
Value
RX 9060 XT 16 GB XFX
48 Avg FPS, 44 1% Low FPS
RX 9060 XT 16 XFX | +300 MHz | + 300 Mem
50 Avg FPS, 45 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3
51 Avg FPS, 45 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
56 Avg FPS, 41 1% Low FPS
RX 9060 XT 16 XFX | +300 MHz | + 300 Mem
RTX 5060 Ti 16 GB Palit Infinity 3
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
Cyberpunk 2077 (1440p RT Ultra) Data
Product
Value
RX 9060 XT 16 GB XFX
30 Avg FPS, 24 1% Low FPS
RX 9060 XT 16 XFX | +300 MHz | + 300 Mem
30 Avg FPS, 25 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3
32 Avg FPS, 26 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
35 Avg FPS, 29 1% Low FPS
RX 9060 XT 16 XFX | +300 MHz | + 300 Mem
RTX 5060 Ti 16 GB Palit Infinity 3
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
F1 24 (1440p Ultra High) Data
Product
Value
RX 9060 XT 16 GB XFX
60 Avg FPS, 46 1% Low FPS
RX 9060 XT 16 XFX | +300 MHz | + 300 Mem
61 Avg FPS, 46 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3
56 Avg FPS, 45 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
62 Avg FPS, 50 1% Low FPS
RX 9060 XT 16 XFX | +300 MHz | + 300 Mem
RTX 5060 Ti 16 GB Palit Infinity 3
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
Homeworld 3 (1440p Epic) Data
Product
Value
RX 9060 XT 16 GB XFX
86 Avg FPS, 65 1% Low FPS
RX 9060 XT 16 XFX | +300 MHz | + 300 Mem
90 Avg FPS, 74 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3
97 Avg FPS, 64 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
104 Avg FPS, 66 1% Low FPS
RX 9060 XT 16 XFX | +300 MHz | + 300 Mem
RTX 5060 Ti 16 GB Palit Infinity 3
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
Metro Exodus Enhanced Edition (1440p Ultra) Data
Product
Value
RX 9060 XT 16 GB XFX
71 Avg FPS, 34 1% Low FPS
RX 9060 XT 16 XFX | +300 MHz | + 300 Mem
74 Avg FPS, 43 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3
72 Avg FPS, 34 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
73 Avg FPS, 31 1% Low FPS
RX 9060 XT 16 XFX | +300 MHz | + 300 Mem
RTX 5060 Ti 16 GB Palit Infinity 3
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
The Talos Principle 2 (1440p Ultra) Data
Product
Value
RX 9060 XT 16 GB XFX
49 Avg FPS, 35 1% Low FPS
RX 9060 XT 16 XFX | +300 MHz | + 300 Mem
51 Avg FPS, 36 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3
51 Avg FPS, 32 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
54 Avg FPS, 35 1% Low FPS
RX 9060 XT 16 XFX | +300 MHz | + 300 Mem
RTX 5060 Ti 16 GB Palit Infinity 3
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
Total War: Warhammer 3 (1440p Ultra) Data
Product
Value
RX 9060 XT 16 GB XFX
62 Avg FPS, 45 1% Low FPS
RX 9060 XT 16 XFX | +300 MHz | + 300 Mem
64 Avg FPS, 47 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3
60 Avg FPS, 44 1% Low FPS
RTX 5060 Ti 16 GB Palit Infinity 3 | +400 MHz
65 Avg FPS, 48 1% Low FPS
At 1440p, it’s much the same story for the 16 GB GPUs as our 8 GB cards. There are gains to be made with a dose of overclocking, but it’s usually not by a whole lot.
Three fps here, four fps there—it’s about what I expected to see from both our 16 GB GPUs, if I’m honest. The 1% low advantage is curbed a little at this resolution, too, although the RX 9060 XT 16 GB proves the exception yet again, as it manages to gain a chonky nine frames in Metro Exodus.
Otherwise, though? There’s no huge gain in performance for any of our cards, although more frames are more frames.
8 GB vs 16 GB overclocked graphics cards
(Image credit: Future)
So, what happens when we combine this data together? Well, to prevent building a couple of massive, hard to read charts, I’ve broken up my findings into two main chunks.
First, let’s take a look at the difference between the RTX 9060 XT 8 GB and 16 GB, at both stock and overclocked speeds.
Here’s a pleasing result to see from the off. In Black Myth Wukong at 1080p High settings, the overclocked RX 9060 XT 8 GB manages to best its 16 GB bigger brother at stock speeds. And even with an overclock, the RX 9060 XT is a mere single frame ahead of it in the 1% lows. The averages? Exactly the same.
Before we all go jumping up and down at the thought of massive savings, though (the 8 GB card is often considerably cheaper than the 16 GB version), here comes Cyberpunk 2077 to ruin the 8 GB card’s day.
Yep, it’s time for an early preview of my One Big Caveat™—certain games will show the flaws of a tight VRAM limit. The overclocked 8 GB AMD card gets within 2 frames of the stock 16 GB GPU on average, but I’d ask you to give some serious consideration to those 1% lows.
(Image credit: Future)
It might not look much on paper, but having played Cyberpunk 2077 on this card while overclocked, I can tell you that it has a tendency to drop frames fairly often when faced with only 8 GB of VRAM. The RX 9060 XT 16 GB isn’t perfect in this regard, either, but it’s a much smoother-feeling experience.
Still, 1080p gains are definitely in evidence here, and sometimes the overclocked RX 9060 XT 8 GB can beat the stock 16 GB version on average. In Homeworld 3, it wins by a whole nine frames—although even the stock 8 GB GPU posted higher figures than the 16 GB version in my testing.
The Homeworld 3 benchmark can be a twitchy thing, after all. I re-ran the tests many, many times, but these are the averages I ended up with. Shrug your shoulders with me, and move on.
At 1440p, the overclocked RX 9060 XT 8 GB comes good once again in Black Myth Wukong and the Talos Principle 2, handily beating the stock 16 GB card in both averages and 1% lows. In fact, in the former, it even manages to best its overclocked score. It’s only by a single frame, though, which is well within benchmark variances.
And yep, here comes Cyberpunk 2077 again to give the AMD card trouble. The overclock allows for a couple more frames in the 1% lows, but the extra VRAM of the RX 9060 XT 16 GB allows it to pull significantly ahead.
Nvidia RTX 5060 Ti – 8 GB vs 16 GB overclocked
RTX 5060 Ti | 8 GB vs 16 GB | 1080p overclocked gaming benchmarks
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
RTX 5060 Ti 16 GB | +400 MHz
Black Myth Wukong (1080p High) Data
Product
Value
RTX 5060 Ti 8 GB
74 Avg FPS, 62 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
80 Avg FPS, 69 1% Low FPS
RTX 5060 Ti 16 GB
75 Avg FPS, 65 1% Low FPS
RTX 5060 Ti 16 GB | +400 MHz
81 Avg FPS, 31 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
RTX 5060 Ti 16 GB | +400 MHz
Cyberpunk 2077 (1080p RT Ultra) Data
Product
Value
RTX 5060 Ti 8 GB
52 Avg FPS, 35 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
56 Avg FPS, 36 1% Low FPS
RTX 5060 Ti 16 GB
51 Avg FPS, 41 1% Low FPS
RTX 5060 Ti 16 GB | +400 MHz
59 Avg FPS, 42 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
RTX 5060 Ti 16 GB | +400 MHz
F1 24 (1080p Ultra high) Data
Product
Value
RTX 5060 Ti 8 GB
90 Avg FPS, 73 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
92 Avg FPS, 75 1% Low FPS
RTX 5060 Ti 16 GB
90 Avg FPS, 71 1% Low FPS
RTX 5060 Ti 16 GB | +400 MHz
94 Avg FPS, 77 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
RTX 5060 Ti 16 GB | +400 MHz
Homeworld 3 (1080p Epic) Data
Product
Value
RTX 5060 Ti 8 GB
108 Avg FPS, 32 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
121 Avg FPS, 55 1% Low FPS
RTX 5060 Ti 16 GB
117 Avg FPS, 65 1% Low FPS
RTX 5060 Ti 16 GB | +400 MHz
123 Avg FPS, 68 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
RTX 5060 Ti 16 GB | +400 MHz
Metro Exodus Enhanced Edition (1080p Ultra) Data
Product
Value
RTX 5060 Ti 8 GB
101 Avg FPS, 65 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
108 Avg FPS, 71 1% Low FPS
RTX 5060 Ti 16 GB
98 Avg FPS, 42 1% Low FPS
RTX 5060 Ti 16 GB | +400 MHz
102 Avg FPS, 42 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
RTX 5060 Ti 16 GB | +400 MHz
The Talos Principle 2 (1080p Ultra) Data
Product
Value
RTX 5060 Ti 8 GB
69 Avg FPS, 37 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
70 Avg FPS, 38 1% Low FPS
RTX 5060 Ti 16 GB
75 Avg FPS, 41 1% Low FPS
RTX 5060 Ti 16 GB | +400 MHz
80 Avg FPS, 47 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
RTX 5060 Ti 16 GB | +400 MHz
Total War: Warhammer 3 (1080p Ultra) Data
Product
Value
RTX 5060 Ti 8 GB
81 Avg FPS, 59 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
85 Avg FPS, 62 1% Low FPS
RTX 5060 Ti 16 GB
83 Avg FPS, 54 1% Low FPS
RTX 5060 Ti 16 GB | +400 MHz
89 Avg FPS, 56 1% Low FPS
But what about the RTX 5060 Ti, I hear you cry? Well, hold on to your hats, because the overclocked 8 GB card puts on quite an impressive showing in comparison to the stock 16 GB GPU.
In fact, it beats the stock RTX 5060 Ti 16 GB in all but one of my tests at 1080p, that outlier being The Talos Principle 2. Otherwise, the extra dose of speed allows the 8 GB card (which is fairly close in performance to its 16 GB variant to begin with at this resolution) to pull quite a few frames ahead on average, even occasionally beating the overclocked RTX 5060 Ti 16 GB.
RTX 5060 Ti | 8 GB vs 16 GB | 1440p overclocked gaming benchmarks
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
RTX 5060 Ti 16 GB | +400 MHz
Black Myth Wukong (1440p High) Data
Product
Value
RTX 5060 Ti 8 GB
50 Avg FPS, 43 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
54 Avg FPS, 47 1% Low FPS
RTX 5060 Ti 16 GB
51 Avg FPS, 45 1% Low FPS
RTX 5060 Ti 16 GB | +400 MHz
56 Avg FPS, 41 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
RTX 5060 Ti 16 GB | +400 MHz
Cyberpunk 2077 (1440p RT Ultra) Data
Product
Value
RTX 5060 Ti 8 GB
33 Avg FPS, 25 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
34 Avg FPS, 26 1% Low FPS
RTX 5060 Ti 16 GB
32 Avg FPS, 26 1% Low FPS
RTX 5060 Ti 16 GB | +400 MHz
35 Avg FPS, 29 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
RTX 5060 Ti 16 GB | +400 MHz
F1 24 (1440p Ultra High) Data
Product
Value
RTX 5060 Ti 8 GB
59 Avg FPS, 47 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
60 Avg FPS, 47 1% Low FPS
RTX 5060 Ti 16 GB
56 Avg FPS, 45 1% Low FPS
RTX 5060 Ti 16 GB | +400 MHz
62 Avg FPS, 50 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
RTX 5060 Ti 16 GB | +400 MHz
Homeworld 3 (1440p Epic) Data
Product
Value
RTX 5060 Ti 8 GB
87 Avg FPS, 42 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
94 Avg FPS, 37 1% Low FPS
RTX 5060 Ti 16 GB
97 Avg FPS, 64 1% Low FPS
RTX 5060 Ti 16 GB | +400 MHz
104 Avg FPS, 66 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
RTX 5060 Ti 16 GB | +400 MHz
Metro Exodus Enhanced Edition (1440p Ultra) Data
Product
Value
RTX 5060 Ti 8 GB
75 Avg FPS, 53 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
80 Avg FPS, 56 1% Low FPS
RTX 5060 Ti 16 GB
72 Avg FPS, 34 1% Low FPS
RTX 5060 Ti 16 GB | +400 MHz
73 Avg FPS, 31 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
RTX 5060 Ti 16 GB | +400 MHz
The Talos Principle 2 (1440p Ultra) Data
Product
Value
RTX 5060 Ti 8 GB
44 Avg FPS, 30 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
46 Avg FPS, 32 1% Low FPS
RTX 5060 Ti 16 GB
51 Avg FPS, 32 1% Low FPS
RTX 5060 Ti 16 GB | +400 MHz
54 Avg FPS, 35 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
RTX 5060 Ti 16 GB | +400 MHz
Total War: Warhammer 3 (1440p Ultra) Data
Product
Value
RTX 5060 Ti 8 GB
57 Avg FPS, 40 1% Low FPS
RTX 5060 Ti 8 GB | +350 MHz | +500 Mem
60 Avg FPS, 44 1% Low FPS
RTX 5060 Ti 16 GB
60 Avg FPS, 44 1% Low FPS
RTX 5060 Ti 16 GB | +400 MHz
65 Avg FPS, 48 1% Low FPS
At 1440p, the Nvidia card continues its overclocked reign of terror. It beats the stock 16 GB card in Black Myth Wukong, Cyberpunk 2077, F1 24, and Metro Exodus, and matches its benches exactly in Total War: Warhammer 3.
The clear winner in most of my tests, however, is the mildly-overclocked RTX 5060 Ti 16 GB. It’s a surprising set of results, and furthers my assessment that the RTX 5060 Ti 8 GB is the graphics card to buy if you can’t afford the higher-VRAM equivalent.
There is, however, a very large caveat to all of these figures. Follow me to the section below.
Analysis, and a very large caveat
(Image credit: Future)
So, what have we learned? Well, these results show that you can indeed overclock a cheaper 8 GB graphics card to bring its performance more in line with the 16 GB variant, and perhaps even exceed it—in the benchmarks. And that’s grand.
But before you put a substantial wad of cash back into your pocket and go running out into the streets to buy an 8 GB graphics card, we need to talk about something important—and that’s the real world.
In-game benchmark tools are useful ways of comparing the performance of graphics cards against each other. And each one of these GPUs has been tested using just such benchmarks. But I’d like to direct you towards my 8 GB vs 16 GB graphics card testing once more, and show you one last chart.
The Last of Us Part 1 – VRAM usage and performance
In-game testing, Ultra settings
RTX 5060 Ti 8 GB Palit Dual
RX 9060 XT 8 GB Asus Prime
RTX 5060 Ti 16 GB Palit Infinity 3
The Last of Us Part 1 (1080p Ultra + Quality upscaling) Data
Product
Value
RTX 5060 Ti 8 GB Palit Dual
89 Avg FPS, 48 1% Low FPS, 6.92 Avg. VRAM
RX 9060 XT 8 GB Asus Prime
79 Avg FPS, 52 1% Low FPS, 6.61 Avg. VRAM
RTX 5060 Ti 16 GB Palit Infinity 3
113 Avg FPS, 92 1% Low FPS, 9.84 Avg. VRAM
RX 9060XT 16 GB XFX
113 Avg FPS, 93 1% Low FPS, 11.47 Avg. VRAM
RTX 5060 Ti 8 GB Palit Dual
RX 9060 XT 8 GB Asus Prime
RTX 5060 Ti 16 GB Palit Infinity 3
The Last of Us Part 1 (1440p Ultra + Quality upscaling) Data
Product
Value
RTX 5060 Ti 8 GB Palit Dual
79 Avg FPS, 44 1% Low FPS, 6.98 Avg. VRAM
RX 9060 XT 8 GB Asus Prime
60 Avg FPS, 42 1% Low FPS, 6.71 Avg. VRAM
RTX 5060 Ti 16 GB Palit Infinity 3
92 Avg FPS, 73 1% Low FPS, 10.39 Avg. VRAM
RX 9060XT 16 GB XFX
89 Avg FPS, 72 1% Low FPS, 12.02 Avg. VRAM
RTX 5060 Ti 8 GB Palit Dual
RX 9060 XT 8 GB Asus Prime
RTX 5060 Ti 16 GB Palit Infinity 3
The Last of Us Part 1 (1440p Ultra+ Quality upscaling + 2x FG) Data
Product
Value
RTX 5060 Ti 8 GB Palit Dual
132 Avg FPS, 61 1% Low FPS, 7.21 Avg. VRAM
RX 9060 XT 8 GB Asus Prime
112 Avg FPS, 61 1% Low FPS, 6.74 Avg. VRAM
RTX 5060 Ti 16 GB Palit Infinity 3
147 Avg FPS, 78 1% Low FPS, 11.09 Avg. VRAM
RX 9060XT 16 GB XFX
149 Avg FPS, 104 1% Low FPS, 12.12 Avg. VRAM
This is my in-game VRAM testing of The Last of Us Part 1, using stock variants of all our tested graphics cards. These results were recorded in an actual level, where I ran around as if I was, y’know, playing video games for a living.
I’ve chosen to show it to you not because it shows overclocked figures (I simply don’t have time in the day to run the OC’d GPUs through all those particular tests too), but because it demonstrates the potential performance difference between the 8 GB and 16 GB GPUs when faced with a VRAM-swallowing game.
And yes, it’s a kinda ugly result. It’s not the only example, either. Resident Evil Requiem also demonstrates a similar performance delta at certain settings, among others.
Transposing the results I’m seeing here with the gains made from overclocking, I very much doubt that, in actual gameplay, the 8 GB cards can be overclocked to deliver the same performance as their 16 GB equivalents, at least in certain games. Benchmarks? It appears so. In-game? Perhaps not.
(Image credit: Sony)
Can overclocking make up some of the gap? Yes, it’s quite likely. But given the sometimes massive difference in performance in certain games at certain settings between the 8 GB cards and the 16 GB variants, I really doubt these overclocking gains will translate to parity when proper VRAM constraints come into play.
Benchmark tools are simply lighter on the VRAM than actual game worlds, and that’s likely what we’re seeing here.
I’ll have to do some more testing to find out for certain, so it’s back to the GPU mines for me. Still, if you own an 8 GB card, or are stretching your budget to afford one, I can confirm that there’s performance left on the table here—at least with the GPUs I have in front of me.
So overclock away, and enjoy those extra frames. It’s pretty simple to do, and there’s a possibility that you will get some extra performance for free. But can you force an 8 GB card to run games just as well as a 16 GB one, once in-game VRAM demands come into play? From what I’ve seen from all my testing, I’d say not.
In this third session of the Let’s Talk Knowledge Engineering series, Ben Taylor, Rainbird CTO and co-founder, is joined by Lucie Hunt, VP Enablement at Rainbird, to explore the practices that help knowledge engineering projects scale from early ideas into reliable production systems.
Together, they look at how strong knowledge architecture, clear graph design, disciplined testing, and structured change management help teams build models that are easier to reuse, maintain, and extend over time. The session focuses on what good looks like in practice, from setting knowledge boundaries and layering expertise through to designing graphs that stay clean and manageable as they grow.
You can register for the remaining sessions in the series here or watch past episodes.
What you’ll learn
Why knowledge architecture is a design discipline, and how setting the right boundaries helps graphs scale and remain maintainable.
How layering knowledge across foundational, domain, policy, and jurisdictional levels improves reuse and reduces duplication.
Why separating knowledge from data matters, and how it enables the same reasoning models to be applied across different systems and use cases.
What practical graph design best practices look like, including naming conventions, reusable concepts, rule design, and graph hygiene.
How testing, versioning, and structured change management help keep knowledge graphs reliable as requirements evolve.
Resources shared in the webinar
Rainbird Studio Community Edition: Experiment, model, and bring decisions to life, visit app.rainbird.ai
Rainbird Academy: Learn the foundations of explainable decision intelligence, visit academy.rainbird.ai
Rainbird Forum: Ask, discuss, and shape the conversation, visit forum.rainbird.ai
Diablo 4‘s second expansion, Lord of Hatred, is on the horizon, but not everything included will be immediately useful.
Aside from the new classes and questline, the Diablo 4 version of the Horadric Cube, first introduced in Diablo 2, is one of the big selling points of the Lord of Hatred DLC. If all goes well, it could shake up endgame itemization in a way that people have been asking for for years. For the uninitiated and in the simplest of terms (I go over it in more detail here), the Cube lets you use items you’d otherwise salvage to upgrade gear or create entirely new items, opening up an unprecedented level of gear customization and dramatically expanding the loot economy.
For the new issue of PC Gamer magazine, associate game director Zaven Haroutunian really hammered home how central the new Cube is to Lord of Hatred. The Diablo 4 iteration is such a big deal that Blizzard packed junk items into Lord of Hatred just to help players make better use of the Cube.
Article continues below
“I think players are going to shock us with what they’re going to do with it,” Haroutunian said, adding that those pesky item drops clogging up your inventory, wasting your precious clicking strength, and waiting to be salvaged will soon have a “second life” thanks to the Horadric Cube. “I will say that with Lord of Hatred, there’s at least one item we’re intentionally putting in there that’s basically useless because we know it’s going to be a target for the cube,” said Haroutunian.
The return of the Cube in Diablo 4 is accompanied by a separate and nearly as transformative loot change that’ll bring low-level loot back to the endgame with one huge caveat: it’ll drop with the chance to have a greater affix attached, yet another design tweak purposefully designed to feed into the Horadric Cube, as it means the chump change you pick up in dungeons will have the chance to become your new main weapon, chest piece, or shield. That should, in theory, make endgame runs a whole lot more interesting, which is presumably high up on Blizzard’s priority list as a lack of depth is one of the main pain points for high-level players. We’ll find out when Lord of Hatred launches on April 28.