Monday, August 4th 2025

AMD Radeon RX 9060 non-XT GPU Specs Surface: 2,048 Cores and Slower 8 GB GDDR6 Memory

AMD is nearing the release of its mid-range RDNA 4 GPU, the Radeon RX 9060 non-XT. According to leaked specifications obtained by VideoCardz, AMD will equip its Radeon RX 9060 GPU SKU with a Navi 44 XL die featuring 2,048 cores, clocked at 2,400 MHz game clock and boosting up to 2,990 MHz during peak boost clock settings. Paired with 18 Gbps GDDR6 memory featuring an 8 GB capacity, it operates on a 128-bit bus, delivering 288 GB/s of bandwidth. Specifications-wise, it appears to be very close to its older sibling, the Radeon RX 9060 XT, with slower GDDR6 memory (18 vs 20 Gbps) and reduced memory bandwidth (288 vs 320 GB/s), enabled by slower 18 Gbps GDDR6 modules. Performance should also be a downgrade from the XT version, as the RX 9060 non-XT features a boost clock about 140 MHz lower, as the XT version boost 3,130 MHz. Additionally, the card also requires a full PCIe 5.0 ×16 lanes.

The value proposition of this card remains unclear, as we are unsure how low AMD can price it to make it appear attractive. The Radeon RX 9060 XT carries an MSRP of $299 for the 8 GB version and $349 for the 16 GB version, suggesting that the $50 price difference between the two could also translate to the non-XT version, and we could even see a $250 price point for the Radeon RX 9060 non-XT. With a launch date expected in Q3 2025, we are left to wonder what value proposition this card will represent and whether AMD's RDNA 4 rollout will be complete. As always, until the card is launched, AMD still has time to change specifications and even the possibility of launching the card.
Source: VideoCardz
Add your own comment

34 Comments on AMD Radeon RX 9060 non-XT GPU Specs Surface: 2,048 Cores and Slower 8 GB GDDR6 Memory

#26
AlB80
jnv11We are not being "infantile maximalists". We are being realistic about today's games which are designed for the PlayStation 5 and were ported to the PC, so they routinely bust the 8 GB VRAM usage limit on video cards. 8GB was barely good enough when most games were ported from the PlayStation 4 to PC instead of the PlayStation 5 to PC with the PS5's increased unified RAM capacity.
Is it "being realistic" and use max texture quality with low settings?
Simple plan
1. Lag or stutter -> reduce texture quality
2. Low FPS -> reduce graphics settings (keep texture quality)
Why? Because it's a cheap graphics card. It doesn't pretend to handle max texture quality.
Posted on Reply
#27
BSim500
If I wanted to play the heaviest 2025 AAA's I'd probably buy at least a +12GB GPU too. At least 16GB VRAM for future proofing. Most people on this forum would. However, I think "enthusiasts" often get so wrapped up in their own "enthusiasm" they forget a VERY large chunk of the market (as in "many people who don't hang round tech forums all day long") don't play AAA games. And the gap between "what gets played the most" vs "what tech sites / Youtube hardware review influencers cherry-pick and constantly howl at the moon over" could not be more great. A lot of low-end gamers won't give a flying f*** how badly the latest unoptimised AAA turd runs on them if they simply aren't interested in playing them. Something that flies over the heads of "overly-enthusiastic enthusiasts" every time.

The 1:2:4 ratio also isn't new at all. Go back 10 years to the mid-2015's and we had 2GB GTX 1050 vs 4GB GTX 1050Ti's vs 3-6GB GTX1060 vs 8GB GTX 1080's in the same gen. Go back another 10 years to 2005 and we had 128MB vs 256MB vs 512MB GeForce 6xxx. In 1999 we had 8MB vs 16MB vs 32MB on Riva 128/TNT. Go back to the EGA to VGA transition era of Diamond, Tseng Labs, S3, Cirrus Logic, etc, brands and it was 256KB vs 512KB vs 1MB. The 1:2:4 ratio has always been there. "Today I learned that low-end GPU's will often have only 1/4 of the VRAM of flagship GPU's is in fact observed going back all the way to MS-DOS era, and that, no, the mere existence of low-end GPU's doesn't 'kill games'" vs the current over the top hyperbolic clickbait-fuelled cringe-fest doing the rounds. The main issue is the pricing. If low-end GPU's cost $150-$200 and mid-range still cost $250-$350, we wouldn't see half the pant-pissing over VRAM we do today, just as didn't see any similar absurd "river of tears" from GTX 1070 owners that 2GB GTX 1050's were "killing my games". It was self evident common sense back then that low-end GPU's were mainly for non AAA gamers. Nothing has changed today except price-inflation fuelled misplaced expectations and a desperation to continue to avoid admitting those prices only are what they are because "Real Gamers" continue to have about as much self-control as Steven Seagal at a burger stand in the higher end, which then drags up everything in the mid-range / low-end too. $300 RTX 3050's didn't set the price of RTX 3090's, it's the other way around.

The other real problem is the normalization of non-existent optimisation of AAA and reliance on "Just Upscale Dude" gaslighting, and that "Real Gamers (tm)" are consistently the most gullible f***s on the planet with who fall for it over & over. "Our barbed wire fence with transparent textures now look worse in 2024 on UE5 using 9GB VRAM" vs "How crisp & sharp same transparent textures on wire fences looked in 2004 on a 1GB VRAM GPU". "Our shiny new premium UE5 AAA looks like this with 9GB VRAM textures on Low" vs what 2015-era low-budget Indie walking simulators managed on CryEngine 4 in 4GB VRAM. The UE5 garbage churned out today is absolutely a software problem than a hardware problem, it's just become so prevalent that "Gamers" who play nothing but AAA's have stopped seeing it for what it is, and find it easier to cope with by declaring low-end GPU's are "holding me back" than admit the AAA gaming industry itself has turned into an intelligence-insulting enshitified train-wreck, overly centralized around using one game engine no-one wants to admit has 'evolved' into a broken mess. "But, but, but unless low-end GPU's come with 65,536GB VRAM, we might have an AAA gaming industry crash!" Bring it on. We sorely need one now more than ever...
Posted on Reply
#28
iameatingjam
BSim500If I wanted to play the heaviest 2025 AAA's I'd probably buy at least a +12GB GPU too. At least 16GB VRAM for future proofing. Most people on this forum would. However, I think "enthusiasts" often get so wrapped up in their own "enthusiasm" they forget a VERY large chunk of the market (as in "many people who don't hang round tech forums all day long") don't play AAA games. And the gap between "what gets played the most" vs "what tech sites / Youtube hardware review influencers cherry-pick and constantly howl at the moon over" could not be more great. A lot of low-end gamers won't give a flying f*** how badly the latest unoptimised AAA turd runs on them if they simply aren't interested in playing them. Something that flies over the heads of "overly-enthusiastic enthusiasts" every time.
You know, I actually don't disagree necessarily, but if thats what these cards are made for, I really feel like that should be made clear on the packaging, like if its for old or indie games, then IDK, put something like 5050 Light or maybe you could have a graphic on a back where it shows how well it performs in what class of games in a xx/5 type system, I don't know, I'm sure they could think of something. Because right now cards are only numbers, and people don't always know what those numbers mean, they only know the brand, the same demographic you are talking about, the non-enthusiast.

And its not like these are trivial amounts of money, this is more for say, 5060/tis, but I guess it could apply here, if you buy a $250-$450 card and it wont play a game released earlier this year without constant frame drops at 1440p or even 1080p, well you know what that means.... developers get punished. Not nvidia, because that would imply its the consumers fault, and as humans we don't like to blame ourselves for making the wrong choice, so its the games fault. I see this constantly constantly constantly on steam forums.
Posted on Reply
#29
Lew Zealand
Vayra86You enjoy buying 8GB cards at new-card money then, and I'll enjoy being the infantile maximalist that doesn't want to burn money like that, eh?

The resell value of this card is zero. Not $50,-, not $25,- in five to seven years time, no. Its $0,-...so....
Exaggeration and rhetoric are useless and boring. How about some data?

Currently the cheapest BIN GTX 1650 4GB on eBay sells for $76. A 4GB card 6 years later, still fetching $76. You will be selling an 8GB card 6 years from now for similar money.

LOL probably more thanks to inflation.
Posted on Reply
#30
jnv11
AlB80Is it "being realistic" and use max texture quality with low settings?
Simple plan
1. Lag or stutter -> reduce texture quality
2. Low FPS -> reduce graphics settings (keep texture quality)
Why? Because it's a cheap graphics card. It doesn't pretend to handle max texture quality.
Your plan fails to account for many causes of lag, stutter, low FPS, and high VRAM consumption. It seems to assume that lag and stutter are caused only by overflowing the VRAM, and low FPS is caused by high graphics settings. There are a lot more causes of those than what appear to be your assumptions.

Lag and stutter could be due to VRAM overflow due to large textures, but could also be caused by the game itself, the GPU architecture or driver, or inadequate hardware. Maybe the game uses effects that consume a lot of VRAM. Maybe the game chooses inappropriate times to do shader compilation.

Low FPS could be due to too high game settings, but could be due to other things. Nvidia and Intel do warp or workgroup-level instruction scheduling on the CPU in their drivers. This approach pays off on powerful CPUs by highly optimizing the software-scheduled code that is eventually sent to the GPU in ways that cannot be done by AMD’s instruction scheduling hardware because the time saved in the GPU is more than the time that a powerful CPU spent to optimize the code that is sent to the GPU, but heavily backfires on weak CPUs which spend more time in the driver to optimize the code versus what is saved in the GPU by having heavily optimized code. AMD uses instruction-scheduling hardware to offload that job from the CPU. However, we have learned that RDNA 3 and earlier AMD GPUs wasted lots of time on bad scheduling compared to RDNA 4. Nevertheless, this bad scheduling hardware in RDNA 3 and earlier Radeon GPUs still saved time when the GPU was paired with a weak CPU by cutting more CPU overhead than the time that was wasted in the GPU by the sub-optimally scheduled code. Also, overflowing the VRAM forces the GPU to swap data with system memory to get the job done, but this is far slower than direct VRAM access, slowing down the FPS. Also, some people have higher resolution monitors to boost productivity when they are not gaming, but that high resolution also increases VRAM consumption and required GPU performance.

There is also the frustration caused by both finding what low settings can work with such a GPU and not having fun due to having drab graphics at those low settings kill game enjoyment if they are bad enough, absolutely destroying the value of such video cards made obsolete by lack of adequate VRAM.
Posted on Reply
#31
Vayra86
Lew ZealandExaggeration and rhetoric are useless and boring. How about some data?

Currently the cheapest BIN GTX 1650 4GB on eBay sells for $76. A 4GB card 6 years later, still fetching $76. You will be selling an 8GB card 6 years from now for similar money.

LOL probably more thanks to inflation.
Would you buy that at 76 bucks? There are a lot of people asking a lot of things, doesnt mean they get it or that it has any merit.
Posted on Reply
#32
Lew Zealand
Vayra86Would you buy that at 76 bucks? There are a lot of people asking a lot of things, doesnt mean they get it or that it has any merit.
It doesn't matter if I buy it. It matters that many someones are buying them, and $76 is merely the cheapest for BIN as the price goes up from there. There's always a market for GPUs cheaper than the minimum for brand new, which is about $170 oof, $195! for an RTX 3050 6GB or ~$250 for a 5050/B570/maybe this 9060.
Posted on Reply
#33
iameatingjam
Lew ZealandIt doesn't matter if I buy it. It matters that many someones are buying them, and $76 is merely the cheapest for BIN as the price goes up from there. There's always a market for GPUs cheaper than the minimum for brand new, which is about $170 oof, $195! for an RTX 3050 6GB or ~$250 for a 5050/B570/maybe this 9060.
You cant always but its really common to find the b580 at msrp or $10 more now. I bought one, actually under msrp locally. And here's one on newegg:

www.newegg.com/asrock-challenger-b580-cl-12go-arc-b580-12gb-graphics-card-double-fans/p/N82E16814930131

Free shipping too. You likely don't have to go to b570 to get $250, especially if you have a little bit of patience. Sorry I know thats not the point of your post just fyi.
Posted on Reply
#34
AlB80
jnv11Your plan fails to account for many causes of lag, stutter, low FPS, and high VRAM consumption. It seems to assume that lag and stutter are caused only by overflowing the VRAM, and low FPS is caused by high graphics settings. There are a lot more causes of those than what appear to be your assumptions.
This plan has been working for many years. And, of course, here, in the "graphics cards" thread, I wrote only about graphics cards.
jnv11There is also the frustration caused by both finding what low settings can work with such a GPU and not having fun due to having drab graphics at those low settings kill game enjoyment if they are bad enough, absolutely destroying the value of such video cards made obsolete by lack of adequate VRAM.
Reducing the texture quality by one or two points does not significantly degrade the image.
Posted on Reply
Add your own comment
Nov 21st, 2026 01:01 PST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts