Sunday, September 20th 2026

AMD Radeon RX 9050 4 GB Benchmarks Surface, 8 GB Model 37% Faster

Hardware Unboxed and Toasty Bros have both tested the mysterious Radeon RX 9050 4 GB, and the results look rough. AMD only sells this version in OEM systems, so both channels had to buy CyberPowerPC prebuilts to get one. In Hardware Unboxed's 18-game test at 1080p Low, the 4 GB card averaged 78 FPS with 55 FPS 1% lows, against 107 FPS and 78 FPS for the RX 9050 8 GB. That makes the 8 GB model about 37% faster. At the lowest settings tested, it was still around 32% ahead, at 147 FPS versus 111 FPS. Both cards share 1,024 Stream Processors and up to a 2.6 GHz boost clock, so the difference comes from memory. As AMD confirmed, the 4 GB model has a 64-bit bus, 144 GB/s of bandwidth, and 16 MB of Infinity Cache, exactly half of what the 8 GB model has.

Toasty Bros opened their prebuilt and found an XFX dual-fan card with a single 8-pin connector, so it looks like ASRock is not the only partner making one. AMD has said it developed the Radeon RX 9050 4 GB version for a specific partner system aimed at entry-level gaming customers, and that it has no plans to sell it outside that OEM PC. The card was first spotted in a CyberPowerPC Gamer Master prebuilt at Best Buy, which was listed at $889.99. Keep in mind that this is still a small sample, with only two outlets testing the card so far, so it's important to wait for more independent results before drawing firm conclusions. Still, AMD launching a card with only 4 GB of VRAM in 2026 is not looking like a great decision so far.
Sources: Hardware Unboxed, Toasty Bros, via Videocardz
Add your own comment

88 Comments on AMD Radeon RX 9050 4 GB Benchmarks Surface, 8 GB Model 37% Faster

#76
dgianstefani
TPU Proofreader
john_Well, sure, okay. So when Nvidia released the MX550 alongside the MX570 on the very same day, presenting it as part of the new architecture when it was actually a card based on an older architecture and should have been named MX470, do you think they didn't do it because they know how business works with OEMs? Do you think they just wanted to fool the public? I believe Nvidia did it because they know the demands of a customer who isn't buying a single card off the shelf, but a million cards at once. So when Nvidia itself plays this exact game, you go right ahead and keep hyper-focusing on AMD while pretending you live in some parallel universe.
It's a literal red herring what NV did or did not do when the name is different.

Those two products you keep bringing up sit at two performance levels and two easy to understand naming "tiers", the bigger "70" one is faster than the "50" and presumably more expensive. While mildly misleading they launched together, it's not the same thing as what's happening here. I suspect you know that.

Do you take issue with the RX580 being an RX480? Two different architectures for the two different names? No, the same one? Is that also relevant to this thread? No.

As I've said multiple times (as you've kindly pointed out), everyone, including me, rightfully bashed NVIDIA when they tried the 4080 12 GB. You implying bias is predictable, but the topic of this thread, is this card. Not a different one.
Posted on Reply
#77
Geofrancis
At this point I wouldnt be surprised if nvidia or amd bring back the "turbo cache memory" where it just allocates 6Gb of system memory to a 2Gb card and just calls it self an 8Gb GPU.
Posted on Reply
#78
dgianstefani
TPU Proofreader
GeofrancisAt this point I wouldnt be surprised if nvidia or amd bring back the "turbo cache memory" where it just allocates 6Gb of system memory to a 2Gb card and just calls it self an 8Gb GPU.
The old Apple "our 8 GB RAM is the same as Windows 16 GB".
Posted on Reply
#79
john_
dgianstefaniIt's a literal red herring what NV did or did not do when the name is different.
No, it simply bothers you when someone points out the obvious. That these two companies operate in the same market, offering products that meet the same needs, dealing with the same customers, and to an extent using similar commercial practices. I'm sorry if reality bothers you.
dgianstefaniThose two products you keep bringing up sit at two performance levels and two easy to understand naming "tiers", the bigger "70" one is faster than the "50" and presumably more expensive. While mildly misleading they launched together, it's not the same thing as what's happening here. I suspect you know that.
Alright, seriously now? No, honestly, is that an argument? I don't know whether I should comment on it or just facepalm. "Slightly misleading"? It was extremely misleading. The fact that they were announced on the same day and belonged to the same series is the absolute definition of deception. And you know what? AMD does it too with their laptop processor naming schemes. Look, the very fact that you're throwing out this—let me not characterize it—shows that right now you don't actually want to have a dialogue. You want to say something, you feel the need to push your own virtual reality, but you're not having a dialogue right now.
dgianstefaniDo you take issue with the RX580 being an RX480? Two different architectures for the two different names? No, the same one? Is that also relevant to this thread? No.
It's a clear refresh with higher clock speeds and some optimizations. It could have easily been released as an RX 485.
Where exactly is your objection? Don't make the mistake of thinking I'm sitting here clearly wasting my time with you because I want to hide something. That's not my mindset. I'm not going to point at a table and insist it's a chair—which is what you've been doing for the last few posts, culminating in those excuses for the MX550.
dgianstefaniAs I've said multiple times (as you've kindly pointed out), everyone, including me, rightfully bashed NVIDIA when they tried the 4080 12 GB. You implying bias is predictable, but the topic of this thread, is this card. Not a different one.
Well, yeah. And when the thread is about some controversial Nvidia policy and I comment there, the response I'll get will be "What are you doing here? You don't even prefer Nvidia hardware."

And people were right to react over the 4080 12GB, because dozens of posts have been written and dozens of charts posted showing how Nvidia tries to present smaller GPU dies in each new series as higher-tier models. People protected their own interests with that reaction, because if they hadn't, the RTX 5070 Ti would have been an RTX 5080 at $999, and the RTX 5080 would have been an RTX 5080 Ti at $1400. Aren't we doing the exact same thing here? Aren't we saying how wrong it is for a model cut down beyond just VRAM to share the same name? Don't we ALL agree that it should at least be called the RX 9040? The question is why you demand that people who agree with you step out into the middle of the public square and beg for forgiveness. Why do you focus on the brand name behind the RX 9040 instead of admitting that everyone does this, and that there should be—just like in the case of the 4080 12GB—a widespread pushback? Especially from tech sites and journalists. Because the two of us arguing here right now achieves nothing other than wasting our time.

And with that, I'm wrapping this up, because as I told you, I realize you're not interested in a dialogue right now. Maybe some other time.
Good evening.
Posted on Reply
#80
dgianstefani
TPU Proofreader
john_No, it simply bothers you when someone points out the obvious. That these two companies operate in the same market, offering products that meet the same needs, dealing with the same customers, and to an extent using similar commercial practices. I'm sorry if reality bothers you.
Nice projection. Always enlightening to have how I think about things explained to me.
john_Alright, seriously now? No, honestly, is that an argument? I don't know whether I should comment on it or just facepalm. "Slightly misleading"? It was extremely misleading. The fact that they were announced on the same day and belonged to the same series is the absolute definition of deception. And you know what? AMD does it too with their laptop processor naming schemes. Look, the very fact that you're throwing out this—let me not characterize it—shows that right now you don't actually want to have a dialogue. You want to say something, you feel the need to push your own virtual reality, but you're not having a dialogue right now.
You don't like the dialogue ≠ no dialogue. I'm simply presenting cases where the rules you apply in one context suddenly don't matter in another. It's predictably inconsistent.
john_It's a clear refresh with higher clock speeds and some optimizations. It could have easily been released as an RX 485.
Where exactly is your objection?
It wasn't released as a 485 now though, was it? Did you bring that up at the time? No, but you'll excuse it while criticising similar behaviour that's equally irrelevant to the thread in the same sentence.
john_Don't make the mistake of thinking I'm sitting here clearly wasting my time with you because I want to hide something. That's not my mindset. I'm not going to point at a table and insist it's a chair—which is what you've been doing for the last few posts, culminating in those excuses for the MX550.
Sure, if that's all you're able to take from it, go ahead. It's clear for other readers, which is the objective.
john_Well, yeah. And when the thread is about some controversial Nvidia policy and I comment there, the response I'll get will be "What are you doing here? You don't even prefer Nvidia hardware."
Not from me, so don't play the game where you talk about responses you get from others in a direct response to me.
john_And people were right to react over the 4080 12GB, because dozens of posts have been written and dozens of charts posted showing how Nvidia tries to present smaller GPU dies in each new series as higher-tier models. People protected their own interests with that reaction, because if they hadn't, the RTX 5070 Ti would have been an RTX 5080 at $999, and the RTX 5080 would have been an RTX 5080 Ti at $1400. Aren't we doing the exact same thing here?
We're trying to. But people keep being defensive by muddying the waters, often by incomplete/misrepresentation of easily verifiable stats, or conflation with irrelevant comparisons, seemingly with intent on trying to normalise something inexcusable.
john_Aren't we saying how wrong it is for a model cut down beyond just VRAM to share the same name? Don't we ALL agree that it should at least be called the RX 9040? The question is why you demand that people who agree with you step out into the middle of the public square and beg for forgiveness. Why do you focus on the brand name behind the RX 9040 instead of admitting that everyone does this,
There's that conflation, mixed with some projection. I'm making no personal attacks. This is a public thread on a public forum, why be defensive?
john_ and that there should be—just like in the case of the 4080 12GB—a widespread pushback? Especially from tech sites and journalists. Because the two of us arguing here right now achieves nothing other than wasting our time.
I don't think trying to maintain a big, as complete as possible picture that's detail orientated when criticising or debating a topic, especially one that continues into extended discussion, is a waste of time. Like I said, normalising cancerous corporate behaviour by saying "everyone does this, it's not unusual", downplaying either intentionally or unintentionally, doesn't help things. I do find it interesting that people get so caught up on others pointing out specific details that are extreme examples of bs, but selectively, depending on the topic.
john_And with that, I'm wrapping this up, because as I told you, I realize you're not interested in a dialogue right now. Maybe some other time. Good evening.
Being polite doesn't change the messaging, there's no moral high ground to be found in a technical discussion, nor should egos or projection be present in the same sentence for that politeness to be genuine. If possible. That's what I aim for anyway.
Posted on Reply
#81
john_
dgianstefaniNice projection.

You don't like the dialogue ≠ no dialogue. I'm simply presenting cases where the rules you apply in one context suddenly don't matter in another. It's predictably inconsistent.

It wasn't released as a 485 now though, was it? Did you bring that up at the time? No, but you'll excuse it.

Sure, if that's all you're able to take from it, go ahead. It's clear for other readers, which is the objective.

Not from me, so don't play the game where you talk about responses you get from others in a direct response to me.

We're trying to. But people keep being defensive by muddying the waters, often by incomplete/misrepresentation of easily verifiable stats, or conflation with irrelevant comparisons, trying to normalise something abnormal.

There's that conflation, mixed with some projection. I'm making no personal attacks. This is a public thread on a public forum, why are you being defensive?

I don't feel like trying to maintain a big picture that's detail orientated when criticising or debating a topic is a waste of time. Like I said, normalising cancerous corporate behaviour by saying "everyone does this, it's not unusual", downplaying either intentionally or unintentionally, doesn't help things. I do find it interesting that people get so caught up on others pointing out specific details that are extreme examples of bs, but only for certain companies.

Being polite doesn't change the messaging, there's no moral high ground to be found in a technical discussion, nor should egos or projection be present. If possible. That's what I aim for anyway.
Posted on Reply
#82
Rulumi
GeofrancisAt this point I wouldnt be surprised if nvidia or amd bring back the "turbo cache memory" where it just allocates 6Gb of system memory to a 2Gb card and just calls it self an 8Gb GPU.
Now that you mention that, when I think well funnily that's kind of what some companies and sellers do with phones and tablets now with their advertising of "4+4 for 8GB memory" and similar nowdays, specially in the lower ends.

I would like to say it isn't that bad of an issue, but I know some people that said they were kind of confused by all of that and only noticed the actual RAM amount after they already purchased the phone.
Posted on Reply
#83
Visible Noise
john_And with that, I'm wrapping this up
Your word is so golden you lasted 45 minutes.
Posted on Reply
#84
wolf
Better Than Native
If they had even kept the 8GB to under 75w ,it could have claimed fastest slot powered card, perhaps helping to mitigate this PR fiasco with the 4GB version. Then you name the 4GB the 9040 and keep it to 75w also, and it could have ridden the coat tails of the bigger brother with effectively no controversy, and now you have two options that satisfy that market and attract more positive PR, the 4GB ~equalling the 3050 6GB, and the 8GB destroying it, and a new crown in that particular segment.

It's part of how the 3050 6GB "got away with it", which it still didn't, it is also not named correctly as it doesn't offer 3050 tier performance, and was rightly criticised at the time as it should have been the 3040, but they were able to tack a win onto it by having it launch as the fastest slot powered card. I can't really find a redeeming quality or win for the 4GB 9050 to claim here.
Posted on Reply
#85
epicsithtrooper
Remember when we were complaining about the price to performance difference between the B580 and B570, we should have realized how great it was that there was a sub 250 dollar card with more than 8GB of Vram.
Posted on Reply
#86
john_
epicsithtrooperRemember when we were complaining about the price to performance difference between the B580 and B570, we should have realized how great it was that there was a sub 250 dollar card with more than 8GB of Vram.
RX 570 8GB was selling under $150 at a period because people couldn't stop buying the sticker on the 6GB even the 3GB GTX 1060.
Posted on Reply
#88
AleXXX666
GeofrancisAll they had to do is call it the 9030 or 9040 and there would be zero issues.
yep, could make 9050 6GB at least.....:sleep::laugh:
john_RX 570 8GB was selling under $150 at a period because people couldn't stop buying the sticker on the 6GB even the 3GB GTX 1060.
but 1060 was still far better lmfao:D:oops:
Posted on Reply
Add your own comment
Nov 21st, 2026 02:39 PST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts