9060xt graphics card or 5060ti?
SolvedFlowOne31 -
Good evening,
I would like to change my old but surprising graphics card (1660 Super 6GB Tuf) for a slightly newer one.
A while ago, I had to update my setup thanks to Windows 11, like many others, but financially I couldn't keep up for the graphics card; it's a big investment these days.
Formerly a pro Nvidia / Intel user, I was rather disappointed with the direction Intel took with their latest CPUs, which in my opinion resemble more overclocked smartphone CPUs than real CPUs with raw power to exploit from back in the day (the "K" series, in particular), so I switched to an AMD 5900X (already old, I know, but it has power if you manage its temperature).
I initially wanted to get a 5060 Ti 16GB, but with all the release problems at the beginning of the year, it quickly cooled my enthusiasm for Nvidia as well. I don't want a card that burns from poorly managed hot spots or have missing ROPs supposedly due to design errors; it's been a few months since we stopped talking about it, but that doesn't mean defective stocks have been pulled from the market...
The AMD equivalent is the 9060 XT 16GB, which interests me, but the price is significantly higher than the 5060 Ti for slightly lower (on paper) performance, and above all, I don't have enough perspective on AMD GPUs (performance/reliability/drivers) to make a decision.
I can't invest in a more powerful GPU than these models, and with the mega-inflation of RAM prices right now, I think the prices of all cards will hit the roof in the coming weeks/months.
So here it is, I need to change my card now; my 1660 Super struggles to keep up with Unreal Engine 5 (not that demanding for rendering) and is not compatible with DLSS and RTX features (thankfully there's FSR for everyone), and honestly, playing on "all minimum" settings doesn't interest me; I might as well do something else.
The latest remakes from Capcom and Konami have rekindled my gamer side, and I want to dive back into these games in good conditions without constantly watching the PC temperatures or playing at 15 fps in 1080p.
What do you think, 9060 XT or 5060 Ti? Is there really a difference in performance? Is one more reliable than the other? A big scam for those on a budget?
My current setup:
Ryzen 9 5900X
32GB of RAM
Asus B-550 Plus TUF Gaming WiFi
GTX 1660 Super 6GB Asus TUF Gaming
Thermaltake 750W PSU (already 6 years old, any recommendations?)
Thank you in advance for your thoughtful opinions; it will greatly influence my final choice.
5 answers
-
flo88 Posted messages 28485 Registration date Status Contributor Last intervention Ambassadeur 5 168
Hey guys! After such a back-and-forth of arguments and counterarguments, what does the client end up choosing?? Lol....
I'm going with an Intel Arc B580....that way, everyone is happy. LMAO.
-
It's good, it cost me more than expected and will have to cost me quite a few overtime hours. But I came across a very good price for a RX 9070 XT OC in Asus Prime (€610) and a TUF Gaming power supply, so my current setup is full AMD/Asus; I hope to be good for a little while.
-
Hello
Right off the bat (in addition to wishing you happy reading), forget about Ray Tracing (RT) on these 2 references: the computing power is not sufficient to exploit it at 100%. At best it will be a bonus but shouldn't be considered as an argument. Both handle it but since we will have to throttle it anyway, let's not be offended by a "concession" on that side.
Also, an important point, the CPU is more than sufficient.
Back to the GPUs, regarding the software technologies, we are more or less equal. There is an image generation technology on both sides; I prefer to remind that it is only good in single-player: it generates latency (3 frames /!\) and it hallucinates (the same concept as in generative AI but "not the same") making it a handicap in multi-player. Having 144 frames per second with 3 frames generated by "prediction" is worse than having 60 frames per second where each frame is the exact current situation of the game. In single-player, it's "your preferences", you gain in "perceived fluidity" at the cost of artifacts.
As for performance and price, you've already said it and it's confirmed (at least for the performance (link)): "the 9060xt 16Go that interests me but the price is much higher than the 5060ti for slightly lower performance": go for the 5060TI.
Bonus, okay we are at 16Go of VRAM on both sides but 448 GB/sec memory on the 5060TI VS 320 GB/s on the 9060XT, I can only recommend the "green" one.
Note 5 things:
- Avoid models with the 12VHPWR connector (documented issue on Youtube by Der8auer and JaysTwoCents but also on Reddit). After all, with 180W max for nVidia and 140W max for AMD it’s not a problem but... principle.
- The math is reassuring. In the worst-case scenario, counting the 180W max of nVidia and removing the 75W provided by the PCIe port, it’s 8.75A, almost exactly the max allowed by this impossible case of the card that works with a connector so poorly designed that 100% of the current passes through a single wire, a wire designed to pull 8.33A continuous (standard at 600W over 6 wires) and there will necessarily be a bit of current distributed over the other 5.
- If you want to avoid any risk, the "good" connector is the 8 pins, often found in 2 x 8 pins in this range.
- I could have skipped this warning here but you never know, "a 5080 can come so quickly" and then it’s not the same lemonade with 360W.
- We are talking here about the "max power specified by the manufacturer". According to nVidia, my 3080TI FE (the most "pure") should stay under 350W but I sometimes measure it at 410W, +17% in my case.
- Nvidia's 5th Gen drivers are solid. AMD's 9xxx gen drivers are too. No worries on that front. Open to information but the bad feedback I saw concerned the 6xxx gen from AMD.
- "poorly managed hot-spots"
- I recommend you to repaste and repad, my card no longer exceeds 57° in air cooling with a case that is indeed excellently ventilated. You can find the pads on Amazon, Kryonaut everywhere and the procedure on other forums or Youtube.
- "missing rops"
- Not seen a single mention on the series below xx80. Open to information.
- "Thermaltake 750w PSU (6 years already, any recommendations?)"
- If it is certified 80 PLUS GOLD or better, it’s good. Bronze or Silver, plan to change, a very good PSU (≥ Gold) has a lifespan of 10 years. I just changed my PSU for this reason (Gold nearing 10 years). The new Cybenetics certification is stricter; you can trust it if available.
- In terms of power, 750W is still plenty, even with OC, considering current and planned specs. Beware, "a 5080 can come so quickly" and then it’s not the same lemonade with 360W ;-)
The MAMAA may not have oil but they have data!
Do you feel my Big Data?
Sacrifice some freedoms for more security and you lose them ALL.
ALL YOUR DATABASE ARE BELONG TO US-
I don't feel like you know these cards very well; we're no longer at the same stage of ray tracing and the flawed pseudo image generations. Unless you’re an engineer in the field, we can't dismiss these technologies as defective, especially for DLSS 4, which has been designed to avoid any downsides.
I'm very interested in the technical side of things, as raw computational power or memory bandwidth should really be taken with a grain of salt when we are able to consider other overlooked details.
A RTX 5050 has more transistors than an Xbox Series X, including both CPU and GPU parts, and I could also compare the architecture of Apple's ultra-powerful M5 for digital processing, which has only 5 teraflops on the GPU FP32 and limited bandwidth.
-
I maintain and I explain my point of view:
Ray Tracing is an excellent technology but very demanding; it requires computing power greater than what is offered by the two references mentioned. In their case, it’s a marketing argument. On a 5070TI or higher, I would not at all have said the same thing.
On the image generation side, it works too, it’s not "shoddy". At worst, we note slight artifacts explaining why it remains relevant in solo mode. In multiplayer, no, for obvious reasons: image generation cannot anticipate the future movement of other players; that’s the whole point of calculating each image! We can temper if the multiplayer is "slow"; adding latency has no impact, but in simulations, shooting, and any other type of fast-paced game, it’s a hindrance.
"A rtx 5050 has more transistors than an Xbox Series X including both CPU and GPU, and I could also compare the architecture of the M5 from Apple, ultra powerful for digital processing with only 5 teraflops on the GPU fp32 and bandwidth that is lacking."
Here it’s comparing apples and oranges; it’s not at all the same software running on a 5050, an Xbox X, or an M5. So not at all the same hardware resources required.
Oh, and I reviewed a few points from my first post before tackling this one, proofreading and all that. I’m done with him, just in case you’ve seen changes since your first read ;-)
-
But the problem is there, you make too many unfounded assumptions, we don't need computing power with the latest generations, we are not on a console without dedicated parts.
The RT cores and Tensor cores are not the same, we can push everything to the max at 1080p 144hz starting from an RTX 5050 for the games that take advantage of it.
We are no longer in the RTX 2xxx or RX 6xxx generation.
The ghosting problems caused by AI are only present with DLSS3 and PS5 Pro's PSSR.
And if they are the same games, so if we can compare, however yes for the M5 it’s not the same usage.
-
"AI Ghosting problems" cannot be eliminated; it's impossible.
Let's play a quick game.
The game normally generates 60 fps fairly cleanly, but you can feel that having 120 would be smoother. The game manages DLSS, we activate it, let's say it pushes to 144. Under the hood, it had to guess 2 extra frames. That's computation time, certainly much less than a full generation, but instead of refreshing the players' states 60 times per second, it only did so 48 times (144/3). Here, we even lose information!
Let's not even take that into account. Let's assume we got 180 FPS with DLSS (not possible since you have to allocate computational resources to the divination that was initially 100% allocated to the calculation, but let's say) and therefore our 60 refreshes per second. As a player, we consider we have our 180 FPS and we play with that in mind. Each time a frame is calculated, the next 2 displayed frames will be guessed frames.
There will be a delay between the 3rd displayed frame (2nd guessed) and the 4th displayed frame (2nd calculated). And this happens every time.
The persistence of vision smooths it out and we don't feel the difference. At worst, we notice players' positions or features that shouldn't be in certain places, etc., but the artifacts are present. But above all, we are still playing at 60 FPS in terms of state refresh, even though the GPU is pushing 180 FPS on the screen!
Since the beginning of this technology and regardless of its advances, it has always been debated for this reason. If calculating a frame is heavy, well, that's how it is, but each frame contains exactly the information it is supposed to have. As soon as you introduce guessing into the equation, no matter how close it may be, it is not reality.
Bonus, fresh news (link). Spoiler, there are still artifacts.
As for the RT cores, it's x2 between a 5070TI and a 5060TI (link).
-
Okay, the Nvidia engineer... You're just speculating, you don't know how technologies work, how this is made possible.
<<
1. Dynamic Multi-Frame Generation (DMFG)
This is the biggest advance for latency. Unlike DLSS 3, which was "fixed" (you either activated or deactivated frame doubling), DLSS 4.5 dynamically adjusts the frame multiplier (up to 6x on the RTX 50-series).
-
Why it's relevant: The system automatically "downgrades" the multiplier if the scene becomes too complex, thereby avoiding the latency spikes we used to experience. This allows for a consistent response time even as the action intensifies.
-
Impact: We can reach peaks like 240 Hz or 360 Hz with input latency remaining close to native mode + Reflex, as the technology generates up to 5 frames for one real image with reduced processing costs on the Blackwell architecture.
2. The Transformer v2 Model (Super Resolution)
DLSS 4.5 uses a new AI model (Presets K, L, M) that processes motion vectors more intelligently.
-
Temporal Stability: By reducing "ghosting" (trails behind objects) and flickering, the brain perceives a clearer image in motion. A more stable image aids in click precision, improving perceived latency (the feeling of control).
-
The downside (RTX 20/30 cards): This new model is very resource-intensive. On older generations, activating the Transformer v2 model can paradoxically increase latency and drop FPS by up to 20% if using the wrong presets (M or L forced), as these cards lack the FP8 accelerated units of the 40 and 50 series.
>>
-
- Avoid models with the 12VHPWR connector (documented issue on Youtube by Der8auer and JaysTwoCents but also on Reddit). After all, with 180W max for nVidia and 140W max for AMD it’s not a problem but... principle.
-
Thanks for the responses, my question turned out to be not as simple as expected given the debate it sparked.
"poorly managed hot-spots"
I recommend repasting and repadding, my card no longer exceeds 57° with air cooling in a case that is admittedly excellently ventilated. You can find the pads on Amazon, the Kryonaut everywhere, and the procedure on other forums or YouTube.
_That's precisely the main reason that makes me reconsider NVIDIA; I know how to do a complete and efficient cooling setup for a GPU (even a full PC). However, I usually do this when the card is at the end of its life (technologically as it always runs at 100%++) or when the control over temperature curves doesn't really change anything, not when buying a new and guaranteed card.
"having missing rops"
I haven't seen a single mention of the series lower than the xx80. Open to information.
_In fact, shortly after, this issue was confirmed on the 5090/5080/5070, so I doubt that the cheaper and more profitable cards are really spared.
Regarding power supplies, I don't pay too much attention to gold/silver/bronze ratings anymore, because in my experience, I've built 2 setups with a low-end "advance" power supply costing 40 bucks that lasted 12 years without issues or failure; I changed it out of caution for a gold one from Corsair (close to 200 euros), which lasted me 3 years, and when it failed, it took my motherboard and graphics card with it... My current "silver" Thermaltake has already proven itself since, in 6 years it's twice as long as the Corsair gold, I only had to replace the fan that failed 2 years ago (thankfully it made a hell of a noise before stopping...). I plan to get the same one again if available, we'll see.
So, yes, I still don't have a real decisive opinion between the two brands, but with all the inconveniences encountered with the 5000 series, I'm starting to lean towards AMD. I wasn't disappointed with my first AMD CPU (current CPU 5900x) and I think they are as experienced as NVIDIA given how long they've been in place. The main problem is that I don't really know what the actual value is; my only experience with an AMD GPU dates back years and it was hell with the drivers compared to GeForce. I started gaming on an 8800GT until my current 1660s, which is also 6 years old, so I only really know NVIDIA.
I would like more opinions on AMD GPUs, especially the 9000 series, or to know if the problems with NVIDIA 5000 have been "resolved".
There are many articles talking about it, I know, but I prefer the opinion of users like you on this forum. A test in a publication or product announcements can't replace real-time use over years of usage.
-
Thanks for the feedback ;-)
- The repa(d/ste)
It's preventive, let's say. Going through it made me gain some degrees but it was very good to begin with. For this point, it really depends on the integrator and the reviews done on it (link). A Founders Edition is fine anyway. A Sapphire at AMD is also good.
I know of 2 Nvidia 5xxx in 5070TI, no reported issues, they're very happy. My card comes from one of them who upgraded.
- ROPs
I just saw that it also affects the 5070ti. I'll mention it to the guys. The 5060TI is not mentioned at all.
- PSU
It's just the two of us, I've also known a faulty PSU. The motherboard went into safety mode but my 6 HDDs+SSDs literally burned. A time when I wasn't regularly updating my cold storage; that has changed.
Since then, I recommend 80+ Gold as a minimum to simplify searches without making someone invest in Platinum or Titanium. Quite a few non-certified PSUs last 20 years, there's no miracle recipe apart from reviews. You're aware, no need for me to go further ;-) In such a forum where we don't know the technical background of the interlocutor (and believe me, very few askers have one), we have to take shortcuts.
- AMD or Nvidia
No personal preference. Before going for the 3080TI, I almost bought a 6950XT; the deal was "too" good. Nowadays, it's only a pure performance/price comparison that makes me choose according to my game catalogue. If AMD is cheaper, go for it, otherwise a green one.
As for feedback on AMD 9xxx or Nvidia 5xxx, it's better to browse forums; here you might only get 3 experiences (if any pop up) and I won't be part of them --'
- Problems
AMD drivers have largely evolved for the better. I still had 2 GPUs in my tower not long ago, the second AMD light to relieve my main screen on an Nvidia. The 2 drivers cohabited without issue, the AMD had no problems. You'll always see some issues popping up on one side or the other but it's not "widespread".
The problems of the 5xxx series seem to less concern the 5060TI due to its range, its release date, and the powers required.
- Shortage
A piece of advice, choose quickly. GPUs will be affected by the RAM chip crisis and shortages are expected in new ones, which will in turn drive up the prices in the second-hand market. Today I could still make a profit with my card! If you can sacrifice some technologies reserved for the latest series, take a look at the previous models. The gap isn't that big with the 4000 series (link), depending on availability and prices, it could be worth it to go for "bigger older", red or green.
-
-
Yes, I'm getting to it now because of the current inflation that is going to drive up the prices of all components requiring RAM. It's already starting, a few models that I was tracking the price of have gone up by up to 50 euros according to the online retailers compared to last month. So, with that and the future arrival of Windows 12 which will change the "standard" in hardware with its permanent AI, requiring new CPUs to fully enjoy the OS, well, it's still speculation, but I think this is the moment to build my last setup even if it's far from optimal.
By the way, I reviewed your recommendations from previous messages regarding the range of graphics cards I'm interested in, price/power ratio, and I've decided to step up a notch. It's a bigger investment but with more longevity and much more flexibility in gaming (hoping to avoid unforeseen technical defects).
Yes, I often check the prices of the RTX 4000 series and the same goes for AMD, which have very good reputations and have proven themselves. Besides, they're starting to be more expensive new than the current models on a lot of sites. I feel like this latest generation of cards isn't really groundbreaking in terms of raw power and technical evolution; anyway, that's just my personal opinion.
AMD or Nvidia
I'll do like you and hunt for what I can find without preconceived notions between AMD and Nvidia (not easy when you're stubborn), to find the best deal. The brand will also play a role in my final decision because the technical issues mentioned earlier seem quite random from one design to another. In any case, the problem will be resolved this weekend :)
Problems with AMD drivers
Glad to see that these problems are resolved; my only experience with an AMD GPU was catastrophic with the drivers (my brother's PC), difficulties installing without errors, bugs with every driver update, crashes when opening the config menu... argh, a nightmare.
You're aware, no need for me to go further ;-)
Yes, I assemble, repair, troubleshoot, and disassemble hardware out of passion for quite a few years now, mine or others' (friends, work, family...) but all this is thanks to you on this forum. I assembled my first PC all by myself with the help and recommendations I found on CCM, a revolution at the time, the first dual-core lol. Even today, I still learn a lot and I love it :)
I also recognized quite a few usernames still active on the site helping others, including yours. Honestly, well done everyone, still as passionate after all these years; that deserves recognition, I think :)
All that's left is for me to find my gem; for better or worse, it all comes down to this weekend.
Thank you for your help and advice, it has really helped me with my GPU dilemma; the hardest part is still to come, getting to the checkout lol.
Have a good evening and a happy new year.
