• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

RTX 5060 8GB performance

Status
Not open for further replies.
True, but the 4060Ti wasn't exactly great either.
And given all of the malicious attitude Nvidia is having towards reviewers with the 5060, it would be a do not buy if I were looking for a midrange card. I'd rather buy used or wait for the 9060XT.
Don't misunderstand me, I'd very much like the current 60 to match the former 70. And to that extent I truly despised both Ada 60's.
But at least the 5060 is a marginally better product in this comparative scenario versus its older sibling as it at least matches the former half-a-tier-up card, which the 4060 absolutely failed to do.
 
Ngreedia - turns out the 5060 is best perf / $ gpu available.

View attachment 400459
LOL, that's hilarious - how is it that the 5060 is close to MSRP in the US? I thought they were struggling through a tariff shitstorm being disguised as a scalpocalypse....
 
I don't remember prior generations of Nvidia GPUs being this troubled and problematic
Early G80 batches suffered from die bump-out (or how do you call it in proper English, I only know it as «отвал чипа» in my mother tongue). Blackwell might be horrible in many ways but at least the GPUs themselves will outlive their usefulness, not the other way around (these G80 cards only lasted for a year or two).
 
I think it’s the shortcoming of 8GB on Blackwell. I believed w1z mentioned that the software stuff that helps you get better FPS (franegen) consumes up to 1GB of VRAM, which, especially at 4K, results in a slideshow. The 5060 (Ti) GPU appears to be fast enough to do more, but it’s VRAM malnourished at 8GB.
 
MSRP means nothing, cheapest B580 is currently $350 in the US. https://pcpartpicker.com/products/video-card/#sort=price&c=585
The 5060 probably won't hit the claimed $299 MSRP either, unless Nvidia is doing some shady tactics to get cards on shelves at MSRP.
I'd trade 5-6% of performance I wouldn't notice unless constantly looking at an FPS counter to get extra VRAM. 8GB just isn't enough on a GPU over $150.
I think people often forget that tech enthusiasts who frequent TPU forums and watch YouTubers are a very, very small portion of the market. The only thing that matters to the ordinary consumer is performance, features, and price.
Most of the market will probably buy it anyway in a pre-built gaming PC or OEM system. The 5060 doesn't have performance, or price, it's too expensive for what it offers over previous gen, the drivers are terrible, and 8GB of VRAM is the worst kind of stagnation as some games need more currently, value minded buyers aren't replacing their GPU every 2 years either.
If you need CUDA so badly for work purposes then you're probably not shopping in the $300 range.
 
I think people often forget that tech enthusiasts who frequent TPU forums and watch YouTubers are a very, very small portion of the market. The only thing that matters to the ordinary consumer is performance, features, and price.
I'd argue the ordinary consumer is concerned with marketing and price a lot more than performance and features. They don't know or understand the features.
 
Don't misunderstand me, I'd very much like the current 60 to match the former 70. And to that extent I truly despised both Ada 60's.
But at least the 5060 is a marginally better product in this comparative scenario versus its older sibling as it at least matches the former half-a-tier-up card, which the 4060 absolutely failed to do.
I think the core issue is Nvidia has identified the "gaming" market is actually split into at least two groups, with very different market behavior. On one hand, there's casual gamers, who buy prebuilt desktops and laptops and use 1080p gaming monitors. These are your COD, Overwatch, Valorant, CS2, Marvel Rivals, etc, or lower-end indie friendslop games. These games don't have high system requirements, so maintaining almost the same performance across this lowest tier over years is the smart choice for Nvidia. Maximize your silicon usage by shrinking the GPU instead of bumping up the performance gen-on-gen, get better power efficiency for laptops, and pump out millions of these GPUs for these casual gamers. When they occasionally want to play a more demanding game like Dark Ages, DLSS and MFG gives them a playable experience. 8GB of VRAM is more than enough for all these casual games.

The second market is the much smaller group of enthusiasts, who have higher resolution monitors, are less accepting of DLSS and frame gen, and play more demanding, cinematic games and use their computers for other things like AI tools or professional software. For them, they get the higher-end GPUs with more memory and much better performance.

TLDR: The 5060 isn't meant for tech enthusiasts like us.

I'd argue the ordinary consumer is concerned with marketing and price a lot more than performance and features. They don't know or understand the features.
The marketing advertises the performance and features...
 
If performance per dollar was the only metric to judge a GPU by, then the GT 1030 would have been the king of all GPUs for a long time, maybe even today, thanks to its $80 MSRP.
No, the gt1030 at 80$ has terrible performance / dollar actually. You realize the 5060 is more than 10 times faster, right?

If the MSRP holds is really not a bad card - could have been way better at $345 with 12gb
A lot of stuff could have been better but the fact is not only does nvidia have the best perf / $ card, it also has the cheapest > 8gb vram cards (alongside intel). So you need vram for cheap, you buy nvidia, you want perf / $, you buy nvidia, you care about the feature set, or RT performance, you get nvidia. It's sad, what the hell is everyone else doing is beyond me.

LOL, that's hilarious - how is it that the 5060 is close to MSRP in the US? I thought they were struggling through a tariff shitstorm being disguised as a scalpocalypse....
Well, it's nvidia, contrary to their competition they don't engage in that "fake msrp promo code" BS everyone else is doing. You can find a 5060, right now, available in stock for 319$. So the performance / $ is even better than in that graph.
 
The marketing advertises the performance and features...
The performance and features that ordinary people can't understand- especially due to the obfuscated marketing.

1747768347904.webp
 
The performance and features that ordinary people can't understand- especially due to the obfuscated marketing.

View attachment 400482
If the the 5070 has 4090 performance, then the 5060 should have 4080 performance.
That intentionally misleading marketing is the kind of BS Nvidia should be sued for, especially since there isn't any asterisk or disclaimer on the slide.
 
Early G80 batches suffered from die bump-out (or how do you call it in proper English, I only know it as «отвал чипа» in my mother tongue). Blackwell might be horrible in many ways but at least the GPUs themselves will outlive their usefulness, not the other way around (these G80 cards only lasted for a year or two).
LOL, I was unaware of that and I wasn't an internal SI at the time so I only built a few dozen PCs a year back then.

My own 8800 GTS (320) died after 18 months, it was replaced with a GTS512 under warranty and that apparently died not long after I gave it to a friend at the grand old age of 3.

Well, it's nvidia, contrary to their competition they don't engage in that "fake msrp promo code" BS everyone else is doing. You can find a 5060, right now, available in stock for 319$. So the performance / $ is even better than in that graph.
If only that applied to the good cards with enough VRAM which are all still commanding large price premiums above their MSRPs. That's true of everyone - Nvidia 5070Ti and up, RDNA4, and even the B580 has insane markup - oddly it's the one the market needs to be cheap most of all, and it's the one with the largest percentage of scalper "tariff" markup right now!

It's also perhaps that even the uninformed punters know enough after 2+ solid years of the media warning against 8GB cards costing over $250, so maybe there's low demand to prevent the supply being overwhelmed (and then scalped).
 
Last edited:
The performance and features that ordinary people can't understand- especially due to the obfuscated marketing.

View attachment 400482

You can't complain about how much Nvidia is pushing AI and fake frames in one breath, and then play dumb and think that when he said DLSS 4 gives the 5070 4090 performance you thought you were actually getting 4090 raster performance on a $600 card on the same node. Either Nvidia is pushing AI or it isn't.
 
No, the gt1030 at 80$ has terrible performance / dollar actually. You realize the 5060 is more than 10 times faster, right?
It is now, yes. I was talking about the past.

My point is that performance per price is a nice metric for cards that are within an acceptable performance range to begin with. How true that is for an 8 GB card anno 2025, I'll leave it up to anyone's judgement.

As for me, like I stated in another thread, if I dug out an old 8 GB card from the closet, I'd see no problem using it in a secondary / indie gaming rig, but I'd never spend 300 for one, regardless of its price to performance data.

I've been eyeing a low profile 4060 for my HTPC for a long time, but I'm not pulling the trigger exactly because of its horrible price.
 
If the the 5070 has 4090 performance, then the 5060 should have 4080 performance.
That intentionally misleading marketing is the kind of BS Nvidia should be sued for, especially since there isn't any asterisk or disclaimer on the slide.
DLSS 4 was better than advertised, actually. The 5070 gets more than twice the 4090 performance with DLSS 4. e.g. Hogwarts Legacy, 1440p ray tracing. 4090 = 85fps, 5070 x4 framegen = 175fps.

Gonna wait for TPU's review to confirm, but yeah, the 5060 will probably have better than 4080 performance with MFG.

As for being misleading, the entire presentation was about AI. If you actually thought you were getting 4090 raster performance on a $550 card on the exact same node, you're too stupid to be allowed on the internet. He said multiple times that MFG gave the 5070 4090 performance.
 
DLSS 4 was better than advertised, actually. The 5070 gets more than twice the 4090 performance with DLSS 4. e.g. Hogwarts Legacy, 1440p ray tracing. 4090 = 85fps, 5070 x4 framegen = 175fps.

Gonna wait for TPU's review to confirm, but yeah, the 5060 will probably have better than 4080 performance with MFG.

As for being misleading, the entire presentation was about AI. If you actually thought you were getting 4090 raster performance on a $550 card on the exact same node, you're too stupid to be allowed on the internet. He said multiple times that MFG gave the 5070 4090 performance.
Yet none of the made up FPS matters because 4X frame gen won't feel like 4X the FPS, unless the frame rate is already high enough to not need multi frame gen.
As for the marketing, the average consumer won't know the difference or care it was raster or fake frames, if you want to call me stupid because the marketing slide is obviously a misleading lie, then cool good for you.
 
Yet none of the made up FPS matters because 4X frame gen won't feel like 4X the FPS, unless the frame rate is already high enough to not need multi frame gen.
As for the marketing, the average consumer won't know the difference or care it was raster or fake frames, if you want to call me stupid because the marketing slide is obviously a misleading lie, then cool good for you.
Exactly. If you can make 60 FPS with the feel of 60 FPS out of 20, sign me up. Otherwise, it's marketing bullshit, and e-peen.
 
It's also perhaps that even the uninformed punters know enough after 2+ solid years of the media warning against 8GB cards costing over $250, so maybe there's low demand to prevent the supply being overwhelmed (and then scalped).
The 4060 mobile 8GB just became the most popular GPU. 6/10 of the most popular GPUs have 8GB VRAM, and 2 of them have 6GB VRAM. https://store.steampowered.com/hwsurvey/videocard/

You live in an echo chamber.
 
Well, it's nvidia, contrary to their competition they don't engage in that "fake msrp promo code" BS everyone else is doing. You can find a 5060, right now, available in stock for 319$. So the performance / $ is even better than in that graph.
hooray we get the rtx 3060Ti/3070 performance 5 years later with the same amount of VRAM for $320 , what an absolute "win" right ... lol ... :laugh:
rtx 4080 says hi with its whooping 80% lead (where are the times when gtx 1060 6gb delivered 980 performance and rtx 2060 did the same with gtx 1080 at $350)
 
Yet none of the made up FPS matters because 4X frame gen won't feel like 4X the FPS, unless the frame rate is already high enough to not need multi frame gen.
Shifting the goalposts is a sure sign you know you're wrong. At first it was Nvidia just was lying, now its "well, it doesn't feel the same, so I don't like it."

Nvidia was right about what MFG delivered. They severely undersold it, actually, They could have said 5070 = 2x 4090, and would have been more accurate. Whether you like the feature or not is an entirely separate question. Personally, I don't like the move towards more upscaling and frame generation, I'd prefer if Nvidia dedicated more die space to RT cores instead of relying on DLSS to compensate. But it's there, and you can't say they were wrong about what it does.

And the general public is buying Nvidia 40 and 50 cards. You obviously aren't, which is of course your choice. You get to decide what's best for your money. But you need to be able to distinguish your personal thoughts and feelings from what the general public obviously wants.
 
Shifting the goalposts is a sure sign you know you're wrong. At first it was Nvidia just was lying, now its "well, it doesn't feel the same, so I don't like it."

Nvidia was right about what MFG delivered. They severely undersold it, actually, They could have said 5070 = 2x 4090, and would have been more accurate. Whether you like the feature or not is an entirely separate question. Personally, I don't like the move towards more upscaling and frame generation, I'd prefer if Nvidia dedicated more die space to RT cores instead of relying on DLSS to compensate. But it's there, and you can't say they were wrong about what it does.

And the general public is buying Nvidia 40 and 50 cards. You obviously aren't, which is of course your choice. You get to decide what's best for your money. But you need to be able to distinguish your personal thoughts and feelings from what the general public obviously wants.
Too bad the laptop 5090 is no faster than the 4090.
 
It is now, yes. I was talking about the past.

My point is that performance per price is a nice metric for cards that are within an acceptable performance range to begin with. How true that is for an 8 GB card anno 2025, I'll leave it up to anyone's judgement.

As for me, like I stated in another thread, if I dug out an old 8 GB card from the closet, I'd see no problem using it in a secondary / indie gaming rig, but I'd never spend 300 for one, regardless of its price to performance data.
Personally my opinion is that 8GB cards will be mostly great in 2025. 90% of the new games run fine on them at 1080p, sometimes even 1440p! It'll be 2026 and 2027 where they're getting thrown in a dumpster as the 12GB and 16GB cards costing barely any more right now keep on being fine for another 3+ years. It's the short-term thinking like this that causes most of the world's economic crashes and depressions. "Investing tomorrow's money in something that likely won't be much use tomorrow".

Cards without enough VRAM are usually totally fine until they're suddenly completely missing the mark for a majority of brand new games in a very short timeframe. See: GTS 450 (512MB), ATi Fury (4GB HBM) GTX 1060 3GB - to name just a few. All fine performers at their date of launch but reduced to scraping along at the entry-level settings within a couple of years while their VRAM-endowed siblings made it to legendary status as GOATs that delivered a usable experience for 2x, 3x, or even 4x as long.
 
Last edited:
Personally my opinion is that 8GB cards will be mostly great in 2025. 90% of the new games run fine on them at 1080p, sometimes even 1440p! It'll be 2026 and 2027 where they're getting thrown in a dumpster as the 12GB and 16GB cards costing barely any more right now keep on being fine for another 3+ years....

Cards without enough VRAM are usually totally fine until they're suddenly completely useless for brand new games in a very short timeframe. See: GTS 450 (512MB), ATi Fury (4GB HBM) GTX 1060 3GB - to name just a few.
At that point get a 6600 and save your money. Where I live this card is 3x the cost of the 6600.
 
Shifting the goalposts is a sure sign you know you're wrong. At first it was Nvidia just was lying, now its "well, it doesn't feel the same, so I don't like it."

Nvidia was right about what MFG delivered. They severely undersold it, actually, They could have said 5070 = 2x 4090, and would have been more accurate. Whether you like the feature or not is an entirely separate question. Personally, I don't like the move towards more upscaling and frame generation, I'd prefer if Nvidia dedicated more die space to RT cores instead of relying on DLSS to compensate. But it's there, and you can't say they were wrong about what it does.

And the general public is buying Nvidia 40 and 50 cards. You obviously aren't, which is of course your choice. You get to decide what's best for your money. But you need to be able to distinguish your personal thoughts and feelings from what the general public obviously wants.
He's not shifting the goalpost. Frame gen 60 fps is not the same as original 60 fps, so you can't compare the two. You cannot put an equation sign between the two in any case. It's not about liking or not liking the feature.
 
Status
Not open for further replies.
Back
Top