NV 4060 / 4060TI reviews

[Mopetar]

They probably saw the 7600 selling decently and the 4060 Ti not selling at all, so they decided to put something up against the 7600.

And I do expect the 4060 to sell far better than the 4060 Ti, since it should effectively be a slightly cheaper 3060 8 GB with DLSS 3.

Is the 7600 actually selling well? That was panned almost as much as the 4060 Ti in reviews and I can't see too many NVidia customers jumping ship for AMD over a $270 7600. Even at $250 I think the card is overpriced and anything with 8 GB has a limited future.

DLSS3 is a non-factor on a 4060. Do you want a card that's going to generate fake frames from a low-res texture that the game engine swapped to because 8 GB of VRAM wasn't enough? Just smear Vaseline on your monitor and you won't be able to tell the difference.

The 4060 might look worse in a lot of benchmarks due to having less VRAM (8GB vs 12GB) and even though the clock speed is higher there's fewer CUDA cores (3072 vs 3584) so even casual buyers are going to look at a spec sheet and wonder why it has less.

 

[MrTeal]

Has smaller L2 though, 24MB vs. 32MB for the 4060Ti, which brings "effective bandwidth" down. Nvidia quoted a higher perf delta for 4060 vs. 3060 than 4060Ti vs 3060Ti, so I would assume the drop in memory subsystem performance is more than matched by the drop in compute.

Yeah, in that Nvidia press deck they had the 4060 Ti as 13.4% faster than a 3060 Ti, while the 4060 was 17.8% faster than a 3060 for non-frame gen games. We'll see how that translates to review roundups for the 4060, but for example in HUB/Techspot's 15 game 1080p average had less improvement with the 4060 Ti 9% faster than the 3060 Ti. If the scaling was the same you'd end up with a 4060 13% faster than a 3060 which is basically tied with a 7600.

4060 vs 3060 is going to be really complicated by the extra 4GB of VRAM, so there's more likely to be situations where lows are way better on the 3060 than 4060, or when IQ on the 4060 is lower than the 3060 because textures aren't being loaded.

 

[Aapje]

Is the 7600 actually selling well? That was panned almost as much as the 4060 Ti in reviews and I can't see too many NVidia customers jumping ship for AMD over a $270 7600.

MLID had some retailer comments saying that the card was selling better than expected, although that is probably a low bar.

In this down market, selling better than a fresh turd is probably counted as a victory.

 

[Mopetar]

MLID had some retailer comments saying that the card was selling better than expected, although that is probably a low bar.

In this down market, selling better than a fresh turd is probably counted as a victory.

I just saw another thread where someone posted sales from MindFactory. The 7600 has lower sales than the 7900XT. The 4060 Ti was even worse, but the 7600 is not selling all that well either.

But then again I'm surprised by the results. Due to the way the math works out you could even say the card is selling infinitely better than I would have expected.

Aim low and achieve as they say.

 

[Aapje]

I am very interested in seeing how well the 4060 will sell later this month. My guess is weak sales compared to the 3060, but higher than the 4070.

 

[jpiniero]

NVIDIA GeForce RTX 4060 has been tested in Cyberpunk 2077, OC gives 1% increase at 1080p Ultra settings - VideoCardz.com

NVIDIA RTX 4060 tested with Cyberpunk 2077 and overclocked Today NVIDIA has lifted the partial embargo on GeForce RTX 4060 graphics cards performance data. NVIDIA RTX 4060 tests, Source: JayzTwoCents NVIDIA had already published official benchmark results featuring RTX 4060 last month, but this...

videocardz.com

Apparently nVidia is letting some Youtubers do some mini previews of the 4060? Anyway it is slower than the 3060 Ti.

 

[gdansk]

And you can get a 3060 Ti for $275 here in the US anyway. But those will dry up soon.
The real advantage is lower power draw. The fake advantage is DLSS3.

 

[MrTeal]

Jayz rewarded for his glowing 4060 Ti review, I guess.

26% above a 3060 is higher than I'd expect even in non-VRAM limited scenarios, but there's probably a reason why Nvidia chose to let an early release of this one specific game. We'll see what happens when we get full independent reviews.

 

[coercitiv]

Jayz rewarded for his glowing 4060 Ti review, I guess.

26% above a 3060 is higher than I'd expect even in non-VRAM limited scenarios, but there's probably a reason why Nvidia chose to let an early release of this one specific game. We'll see what happens when we get full independent reviews.

First they bench the card and measure 81 average FPS on 1080 Ultra without RT/FG/DLSS, then they show gameplay where they drive through the city, using the same settings, and repeatedly dip bellow 60 FPS. Must be the weather.

 

[MrTeal]

First they bench the card and measure 81 average FPS on 1080 Ultra without RT/FG/DLSS, then they show gameplay where they drive through the city, using the same settings, and repeatedly dip bellow 60 FPS. Must be the weather.

Yeah, that's not even small dips. While he was driving there it got down to pretty consistent 50-low 60s FPS for 20 seconds.

 

[Mopetar]

It would be useful to look back at how the 4060 Ti stacked up against the 3060 Ti I. This title to see if it may just be an outlier.

TPU had it at 13.3% / 11.7% / 9.0% above the 3060 Ti at 1080 / 1440 / 4K so this title is better than average for the Lovelace vs Ampere comparison than a lot.

However, anything that's sensitive to either bandwidth or VRAM capacity is going to perform worse. The 4060 has 75% of the bandwidth as the 3060 which isn't quite as bad as the 64% with the Ti, but we already know how quickly the extra L2 cache falls off, so even titles where the 4060 wins at 1080p can turn into losses at higher resolutions.

It might look okay next to the 7600, but that's not really saying much. We've already seen the 12 GB 3060 beat the 8 GB 4060 Ti and 4070 in select benchmarks, so there's going to be a repeat here. The biggest saving grace for the 4060 is that it's not really any more bandwidth or memory starved than the 4060 Ti so it can perform comparably well in any situation where that card was significantly bottlenecks by the memory.

 

[Ranulf]

When Nvidia is singing the praises of low power usage you know they're in trouble performance wise.

 

[DAPUNISHER]

I am having many lulz about how they are only allowed to show Nvidia's bought and paid for tech demo, I mean cyberpunk'd. Now you understand why, if you didn't before, that Alex, the Nvidia marketing human caterpillar segment, made that video description saying "Path tracing on a triple-A videogame? On THE triple-A videogame? "

Daniel-San says they are only allowed to show you Punk'd, and only at, and I quote, "These exact settings."

If we were supposed to grab our pitchforks because AMD allegedly kept DLSS out of games. What is the proper reaction to Nvidia proving that they basically implemented the Nvidia partner program after all? 🤫 I expect we will find out shortly as the marching orders are given to the minions on how to astroturf this one.

 

[DeathReborn]

First they bench the card and measure 81 average FPS on 1080 Ultra without RT/FG/DLSS, then they show gameplay where they drive through the city, using the same settings, and repeatedly dip bellow 60 FPS. Must be the weather.

Their own 7600 review has the 7600 on 86fps and they say they used the same settings for both. They do say the benchmark they use is higher FPS than some parts of the game, so not unusual if you pay attention to the video.

 

[AtenRa]

When Nvidia is singing the praises of low power usage you know they're in trouble performance wise.

Where is this from ??

 

[Saylick]

When Nvidia is singing the praises of low power usage you know they're in trouble performance wise.

Good lord. That's the pitch? Upgrade from the 3060 to the 4060 and spend 20 hours of PC gaming per week, every week, for 4 years straight and you just might save $50 over that time frame.

C'mon Nvidia. You know you can simply just lower the price of the 4050 4060 by $50 to $249 and save us all the trouble, right?

 

[Ranulf]

Where is this from ??

I saw it on Videocardz:

NVIDIA claims GeForce RTX 4060 is 20% faster than RTX 3060 without Frame Generation - VideoCardz.com

NVIDIA RTX 4060 is 20% faster than RTX 3060 without Frame Generation On the NVIDIA website there is a new post covering the upcoming launch of the entry-level GeForce RTX 4060 graphics card. The RTX 4060 launches on June 29th, exactly a week from now. The release date has been pushed forward...

videocardz.com

Official news post at Nvidia:

GeForce RTX 4060 Launching June 29th, Starting At $299

1080p gamers get a substantial upgrade with the new GeForce RTX 4060, powered by the NVIDIA Ada Lovelace architecture, and AI-accelerated by DLSS 3. Play today’s best games at high frame rates and fantastic levels of detail, enhanced with DLSS in over 300 games and apps.

www.nvidia.com

 

[igor_kavinski]

What they fail to mention is that the 3060 has extra 4GB VRAM, at least the original ones do. The 4060 is a 4050 in disguise, folks. I hope they suffer massive losses as stock just sits on the shelves as more and more people realize that this shouldn't sell for more than $219.

 

[moonbogg]

I blame the people who bought 3070 ti's for $900 and bragged about how good of a deal they got during the mining craze.

 

[coercitiv]

They do say the benchmark they use is higher FPS than some parts of the game, so not unusual if you pay attention to the video.

It's not unusual when it's just one game vs. a roster of games, or one review among many. This time their benchmark isn't good enough to form an opinion about the performance needed to play the game.

I'm not pointing the finger at Jay, he's the one showing a gameplay session and doing a better job at informing the audience. I'm signaling that these previews should not be treated as mini reviews.

 

[Aapje]

It's really Nvidia that is manipulating the market here by getting the reviewers to make these mini-reviews of a very restrictive scenario that paints the card in the best light (no VRAM issues, DLSS 3 is in the game, etc).

It reeks of increasing desperation among their marketing team, who are set up for failure by Jensen and now probably are not going to get their bonus.

 

[Heartbreaker]

First they bench the card and measure 81 average FPS on 1080 Ultra without RT/FG/DLSS, then they show gameplay where they drive through the city, using the same settings, and repeatedly dip bellow 60 FPS. Must be the weather.

I don't understand why anyone follows Jayz2C. He is so unreliable.

Anyway all these early previews with strings attached are always highly suspect and pointless.

 

[Aapje]

I don't understand why anyone follows Jayz2C. He is so unreliable.

He's entertaining. You really need to understand that people tend to react better to charisma and big tit... inclusivity than facts and quality.

For me he's as much of a hate-view as anything. He's just so earnestly lazy, incompetent and short-sighted. Watching him is like watching a cat do crazy stuff with a cardboard box. You don't get rid of the cat because he act inane, you love him for it.

 

[Saylick]

Looks like Nvidia isn't exactly paying certain content creators to do mini-reviews but more of a relaxed NDA where the incentive is potentially having content out before others, albeit highly limited in what Nvidia lets you test. Content creators then have cover by stating that they weren't paid, which is technically true.

https://www.reddit.com/r/hardware/comments/14k0i9b

 

[BFG10K]

HWUB turned down the offer to be NV's mouthpiece.