Nvidia announces 5060 and 5060 ti

[Ha-Satan]

Summary

Nvidia announced their 5060 8GB ($299), 5060 ti 8GB ($379), and 5060 ti 16GB ($429)

 

Quotes

Quote

The RTX 5060 will launch with an MSRP at $299. Then there's the RTX 5060 Ti with 8 GB of VRAM at $379 (£349), and above that the RTX 5060 Ti with 16 GB of VRAM at $429(£399).

 

My thoughts

To be honest, in the current market (just looking at MSRP and ignoring tariffs/scalper markups) I actually don't hate the 5060 ti 16GB. While it still feels more expensive than it should be, the claimed 20% performance improvement over the 4060 ti puts it into the general ballpark of the 4070, but with more VRAM. $429 is actually relatively compelling against the typical selling price of a 3080 or 6800 XT on ebay, which might also be in the same performance tier, but with less VRAM and/or less compelling feature sets.

 

Sources

https://www.pcgamer.com/hardware/graphics-cards/nvidia-rtx-5060-ti-and-rtx-5060-announced-starting-from-usd299-we-can-work-with-our-partners-to-get-these-out-at-reasonable-prices-which-we-are-doing/

[Agall]

7 minutes ago, Ha-Satan said:

To be honest, in the current market (just looking at MSRP and ignoring tariffs/scalper markups) I actually don't hate the 5060 ti 16GB.

Considering the 4060ti 16GB was MRSP $500, $430 for the next generation seems pretty good. The upgrade to GDDR7 will probably make the 16GB variant not as dumb compared to the 4060ti 16GB's GDDR6.

 

I agree overall, the 5060ti 16GB might end up being a good option if available at MSRP.

[Agall]

Thank the Lord that Gigabyte is being the good guy here and making the PCIe slot make sense. Their cards might end up the best out of the series from that alone:

 

 

GIGABYTE WindForce GeForce RTX 5060 Ti Graphics Card GV-N506TWF2-16GD - Newegg.com

 

Nothing more infuriating than having a card that doesn't work in some SFF scenarios because they arbitrarily made the PCIe full length instead of to what the card is electrically wired to.

 

Its also got an interesting VGA aux power connection location that could be really cool:

 

 

Imagine how adorable a waterblock for this tiny thing would be.

[porina]

Reviews tomorrow then? Assuming OP source article claiming 20% uplift is over 4060 Ti, then it would sit somewhere between 3070 Ti and 4070, not quite reaching the latter. AMD side, that could put it ahead of 6800 and 7700 XT. Worst case based on compute only it could be slightly behind, but that doesn't take into consideration bandwidth impacts.

 

13 minutes ago, Agall said:

Nothing more infuriating than having a card that doesn't work in some SFF scenarios because they arbitrarily made the PCIe full length instead of to what the card is electrically wired to.

Thought most desktop ITX boards were x16 so wouldn't matter? I'd rather have full mechanical length for physical strength. Unless you're looking at MODT ones that are natively limited to x8 or lower?

[jaslion]

17 minutes ago, Agall said:

Nothing more infuriating than having a card that doesn't work in some SFF scenarios because they arbitrarily made the PCIe full length instead of to what the card is electrically wired to.

Where would you run in to this where you also have a psu powerfull enough to power these? Other than like sff desktops that cant even have a psu powerfull enough to run it I haven't run into x8 sized connectors much. I like the non wired up x16 connector as it does add some rigidity. Another issue I see with the x8 wired ones is well for basically every single computer that these will be going in to they now just don't latch meaning they can get loose much easier.

 

It is funny how tiny they look and I am curious how these would stack up against the mobile counterparts because I really feel like this is just going to be the same chip as the 5060m and 5060tim :p. Especially with how tiny this thing is.

[Agall]

18 minutes ago, porina said:

Thought most desktop ITX boards were x16 so wouldn't matter? I'd rather have full mechanical length for physical strength. Unless you're looking at MODT ones that are natively limited to x8 or lower?

I don't think that matters when the PCB is only like 4" long and the majority of the card is flow through.

 

13 minutes ago, jaslion said:

Where would you run in to this where you also have a psu powerfull enough to power these? Other than like sff desktops that cant even have a psu powerfull enough to run it I haven't run into x8 sized connectors much. I like the non wired up x16 connector as it does add some rigidity. Another issue I see with the x8 wired ones is well for basically every single computer that these will be going in to they now just don't latch meaning they can get loose much easier.

 

It is funny how tiny they look and I am curious how these would stack up against the mobile counterparts because I really feel like this is just going to be the same chip as the 5060m and 5060tim :p. Especially with how tiny this thing is.

The most recent example of this is some servers, one that I had to add a GT 1030 to and it only had an 8x slot. That GT 1030 having an 8x slot already, but the card I originally bought for it didn't, even though it was wired for only electrically 8x but physically 16x.

I remember years ago seeing this problem with some ATX boards where the lower PCIe slots had open backed 1x or 4x but couldn't fit a full 16x slot card, but could an 8x. Its a conclusion I came to a few years ago after running into this problem on various ATX boards and SFF rigs that's only been validated as better than the latter.

 

13 minutes ago, jaslion said:

Another issue I see with the x8 wired ones is well for basically every single computer that these will be going in to they now just don't latch meaning they can get loose much easier.

I don't think this is a big deal. Bolting the card down at the PCIe bracket while also having the rear I/O shield's tabs press fitted into the case makes the PCIe slot latch unnecessary with how small this card is. I wouldn't doubt if the PCIe slot screws provide most of their tension on the first 8 lanes of the PCIe slot as is.

 

It all makes me think of how ridiculous server rack mounted stuff can be where you're holding up a ton of expensive weight with just 4 bolts into the front.

[porina]

31 minutes ago, Agall said:

I don't think that matters when the PCB is only like 4" long and the majority of the card is flow through.

I can't use your earlier newegg link since it geolocates and tells me it doesn't exist where I am. Just found the product on Gigabyte's site. I now notice it looks like their 3 fan cards are x16, and 2 fan cards are x8, so you got some choice for your niche of a niche.

[WereCat]

51 minutes ago, porina said:

Reviews tomorrow then? 

According to HUB only 16GB 5060ti will be reviewed as NVIDIA and AIB wont sample 8GB for reviews. 

 

And 5060 will not be sampled for reviews either, it will just release on May in a random day. 

[porina]

1 minute ago, WereCat said:

According to HUB only 16GB 5060ti will be reviewed as NVIDIA and AIB wont sample 8GB for reviews. 

I did mean 5060 Ti only since that is the one going on sale soon. Unless there is some surprise there, they'll perform the same where not VRAM limited. It'll give so called reviewers an excuse to kick the dead horse once again when they get the 8GB by other means later.

 

1 minute ago, WereCat said:

And 5060 will not be sampled for reviews either, it will just release on May in a random day. 

Not at all, or not until closer to the time?

[WereCat]

2 minutes ago, porina said:

 

 

Not at all, or not until closer to the time?

Apparently not at all. Only reviews will be from people who can buy it or get via other means.

[Millios]

19 minutes ago, WereCat said:

According to HUB only 16GB 5060ti will be reviewed as NVIDIA and AIB wont sample 8GB for reviews. 

 

And 5060 will not be sampled for reviews either, it will just release on May in a random day. 

Sounds like team green has no hope in it doing good, at least to me

That actually tho will be a good thing in the long run

[jordanbuilds1]

theres one thing i care about... how big is the node!!!

[Agall]

24 minutes ago, porina said:

I can't use your earlier newegg link since it geolocates and tells me it doesn't exist where I am. Just found the product on Gigabyte's site. I now notice it looks like their 3 fan cards are x16, and 2 fan cards are x8, so you got some choice for your niche of a niche.

I'm sad to see no low profile option from the start. We'll likely have to wait a few months to get those as an option. Single slot seems possible too since they could take this existing design and have it with only 1 fan.

 

We're all probably aware by now with how down binned the xx60 cards are. At least what's exciting about the xx60 cards now is the small die size that allows for interesting designs. That diversity to me is where the xx60 series shines now.

 

Still hoping we see an RTX 5050 with a sub 75W TDP for PCIe only power.

[Agall]

1 hour ago, porina said:

I'd rather have full mechanical length for physical strength. Unless you're looking at MODT ones that are natively limited to x8 or lower?

Oh, thought I'd reiterate on this, that I include 4x cards into this equation which are far more limiting. The RX 6400 and 6500XT being the worst example where it was a 4x card but basically all models had a full 16x slot. 4x being a far more common restriction compared to 8x, but I'm glad to see 8x having a proper physical configuration. For me, it limits the overall versatility of the card unnecessarily. 

[porina]

3 minutes ago, Agall said:

For me, it limits the overall versatility of the card unnecessarily. 

I'm guessing the majority of the target use case will have a x16 mechanical slot so it doesn't matter. In the case of the Gigabyte card you reference, maybe they're thinking as they managed to get the PCB small enough, why spend the extra cost on a longer board. Like I said, it looks like only the 2 fan models get a short connector, while the 3 fan ones remain at x16. So they're not using the same board. What else is different?

[Ha-Satan]

4 minutes ago, porina said:

Like I said, it looks like only the 2 fan models get a short connector, while the 3 fan ones remain at x16. So they're not using the same board

Maybe I'm stupid and don't understand GPU design well enough, but would it be possible to re-use the PCB for a higher-end card on a lower-end card? Like, maybe you remove some of the power circuitry components but otherwise leave the traces and everything the same?

[porina]

Just now, Ha-Satan said:

would it be possible to re-use the PCB for a higher-end card on a lower-end card?

Nvidia have done that before, just don't test me on which ones work where. They need to make the footprint compatible. Beyond some difference it is worth making different PCBs for different tiers.

[starsmine]

430 USD being the cheapest viable card in their line up is wild
yes the 5060ti 8gb/ 5060 exist. those cards are e-waste. They break at running games at 1080p medium that came out over a year ago. They will break on more games coming out. 

Not delaying the 5060 for the inevitable refresh with 3GB chips in a couple of months is so annoying. 

Like they are suppressing reviews of the 8gb cards? that is a wild unhinged thing to do. 

Also If nvidia could stop lying about the FG implications, that would be great. Like this SHOULD be a kind of FTC violation




Also this chart stinks to high heaven. How did Hogwarts legacy go from 35fps to 45 fps and latency went from 125ms to 50ms? (between 3060 and 4060)
THIS MATH DOESNT ADD UP.

they did it again with the 5060 chart

Hogwarts Legacy goes from 45 fps to 60 but latency went from 110 to 40ms? THIS MATH DOESNT MATH.

DId they just not turn on reflex on the 30 series cards for any of these tests?

Like it is one thing to have done what AMD did with RDNA2 and ran those tests on unstable drivers that have a hazard on the launch slide and that "mislead" people.
But nvidia is just straight up lieing in these charts about the performance of their card, and there is no FTC to stop them.

[FrostyyTech]

5060 ti 8GB? Why does nvidia still think 8GB vram is enough?

[starsmine]

8 minutes ago, FrostyyTech said:

5060 ti 8GB? Why does nvidia still think 8GB vram is enough?

I don't think they do think its enough, given that they told AIBs not to seed any for reviews. 

(128/32 = 4, 4*2 = 8)

They should have just waited for the 3GB cards to release a 12GB version for their budget model for that price. the 5060ti 8GB may be a very short lived card but will find itself in OEM builds to advertise they have a 5060ti in their prebuilt. 

[jordanbuilds1]

is this on a 3nm chip?

[starsmine]

4 minutes ago, jordanbuilds1 said:

is this on a 3nm chip?

No, the special enhanced nvidia version on N5/N4P called 4N (Which should be called like 4NP to be less confusing but it aint). Same as the rest of blackwell (and ada). AKA no die shrink. 

[FrostyyTech]

3 minutes ago, starsmine said:

I don't think they do think its enough, given that they told AIBs not to seed any for reviews. 

(128/32 = 4, 4*2 = 8)

They should have just waited for the 3GB cards to release a 12GB version for their budget model for that price. the 5060ti may be a very short lived card but will find itself in OEM builds to advertise they have a 5060ti in their prebuilt. 

5060ti 16GB could actually very well be an amazing card, but 8GB model? No. I was playing forza horizon 5 the other day on max settings and was using 9gb vram. This card doesn’t even have that. 

[jordanbuilds1]

1 minute ago, starsmine said:

No, the special enhanced nvidia version on N5/N4P called 4N (Which should be called like 4NP to be less confusing but it aint). Same as the rest of blackwell (and ada)

welp, ig were gonna see if nvidia made a mistake or not.

[Ha-Satan]

9 minutes ago, FrostyyTech said:

5060 ti 8GB? Why does nvidia still think 8GB vram is enough?

Why would they incur more cost by putting more VRAM on it when 8GB clearly doesn't stop people from buying it? People yelled about the 4060 only being 8GB when it launched, and clearly that made it sell like shit, right?
 

Nope, it's the 2nd-most-used card on the Steam hardware survey, and only a fraction of a percent behind the most-used card, the 3060.

 

Nvidia will not give cards in this price bracket more VRAM until they actually sell poorly.