News Preliminary GeForce RTX 4060 Ti Specs Leak: 220W and a Short PCB

[Admin]

Nvidia's GeForce RTX 4060 Ti specs have allegedly leaked, and point to an AD106 GPU with 4352 CUDA cores.

Preliminary GeForce RTX 4060 Ti Specs Leak: 220W and a Short PCB : Read more

 

[Elusive Ruse]

More evidence that Nvidia has abandoned the mid and low range GPU market. 128bit bus is absolutely ridiculous.

 

[Giroro]

It looks like it won't have the memory bandwidth needed to play games with ultra textures, nor at resolutions higher than 1080P.
I bet they charge $800 for it.

 

[oofdragon]

Lol will it at least match a 3070Ti?

 

[emike09]

It'll be just barely better than the 3060 Ti, but with DLSS 3 so it's faster, and will cost only you $800!

 

[Amdlova]

128 bit bus nightmare hype train... Can't be worst. (4050ti 64bit edition)

 

[Ar558]

This is beyond a Joke, we have gone in two generations from a £200 60 Class card, to a likely £500+ 60 Class card which has worse specs than it's predecessor. nVidia think if they only make the 90 and 80 class cards decent and the rest <Mod Edit> they can get people to pay 4 figures. They are sadly mistaken. The money is in the mainstream not the Halo products. At this rate AMD and even Intel will be more relevant to 99% of consumers. Jenson needs to remember while he might be ok, the Cost of living Crisis across europe means most people are struggling to heat their homes and his market is shrinking everyday. This is self-inflicted suicide if they want to be anything other than a Rolls Royce of GPU's, they may have Kudos but they make very little money

 

[BX4096]

At this rate AMD and even Intel will be more relevant to 99% of consumers.

Fine by me. The sooner Ngreedia gets its much needed reality check, the better it'll be for the rest of us.

 

[InvalidError]

All welcome Nvidia's new $700 entry!

In a sane competitive market, 128bits would be at the ~$200 price point and 50-tier.

 

[tennis2]

This is beyond a Joke, we have gone in two generations from a £200 60 Class card, to a likely £500+ 60 Class card which has worse specs than it's predecessor.

Erm....the 2060 (vanilla) was $350 MSRP at launch. Even the 2060KO was $300

3060 was $330 MSRP whilst being ~25% faster than the 2060.

 

[nitrium]

I'm looking to upgrade my aging RTX 2060. Probably it will be replaced with the upcoming RTX 4060 (presumably 12GB), but really it all depends on what it costs and how much faster it is than my existing GPU. So far I'm not super impressed with the price/performance of the 4000 series.

 

[user7007]

128bit bus and 8gb sound like a 4050. It has enough cores to be a 4060 though. Guess we'll see how it performs. It sounds like it should be cheap to manufacture but if they follow current trends it'll be expensive for what you get. I wonder if we'll see a 64 bit bus 4050 or 4030 card.

AMD may have an opportunity in the midrange if nvidia prices it high.

 

[edzieba]

Always a good giggle listening to people complain about bus width who clearly have no clue what a bus is and what its bit width does,. but are sure that Smaller Number Bad so Nvidia Bad.

Remember the GTX 970 RMADAC partitioning 'issue', that was so much of a major issue that nobody even noticed it existed for a good year after the card launched, was extensively independently benchmarked, and remained the price/performance peak for its generation? The internet hate machine got a good rager on for that, with complete lack of understanding of what was actually going on not being any impediment - the card was not 'missing' 500mb, it just had a single RAMDAC channel that was unidirectional and not bidirectional, with driver-side memory readdressing to avoid it having any practical performance impact (the whole reason GPUs use GDDR rather than DDR after all is because they perform mostly large file reads, occasional large file writes, and basically no small file random access).

As always: buy components based on actual benchmarked performance, not specsheets or branding. If it offers adequate performance at a price you're willing to pay, then great, if it doesn't don't buy it. Giving a pair of foetid dingo's kidneys whether e.g. the sticker on the side of the card says "4070Ti" or "4080" is ludicrous if the performance and price remain identical.

 

[kyzarvs]

Erm....the 2060 (vanilla) was $350 MSRP at launch. Even the 2060KO was $300

3060 was $330 MSRP whilst being ~25% faster than the 2060.

£ =/= $
At the time, when the exchange rate was $1.40 then it was about 200 quid.

 

[The Historical Fidelity]

Always a good giggle listening to people complain about bus width who clearly have no clue what a bus is and what its bit width does,. but are sure that Smaller Number Bad so Nvidia Bad.

Remember the GTX 970 RMADAC partitioning 'issue', that was so much of a major issue that nobody even noticed it existed for a good year after the card launched, was extensively independently benchmarked, and remained the price/performance peak for its generation? The internet hate machine got a good rager on for that, with complete lack of understanding of what was actually going on not being any impediment - the card was not 'missing' 500mb, it just had a single RAMDAC channel that was unidirectional and not bidirectional, with driver-side memory readdressing to avoid it having any practical performance impact (the whole reason GPUs use GDDR rather than DDR after all is because they perform mostly large file reads, occasional large file writes, and basically no small file random access).

As always: buy components based on actual benchmarked performance, not specsheets or branding. If it offers adequate performance at a price you're willing to pay, then great, if it doesn't don't buy it. Giving a pair of foetid dingo's kidneys whether e.g. the sticker on the side of the card says "4070Ti" or "4080" is ludicrous if the performance and price remain identical.

You obviously don’t understand what you are talking about. You are saying that the 4060 Ti memory specs being significantly worse compared to 3060 Ti doesn’t mean anything? Simple math says otherwise: 4060 Ti has 128 bit bus with 18 Gbps memory for 288 GB/s bandwidth vs 3060 Ti has 256 bit bus with 14 Gbps memory for 448 GB/s bandwidth. That means the 4060 Ti has only 64% of the bandwidth of the 3060 Ti. Taking into account the 4060 Ti will have only 89% of the shader cores of the 3060 Ti and the reduced memory bandwidth, I’d say we’d be lucky to get a 60 Ti grade 4000 series card with maximum 10-15% better performance gen for gen, and the price will be 10-15% more given the 4000 series pricing trend so expect a $449 price tag.

 

[hannibal]

It'll be just barely better than the 3060 Ti, but with DLSS 3 so it's faster, and will cost only you $800!

It looks like it won't have the memory bandwidth needed to play games with ultra textures, nor at resolutions higher than 1080P.
I bet they charge $800 for it.

Well 3060 is 1080p GPU. Just fine... And new low end may be $500+

 

[peachpuff]

It looks like it won't have the memory bandwidth needed to play games with ultra textures, nor at resolutions higher than 1080P.
I bet they charge $800 for it.

Nonsense... $799.99 🤣

 

[Deleted member 14196]

To Solve this problem the demand needs to go away until that does this pricing issues won’t stop

 

[edzieba]

You obviously don’t understand what you are talking about. You are saying that the 4060 Ti memory specs being significantly worse compared to 3060 Ti doesn’t mean anything? Simple math says otherwise: 4060 Ti has 128 bit bus with 18 Gbps memory for 288 GB/s bandwidth vs 3060 Ti has 256 bit bus with 14 Gbps memory for 448 GB/s bandwidth. That means the 4060 Ti has only 64% of the bandwidth of the 3060 Ti. Taking into account the 4060 Ti will have only 89% of the shader cores of the 3060 Ti and the reduced memory bandwidth, I’d say we’d be lucky to get a 60 Ti grade 4000 series card with maximum 10-15% better performance gen for gen, and the price will be 10-15% more given the 4000 series pricing trend so expect a $449 price tag.

And here we have a shining example of "smaller number bad" in practice. Without also direct testing of texture compression differences between GA103 and AD106 (reducing bandwidth demand), cache behaviour differences between GA103 and AD206 (relocating read/write bottlenecks to a different interface), and whether GA103 was even memory bandwidth bottlenecked in the first place, concluding that a smaller number = worse real-world performance is at best premature.

But of course, internet commentators know better than Nvidia's engineers, and making number bigger means making GPU more betterer (and die area for the extra PHY lanes is just free and has no impact on die cost or power budget).

 

[jp7189]

At least the tiers are strikingly differentiated this gen. This card will likely launch for $500, and will make the 4090 look like a deal with 4x more cores, but less than 4x the price. Thats starting to feel like the point of this gen, and quite different from any other launch in recent memory where the halo is twice the price for a 3-5% performance boost.

 

[dopemoney]

Always a good giggle listening to people complain about bus width who clearly have no clue what a bus is and what its bit width does,. but are sure that Smaller Number Bad so Nvidia Bad.

Remember the GTX 970 RMADAC partitioning 'issue', that was so much of a major issue that nobody even noticed it existed for a good year after the card launched, was extensively independently benchmarked, and remained the price/performance peak for its generation? The internet hate machine got a good rager on for that, with complete lack of understanding of what was actually going on not being any impediment - the card was not 'missing' 500mb, it just had a single RAMDAC channel that was unidirectional and not bidirectional, with driver-side memory readdressing to avoid it having any practical performance impact (the whole reason GPUs use GDDR rather than DDR after all is because they perform mostly large file reads, occasional large file writes, and basically no small file random access).

As always: buy components based on actual benchmarked performance, not specsheets or branding. If it offers adequate performance at a price you're willing to pay, then great, if it doesn't don't buy it. Giving a pair of foetid dingo's kidneys whether e.g. the sticker on the side of the card says "4070Ti" or "4080" is ludicrous if the performance and price remain identical.

yeah, what he said. buy based on performance, not specs. specs are sexy, sure, but when it's GO time, you really just want it to work and work well. never forget...the launch price of the Titan X "Pascal" was $1,200...in Aug. 2016...not really surprising that we continue to see overpriced gpus.

 

[Ar558]

Erm....the 2060 (vanilla) was $350 MSRP at launch. Even the 2060KO was $300

3060 was $330 MSRP whilst being ~25% faster than the 2060.

Sorry the 2060 and 1660 were out simultaneously so I was thinking of that (I own the latter). The 3060 might have had a MSRP of that but for most of it's lifespan even established retailers were charging almost double so MSRP doesn't mean much.

 

[Ar558]

Always a good giggle listening to people complain about bus width who clearly have no clue what a bus is and what its bit width does,. but are sure that Smaller Number Bad so Nvidia Bad.

Remember the GTX 970 RMADAC partitioning 'issue', that was so much of a major issue that nobody even noticed it existed for a good year after the card launched, was extensively independently benchmarked, and remained the price/performance peak for its generation? The internet hate machine got a good rager on for that, with complete lack of understanding of what was actually going on not being any impediment - the card was not 'missing' 500mb, it just had a single RAMDAC channel that was unidirectional and not bidirectional, with driver-side memory readdressing to avoid it having any practical performance impact (the whole reason GPUs use GDDR rather than DDR after all is because they perform mostly large file reads, occasional large file writes, and basically no small file random access).

As always: buy components based on actual benchmarked performance, not specsheets or branding. If it offers adequate performance at a price you're willing to pay, then great, if it doesn't don't buy it. Giving a pair of foetid dingo's kidneys whether e.g. the sticker on the side of the card says "4070Ti" or "4080" is ludicrous if the performance and price remain identical.

While I agree in general, paying more for less memory bandwidth and no more VRAM (especially when 8GB is not enough in more and more circumstances) is still a bad deal. I'm sure the Core Clocks and Cache will allow a slightly better performance than the 30 series but if they had done the same keeping the bandwidth the same the performance increase could have been more but they are delibrately hamstringing it so they can do that for 50 Series and try and get another xxx Cash from people

 

[Ar558]

£ =/= $
At the time, when the exchange rate was $1.40 then it was about 200 quid.

Unfortunately it doesn't really work like that, the cost has almost always been the same in dollars as pounds even when the exchange rate was >1.5, now its 1.2 the £ cost is higher than $.

 

[Elusive Ruse]

Always a good giggle listening to people complain about bus width who clearly have no clue what a bus is and what its bit width does,. but are sure that Smaller Number Bad so Nvidia Bad.

Remember the GTX 970 RMADAC partitioning 'issue', that was so much of a major issue that nobody even noticed it existed for a good year after the card launched, was extensively independently benchmarked, and remained the price/performance peak for its generation? The internet hate machine got a good rager on for that, with complete lack of understanding of what was actually going on not being any impediment - the card was not 'missing' 500mb, it just had a single RAMDAC channel that was unidirectional and not bidirectional, with driver-side memory readdressing to avoid it having any practical performance impact (the whole reason GPUs use GDDR rather than DDR after all is because they perform mostly large file reads, occasional large file writes, and basically no small file random access).

As always: buy components based on actual benchmarked performance, not specsheets or branding. If it offers adequate performance at a price you're willing to pay, then great, if it doesn't don't buy it. Giving a pair of foetid dingo's kidneys whether e.g. the sticker on the side of the card says "4070Ti" or "4080" is ludicrous if the performance and price remain identical.

Always a good laugh when someone starts their "be reasonable guys" post with a dash of condescension and expects to be taken seriously after.

Specs are a good indicator of performance no matter what, especially when we are comparing a successor product to its predecessor; so pretending people are being idiots for predicting a weak immediate replacement to the 3060Ti based on the rumoured specs is intellectually disingenuous.

Now, I could be wrong and maybe Nvidia engineers have made a leap in let's say delta colour correction and blows the 3060Ti out of the water. Yet at this stage, being concerned and upset at these numbers are far away from being laughable. Let's wait and see, if the 4060Ti with these specs proves me wrong I will be here and own up to it.