Nvidia is at it again: 3060 8GB released, up to 35% slower than its 12GB counterpart

[Marioguy0]

23 hours ago, RONOTHAN## said:

RX 580 4GB and RX 580 8GB were the exact same card as the name implies, they just had different VRAM amounts. 

Actually, this is not entirely true.  Most 480/580 4GB variants shipped with a lower memory clock of 1750MHz compared to the 8GB variants usually having a 2000MHz memory clock.

 

It may not be huge, but it is a downgrade.  I do agree with your overall sentiment that green team is the worse offender in this area, though.

[WereCat]

10 hours ago, porina said:

Random thoughts: We had a kinda reverse situation. Where was the outrage when nvidia released the 3080 12GB after the 10GB? The changed the vram quantity and width there too, and even the core count! Where were the calls to name it a 4085 or something? I'll give in that case, the performance between the two was a lot closer. If we assume the 3080 was balanced for 10GB and associated width to start with, adding more doesn't really give as much benefit. If 3060 was tuned for 12GB then reducing to 8GB could be more of a constraint, although it may be partially offset if they choose to find some faster vram for it.

 

I still think the true price for the 8GB is not currently reflected by the few random 3rd party listings via Newegg. If we assume its overall performance sits between 3050 and 3060 12GB, and likewise is priced as such, is balance not restored to the Force?

When 3080 12GB released the prices were so bad on everything and yet still everything sold out so you were lucky to grab a card. Outrage was there but it was lost in the even bigger outrage of shortages and scalping prices.

There was also outrage because of LHR, you got LHR 3080, non-LHR 3080, LHR 3080 12GB, LHR 3080ti ... the performance difference between all these cards is so small that it's just slap to the face for anyone who wanted to pay the extra over 3080. Even 3090 was a bad value vs 3080 if you did not need the VRAM. And 3090 had it's own share of issues with VRAM running hot as heck under the backplate because oi inadequate cooling.

But as I said, prices were out of whack so people got what they could when available which is not case this time and you actually have many options to choose from at decent prices in that price range especially.

[Poinkachu]

17 minutes ago, WereCat said:

When 3080 12GB released the prices were so bad on everything and yet still everything sold out so you were lucky to grab a card. Outrage was there but it was lost in the even bigger outrage of shortages and scalping prices.

There was also outrage because of LHR, you got LHR 3080, non-LHR 3080, LHR 3080 12GB, LHR 3080ti ... the performance difference between all these cards is so small that it's just slap to the face for anyone who wanted to pay the extra over 3080. Even 3090 was a bad value vs 3080 if you did not need the VRAM. And 3090 had it's own share of issues with VRAM running hot as heck under the backplate because oi inadequate cooling.

But as I said, prices were out of whack so people got what they could when available which is not case this time and you actually have many options to choose from at decent prices in that price range especially.

For me at the very least 3080 12gb was released while we were still in the RTX3000 release period.

This thing...... well...

[leadeater]

5 hours ago, Marioguy0 said:

Actually, this is not entirely true.  Most 480/580 4GB variants shipped with a lower memory clock of 1750MHz compared to the 8GB variants usually having a 2000MHz memory clock.

 

It may not be huge, but it is a downgrade.  I do agree with your overall sentiment that green team is the worse offender in this area, though.

I doubt there was much choice in the matter, the 4Gb modules would be mass produced at 7Gbps per pin and the 8Gb modules mass produced at 8Gbps per pin. Probably would have cost more to go with slower 8Gb modules 

[mr moose]

So what we have are two products, one clearly marked 8G and one clearly marked 12G.  The 8 gig performs worse in benchmarks than the 12G.  the one that performs better has higher numbers for dumb consumers.  Currently they cost the same.

 

We as enthusiasts know why the 8G variant performs worse, general consumers do not. At best they have done some research and know it performs worse than the 12G at worst they don't know what any of the numbers mean and are just buying what looks prettier.  In either case Nvidia has done nothing to lead those consumer to paying more for less (like they did with the 1030).

 

 

 

 

 

 

 

 

[leadeater]

39 minutes ago, mr moose said:

So what we have are two products, one clearly marked 8G and one clearly marked 12G.  The 8 gig performs worse in benchmarks than the 12G.  the one that performs better has higher numbers for dumb consumers.  Currently they cost the same.

 

We as enthusiasts know why the 8G variant performs worse, general consumers do not. At best they have done some research and know it performs worse than the 12G at worst they don't know what any of the numbers mean and are just buying what looks prettier.  In either case Nvidia has done nothing to lead those consumer to paying more for less (like they did with the 1030).

And any of that makes this ok how?

 

Back to the old but the number at the end actually means something to the uninformed. No, no it doesn't. Like it didn't for the RTX 4080.

 

People know the first part, rely on the first part, talk about the first part. How many people ever really talk about more than that, see GTX 1060 3GB vs 6GB. That was already painful enough and at least that was less of a kick in the shins.

 

There is zero excuse for not calling it an RTX 3050 Ti, could have been a very positively received product if so. Nvidia didn't do that, too bad for them.

[mr moose]

7 minutes ago, leadeater said:

And any of that makes this ok how?

 

Back to the old but the number at the end actually means something to the uninformed. No, no it doesn't .Like it didn't for the RTX 4080.

 

People know the first part, rely on the first part, talk about the first part. How many people ever really talk about more than that, see GTX 1060 3GB vs 6GB. That was already painful enough and at least that was less of a kick in the shins.

 

There is zero excuse for not calling it an RTX 3050 Ti, could have been a very positively received product if so. Nvidia didn't do that, too bad for them.

Then can we also stop the "bigger number insinuates better for dumb consumers" arguments we get around here? Because if that is true then this whole thread is contradicting that position. For the record I believe that bigger numbers makes products look better when they aren't, which is why I don't have a problem when you have two products and the one with the bigger number does actually performs better.

 

At the end of the day it doesn't matter what they call these things, if they have different and clear identifiers then there is obviously a difference in the product which should make nearly all prospective buyers ask what that difference is.  It seems arbitrary to decide that only one part of the name means anything to a subset of consumers. 

[staticpage]

NVIDIA GPU names all across the whole product line are confusing now even the laptop GPUs are named the same. I have not closely followed any hardware releases since 2017 or so because prices went sky high I just moved on to other hobbies. I remember being able to randomly wonder into a microcenter any day and get one of the best GPU they offered for a weeks wadges.

Early last year I visited family out of town for the weekend and ended up having to stay for a few months. I needed a laptop to do work, so I'm standing in a unfamiliar microcenter trying to figure out how and what a RTX 3050ti (laptop) compares to none laptop GPU as that's what I'm having to compare to. One might think that a RTX 3000 series name sake card -even when it's a (Mobile)- would compare to something last gen RTX (Turing) but nope it seams you have to go all the way back to 2016 (Pascal) to find a direct performance comparison GTX 1060-6gb is what I have found and tested. This is not the full story either the RTX 3050ti (Mobile) has a few variants its self that are not even labeled clearly and its a big one that effects the performance more than any names. Wattage based on its cooling solution, there is a 35W RTX 3050ti (Mobile) and the variants go all way up to 75w. Even the sales person at microcenter knew none of this as that store is now more about corporate sales than customer service its not a surprise to me.

I sat at one of the stores $15,000 mac pros and researched this for about an hour and that mac felt so nice I almost said fuck it give me a MacBook Pro.

 

I want to see some type of metric naming scheme based on performance.

[leadeater]

25 minutes ago, mr moose said:

Then can we also stop the "bigger number insinuates better for dumb consumers" arguments we get around here?

No, because you think 8GB vs 12GB is sufficient but that doesn't make you correct. What we have is a good, working, established and known naming scheme that could have been adhered to but wasn't and the case of that is much stronger than "but 8GB".

 

25 minutes ago, mr moose said:

At the end of the day it doesn't matter what they call these things

Yes it does. As a consumer protection law advocate yourself you know this. This is why your position is counter to your own beliefs which makes it al the more odd.

 

I invite you and everyone to Google search "GTX 1060 review". I rest my case.

 

It matters, it has always mattered and it matters more as the difference in product performance increases.

[Quackers101]

5 minutes ago, staticpage said:

NVIDIA GPU names all across the whole product line are confusing now even the laptop GPUs are named the same. I have not closely followed any hardware releases since 2017 or so because prices went sky high I just moved on to other hobbies. I remember being able to randomly wonder into a microcenter any day and get one of the best GPU they offered for a weeks wadges.

 

 Wattage based on its cooling solution, there is a 35W RTX 3050ti (Mobile) and the variants go all way up to 75w. Even the sales person at microcenter knew none of this as that store is now more about corporate sales than customer service its not a surprise to me.

 

I want to see some type of metric naming scheme based on performance.

Not just Nvidia, all of them. And yeah its a mess for consumers. Just like the BS for phones, getting a whole different chip/brain for your device that could have vastly different features and content in it and performance differences.

 

At least the regular to TI and SUPER models kind of works.

As for performance metric, so some kind of standard average of its 1. raster performance 2. raytrace performance 3. AI workloads?

So that you have an chip with RTX 3000 with these scores x,y,z and something about power?

[mr moose]

3 minutes ago, leadeater said:

No, because you think 8GB vs 12GB is sufficient but that doesn't make you correct. What we have is a good, working, established and known naming scheme that could have been adhered to but wasn't and the case of that is much stronger than "but 8GB".

That only works if you assume people don't read the entire name or don't consider the entire name to be relevant.  Your argument rests on the premise that not only do the average consumers know what the first 4 numbers mean and nothing else, but that they ignore the rest of the name too.  I simply don't buy it.    When people are faced with two options that are similar bar for a 1 number, they tend to ask what that number means.  If the guy in the store can;t answer that then they aren't getting good advice anyway (this is not nvidia's fault), and if they can answer then the consumers knows as much as we do.

 

3 minutes ago, leadeater said:

Yes it does. As a consumer protection law advocate yourself you know this. This is why your position is counter to your own beliefs which makes it al the more odd.

But there is a clear and obvious difference in the name,  so it is not counter to my advocation of consumer protection. if they both had the same name and one was slower due to bandwidth then I would be all up in nvidia's shit about it.

 

 

[leadeater]

10 minutes ago, mr moose said:

That only works if you assume people don't read the entire name or don't consider the entire name to be relevant.

You still miss the issue. It's not a matter of if they read the whole thing or not it's a matter of if "8GB" is of enough significance at all. You think that it's enough, but are you correct? Is that really going to stop people not misunderstanding or making the assumption that it's still an RTX 3060 as it literally says it is.

 

Again refer back to the GTX 1060 issue. I can already prove this is a problem.

 

10 minutes ago, mr moose said:

But there is a clear and obvious difference in the name

No see you should say you think it's a clear and obvious enough difference. You think.

 

I think there is an existing decade long established and understood naming scheme Nvidia already has that could have been followed that would have a lesser chance of consumer getting confused or mislead.

 

Now if you want to make a counter argument then I would try and make a case for how RTX 3060 8GB is equally clear as RTX 3050 Ti.

 

It's actually like you are arguing that existing knowledge and understanding doesn't matter. What we call things doesn't matter. Imagine trying to teach Physics but using your own names for everything then arguing "but there is a difference in names for each thing so it cannot be confusing". Established norm do actually matter and people do rely on them.

 

  

10 minutes ago, mr moose said:

if they both had the same name and one was slower due to bandwidth then I would be all up in nvidia's shit about it.

Oh but they do, they do. They both have RTX 3060 in the name and that is not a decision that should have been made nor supported. You should be advocating for better not accepting pathetic corporate first choices because it helps them more than us. It really was not that difficult to have named it RTX 3050 Ti.

 

If you at all in any way believe RTX 3050 Ti would have been a better choice then you should stop defending the current name.

[Quackers101]

as said above, this is not just the name itself.

But also how stores or the sellers will deal with this information.

As some stores hide some of the fact about its VRAM amount or don't think its that important and might fall under the spec list than the product name shown.

For people that know nothing about what is that important or how one or the other will affect the product they buy. Also that the price will likely change.

[staticpage]

5 minutes ago, Quackers101 said:

Not just Nvidia, all of them.

 

As for performance metric, so some kind of standard average of its 1. raster performance 2. raytrace performance 3. AI workloads?

So that you have an chip with RTX 3000 with these scores x,y,z and something about power?

That is sad to think about. I remember even back in Haswell days having to research Intel cpus for a good long time to figure it out. Quickly looking at the new names now WTF Intel.

 

The naming having a 3 number identifier sounds like a good start. I bet if they had an advertising name like that the low end RTX cards would not sell so well though. As I have found out with the RTX 3050ti mobile I bought the raytrace performance is useless. Before I bought it at the store I found that the raster performance was good enough for my work needs and I was willing to pay more for RTX cores to try them out I thought "it has RTX cores so I can try a little raytraceing visuals as well". That is not the case IDK why that chip even has RTX cores it can't handle any raytraceing. If I where to do it again I would have spent more money on a better laptop gpu. Say a 3070 or 3080 even. I feel like the poor naming made everyone lose out on a up sale.

 

[bramturismo]

At this rate, I think nVidia should switch their model namings to what the Japanese did with their cars in the 90s. That way, we will have an RTX 3060 1.4 GLS FE Super Turbo 4-door VVT-I.

[xAcid9]

3 hours ago, bramturismo said:

At this rate, I think nVidia should switch their model namings to what the Japanese did with their cars in the 90s. That way, we will have an RTX 3060 1.4 GLS FE Super Turbo 4-door VVT-I.

Just go back to the old days, RTX 3060 SE a.k.a slow edition and put lower price tag than the 12GB but no, that'll be to easy to understand. 

 

Nvidia intention here is to confuse the customer with the naming and everyone will blame the user for not doing proper research when they bought the wrong 3060.

They not only confuse the customer, they also confuse the retailer with their non-up-to-date shit pricing. Same price as 12GB? Wut

 

They have been doing this for the longest time, much more often than AMD. Anyone remember GTX 560 Ti 448 cores? 

[Fasterthannothing]

17 hours ago, leadeater said:

Ah for years AMD have made sure their different VRAM variants used the same bus width so performance was in fact the same. Nvidia might, and has, messed around with this before and multiple times but like the 4080 situation repeated bad behavior doesn't make it ok. Just proves they don't give a damn.

AMD pulled this same thing with the  Radeon HD 4860 and 4870 (4860 was faster) and everyone was fine with the 3080 12gb soo idk seems like this is bad but not unheard of

[LAwLz]

Making a product name 5 or even 6 letters long is clearly Nvidia trying to profit on gamers' inability to read and short attention span.

We can't expect gamers to be able to read more than 4 numbers in a row.

[Quackers101]

10 minutes ago, LAwLz said:

Making a product name 5 or even 6 letters long is clearly Nvidia trying to profit on gamers' inability to read and short attention span.

We can't expect gamers to be able to read more than 4 numbers in a row.

? very much depends on what all of them mean and how its used. just like we have 3000 series with ti and super (if that comes back).

Some used for marketing reasons too. good luck if something was 3xtxttx vs 3xttxxt and multiple variations of it.

Edited December 4, 2022 by Quackers101

[mr moose]

22 hours ago, leadeater said:

You still miss the issue. It's not a matter of if they read the whole thing or not it's a matter of if "8GB" is of enough significance at all. You think that it's enough, but are you correct? Is that really going to stop people not misunderstanding or making the assumption that it's still an RTX 3060 as it literally says it is.

If you want to know if 8G is enough then you look up some benchmarks and reviews.  Just like if you want to know what the difference is between the XX50 and the XX50ti.  There is absolutely no difference to the unknowing consumer whether you call it the fart-o-matic or a fart-o-matic 2000,  the consumer will see the difference in the name as unless they like taking risks would naturally ask which one is better.  If any consumer sees a 3060 8G on the shelf they know it is a 3060 GPU with 8G of ram (no lies there),  if they want to know if that's a good deal or not because the one next to it is identical bar for having 12G of ram then they need to ask or google it or jump on a forum.  

 

22 hours ago, leadeater said:

Again refer back to the GTX 1060 issue. I can already prove this is a problem.

I don't see what that graph is supposed to prove,  it is a 1060 in a review showing performance results.

22 hours ago, leadeater said:

No see you should say you think it's a clear and obvious enough difference. You think.

 

I think there is an existing decade long established and understood naming scheme Nvidia already has that could have been followed that would have a lesser chance of consumer getting confused or mislead.

Just because they perform different doesn't mean they have to have some sort of caveat in the name.  The name just needs to be unique for each variant of the card.  In this case it is.

22 hours ago, leadeater said:

Now if you want to make a counter argument then I would try and make a case for how RTX 3060 8GB is equally clear as RTX 3050 Ti.

 

It's actually like you are arguing that existing knowledge and understanding doesn't matter. What we call things doesn't matter. Imagine trying to teach Physics but using your own names for everything then arguing "but there is a difference in names for each thing so it cannot be confusing". Established norm do actually matter and people do rely on them.

 

No, I am arguing that what the name is doesn't matter (i.e they could call it the hoovermatic 8G or the runnyfart 12G), what matters is that the name is unique to each variant. If you know nothing about a product then the most important thing you need is for every product variant to have a clearly different name.  Like in this case where one is marked as having a different amount of ram.  Once you know there is a difference then you can find out what that difference is. 

 

22 hours ago, leadeater said:

  

Oh but they do, they do. They both have RTX 3060 in the name and that is not a decision that should have been made nor supported. You should be advocating for better not accepting pathetic corporate first choices because it helps them more than us. It really was not that difficult to have named it RTX 3050 Ti.

 

Your still stuck on only looking at half the name the name is actually RTX3060 8G,  it doesn't stop at 3060.  That is really important and something that all consumer will notice.

22 hours ago, leadeater said:

If you at all in any way believe RTX 3050 Ti would have been a better choice then you should stop defending the current name.

I don;t give a fuck about what they call it, so long as it is different for each variant.

 

17 hours ago, Quackers101 said:

? very much depends on what all of them mean and how its used. just like we have 3000 series with ti and super (if that comes back).

Some used for marketing reasons too. good luck if something was 3xtxttx vs 3xttxxt and multiple variations of it.

If we are talking about consumers with little to no knowledge of GPU's then as I said above, all they need is to know that each variant is different.  Be that using the ram amount, the GPU prefix (i.e RTX or GTX) or even an postfix like TI.  If a consumer can identify the product they are looking at both in the store and on a benchmark table then they have all the information they need to both ask questions about it (i.e is this better for X workloads) and to compare to other products.

 

 

 

To the thread in general,  the only time a name should be a problem is if you have two different products with the same name and no obvious way to tell them apart on benchmarks or in a store.   Like the 1030 was, both cards had the same name but had different type of ram making one card a piece of shit while the other was so bad it was barely worth the listing. That is bad naming.  

 

 

[WereCat]

On 12/4/2022 at 1:11 PM, mr moose said:

That only works if you assume people don't read the entire name or don't consider the entire name to be relevant.  Your argument rests on the premise that not only do the average consumers know what the first 4 numbers mean and nothing else, but that they ignore the rest of the name too.  I simply don't buy it.    When people are faced with two options that are similar bar for a 1 number, they tend to ask what that number means.  If the guy in the store can;t answer that then they aren't getting good advice anyway (this is not nvidia's fault), and if they can answer then the consumers knows as much as we do.

 

But there is a clear and obvious difference in the name,  so it is not counter to my advocation of consumer protection. if they both had the same name and one was slower due to bandwidth then I would be all up in nvidia's shit about it.

 

 

There is not a clear difference in the name. 

 

As was pointed out in the video and my previous comment... the card is named RTX 3060 on the box. 

 

There's no RTX 3060 12GB or RTX 3060 8GB.

The VRAM amount is located on the box but it's not part of the name of the card and can be easily overlooked. 

[Blademaster91]

2 hours ago, mr moose said:

If you want to know if 8G is enough then you look up some benchmarks and reviews.

As others have already pointed out not everyone is going to look at benchmarks and reviews or will know the difference.

2 hours ago, mr moose said:

There is absolutely no difference to the unknowing consumer whether you call it the fart-o-matic or a fart-o-matic 2000,  the consumer will see the difference in the name as unless they like taking risks would naturally ask which one is better.  If any consumer sees a 3060 8G on the shelf they know it is a 3060 GPU with 8G of ram (no lies there),  if they want to know if that's a good deal or not because the one next to it is identical bar for having 12G of ram then they need to ask or google it or jump on a forum.  

So because the uninformed consumer might not know the difference, its ok for them to be misled by product naming or ripped off buying a worse graphics card?

If the naming were clear that the 3060 8GB was a different card then a consumer wouldn't need to google or ask a forum if the 3060 8GB is worth buying compared to a 3060 12GB.

2 hours ago, mr moose said:

I don't see what that graph is supposed to prove,  it is a 1060 in a review showing performance results.

You're missing the point, the 1060 3GB and 1060 6GB perform differently, but both of those are in the same xx60 performance tier.

2 hours ago, mr moose said:

Just because they perform different doesn't mean they have to have some sort of caveat in the name.

It does though, thats how the naming is supposed to work with graphics cards and cpu's, except when a company wants it to be intentionally confusing.

2 hours ago, mr moose said:

If we are talking about consumers with little to no knowledge of GPU's then as I said above, all they need is to know that each variant is different.  Be that using the ram amount, the GPU prefix (i.e RTX or GTX) or even an postfix like TI.  If a consumer can identify the product they are looking at both in the store and on a benchmark table then they have all the information they need to both ask questions about it (i.e is this better for X workloads) and to compare to other products.

But just looking at the box there isn't anything that would obviously point out the difference in either version, the VRAM capacity alone shouldn't be used to differentiate a graphics card, because there is the 1060 3GB and 1060 6GB example with how different a card can be, and the 3080 10GB and 3080 12GB which isn't enough of a difference to call the 3080 12GB a 3080 Ti or a 3080 Super.

2 hours ago, mr moose said:

To the thread in general,  the only time a name should be a problem is if you have two different products with the same name and no obvious way to tell them apart on benchmarks or in a store.   Like the 1030 was, both cards had the same name but had different type of ram making one card a piece of shit while the other was so bad it was barely worth the listing. That is bad naming.  

That is the problem though, if someone is only looking at the product naming then they won't know the difference. I really don't see why people are defending Nvidia for purposely confusing the consumer into buying a worse graphics card, the consumer won't know they're getting ripped off until they wonder why a 3060 with 12GB or VRAM is faster.

[Kobz360]

42 minutes ago, WereCat said:

There is not a clear difference in the name. 

 

As was pointed out in the video and my previous comment... the card is named RTX 3060 on the box. 

 

There's no RTX 3060 12GB or RTX 3060 8GB.

The VRAM amount is located on the box but it's not part of the name of the card and can be easily overlooked. 

Correct. Even on NVIDIA's own website, they don't clearly differentiate them as much as some people here seem to think. If it was so obvious, they would ensure that the "3060 Family" clearly split into 3060 Ti, 3060 12GB and 3060 8GB, but they do not. This is on top of the great point you've already mate. 

 

Absolutely crazy how people are defending this... it's obviously not illegal or anything like that, but it's clearly unethical. 

 

 

How are we supposed to know if this chart is for the 3060 12GB or 3060 8GB? Or we're supposed to be magicians? What's even funnier is that the 1060 is also listed there which also had a controversy of its own lmfao 

[sof006]

I've said it before and i'll say it again, when my 3080 starts to become noticably bad for games in the future i'll either be making the jump to Intel or AMD for my next card.

 

Ain't no way i'm supporting Nvidia and their anti-consumer antics

[Stahlmann]

2 hours ago, sof006 said:

I've said it before and i'll say it again, when my 3080 starts to become noticably bad for games in the future i'll either be making the jump to Intel or AMD for my next card.

 

Ain't no way i'm supporting Nvidia and their anti-consumer antics

I'm in a similar situation. But my main problem is that i play a lot of ray traced games nowadays. I don't want to "upgrade" to a 7900 XT or 8900 XT for example while they can't get their ray tracing performance to the same level as several year old Nvidia cards. And ray tracing will only get more support going forward, being baked into popular next-gen game engines and all that.