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

[mr moose]

22 hours 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. 

I don't know what the stores are like in your area, but here in Australia it doesn't matter which shop you go to the cards are ALWAYS listed by GPU type and Ram size along side the brand and that brands variants.    I have not been able to find a single listing for a GPU that does not have the memory in the name.

 

21 hours ago, Blademaster91 said:

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

Then there is no point in have a unique name for each product then.  If a consumer is not even going to look them up or ask when there is a unique identifier then not having one won't improve that.

21 hours ago, Blademaster91 said:

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?

It's not misleading, the card is sold as an 8G or 12G,  no one knows how well a card performs without looking at benchmarks or buying the thing and testing it themselves, so the product having a unique identifier is the only important part of the name.

 

21 hours ago, Blademaster91 said:

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.

The naming is clear, one has more ram, if a consumer doesn't know that more ram equals better then they should be doing research. 

21 hours ago, Blademaster91 said:

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

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.

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.

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.

At this point it seems you and some others are arguing that the name should indicate a performance metric.  No GPU has ever done that, in fact I am having trouble thinking of a pc component that does that.  You are getting a 3060 with X ram, if you want to know how good that is you have to look it up just like every other component in your pc.  Hell the only reason we know there is a difference is because someone tested them and we read the review.  if you or anyone else goes and buys a GPU based solely on what they think the name should mean then more fool them.  You don;t buy a car just because it has GT in the name do you? what about the numbers they tack on the end?  no you test drive, you compare and read reviews and ask on forums what owners think.  This is no different, we have a unique way to differentiate these cards and that is all you need so you can work out which one is best for you.

 

The other thing people are forgetting in this discussion is that the ram size never tells you anything about performance,  it never has and likely never will.  Suddenly expecting or demanding they make it mean something in comparison to other products on the market is not going to change that.

[mr moose]

15 minutes ago, WereCat said:

 

 

RX 6800 vs RX 6800 XT implies they are different tiers of cards

No, to the untrained it implies nothing more than a difference.  What that difference is needs to be worked out.  For all we know as pleb moron consumers is that XT might mean it has racing stripes.

 

15 minutes ago, WereCat said:

RTX 3060 8GB vs RTX 3060 12GB implies they are the same tier of card with different amount of VRAM.

 

 

But it does imply a difference and if you as a consumer don't know to check out what that difference is then the issue is with you. 

As I said in my last post, ram has never been an indicator of performance, if they put in more ram but it was some dodgy cheap ddr4 or even kept the ram amount the same but used dodgy cheap shit and didn't tell you then there would be a case for misrepresenting the product, but this is going the other way, it is less ram so the marketing angle would suggest a lower tier card.  How much lower is what the consumer has to research,  because no naming scheme will tell you how well a product performs and never can.

 

 

 

[Fasterthannothing]

remove

[LAwLz]

2 hours ago, WereCat said:

Excuse me, what?

 

RX 6800 vs RX 6800 XT implies they are different tiers of cards

RTX 3060 8GB vs RTX 3060 12GB implies they are the same tier of card with different amount of VRAM.

How about we ask my mom which card is faster, the "RX 6800" vs "RX 6800 XT".

My guess is that she will have no idea and guess at random. 

 

If I ask her which card is faster, the "RTX 3060 8GB" or "RTX 3060 12GB" my guess is that she would pick the 12GB model, and she would be right.

 

 

This example only makes sense to you because you are used to it and understand the naming schemes.

I feel like you are talking out of both sides of your mouth. Whenever the spotlight gets aimed at some other weird naming convention you go "well I understand that so it's not an issue" but as soon as we bring up this instance you go "but we have to think of this from the POV of someone who doesn't know anything". You can't have it both ways. 

If 3060 8GB and 3060 12GB are too similar and might fool people then I would argue 6800 and 6800 XT are too similar and might fool people. It's not like AMD couldn't have named them 6800 and 6850, right? 

[Avocado Diaboli]

Nvidia (and every other manufacturer) should stop using numbers altogether. 3060 is a higher number than 2080. Bigger number implies better performance, because people are stupid and only look at one data point to make purchase decisions. /s

[starsmine]

Lets be completely honest. 
Anyone buying a 3060 12GB does not care about price performance value anyways. 

The 6700xt is more powerful then the 3060ti, costs the same as a 3060, and comes with two free games that are 60 dollars each

[leadeater]

You all know this issue isn't that big right? You all know the name is bad and a better one could have easily been chosen...

 

The bigger issue is coming here voicing the opinion that it's not a problem which is a opinion you can have, just also be self aware enough to know this and that you are actively supporting bad product naming in a style that won't always be correct, hasn't always been correct and leads to poor purchasing decisions and creates the prevalence of misinformation..

 

If you are fine with all of the above then that is your choice, just know you are more of the problem than Nvidia is in this situation.

[LAwLz]

1 hour ago, leadeater said:

You all know this issue isn't that big right? You all know the name is bad and a better one could have easily been chosen...

 

The bigger issue is coming here voicing the opinion that it's not a problem which is a opinion you can have, just also be self aware enough to know this and that you are actively supporting bad product naming in a style that won't always be correct, hasn't always been correct and leads to poor purchasing decisions and creates the prevalence of misinformation..

 

If you are fine with all of the above then that is your choice, just know you are more of the problem than Nvidia is in this situation.

I think the naming is bad, but I also don't think this is an issue because it will affect next to nobody.

I just don't like the inconsistency where this very minor issue gets multiple news articles written about it, this thread with almost 150 posts (30 of which are from you) and a bunch of people calling for Nvidia to be sued, how they will never buy from Nvidia, etc.

 

The reactions are so over the top. But when other products have bad names nobody cares. Hell, even names I think are bad from Nvidia didn't get this much attention. 

GTX 1660 vs GTX 1660 Ti vs GTX 1660 Super? Nahh, not at all confusing and doesn't need to be pointed out. I am sure nobody will be confused by that.

RX 6800 vs RX 6800 TX? Totally not confusing at all. No way we should point out that they could have named it the RX 6850 instead to make things easier to understand. Nope, no way we should point that out. That product gets a free pass.

 

The 3060 8GB is slower than the 3060 12GB? This is the worst thing ever! This is proof that Nvidia are the most evil company in the entire world!

Nobody will ever understand this product name! We can't expect people to read more than 4 numbers or look up benchmarks!

 

 

If I had to come up with a naming scheme for graphics cards, I would base the number on the GPU die. All cards that have the same die gets the same name. If there are differences such as different clocks or different memory, that gets added as a suffix. This is how GPUs in general have been named, both from AMD and Nvidia. In pretty much every generation we have had the same GPU paired with different memory, and it has always been a suffix. It hasn't been an issue because people need to look up benchmarks to understand what performance they will get to begin with, so they will see the difference in performance. If you are only googling half the product name then you're on your own. If we start ruling that product names can't contain suffixes because people might not read them then all of a sudden pretty much all product names are misleading. Suddenly we should have an issue with companies putting "OC" on cards as well because someone might look up the performance of an OC card and then buy a non-OC card and thus get fooled into buying a certain brand, because they ignore the "OC" suffix. 

 

If the argument is that the name should reflect a certain performance then we also get into issues with things like OC cards. Should the 4080 OC suddenly be called the 4085 because it performs better than the 4080? If not, why should the 3060 8GB be called the 3050 Ti just because it happens to perform worse than the 3060 12GB? 

[Dabombinable]

It'd really have made a great 3050ti. Considering with Kepler the GTX 650ti was a cut down GTX 660. Though in more than memory bus.

@LAwLz

Nvidia use to distinguish between models with cut down memory bus/ Eg. Riva TNT2 VantaLT/Vanta/M64 and then the TNT2/TNT2 Pro/TNT2 Ultra. And shock horror - the performance difference between a same clocked M64 and standard TNT2 was substantial. The new 3060's name is shit as it is a far slower product than the first 3060.

 

Fuck just look at the Geforce 2 MX200 and MX400. 64bit vs 128bit. Same die. Not given the same model name.

[MageTank]

6 minutes ago, LAwLz said:

The reactions are so over the top. But when other products have bad names nobody cares. Hell, even names I think are bad from Nvidia didn't get this much attention. 

GTX 1660 vs GTX 1660 Ti vs GTX 1660 Super? Nahh, not at all confusing and doesn't need to be pointed out. I am sure nobody will be confused by that.

RX 6800 vs RX 6800 TX? Totally not confusing at all. No way we should point out that they could have named it the RX 6850 instead to make things easier to understand. Nope, no way we should point that out. That product gets a free pass.

I do not disagree with you, but I do want to say that in @leadeater's defense, I have seen him complain about these very things. I've also seen my fair share of new builders on this very forum be confused by the GTX 1660 series so that alone proves it impacts those that may not know where to source proper reviews of these products. It doesn't help that most people typically google "GTX 1660 Fortnite FPS" and end up with some kid on Youtube that has no controls in their testing methodology using a 5 year old processor reviewing budget GPU's and bottlenecking them to near consistent performance.

 

Based on what I am reading, both you and Leadeater appear to be arguing for the same thing (though with differing levels of passion). Seems like we all agree that the current naming conventions suck across the board and not just for Nvidia. Anyone that remembers me during the Zen chipset launches knows I was not a fan of AMD's chipset naming convention being intentionally similar to Intel's but only being off by a single digit, resulting in mass confusion and people purchasing incompatible boards with the wrong vendors processor. This issue does not plague only the GPU segment and needs to be addressed across the industry.

 

My two cents on this 3060 8GB situation (not that anyone asked for it) is that Nvidia is well within their right to name the products as they deem appropriate. Personally I think it is shooting themselves in the foot, but who am I to get in the way of that bullet? If customers suffer the consequences, that is on Nvidia. Now the unfortunate collateral damage is those that work in computer retail and tech support segments that have to support these customers and are often at the forefront of the anger that stems from this issue. If those very retail chains and downstream partners get enough complaints, the burden to address this will surely fall on Nvidia and all other companies that participate in this nonsensical naming scheme. I have my doubts that these tech companies are reading these forums and achieving sudden states of enlightenment as if they didn't know this was a problem before.

 

Those of you on a crusade to save the ill-informed from making a mistake during purchase, put that energy towards something that will actually matter. Yelling at Nvidia and the other companies from a tech forum they don't frequent likely won't reach their ears, but you can certainly reach the ears of their customers if you create guides, write reviews or get the word out to reviewers that can reach those customers.

 

17 minutes ago, LAwLz said:

The 3060 8GB is slower than the 3060 12GB? This is the worst thing ever! This is proof that Nvidia are the most evil company in the entire world!

Nobody will ever understand this product name! We can't expect people to read more than 4 numbers or look up benchmarks!

I don't think people would have a problem with this if retail stores publicly advertised ROP/TMU/SM counts on their specification pages, but almost no retailer has a fleshed out specification tab for GPU's. If I am Joe Everyman and I google "8GB vs 12GB VRAM" I'll likely find an explanation about frame buffer sizes and impact of running out of VRAM, but nothing about similarly classed cards having completely different SM configurations or bus widths which would have a far more significant impact on performance than simply having less VRAM at a given resolution.

 

Now personally I don't put this blame entirely on Nvidia. Could Nvidia give us a better name? Say, an RTX 3055? Sure, nothing would logistically stop them from doing so. However, if retailers would flesh out the product pages for the products they are selling, the "compare this item" buttons on their websites might actually matter. I'd certainly like Nvidia to advertise these in-depth specs too. Sad that I have to go to TPU just to see that breakdown when nothing prevents them from making that advertisement themselves. Motherboard manufacturers get down to the nitty gritty and will tell you what each PCIe slot/M.2 slot is wired to, whether lane bifurcation is supported, etc. It's time GPU's catch up, lol.

 

25 minutes ago, LAwLz said:

If the argument is that the name should reflect a certain performance then we also get into issues with things like OC cards. Should the 4080 OC suddenly be called the 4085 because it performs better than the 4080? If not, why should the 3060 8GB be called the 3050 Ti just because it happens to perform worse than the 3060 12GB? 

I get your point here, however I think there is a point of clarification here that must be addressed. When a card is called "OC", it is because the product has been modified (in this example, overclocked beyond boost specifications). If a card is called "Mini", it is because the PCB is custom and designed to be smaller. If a card is called "8GB" vs "12GB", my assumption (based on this pattern) is that the VRAM capacity between the two cards are different. Now changing things in addition to this is where things get confusing and I think this is where people have a point of contention. They are not wrong for feeling this way, just like you are not wrong for expecting prospective buyers to research products they intend to spend their hard earned money on.

 

Now I do disagree with anyone that believes these suffixes shouldn't exist in general, you have to give these companies something to market with and they exist to describe the physical condition of the product. If every card was called "RTX 3080" on a website but looked differently, it would be just as problematic for customers as you'd have no means of finding the specific card aesthetic you are looking for, or end up overspending on a subpar 3080 when an overclocked version is similarly priced or cheaper.

 

People must also understand that there is always inherent risk with product nomenclature that is bound to confuse the lesser-informed. When my company launched a system containing the RTX 2070 Super Mini from ZOTAC, we had so many customer complaints about us using a "RTX 2070 Super Mini" instead of an RTX 2070 Super, despite it being the exact same card, lol. At some point, we must all agree that no matter how hard we try as a people to make something idiot proof, they will always build a better idiot.

[leadeater]

47 minutes ago, LAwLz said:

I think the naming is bad, but I also don't think this is an issue because it will affect next to nobody.

I just don't like the inconsistency where this very minor issue gets multiple news articles written about it, this thread with almost 150 posts (30 of which are from you) and a bunch of people calling for Nvidia to be sued, how they will never buy from Nvidia, etc.

There are so many replies literally because of you and others like you knowingly arguing against what IS a bad product name that you know could be better because you want to defend against "pile on Nvidia bad" for goodness knows what reason. 

 

The name is bad, the reasoning you want to try and defend it is bad and is also historically not even correct.

 

So if we all know it's bad and could have been better then why are we here... Right? So lets just move on. Like I said you're being more of a problem than Nvidia has been in this situation. They have a right as pointed out to name their products like crap and we have a right to say it's a bad name. You don't have to jump off the deep end to be contrarian and the defender of all who are "unjustly" being slandered. What is so problematic is you know the name could have been better chosen, so what on earth are you even doing? 

 

My strong reaction is specifically at YOU, not Nvidia. 

[leadeater]

1 hour ago, LAwLz said:

The reactions are so over the top. But when other products have bad names nobody cares. Hell, even names I think are bad from Nvidia didn't get this much attention. 

GTX 1660 vs GTX 1660 Ti vs GTX 1660 Super? Nahh, not at all confusing and doesn't need to be pointed out. I am sure nobody will be confused by that.

RX 6800 vs RX 6800 TX? Totally not confusing at all. No way we should point out that they could have named it the RX 6850 instead to make things easier to understand. Nope, no way we should point that out. That product gets a free pass.

 

On 12/4/2022 at 1:36 AM, leadeater said:

Similarly I also find Ti and Super unnecessary and stupid but at least there is historic understanding that these denote performance.

 

Have you finished being incorrect?

 

Have you finished not reading the actual arguments and reasons put forward because it's very obvious you are not reading them.

 

Once I see you at all address established standardized naming then and only then will you be in a position to have this discussion. Harping on about not being able to read more than 4 numbers makes you look as if you're incapable of common debate etiquette. If you aren't going to acknowledge points raised then you have no purpose being in the debate at all. '

 

Similarly you are going on about performance being what should be used for different modeling naming when I very clearly talked about GPU die core configuration and memory bus which effect performance, and it's those origin factors for the performance difference that are what is behind model naming not specifically and only performance.

 

So that's my offer, come to the debate table prepared or don't come.

[sof006]

On 12/5/2022 at 3:37 PM, Stahlmann said:

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.

While that is true, the uptake on anything Ray Traced is still super slow and i've looked at the performance differences and in reality AMD isn't that far behind. I bet the new AMD cards will be a massive improvement.

 

 

[LAwLz]

44 minutes ago, leadeater said:

-snip-

You sound pretty emotional and upset about this. Maybe you should take a while off, cool down and come back to this topic when you have calmed down a little?

 

 

I don't think your argument of "this is okay to do because of history" is a valid argument. For example AMD only started using XT on their 5000 series of GPUs, and yet nobody had any issues with that. I tried looking on this forum and after 15 minutes of searching I could not find anyone who raised even a single objection to that poor naming scheme, and I think that is worse than 8GB vs 12GB, because at least the average Joe will most likely reach the correct conclusion (for the wrong reason) in that regard.

 

 

So do I understand you correctly. You think that GPU numbers should be based on the GPU die in combination with the memory bus. Correct? If either of those things change, then it should have a new name. If neither of those are changed by something else is changed, then it's fine to retain the same product number and just add a suffix. Correct?

[starsmine]

And if you are to argue SE is better, it should be known that people correctly assume that means "Special Edition"
It does not mean slow edition, thats just a nick name people who already know about the card know about.

most would assume special means faster

That misleads consumers more then 8GB would, as less ram makes you assume lesser. 

[bramturismo]

20 minutes ago, starsmine said:

And if you are to argue SE is better, it should be known that people correctly assume that means "Special Edition"
It does not mean slow edition, thats just a nick name people who already know about the card know about.

most would assume special means faster

That misleads consumers more then 8GB would, as less ram makes you assume lesser. 

I'd like to chime in here and say that I know absolutely squat about what an iPhone SE is, other than it being a cheaper version.

[rikitikitavi]

Since most disappointment (afaik) comes from the gaming community, there should be a new naming scheme forgoing all useless 6600xt/3060super bs... it is called CyberpunkBenchmarkAv1440pMax or whatever, so Nvidia 250 and AMD 300 - no more confusion lmao. You buy FPS, nothing else matters.

[Dabombinable]

9 hours ago, starsmine said:

And if you are to argue SE is better, it should be known that people correctly assume that means "Special Edition"
It does not mean slow edition, thats just a nick name people who already know about the card know about.

most would assume special means faster

That misleads consumers more then 8GB would, as less ram makes you assume lesser. 

LE is what Nvidia used to use, as well as XT... SE however was widely used by Ati and even to some who weren't tech savvy it became also known as "Shit Edition"

[Stahlmann]

12 hours ago, sof006 said:

While that is true, the uptake on anything Ray Traced is still super slow and i've looked at the performance differences and in reality AMD isn't that far behind. I bet the new AMD cards will be a massive improvement.

AMD isn't that far behind?

 

Take the 3080 10G and 6800 XT for example. In rasterization they're roughly the same which is why i'm making this comparison. Now turn on ray tracing effects and the 3080 will always be significantly ahead, up to 50% depending on the game and the extent of the ray tracing implementation. That's a huge difference, normally such a difference would place the Nvidia card multiple performance tiers above the AMD card when we'd talk about raster.

 

Until AMD cards can hold a candle to Nvidia's ray tracing performance i won't buy their GPU's. I don't buy stuff from a brand just because they currently have less controversy going on. If i'd stop buying stuff from every brand i've had a bad experience with or heard bad things about, then frankly i wouldn't even be able to build a PC anymore. I shop by what products makes the most sense for me. And you should too. In the end none of these companies we're talking about care about our well being. They're in it for the money, and that's it.

 

In a few years down the line we will reach a point where practically every GPU is "good enough" at raster performance. And at this point the dividing factor will be ray tracing performance. Just like nowadays no one bothers to benchmark 2D games anymore, because everyone knows practically every GPU can run them no problem. That's where we're heading - like it or not. And if AMD can't pull it together by then, they will fall behind and lose most of the market share they're trying to gain right now.

[LAwLz]

5 hours ago, HenrySalayne said:

Textbook example of the fallacy of relative privation...

No it's not.

 

In this thread, I am saying "I don't think this is an issue so I don't get why it gets so much attention. If we buy the argument that people present then shouldn't that argument also apply to situation X, Y and Z?"

 

It would only be a fallacy of relative privation if I agreed that this was an issue, but told people to ignore it and talk about something else instead.

 

 

I am pointing out hypocrisy because so far I haven't seen anyone complain about the things I highlighted. I also think the level of attention things get are way out of proportion.

[LAwLz]

14 hours ago, MageTank said:

I do not disagree with you, but I do want to say that in @leadeater's defense, I have seen him complain about these very things.

I think you are getting leadeater mixed up with someone else.

I have gone through his posts and I have not found a single one where he says the XT branding or the 1660 naming are issues. He did mention Ti and Super (but not XT) earlier in this particular thread but gave them a pass because "history".

 

 

 

14 hours ago, MageTank said:

Based on what I am reading, both you and Leadeater appear to be arguing for the same thing (though with differing levels of passion). Seems like we all agree that the current naming conventions suck across the board and not just for Nvidia. Anyone that remembers me during the Zen chipset launches knows I was not a fan of AMD's chipset naming convention being intentionally similar to Intel's but only being off by a single digit, resulting in mass confusion and people purchasing incompatible boards with the wrong vendors processor. This issue does not plague only the GPU segment and needs to be addressed across the industry.

I don't think we are.

Leadeater thinks that Nvidia are:

• Creating an issue.
• Trying to mislead and rip people off.
• That they don't give a damn.
• Putting in effort in an attempt to misinform customers.
• Says that people should not have to rely on suffixes to understand performance. Only the first part of a name matters.
• Thinks that it should be called the RTX 3050 Ti.
• Thinks that this may potentially be illegal or should be illegal.

 

I think that:

• This isn't an issue.
• That next to no one will be negatively affected by this naming scheme.
• That an enthusiast, which let's be honest are most people buying computer parts off the shelf, will most likely find reviews and get an accurate depiction of the performance they will get.
• That an average Joe looking to buy a 3060 will assume the 8GB model performs worse than the 12GB model because "larger number is faster", which in this particular case is correct, but for the wrong reasons. Thus will not actually cause any harm.
• This card shouldn't be called the 3050 Ti because I think model numbers should be based on the GPU die, not performance. If we start arguing that performance is what defines if something is a 3060 or 3070 then we will probably end up in situations where mini variants of a card and really high end overclocked versions of a card will end up being named different things because "they perform differently". A factory overclocked 3060 shouldn't be called a 3060 Ti or 3070 just because it might perform within spitting distance of a very poor 3060 Ti. Performance should not dictate product names. Memory shouldn't dictate it either because if we go with that logic we end up with stuff like the RX 580 4GB and RX 580 8GB needing different names. To me, the only logical way to have any consistency is to say that the GPU die is what matters, not performance or memory. My suggestion also allows for consistency between laptop and desktop models. 

 

 

14 hours ago, MageTank said:

My two cents on this 3060 8GB situation (not that anyone asked for it) is that Nvidia is well within their right to name the products as they deem appropriate.

I don't really like this argument. I typically find the whole "well they are legally allowed to do this" argument to be kind of a cop out. I know you elaborate more and give your own opinion later in the post, but I just wanted to add this little remark.

 

 

14 hours ago, MageTank said:

Those of you on a crusade to save the ill-informed from making a mistake during purchase, put that energy towards something that will actually matter. Yelling at Nvidia and the other companies from a tech forum they don't frequent likely won't reach their ears, but you can certainly reach the ears of their customers if you create guides, write reviews or get the word out to reviewers that can reach those customers.

My stance that I don't think these potential "ill-informed customers" actually exist, and if they do exist they are probably a very small group that might as well fall into the "ignorant beyond saving" group you mention later in your post.

 

 

14 hours ago, MageTank said:

I don't think people would have a problem with this if retail stores publicly advertised ROP/TMU/SM counts on their specification pages, but almost no retailer has a fleshed out specification tab for GPU's. If I am Joe Everyman and I google "8GB vs 12GB VRAM" I'll likely find an explanation about frame buffer sizes and impact of running out of VRAM, but nothing about similarly classed cards having completely different SM configurations or bus widths which would have a far more significant impact on performance than simply having less VRAM at a given resolution.

I just tried this and the information I got when googling "8GB vs 12GB" was actually fairly decent and would have pointed me, as an uneducated customer, towards the 12GB card. Especially if it is true that the price is the same for both cards which some have claimed.

I think expecting the average Joe with next to no understanding of computers to buy PC parts is already a fairly big stretch, but to also expect them to research how frame buffers work is an even bigger stretch. My guess is that 99% of the average Joes who don't look up benchmarks will go "oh, 12 is a bigger number so it's probably better", which in this case is true.

 

 

14 hours ago, MageTank said:

Now I do disagree with anyone that believes these suffixes shouldn't exist in general, you have to give these companies something to market with and they exist to describe the physical condition of the product. If every card was called "RTX 3080" on a website but looked differently, it would be just as problematic for customers as you'd have no means of finding the specific card aesthetic you are looking for, or end up overspending on a subpar 3080 when an overclocked version is similarly priced or cheaper.

Agree, which is why I am against statements like this:  

On 12/4/2022 at 12:27 PM, 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.

 

I also find the whole "only the first part of a name matters!" completely illogical when the suggested solution is to call the card the 3050 Ti, which has another suffix on it. A suffix that I think even less ill-informed customers would understand.

 

I think it is easier for the average Joe to understand "12GB means it is faster than 8GB" than it is to understand "Ti means it is faster than non-Ti", but both cases hinges on the idea that customers read more than the first part of the name, which the same people arguing for "add Ti to the end" says they don't. It makes no sense to me.

[MageTank]

6 hours ago, LAwLz said:

I think you are getting leadeater mixed up with someone else.

I have gone through his posts and I have not found a single one where he says the XT branding or the 1660 naming are issues. He did mention Ti and Super (but not XT) earlier in this particular thread but gave them a pass because "history".

You're right. My dyslexia strikes again. It was back during the 1060 3GB/6GB debacle:

 

6 hours ago, LAwLz said:

I don't think we are.

Leadeater thinks that Nvidia are:

• Creating an issue. 
• Trying to mislead and rip people off.
• That they don't give a damn.
• Putting in effort in an attempt to misinform customers.
• Says that people should not have to rely on suffixes to understand performance. Only the first part of a name matters.
• Thinks that it should be called the RTX 3050 Ti.
• Thinks that this may potentially be illegal or should be illegal.

 

I think that:

• This isn't an issue.
• That next to no one will be negatively affected by this naming scheme.
• That an enthusiast, which let's be honest are most people buying computer parts off the shelf, will most likely find reviews and get an accurate depiction of the performance they will get.
• That an average Joe looking to buy a 3060 will assume the 8GB model performs worse than the 12GB model because "larger number is faster", which in this particular case is correct, but for the wrong reasons. Thus will not actually cause any harm.
• This card shouldn't be called the 3050 Ti because I think model numbers should be based on the GPU die, not performance. If we start arguing that performance is what defines if something is a 3060 or 3070 then we will probably end up in situations where mini variants of a card and really high end overclocked versions of a card will end up being named different things because "they perform differently". A factory overclocked 3060 shouldn't be called a 3060 Ti or 3070 just because it might perform within spitting distance of a very poor 3060 Ti. Performance should not dictate product names. Memory shouldn't dictate it either because if we go with that logic we end up with stuff like the RX 580 4GB and RX 580 8GB needing different names. To me, the only logical way to have any consistency is to say that the GPU die is what matters, not performance or memory. My suggestion also allows for consistency between laptop and desktop models. 

This is what I meant by "differing levels of passion". Both of you have very differing opinions about Nvidia's intent and what impact the naming convention has on customers, but both seem to agree that the naming convention needs improvement and should change. At the core of both arguments lies the same solution, regardless of how different your plans are to arrive at that point. If Nvidia implemented your idea to base the naming off the die, I doubt Leadeater would have a problem with that and it would likely resolve the points you highlighted from his post.

 

Now if you want my personal opinion on the matter, I do agree and disagree with both of you on several of the points noted in your summary.

• Nvidia is indeed creating an issue. However, it's NOT just Nvidia creating an issue. GPU descriptions in general from ALL sides are annoying to deal with and processors are in the same boat now too. I am not even talking about potentially misleading customers (intent is hard to prove), I am simply talking from a support perspective. If I am a game developer and I list my minimum specifications as "GTX 1060" but did so when the 6GB model was launched, is the burden on me to go back and change my minimum specifications to clarify the 6GB model because I couldn't anticipate Nvidia releasing a 1060 again with slightly reduced specs? If I am Microsoft and I say Windows 11 supports the Athlon 3000G (Picasso) but AMD re-releases the Athlon 3000G with 14nm Raven Ridge, is it on me to go back to clarify that we support the 12nm version, not the 14nm version? I dislike having problematic naming conventions forcing support teams to go back and adjust things just because we couldn't slightly modify the part number on launch. Not to mention, what customer is going to check between part numbers YD3000C6M2OFH and YD3000C6M2OFB to determine if they have the correct Picasso vs Raven Ridge CPU?
• I disagree that next to no one will be negatively affected by the naming scheme, but I don't think the impact is as large as what is being portrayed. Again, we already had people on this very forum asking about the differences between the GTX 1660, Ti and Super to discern which was fastest. I also know first hand that enthusiasts are not the only people buying hardware off shelves. I've shopped at my local Micro Center store dozens of times (I basically live in their water cooling section rent free) and I hear plenty of moms asking about parts to buy for their kids birthday or holidays. Now, where I do diverge from Leadeater's beliefs is that I think this is on the store staff to be trained to qualify the customer and get them the best hardware for their dollar. Component naming conventions should not have an impact if the sales team is well educated (which they should be). One could argue that some people may not seek help and simply buy based on box art and what "looks" good, but I am also in the camp that consumers have to try to educate themselves before spending their hard earned money.
• I agree with you that the average customer will likely assume 8GB is a lesser number and will therefore automatically assume it is worse. However, you have to remember that people in the market for budget-mid tier hardware are likely trying to save some money and will probably just google "8GB vs 12GB VRAM" and not be informed of the different bus widths or in some situations, differences in SM/TMU/ROP counts. Now most places offer a fair return/exchange policy so I cannot image someone will be permanently affected by choosing the wrong card, so I'll concede this is likely not a big deal.
• I completely agree with you on naming the GPU's after their dies. Honestly, it would make rebrands much easier to spot and finding better value from cards based on what they are cut down from.
• Leadeater is correct that intentionally misleading customers is illegal, the problem is proving intent. Intent to deceive is tricky because it implies you are aware what you are doing is deceptive (which they probably are) and that what you are doing is indeed deceptive (which some can argue for and against). Do I personally believe Nvidia is being deceptive? Not really, at least not any more than the other companies playing this game (AMD's chipset marketing team for example). The reason behind my line of thinking is that the name "RTX 3060" doesn't mean anything in and of itself. If you take this to court and ask a judge or jury what does "RTX 3060 mean", they likely won't have an answer. By the letter of the law (at least here in the states), Nvidia would not be subject to any litigation based on their actions. Can we say it's sketchy and shady behavior? Sure, I am definitely in that camp because there is literally nothing stopping them from coming up with a new name, but I don't think what they are doing is inherently illegal, at least not yet. I would argue Intel was in far worse of a position when they had the 7640X and 7740X on the X299 platform as MOST customers would not know based on the naming conventions that using those processors on X299 would kill off memory channels and PCIe lanes, yet Intel got through that just fine.

 

7 hours ago, LAwLz said:

I don't really like this argument. I typically find the whole "well they are legally allowed to do this" argument to be kind of a cop out. I know you elaborate more and give your own opinion later in the post, but I just wanted to add this little remark.

Fair enough. Capitalism is etched into my brain, this line of thinking is common here, lol.

7 hours ago, LAwLz said:

I just tried this and the information I got when googling "8GB vs 12GB" was actually fairly decent and would have pointed me, as an uneducated customer, towards the 12GB card. Especially if it is true that the price is the same for both cards which some have claimed.

I think expecting the average Joe with next to no understanding of computers to buy PC parts is already a fairly big stretch, but to also expect them to research how frame buffers work is an even bigger stretch. My guess is that 99% of the average Joes who don't look up benchmarks will go "oh, 12 is a bigger number so it's probably better", which in this case is true.

When I googled 8GB vs 12GB VRAM, it pointed me towards a reddit thread recommending the 8GB 3060 Ti: 

 

Now we could assume that once this card launches, there will be better information out to inform potential buyers, but as it stands, I could potentially be mislead based on the logic followed in that reddit thread.

 

Now let me preface my bias by saying I am in the PC integration business so I am very guilty of preying on the ignorance of customers to sell my systems. If everyone knew how to build a PC, I'd be out of a job. Now with that said, I do not believe it is uncommon for average joes to buy individual PC components. I mentioned it in my wall of text above, but people do so as gifts all of the time. Now they often rely on the expertise of sales associates, so I would hope that they are in good hands if they are doing so. If you are walking into a computer store with the mentality that you know more than the sales rep and you end up making the wrong purchase, you deserved that karma, lol.

 

7 hours ago, LAwLz said:

I also find the whole "only the first part of a name matters!" completely illogical when the suggested solution is to call the card the 3050 Ti, which has another suffix on it. A suffix that I think even less ill-informed customers would understand.

 

I think it is easier for the average Joe to understand "12GB means it is faster than 8GB" than it is to understand "Ti means it is faster than non-Ti", but both cases hinges on the idea that customers read more than the first part of the name, which the same people arguing for "add Ti to the end" says they don't. It makes no sense to me.

I agree, which is why I think your idea of calling the GPU's by their dies is the most logical solution to this problem. That is, until companies decide they want to troll each other with very similar GPU names like AMD did with the motherboard chipsets. Then we have a right to raise some pitchforks, lol.

[LAwLz]

17 hours ago, MageTank said:

I agree, which is why I think your idea of calling the GPU's by their dies is the most logical solution to this problem. That is, until companies decide they want to troll each other with very similar GPU names like AMD did with the motherboard chipsets. Then we have a right to raise some pitchforks, lol.

The thing is though, both of these cards (the 3060 8GB and 3060 12GB) seem to have the same die.

Both of them uses the GA104 die (edit: sorry, GA106), which is why I think it makes sense to call them both the 3060.

[MageTank]

Just now, LAwLz said:

The thing is though, both of these cards (the 3060 8GB and 3060 12GB) seem to have the same die.

Both of them uses the GA104 die, which is why I think it makes sense to call them both the 3060.

Would be very unlikely if both had the exact same full die. Typically when Nvidia cuts down a card by ANY measure, it gets a slight change. They'd call it the "GA104-150-A1" for the 12GB, and likely cut this 3060 8GB down into something else, like a GA104-125-A1 or something along those lines. Based on the fact that its 8GB VRAM and also exactly half the bus width of an 8GB 3070, I am going to assume this is Nvidia cutting down some very defective 3070's and fusing off half the memory controllers and a significant portion of the SM's.

 

It is very possible I may have misinterpreted your desire to name these after the dies, as I was thinking the full die, not just GA104 vs GA102, etc. Though I also understand nobody is going to want to walk into a PC store and ask if any GA104-125-A1's are in-stock. Especially with my hillbilly accent, that would be a mouthful.

 

Perhaps the solution would be both a change to the die naming convention and then naming the cards after the revised die naming convention? Using Nvidia as an example, they could do a G for Geforce or Gaming, whatever the G stands for. Follow that letter up with the architecture (though this gets confusing when you have Ada and Ampere), followed by a number. Let's say 1-5 to keep it simple, 1 being their budget number, 5 being their halo tier products. You want a 3080 Ti? You have a GA550. A 3080? GA500. A 3070? GA400. 3060? GA300. If you want to add products in between, modify the last two digit numbers. If you keep this rule consistent, people will always know the performance class of the product based on the architecture and numeric performance class. As long as you do not make a GA300 that is slower than a GA300, but call it GA350, you'll be fine, lol.

 

This also opens up better naming conventions for Nvidia's other product stack. Quadro cards from the Ampere architecture could be called QA500,QA400, etc. The same with their Tesla series (if those still exist). 

[LAwLz]

29 minutes ago, MageTank said:

Would be very unlikely if both had the exact same full die. Typically when Nvidia cuts down a card by ANY measure, it gets a slight change. They'd call it the "GA104-150-A1" for the 12GB, and likely cut this 3060 8GB down into something else, like a GA104-125-A1 or something along those lines. Based on the fact that its 8GB VRAM and also exactly half the bus width of an 8GB 3070, I am going to assume this is Nvidia cutting down some very defective 3070's and fusing off half the memory controllers and a significant portion of the SM's.

 

It is very possible I may have misinterpreted your desire to name these after the dies, as I was thinking the full die, not just GA104 vs GA102, etc. Though I also understand nobody is going to want to walk into a PC store and ask if any GA104-125-A1's are in-stock. Especially with my hillbilly accent, that would be a mouthful.

 

Perhaps the solution would be both a change to the die naming convention and then naming the cards after the revised die naming convention? Using Nvidia as an example, they could do a G for Geforce or Gaming, whatever the G stands for. Follow that letter up with the architecture (though this gets confusing when you have Ada and Ampere), followed by a number. Let's say 1-5 to keep it simple, 1 being their budget number, 5 being their halo tier products. You want a 3080 Ti? You have a GA550. A 3080? GA500. A 3070? GA400. 3060? GA300. If you want to add products in between, modify the last two digit numbers. If you keep this rule consistent, people will always know the performance class of the product based on the architecture and numeric performance class. As long as you do not make a GA300 that is slower than a GA300, but call it GA350, you'll be fine, lol.

 

This also opens up better naming conventions for Nvidia's other product stack. Quadro cards from the Ampere architecture could be called QA500,QA400, etc. The same with their Tesla series (if those still exist). 

I am fairly sure it is the same.

On TechPowerUp they are listed as GA-106-302-A1 and GA106-300-A1, but when you look around on Google it seems like even some newer 12GB cards uses the 302 die.

It seems like the 302 die revision was done because of mining-related things, and all new cards uses the 302 die. The same die the 3060 8GB is reported as using.

 

Maybe I am completely wrong, but couldn't it be the case that the 8GB card simply has fewer memory chips connected, and as a result not the entire memory bus is used?

 

The 12GB model has 6 memory chips, 2GB each, and a 32 bit bus each connecting them to the GPU. That results i 12GB of memory connected on a 192 bit wide bus.

The 8GB model could have 4 memory chips, 2GB each, and a 32 bit bus each connecting them to the GPU. That results i 8GB of memory connected on a 128 bit wide bus.

 

The GPU remains unchanged.

 

 

Considering the lack of press release and seemingly lack of availability, this might just be an OEM card that some vendors requested. I find it hard to believe that Nvidia would create a a die just for this seemingly low volume part. Seems more likely that they would just let ABIs not connect all the memory chips if they didn't want to.