Do you do not forget that time when Nvidia “unlaunched” a product, stating that regardless of being a improbable graphics card, it wasn’t named proper?… “Having two GPUs with the identical RTX 4080 designation is complicated.” That is proper, we’re speaking concerning the “unlaunching” of the 12GB RTX 4080 about two months in the past.
For these of you lower than pace, when Nvidia introduced their newest GeForce 40 sequence, they did so with three merchandise: the flagship RTX 4090, the RTX 4080 16GB and RTX 4080 12GB. The issue was that the 2 RTX 4080 variants had been very totally different and people variations weren’t restricted to reminiscence capability because the naming scheme was hinting.
That is as a result of in an effort to reduce down the reminiscence capability with out utilizing decrease density chips, you additionally want to cut back the reminiscence bus width. Utilizing totally different silicon that may end in considerably totally different efficiency was not an excellent transfer, however this type of dodgy habits wasn’t new. Nvidia had carried out it earlier than, for instance with the DDR4 model of the GeForce GT 1030.
The massive mistake was saying the RTX 4080 12GB and that individual anti-consumer transfer to the world within the greatest occasion Nvidia had held in years. Historically how they wish to deal with dodgy product releases is to sneak them out at a later date through a silent launch, which means far fewer individuals discover and even fewer appear to care.
Properly, that is precisely what they’ve carried out with the 8GB variant of the GeForce RTX 3060. Satirically, they launched this product on October 27, simply 12 days after unlaunching the 12GB 4080 for having a complicated title.
Though it is true that each the 8GB and 12GB variants of the RTX 3060 are based mostly on the GA106 silicon — so it’s technically the identical GPU — the modifications nonetheless go nicely past an adjustment to reminiscence capability. The reminiscence subsystem of the 8GB mannequin has been drastically decreased, the L2 cache is 50% bigger for the 12GB mannequin as is the bus width. All in all, bandwidth is 61% larger within the authentic RTX 3060 mannequin producing 360 GB/s whereas the “new” 8GB variant has been slashed to simply 224 GB/s — the identical bandwidth because the RTX 3050.
Finally, this implies the 8GB RTX 3060 goes to be slower than the 12GB mannequin — and spoiler alert, it is ~15% slower on common — however in some situations we’re seeing margins of over 30%, so it is a sufficiently big deal for us to convey it to your consideration.
One further word, Nvidia hasn’t made any official changes to MSRPs, so the 8GB RTX 3060 nonetheless is available in at $330. Trying over at Newegg, we will see that the 12GB RTX 3060 begins at $350, with quite a lot of fashions accessible for between $350 and $380. In the meantime, probably the most reasonably priced 8GB variant comes from Inno3D and it begins at $385, although there are some Asus fashions on backorder at $360 and $370. Level is, when it comes to pricing, there is not any distinguishing between the unique 12GB fashions and these new gimped 8GB variations.
In Australia we purchased a Galax mannequin that we bought for a ten% low cost from the unique 12GB model, and also you would possibly suppose, “Properly that is okay,” but it surely’s not. The GeForce RTX 3060 was launched nearly 2 years in the past, everyone seems to be accustomed to the way it performs, and people who aren’t can rapidly seek the advice of the bevvy of on-line benchmarks. In the meantime, this new 8GB mannequin is nowhere to be discovered as there is not a evaluation program, it is as much as reviewers to buy the cardboard and inform customers.
Actually, you may not even know there was an 8GB model or just assumed the RTX 3060 solely got here with 8GB of VRAM, simply because the 3060 Ti does. The entice right here is that unsuspecting customers may very well be baited into paying RTX 3060 12GB cash, or very close to so for a product that is considerably slower.
We requested Nvidia for an official response about why the 8GB model was launched and so they merely stated it was to supply prospects extra choices. Additionally they acknowledged that the 12GB model was going to stay accessible in parallel. And simply to be clear, that is an Nvidia sanctioned product and never their companions going rogue, fairly it is attainable companions are obligated to supply it.
Nvidia seems to be trying to filter their stock of lower-end Ampere GPUs, and are little question eager to maximise income as they accomplish that. This additionally permits them to promote faulty GA106 silicon for excess of they might get promoting it as an RTX 3050.
Our subject is in fact that we now have two merchandise referred to as “GeForce RTX 3060” and so they do not ship the identical degree of efficiency and in lots of situations it is not even shut. And we’re not speaking about reminiscence certain testing — as in exceeding the 8GB VRAM buffer — we’re speaking about efficiency when remaining nicely throughout the 8GB buffer.
So let’s check out that, we’ll go over a couple of video games after which present you a 12 sport common. For amassing these outcomes we used contemporary information with each 12GB and 8GB RTX 3060 playing cards utilizing the newest show drivers. Different GPUs are included for comparability functions.
Benchmarks
Beginning with Cyberpunk 2077, we discover that the 12GB mannequin is 13% sooner at 1080p when in comparison with the brand new 8GB mannequin, not an enormous distinction but it surely’s definitely noteworthy.
That margin grows to 18% at 1440p and now we’re wanting a two totally different tiers of efficiency and a noticeably higher gaming expertise with the unique 12GB mannequin.
Shifting on to Dying Mild 2, right here we’ve a sport that’s clearly reminiscence heavy and as such the 8GB mannequin limits the efficiency of the RTX 3060 GPU.
The unique card was an enormous 35% sooner at 1080p, that is an incredible efficiency distinction for 2 merchandise sharing the identical title, and it is the identical story at 1440p. Actually, right here we go from a really console-like 34fps with the 8GB model to a a lot smoother 46 fps utilizing the unique 3060.
Halo Infinite sees the 12GB mannequin providing 21% larger efficiency at 1080p, going from 63 fps to 76 fps, one other huge efficiency distinction for 2 merchandise that share the identical title. The margin stays related at 1440p, the place the unique 12GB 3060 was 19% sooner.
Horizon Zero Daybreak sees a smaller 16% efficiency enchancment for the 12GB mannequin over the brand new 8GB model, although that margin did develop to twenty% at 1440p.
Each outcomes are unhealthy as a double-digit change in efficiency is unacceptable, particularly given the distinction in reminiscence capability performs no position right here.
We’re additionally 22% larger efficiency for the 12GB variant in Shadow of the Tomb Raider, as the brand new 8GB mannequin drops efficiency to 85 fps. The margin is analogous at 1440p, the place the 12GB model is 20% sooner, rendering an extra 12 fps.
Efficiency Abstract
That is sufficient with the person sport outcomes, you possibly can simply get the concept. On common you possibly can anticipate 17% higher efficiency with the 12GB mannequin at 1080p, however as we noticed that margin may be as excessive as 35%…
It is a related story at 1440p, when working throughout the 8GB buffer the 12GB mannequin was nonetheless on common 18% sooner because it enjoys considerably extra reminiscence bandwidth.
What We Discovered
The brand new 8GB model of the GeForce RTX 3060 is noticeably slower than the unique, and in quite a lot of situations was considerably slower in our testing. In our opinion it is a horrible anti-consumer transfer by Nvidia, it serves little greater than to line their pockets whereas deceptive and probably ripping off prospects.
It is true that pricing may very well be improved, at which level it may not be unhealthy when it comes to worth. In any case, for instance Nvidia drops the value relative to the 12GB model by 20%, and even 30%, then the 8GB RTX 3060 would begin wanting fairly nice when it comes to worth, not less than in comparison with Nvidia’s personal product stack.
We’d agree with that evaluation had it been referred to as, for instance the “GeForce RTX 3060 LE 8GB” one thing that makes it clear to potential consumers this is not your typical RTX 3060. In that case, we would not actually have an issue with this cut-down mannequin and whereas we might nonetheless desire a reputation change completely, we might not less than settle for a extra trustworthy try at making it clear to consumers that it is a totally different model.
To exhibit simply how tough it’s for the uninitiated to even inform which mannequin they’re shopping for, or that there’s even a special mannequin, here is the field for our authentic RTX 3060 and the brand new 8GB field…
These working below the idea that an RTX 3060 is an RTX 3060 would possibly see an 8GB mannequin promoting for an inexpensive low cost and suppose they’ve stumbled upon a discount, snap it up, solely to appreciate their discount is sort of a bit slower than it could have been in any other case. And for these none the wiser it simply means you might have carried out higher, and presumably gone additional between upgrades.
Sadly this type of habits will proceed except Nvidia faces sufficient backlash from customers and the fanatic group. We’re all the time going to combat in opposition to this type of anti-consumer habits from firms, whether or not or not it’s Nvidia, AMD, Intel, LG, Samsung, MSI, Asus and so forth, many of those corporations we have taken to activity up to now, and we’ll proceed to take action.
Relaxation assured, if AMD was to tug an identical stunt we might go after them simply as onerous, very similar to once we devoted a number of articles to name them out over the plans to axe Zen 3 help for 300 and 400 sequence motherboards. We had been having none of their excuses, we did not purchase into the technical causes for why they needed to finish help. Put merely, they misled customers, we went after them, and so they corrected course, and this was made attainable because of robust push again from the group.
We do not anticipate to see the identical factor occur right here with Nvidia, however we have carried out our bit, you have now been made conscious of what the 8GB RTX 3060 is, and you already know to keep away from it at present asking costs.
