Continuing onwards from Intel’s laughable presentation we move towards NVIDIA to seen what kind of wares Jensen Huang is trying to peddle this time around.
Starting off nice and quick, NVIDIA have announced its entry-level product in the RTX 3000 lineup that consumers can’t actually buy: The RTX 3060.
The GeForce RTX 3060 offers all the gimmicks of Ampere at a supposed modest price, with this new graphics card having the MSRP of just $329 USD and will be made available from late February
However $329 isn’t actually going to be its price-tag, I’m sure we’ve all been made clear on that regard with both NVIDIA and AMD graphics cards released this generation being hardly available to purchase in which case retailers or even AIB partners have increased the overall retail price significantly, and dumb fucks are still paying for it. You see, in this modern era merely buying a graphics card at $RRP would actually be considered a bargain.
And with suspicion arising on team green for selling products directly to miners doesn’t really bode well for anything PC related at the moment.
The GeForce RTX 3060 comes equipped with just a lowly 26SM’s giving a CUDA core count of 3584 (1792) along with further cuts to its RT cores and worthless Tensor cores as well.
The X60 series from NVIDIA is by far the most popular, being an entry / mid-range product that once was actually affordable to purchase it’s no wonder why so many people opt to buy hardware within the reasonable price bracket of $249, or in this case thanks to garbage business practices and retarded consoomers. What once used to be $249 with the GTX 1060 has now become a $329+ staple.
Maybe NVIDIA should castrate Ampere like they did to Turing and release worthless garbage GTX componentry again. If it’ll make a buck I’m sure Jensen would be more than happy to release a SKU every $49.
12GB of VRAM is actually impressive, and just goes to show how little respect NVIDIA has for their high-end customers who’ve since been stunted with inferior quantities of onboard memory. As the 3070 and 3080 have less overall memory compared to this entry-level card.
However there needs to be a leach on this “beast” so the RTX 3060 and there is, the RTX 3060 is effectively neutered by a 192-bit memory interface however to keep the bandwidth more inline with the 3060Ti it actually offers faster memory chips.
However in the only metrics of performance we have for this entry level Ampere showcases its performance relative to previous generation hardware of its given SKU designation such as the Turing based RTX 2060 and Pascal based GTX 1060 which obviously falls way below expectations.
Like always, majority of the testing conducted happens to be on titles that prioritize NVIDIA’s lovely gimmicks such as DLSS and RTX Ray Tracing, which in comparison to the RTX 2060 which actually offered 30 SM’s minus the half-rate FP32 it’s not really all that exciting, certainly not worthy of dumping Turing for Ampere which has been the case for majority of the Ampere lineup unfortunately.
With testing conducted at a lowly 1080p resolution it’s unsurprising that NVIDIA have conducted these RTX ON tests with DLSS being enabled simultaneously at even such a low resolution these low-end RTX graphics cards simply do not have the RT horsepower to provide meaningful performance with Ray Tracing enabled.
Hence the requirement to enable DLSS as such a low native resolution.
RT bullshit aside the GTX 1060 is woefully unequipped to touch either the 2060 or the 3060, no surprises there considering it does utilize an effectively different architectural approach than any Volta based derivatives and merely has only 6GB of GDDR6 memory to spare. Comparative to the RTX 3060 on the only two games you can accurately compare from an eyeball perspective the 3060 simply flogs the Pascal based GPU by 102.5%.
However, compared to its previous generation the 3060 doesn’t look to be all that impressive unless of course you’re utilizing RTX and DLSS extensively which are practically featured in a limited amount of titles and are complete bullshit as you can’t enable RTX without DLSS being enabled lest you don’t actually enjoy gaming with a modest framerate, even at 1080p.
From an eyeball since there are no real figurative numbers displayed in Total War: Three Kindoms the 3060 looks to be only 8.62% faster than the RTX 2060, while in The Division 2 it’s as high as 17.14%.
Comparing RTX and DLSS being enabled at the same time, both cards offer garbage performance figures with RTX ON natively however I’m more than certain that Ampere can certainly handle the task more so than Turing ever could. Shocker I know.
With SOTR performance leaning around the 14.51% mark, comparing a newer title such as Watch Dogs Legion shows us the biggest performance gap at 52.5% while Fortnite and Cyberpozz 2077 show 32.69% and 27.08% respectively which means that the GeForce RTX 3060 is on average 25.42% faster than its predecessor.
Which seems like a lot, but that’s because it is. However the average when comparing the only two instances where RTX and DLSS aren’t enabled shows just a 11.56% gain in favor of the 3060 which isn’t anything to write home about at all.
It’s honestly quite pathetic if such an entry-level component can’t be previewed with RTX enabled at 1080p resolution without having to resort to artifact inducing tactics like DLSS. And even then the performance figures vary everywhere from 60-70 fps at that resolution.
Give it time for new titles to release to put Turing back into its place just like Pascal has been rendered obsolete the day it became EOL.
Moving onwards, the previous leaks of mobility GPUs of the RTX 3070 and its further gimped cousin have come into fruition. With NVIDIA announcing three mobility graphics cards, such as a mobile variant of what bullshit NVIDIA has the audacity to call a 3080, a cucked 3070 and an RTX 3060 with 30SM’s and 6GB of VRAM because I just fucking can’t anymore with this company.
Seems NVIDIA will not and can not allow the goyim to have quantities of VRAM without a catch.
Starting off is the supposed “flapship” of the desktop “flagship” which wasn’t the flagship but is now based on GA104 which is in lines with a desktop 3070, the GeForce RTX 3080 Mobile GPU.
Featuring the same 48 SM’s as previously discussed providing 6144 (3072) CUDA Cores it offers a base frequency of 1245MHz and a boost clock of 1710MHz and is equipped with 16GB of standard GDDR6 memory on a 256-bit memory interface. It’s unknown whether NVIDIA are utilizing 12Gbps or 14Gbps memory chips but even still this RTX 3070 at heart offers double the VRAM of its desktop equivalent and 6GB more than the real RTX 3080.
What the fuck is Jensen thinking?
The RTX “3080” Mobile will come in Low-IQ and Max-P variants individually tailored to the specifications of the partner with TGP ranging anywhere from 80W to 150W. NVIDIA claims that notebooks with the RTX “3080” will be starting from just $1999.
AND IT COMES WITH A FUCKING TV.
Secondly we have the RTX 3070 mobile which has a further 6 SM’s disabled compared to its desktop counterpart equaling 40 SM’s in total giving you 5120 (2560) CUDA Cores with its 256-bit memory bus it comes equipped with 8GB of GDDR6 memory with base and boost frequencies of 1290MHz and 1620MHz. This is essentially a castrated “3080” and I’m sure if you were to put this hot potato inside a fridge you’d probably get roughly similar performance out of it.
With Max-Hue cards coming with a 80W TGP and Max-P cards coming with “up to” 125W it’s slated that RTX 3070 laptops can be had starting from $1299.
Lastly we have the RTX 3060 mobile which is a true testament to how indecisive NVIDIA are when it comes to VRAM allocation and fucking over consumers at every which way they possibly can.
This here mobile RTX 3060 actually offers 30 SM’s giving you 3840 (1920) CUDA Cores which is MORE than the standard desktop variant but don’t be fooled, as this fucking disgrace only comes equipped with 6GB of VRAM which is far below the requirement for modern gaming, even at 1080P.
You can just forget about even considering anything RTX with this piece of shit as 6GB is nowhere near enough aside from the lackluster RT performance the GPU itself has.
The 6GB of GDDR6 runs with the same 192-bit memory interface as the desktop card but run at 12Gbps resulting in a effective bandwidth of just 288 GB/s with base and boost frequencies topping out at 1283MHz and 1703MHz.
With TGPs ranging anywhere from 60W to 115W this will certainly be the most efficient Ampere based mobile card, and also by far the worst.
NVIDIA themselves are desperate to gain traction and attention for that matter on their mobile efforts with the claim of the mobile RTX 3060 being 1.3X faster than a Soyny PlaugeStation 5 which gives off the same levels of satisfaction by pretending to be retarded and winning gold in the special Olympics.
However as the hardware inside the notebook ages, the same cannot be said for consoles and their vastly superior optimization. In a matter of years with the current crop of consoles are halfway through their lifecycle I can guarantee you that even the PS5 would reign supreme over this 6GB turd.
All the while the Xbox Series X is more than certainly faster than the RTX 3060 mobile with both the PS5 and XBOX Series X fighting it out for the price of $499 (if you could actually buy them) it would make sense considering that laptops are just consoles with a free TV.
Targeting 1080p 90fps with prices ranging from $999 and above mobile devices utilizing the RTX 3060 GPU aren’t all that competitive to a console as it too offers everything you’d need barring the screen.