- Joined
- Jan 20, 2004
- Messages
- 34,827
- Reaction score
- 27,887
Nvidia announces the $399 RTX 3060 Ti —and we’ve tested it
At $399 this thing absolutely shits on the value of the RX 6800 at its $579 MSRP (when it becomes available and actually reaches this price).
*Note: the Memory Clock here is quoting Ampere's GDDR6 bandwidth. That's not what I'm used to seeing quoted under that label. More traditionally speaking, the Memory Clock is 1750 MHz for both (which translates to an effective transfer rate of 14 Gbps since it is GDDR6.
The 3DMark gives a generally accurate idea of relative gaming power:
TPU's gaming roundup showed an even smaller overall discrepancy @1440p of -20% for the 3060 Ti. Note that it's actually outperforming the RTX 2080 Super:
https://www.techpowerup.com/review/nvidia-geforce-rtx-3060-ti-founders-edition/
Although in the rare case you can turn on DLSS 2.0, a list that is growing over time, the 3060 Ti suddenly stomps the RX 6800 out (even in Watch Dogs which was a title where the RX 6800 otherwise enjoyed an inordinately large advantage):
I have no clue what the fuck AMD is doing. They got a little high on themselves. They must have had lingering beer goggles from their Zen 3 celebration, and mistook Huang Jensen rolling up in his NVIDIA-embossed leather jacket for an Intel executive prior to their RDNA 2.0 launch, because their GPUs are a bit of a joke at the prices they're asking. Memo to Lisa: you're still the far-trailing underdog in the GPU game.
I hear Samsung giving Nvidia a sweet deal to use their process node. Nvidia took a gamble running with Samsung "though they had previously had good luck with the 1050ti". I don't know what Samsung run rate is or failure percentage but I imagine they are working out the issues.
It's very possible that AMD paying a higher price from TMSC to secure parts while Apple is sucking the production time on 5nm M1 chips. This is why I complain so much about the lack of competitors in the Foundries space. IBM help launch TMSC with 10 nm process and now they are running big.
The US government has to push to create more foundries to pump out competitive processes because TMSC cannot grow fast enough to do it all. Intel been failing likely due to not wanting to spend 50 billion dollars going to 7nm.
We need more capacity to handle the demands because Nvidia keeps selling out everything they make it seems they could double production and still be making people pissed off as well as AMD.
