OK, 16gb liau, no more 8gb.


Actual pcb is half the GPU heat sink.


2 heatsinks with 4 pipes.


Why? Why put the power so close to the bracket?


Back plate is metal. Unmarked warranty sticker on screw.

Then?

so slim, my 3070 more bulky

How much? Msrp?

will be better buy than coloful 5060ti nb duo (with only 2 heatpipes). fark colorful refused to answer my email, all i asked were test data from their end(temperature, fan speed)

No bad. 4k, 14700f, 128GB RAM. Max settings with DLSS at Quality.

This post has been edited by samftrmd: Jun 19 2025, 02:33 PM

Lucky the GPU cable is long enough.

Wah, mine confirm cannot reach that 8-pin connector.

Overpriced Jensens shit, cant even match up much less beat 4070.

It is better if you run AI models.
- You can run NVIDIA Broadcast for noise cancellation when playing games while chatting. The difference of voice output is obvious. That model alone will eat up 2GB of VRAM.
- You can fit in more context when running LLM. 12GB VRAM is pretty much the limit for running a okay-ish model, i.e. 12B Q4+small context will hit it, since Windows alone also consume up quite a bit of VRAM. 16GB VRAM can let you fit in 14B model and slightly more context.

This post has been edited by kingkingyyk: Jun 20 2025, 09:30 AM

Might as well call it an AI card if it performs better in AI than it does in gaming

In cases where 12GB VRAM can fit in the model, 4070 will perform better due to higher memory bandwidth.

TS,
how is PNY distributor in Malaysia and APAC?
RMA fast or slow?

EVGA stop making GPUS after 4000series. They seems to be winding down their other businesses.

lol I just got MSI punya

bought impulsively now after reading reviews kind of regret already

A forumer just recently got his PNY card back after >2mths RMA process. Fcuk PNY!