Why does Nvidia insist on using a 256-bit bus when even low-end AMD cards use 512-bit?

Why does Nvidia insist on using a 256-bit bus when even low-end AMD cards use 512-bit?

>>>/sqt/
did this really warrant an entire thread?

well because its probably not related to the performance, or not a bottleneck. not trying to shill for nvidia tho ,maybe amd cards of similar specs are better in some programs thanks to it, but we would need benchmarks on non gimping programs/drivers.

because bus width is generally meaningless and design-dependent?

fuck off with your generalshit

>nvidiots actually believe this

Because 256-bit GDDR5X at the lowest binning (10GHz) still makes around 500GB/s effective on Nvidia cards.

AMD was forced to use a 512 bus because of their memory controller design or some hardware design error or something. They had to make up for something I can't remember and were forced to use a larger bus than required. Sorry I cant be more specific, I forgot the details.

I bet you think more Ghz makes a processor faster, too

My guess would be that it's a simple tradeoff really...

To achieve certain bandwidth to memory ether requires more lanes and thus a bigger memory controller or one with a higher clock rate that as a result runs hotter. With Nvidia having the upper hand in efficiency they can easily afford to take the latter approach.

Well he is correct here, a wider bus does allow for more data to be pushed trough each clock cycle and the width of a bus does not dictate what clock rate it has to be run at. There's no reason a 512 lane wide bus has to be run at the same clock rate as a bus half it's width.

Okay shill

No source right? :^)

explain why they used a size that no modern or upcoming card can saturate then. Use your hard, I cant spoonfeed you.

aka maths

8 gigs vram

Also 512-bit bus because 4k

Him asking for you to back up your claims with a source is hardly spoonfeeding.
>damage control

The GTX 280 was 512 bits. What's up with that? They've been shrinking ever since.

Bigger bus require bigger memory controllers which generate more heat and require more power.

Every Nvidia GPU would benefit from a bigger bus but would totally wreck Nvidia's GPU pricing line up as some cards would be too strong paired with AMD cards barely competing, people would ending up buying the cheapest card available.

is more a marketing thing to avoid cannibalizing their sales.

>this is what nvidiots actually believe

thanks for the kek.

gtx 1080 has 8 gigs and bitrate 10gbps

Gddr5

Different memory controller


>Le 6 gorillion ghz
>Same speed as lower clocked amd

A good example is HBM. Much more efficient at a lower clock rate due to a huge bus width. Unfortunately bus width isn't the major player in the FPS performance of a card as much as core arch and speed as well as other sympathetic systems such as CPU perf/arch and HDD speed.

planned obselence