What are the efficiency of two graphics cards?

[doopdoop]

This article was written through a translator. Please excuse me.

I upgraded from gtx 970 4g to 1080ti 11g and it brought about 2.5 times the training speed.

But I found a better graphics card because the time was limited.

So the countermeasure found is to have another 1080ti or buy 2080ti.

Two 1,080 ti are about $800 and one 2080 ti is about $1,100.

Which will help improve performance for the price?

And a higher level of RAM than a 16 gigabyte RAM helps improve the speed of training?

And is 3600x or 3700x more helpful than Ryzen 1800x? I can pay $300 for this cpu upgrade.

thank you for all

[tochan]

here are some results with my 1080, 2080, 1080+2080 and 2x2080 setup.

viewtopic.php?f=16&t=130

For training and convert: 2 cards are much faster than one 2080ti i think, but for extract you can use only one GPU.

For training you can double the batch size with 2 cards(2x GPU RAM).

When you play games ore other use other "one GPU thinks" I would buy one TI RTX card.

I buy my 2x2080 RTX for 1,100... one 2080ti is 30% faster in Deeplearning. 2x2080 double the performance
https://lambdalabs.com/blog/2080-ti-dee ... enchmarks/

I'm happy with my 2x2080 RTX cards