Review: GeForce RTX 3080 Founder edition

In Switzerland, price range from 750USD to 1100 USD, with most of them sitting around 850-900 USD, availability is 2-4weeks..

 

i don't mind paying for the FE's superior cooler.

 

This is cheapest ones on scan now in case anyone can't load it due to overloading... all gone and PREORDER only again!

 

The more expensive ones... I can't load overclockers at all either.

 

Check this out... overclockers.co.uk error failure due to overload...

Fatal error: Uncaught Zend_Db_Adapter_Exception: SQLSTATE[HY000] [1226] User '******' has exceeded the 'max_user_connections' resource (current value: 480)

480 user connections at once!

Their full error message from server is a goldmine for hackers actually... they shouldn't be showing all that!

 

I'm not even buying one but its funny to see the sites all crumble like this....... **insert evil laugh**

All jokes aside, I don't even think I've seen this happen before with a hardware launch and I've been on these sites since '97-98 and actually buying and building since 2001.

 

There are no actual cards in stock around here, but the price they're asking for the few models that are showing up is in the 900+ euro range for the 3080... guess it will be a while till these become "affordable".

 

Asus 3080 TUF Gaming cooler thermal imaging shots:
https://www.guru3d.com/articles_pages/asus_geforce_rtx_3080_tuf_gaming_review,8.html
vs FE
https://www.guru3d.com/articles_pages/geforce_rtx_3080_founder_review,9.html

 

ASUS did great job with TUF - I'd wish that they'd tone down the wannabe gamer look (not a fan of those tyre marks etc.) and learn from FE card. I'd buy one but since availability will suck for foreseeable future, I'll definitely wait for 20GB version. Also who knows if RDNA2 turns out to be really great (though, not a fan of their drivers... and features).

 

I don't understand why AIB's don't go for a higher quality design - AMD have seemingly opted to go for the 90's transformer toy look on their reference cards too, which is a massive shame given how nice the 5700 looked.

 

I thought the TUF looked pretty nondescript really, not flashy at all, especially compared to all the other AIBs. The FE does look better but the TUF also runs cooler and faster than FE so there’s that.

 

Why is everyone saying you can't compare the 3080 to the 2080ti? Because of pricing? Let's be honest, if Nvidia knew that the RDNA2 architecture was meh, they would probably charge you upwards of 1000USD and some of you would be thanking Jensen for being a merciful god.
And comparison wise, on paper there's nothing from the previous generation that you can remotely compare it to. Even against the top of the line 2080ti, the 3080 has double shader core count(!), same rt cores, ROP and TU's, but half the tensor cores. This all equates to more than double shader and rt performance. Still on paper, the cards have a similar TGP (even though that is worth what is worth) that is 320W. Sounds phenomenal. If we would take these specs as facts it would mean Nvidia more than doubled performance at the same power envelope. Unfortunately you look at benchmarks and it paints a different picture.
Don't get me wrong, I think it's a beast of a card, king of the hill until 3090 is released and seeing it pump those numbers in 4k makes me cry in poor. What I'm pointing out here is that the performance isn't close to the theoretical ability of the card even if we don't factor in the increase in power draw. When you factor it in, you can see it providing a significant performance increase with an almost as significant power draw increase. All that combined and the improvement isn't that big.

To conclude let's do this exercise: if we were to believe marketing and say that a node shrink is usually a 50% performance increase in the same TDP, which is not, and you can't even compare it since it's different manufacturers, but let's just imagine the 2080ti chip was shrinked to 8nm process, maintained performance but reduced power draw by let's say 20% (it's not an unreasonable number). That would put the 2080ti at around 210W. Now if we redid the comparison, that would mean the 3080 is consuming more than 50% more energy with an average of 30% more performance. You see the problem here?
There's either a problem with Samsung's 8nm, or even worse with Ampere itself. It's quite hard to understand what it is, on one side you see the chip is bigger and has double shader cores and the power draw only slightly increases so there has to be some improvement in efficiency somewhere, but on the other side it seems they had to throw a lot of shader cores into it to be able to surpass the previous Turing generation which also says it's inefficient. Soooooo confusing.

Honestly, I'm expecting someone to say that Nvidia counted these shader cores differently to previous gen and each real core is counted like 2 or 4x because of its ability to multi-task. But if that would be the case, how do you explain the increase of almost 10 bil transistors?! My head is going to explode!!! HELP

 

Well, this solidified it. bowing out of PC gaming. Was on the fence about this card, seeing the lack of availability and high prices just like they did with the 2080 has been then straw to break the camels back so to speak. the reading dmc 5 won't be bringing ray tracing to pc "because it takes advantage of processing power only found on ps5 and xbox series x" that seems like a precursor for many companies pulling that. PC gaming is too expensive for the shaft it gets. It will always hold a dear place with me, I'll stick to the console sub forum.


....until amd releases their 6000 series lmao.

 

Despite the power draw, these cards are easy to cool. Due to the small PCB which allows heat pipes to extend beyond it and free from enclosed spaces.

 

Check back in 2-3 months.

 

With some tuning they could have made it much better looking.

 

Nice paper launch. What a fraud that Huang guy, lol!

 

You don't understand what a paper launch is...do you....

 

A release of a product, especially a computer component, in extremely limited quantities, making it very difficult for consumers to get their hands on. The purpose of this is generally for a company to be able to say "we have the fastest chip", before they can actually produce large numbers of them.

Please just don't comment.