• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

NVIDIA ‘Ampere’ 8nm Graphics Cards

ZOTAC-GeForce-RTX-3090-Trinity-Graphics-Card_1-1480x755.jpg


Triple 8 pin, this thing looks a monster! Always liked Zotac cards this might be the one I go for.

A card so big they can only fit it into CAD atm :D :p
 
Can you translate this pic for laymen like me? Don't have a frame of reference to understand it.

Modern GPU's run at higher clock speeds when they are cooler. I have never used that software, but my guess is that 45 degrees is due to a light load...so 45c = boosting pretty high. (1933)

The clocks will most likely be lower at higher temps, and the temps will probably be higher at higher loads.
 

Hilarious.

Modern GPU's run at higher clock speeds when they are cooler. I have never used that software, but my guess is that 45 degrees is due to a light load...so 45c = boosting pretty high. (1933)

The clocks will most likely be lower at higher temps, and the temps will probably be higher at higher loads.

Ah okay thanks. I thought the Firestorm pic was showing some benchmark figures/score, and hence would have been meaningless to me without a comparative score.

Given the focus Nvidia put on the GPU cooler video they released and this somewhat unique design on the founders edition cards, have we got it wrong then in terms of the heat i.e. it doesn't need a fancy new cooler because its so hot, but rather it runs (relatively) cool thanks for the fancy new cooler?
 
Hilarious.



Ah okay thanks. I thought the Firestorm pic was showing some benchmark figures/score, and hence would have been meaningless to me without a comparative score.

Given the focus Nvidia put on the GPU cooler video they released and this somewhat unique design on the founders edition cards, have we got it wrong then in terms of the heat i.e. it doesn't need a fancy new cooler because its so hot, but rather it runs (relatively) cool thanks for the fancy new cooler?

We won't know until we get 3rd party reviews, but the "official" coolers look like too much of an added expense just to get a little extra clock speed. (100mhz maybe?)

My gut feeling is that Nvidia is pushing the hardware...HARD.
 
It's generally thought that Samsung 8nm is a bit of a misnomer, and it's closer to a half-node off of 7nm TSMC in terms of advancement. I'm gonna bet that they're producing the 3070 and below on Samsung, and the 2 above on TSMC (initial results must NOT have been good!).
 
It's generally thought that Samsung 8nm is a bit of a misnomer, and it's closer to a half-node off of 7nm TSMC in terms of advancement. I'm gonna bet that they're producing the 3070 and below on Samsung, and the 2 above on TSMC (initial results must NOT have been good!).

Exactly what I thought when I saw the latest leak.

hopefully they’re pushing these hard because they know AMD are bringing something to the table.. wishful thinking hat on of course
 
Just because it's cut down does not mean all disabled cores had defects - it's just transistors can use more power than some more luckier ones, so you can't use all cores at same high frequency even if there were no defects, that's why (I think) power usage isn't that massive between two variants of the same chip.

If there is such a massive variation then again that indicates yield problems though.

Its much more cut down than the Pascal and Turing top two tiers. The original Pascal Titan and GTX1080TI had the same number of cores,the GTX1080TI only had 6.6% less cores than the Titan Xp. The RTX2080TI only had 5.5% less cores than the Titan RTX.

If the leaked specifications are true,the RTX3080 has 17% less cores than the RTX3090. But look at the leaked clockspeeds,they are basically the same.

The RTX3090 is probably basically a Quadro rebadged as a gaming card,with the top 10% of dies I suspect. It wouldn't surprise me if it is expensive,and supply is hit and miss at the launch.

The RTX3080 is probably using poor quality,leaky silicon,so Nvidia has to bumped up voltage to maintain clockspeeds. Its why despite having far less RAM,it still needs to be cut down so much.

Unless OFC,the stated board powers are a bit of a lie.

You have to consider this is Samsung,and they don't have as much experience with large dies as TSMC. As a company almost all of what they make on leading edge nodes,is mobile SOCs.At over 600MM2 this will be the biggest chip ever made on a leading edge Samsung node.

You know what the previous biggest chip made on a Samsung derived leading node was?? Vega 10 at nearly 500mm2.
 
Last edited:
It's generally thought that Samsung 8nm is a bit of a misnomer, and it's closer to a half-node off of 7nm TSMC in terms of advancement. I'm gonna bet that they're producing the 3070 and below on Samsung, and the 2 above on TSMC (initial results must NOT have been good!).

They have definitely involved Samsung in the process but what will actually hit the market is another matter. One claim from a unknown source was 8nm on lower cards, TSMC on higher but I have a decent source says the other way around and a recent leak, from a previously right source, says TSMC on most/all products.
 
Back
Top Bottom