Mining performance of Nvidia RTX 2080 Series

[RivAngE]

Thank you Luke.

Too bad the Lyra2REv2 will be ASIC mined now...
From another benchmark, I noticed that the HEX algo (XDNA coin) using the Enemy miner 1.20 (newer version is 1.21) offered a big boost with the 2080ti over the 1080ti, you might want to try this algo too.
As for the Ethash, which miner did you use? Claymore or something else?

[1715586050]

1715586050

[giagge]

Read this:

https://www.phoronix.com/scan.php?page=article&item=nvidia-rtx2080ti-compute&num=1

.

[giagge]

Hi guys, it's Luke from Cudo Miner here. We've just reviewed the latest GEFORCE RTX 2080 and compared its mining performance against the GTX 1080 on 4 different algos. Just wanted to share the video we've produced to show you the results. 📹

You can view the video here: https://www.cudominer.com/rtx2080benchmark/

[Too long; didn't watch😉] Here are the results:

Hashrate

Ethash

NVidia GTX 1080 Founders OC: 20.2 Mh/s (28.7 with Ethpill)    

Gigabyte RTX 2080 OC: 35.3 Mh/s (+75%)

Cryptonight Variant 1

NVidia GTX 1080 Founders OC: 507 h/s  

Gigabyte RTX 2080 OC: 769 h/s (+50%)

Lyra2REv2

NVidia GTX 1080 Founders OC: 42.1 Mh/s  

Gigabyte RTX 2080 OC: 72.1 Mh/s (+70%)

Equihash

NVidia GTX 1080 Founders OC: 391 Sol/s  

Gigabyte RTX 2080 OC: 566 Sol/s (+45%)


See the following images for tables and graphs: https://imgur.com/a/ZvPdYcohttps://imgur.com/a/iQmZkdT


What are your thoughts on the results? Has anyone else tested the RTX 2080 on their rig? What hashrate did you get?Cheers!

I have 30mhs on eth with claymore on nvidia gtx 1070 (ethpills dont work on gddr5 and have samsung mem ) with PL 55 clock +150 Mem +600 .

[dragonmike]

There's not a single post here that isn't comparing apples to oranges currently.

To be 100% useful, you'd have to compare a 1080 with pill vs a 2080 with pill on ETH (not possible as the latter pill hasn't been released)
You'd have to compare FE cards vs FE cards. An EVGA 1080 SC does 37MH/s on ethash with the pill but the FE does only 28? Extrapolating this, why would one compare a Gigabyte 2080 vs a 1080 FE?

So yeah. These "comparisons" are all tosh.

[gsrcrxsi314]

There's not a single post here that isn't comparing apples to oranges currently.

To be 100% useful, you'd have to compare a 1080 with pill vs a 2080 with pill on ETH (not possible as the latter pill hasn't been released)
You'd have to compare FE cards vs FE cards. An EVGA 1080 SC does 37MH/s on ethash with the pill but the FE does only 28? Extrapolating this, why would one compare a Gigabyte 2080 vs a 1080 FE?

So yeah. These "comparisons" are all tosh.

Personally I think the best comparison would be 1080/1080ti w/ ETHLargement pill to 2080/2080ti without pill.

This compares what is possible RIGHT NOW. If you’re trying to run a 1080/1080ti on memory dependent algos without the pill you’re just leaving performance on the table for no reason.

You shouldn’t assume that a GDDR6 pill will materialize. At least not for free. The current pill works for GDDR5X only, and it has been reported that a GDDR5 version exists, but available only for some amount of money to farm owners. I suspect that IF (huge IF) a GDDR6 version of the pill comes out, it will be a private for sale version only. Another point leaning to no pill for GDDR6 would be that the devs seem to already have their hands full with other more pressing projects. Mineority and FPGAs. My bet is that it just never happens.

[nsummy]

There's not a single post here that isn't comparing apples to oranges currently.

To be 100% useful, you'd have to compare a 1080 with pill vs a 2080 with pill on ETH (not possible as the latter pill hasn't been released)
You'd have to compare FE cards vs FE cards. An EVGA 1080 SC does 37MH/s on ethash with the pill but the FE does only 28? Extrapolating this, why would one compare a Gigabyte 2080 vs a 1080 FE?

So yeah. These "comparisons" are all tosh.

A.  I have never been able to get the pill to work on my 1080 cards.  Yes I know it works for most people but using a 3rd party program doesn't show true performance comparison.  It would be like comparing a 2018 Corvette with a 2019 Corvette, except the 2018 has been modified. 

B.  The 2080 FE cards are already overclocked.  The performance of the Gigabyte is probably similar.  I don't think its possible to even buy a stock clock 2080 right now.

[Pono]

Lyra2REv2
NVidia GTX 1080 Founders OC: 42.1 Mh/s   
Gigabyte RTX 2080 OC: 72.1 Mh/s (+70%)

Equihash
NVidia GTX 1080 Founders OC: 391 Sol/s   
Gigabyte RTX 2080 OC: 566 Sol/s (+45%)

Thanks for the video, well done. There are ASICs for both Lyra2REv2 and Equihash and hashrate for Lyra2REv2-based coins is going through the roof thanks to Dayun Zig Z1 so we might forget about those two algos completely. To be honest I do not see RTX line as a good fit for mining given its price.

[danivan7]

I am seeing estimated mining performance figures of RTX 2080Ti at 69MH/s. Please see source here: https://www.bitnand.com/product-page/nvidia-geforce-rtx-2080ti-fe

Does that make sense to you? Let's compare GTX 1080Ti vs RTX 2080Ti memory performance here.

GTX 1080Ti vs RTX 2080Ti
Memory type: GDDR5X vs GDDR6
Memory speed: 11Gbps vs 14Gbps (+27%)
Size: 11GB vs 11GB (same)
Interface width: 352-bit vs 352-bit (same)
Memory bandwidth: 484GB/s vs 616GB/s (+27%)

Given ETHash is a memory intensive algorithm that does not depend much on the GPU itself, the difference in memory performance should be a good predictor of RTX 2080 series performance.

price for this GPU is to expensive, and in other side,
eth price was fall into deep.
so this GPU not giving solution about profit for mining ETH.
I think can use for mining other coin with high ROI

[giagge]

.

[revenant2017]

.

If this card is not as equal as the performance of 1080, then i'm afraid to say this is the end of GPU mining. Also this card can be good for VR and 1440p gsync.

[adaseb]

Given the huge price point of these GPUs, I am wondering if Nvidia will have issues getting them sold for gaming uses.

Basically there are tons of 1070,1080,1080ti on the market right now. Probably more will hit the market in one month when ETH reduces its block reward from 3 to 2 ETH.

I am pretty sure a gamer is better off buying 2x of the 1080 second hand rather than a new 2080ti and just SLI them together. It would use more power sure, but since its a gaming rig it doesn't get used constantely so power shouldn't be an issue.

For mining, its definately not worth it. Since there are tons of second hand RX 470/570 going for cheap on eBay right now.

[Marvell2]

Given the huge price point of these GPUs, I am wondering if Nvidia will have issues getting them sold for gaming uses.

Basically there are tons of 1070,1080,1080ti on the market right now. Probably more will hit the market in one month when ETH reduces its block reward from 3 to 2 ETH.

I am pretty sure a gamer is better off buying 2x of the 1080 second hand rather than a new 2080ti and just SLI them together. It would use more power sure, but since its a gaming rig it doesn't get used constantely so power shouldn't be an issue.

For mining, its definately not worth it. Since there are tons of second hand RX 470/570 going for cheap on eBay right now.

ppl are not buying nvidia gear to mine eth , Ive not mined it even with my 1070s for like 5 months and
don’t use that dumb eth has so many gpus on it arguments either.  50 to 60 percent of eth and eth forks hash
is asic based , it barely afffected us when eth dropped to $170 cuz we haven’t been mining it anyways

[Marvell2]

stop fucking comparing the 2080 to the 1080 , no one who is fucking sane would buy a 1080 vs a 1080ti

when the 1080ti costs LESS than the 2080 and performs better than the 1080 by alot.


where are these idoit newbies coming from benching a 1080 , no one gives a fuck about a 1080 if they are even considering a 2080... not GAMERS or MINERS. NOONE

stop it with these bullshit 1080 benches

[greyday]

Ok, my 2080 is up and running. In Ethos, with no tinkering, it's running 39.8 @ 207W. Will be playing with settings to drop that wattage.

With ethpill and tweaks, I get between 49-51 with my 1080ti at about 175W.

2080ti is, of course, delayed, so no numbers there yet.

I decided to cancel my other 2080 (which is also delayed) and will be picking up a 2070 for comparison, but I would say that once miners are coded for these they may be truly competitive. Until then I'd rather spend the money on another 1080ti unless I was building a gaming rig/media center.

[dragonmike]

stop fucking comparing the 2080 to the 1080 , no one who is fucking sane would buy a 1080 vs a 1080ti

when the 1080ti costs LESS than the 2080 and performs better than the 1080 by alot.


where are these idoit newbies coming from benching a 1080 , no one gives a fuck about a 1080 if they are even considering a 2080... not GAMERS or MINERS. NOONE

stop it with these bullshit 1080 benches

It's like everybody's retarded.

EThash is a memory-bound algo that's never favoured nVidia until sombody managed to artificially lower mem timings for a small selection of GPU models.

Surely if you wanted to compare raw hash power of new generations of nVidia cards you WOULD NOT use Ethash as battleground? Yet everybody here in this thread is doing just that.

Use a core-intensive algo if you're gonna compare, plonkers.

[RivAngE]

stop fucking comparing the 2080 to the 1080 , no one who is fucking sane would buy a 1080 vs a 1080ti

when the 1080ti costs LESS than the 2080 and performs better than the 1080 by alot.


where are these idoit newbies coming from benching a 1080 , no one gives a fuck about a 1080 if they are even considering a 2080... not GAMERS or MINERS. NOONE

stop it with these bullshit 1080 benches

It's probably because the 2080 is easier to get at the moment, so people think that the best comparison would be with the 1080. However your point is valid, I'm just saying what I think people are thinking!

[Marvell2]

stop fucking comparing the 2080 to the 1080 , no one who is fucking sane would buy a 1080 vs a 1080ti

when the 1080ti costs LESS than the 2080 and performs better than the 1080 by alot.


where are these idoit newbies coming from benching a 1080 , no one gives a fuck about a 1080 if they are even considering a 2080... not GAMERS or MINERS. NOONE

stop it with these bullshit 1080 benches

It's probably because the 2080 is easier to get at the moment, so people think that the best comparison would be with the 1080. However your point is valid, I'm just saying what I think people are thinking!

Um no in the usa at least you can buy 1080tis all day for $729 or less
https://m.newegg.com/ProductList/keywordsearch?keyword=1080ti

cheaper used if you go on amazon, there are plenty of 1080tis available so its not that

I really don’t understand the logic @dragonmike exactly , its like the forum has gone full retard? Ethhash? on Nividia ? Even 1080tis with the pill are not even mining ethash algos eith ethereum being under $300

[NiklasFalk]

There is no point of comparing the 2080 to anything in a "controlled" test. We all know what the Nvidia cards we have do in our own hands. The interesting thing is what can you squeeze out from a 2080.

Values of < 25 Mh/s on Ethash (e.g. as mentioned for a 1080) is at or lower than my 1060 3G (9x makes 210Mh/s dead stable), i.e. worthless piece of crap.

If a 2080 cannot beat 3x1060 (micron) in any algo, it's only useful for compact rigs or bragging rights.
70 Mh/s Ethash (210w)
20 Mh/s ProgPow (195w)
70 Sol/s Equihash 144 (180w)
3 card values average over 9 cards.

Nvidia have plenty of Pascal chips in stock, why would they have any interest in selling the 20#0 cards, it's mostly marketing to get some attention (and priced to not disturb the Pascal sales).

[TrailingStop]

Hi,

not sure if I got your table correct, but do you get 20 Mh/s ProgPoW out of a single 1060??? Do you mind to share your power and OC settings? I run 1070 @ 50% power and hardly get more than 12 Mh/s.

[NiklasFalk]

Hi,

not sure if I got your table correct, but do you get 20 Mh/s ProgPoW out of a single 1060???

Just wonder...

Sorry!, that was for all 9... so it should be 20 for 3.