Topic: CryptoDredge 0.27.0 — CUDA Miner - page 32. (Read 86186 times)

can you test your rtxs on phi2 and lyra2z algo

How long is it going to take to you to make up the extra 300 dollars per card. I bet you could calculate that in years.

I have not asked for your opinion. The 2080 RTX is faster than 1080ti and with lower power consumption, and that the miners are not optimized.

If you do not have one or several, do not speak as if you knew them. I get better HASH with the miners without updating, but it crashes a lot, and I spend less light. You should study better in which you invest your money. I have sold 6 1080 and bought 6 rtx 2080. It is silly not to see that it spends less and that it produces more.

And it's not a bubble newie dear, the 2080ti are sold out weeks ago, that's why shipping has been delayed. But I assure you, that the normal 2080, equal and exceeds the 1080ti in HASH for $ 790, and with 20% less consumption.

I do the accounts to you ?? . Or make them yourself, more hash and less consumption = more profit

A few months ago the 1080ti cost $ 1200 and now they are for $ 500 second-hand, but for a little more I have more power. I have limited power, so the less they consume, the more cards I can put, and they produce more HASH

Please save your acid comments for yourself, if you do not have market knowledge.

You think that the 2080TI to 1300 $ for the gamers have been exhausted ??, awkward, miners have bought it

I profitably calculate it in a way that you do not understand, I use it mainly to mine coins for masternodes and if it's not worth it, I'm the most profitable with Awesome miner and then I buy the coins with those profits.

But please, do not be so smart, which certainly is not because of your stupid comment.

Hi,

About the API again.. I've read posts on first couple of pages, and I see you said API was fixed, but I keep getting ERR_EMPTY_RESPONSE when I try to access the API. It was set in command line with 0.0.0.0:4000. I am trying to access it from different machine in LAN. CryptoDredge is running on Ubuntu 16.04.5 and CUDA 9.2. I see the output in terminal after the miner starts, that the API server started on given address. I've turned off firewall so closed ports or network isn't the case here. Any help or guidance would be really appreciated.

Thanks

Sorry to break your bubble bro, not many people are buying RTX for mining. So your alot of miners numbers might a bit off
You tried to calculate your capital recovery on those cards? me thinks not

The new RTX series from nvidia went on sale, I already have RIGs with these cards and like me too many miners with large farms.

It would be convenient a version with optimization for RTX, since it uses a different driver, cuda 10, different memory etc ...

You can get great improvements to HASH when there is this update.

Right now there is no miner optimized for RTX, and I guess the first one to do it, suddenly get many miners hit using their software.

Right now Cryptodedge works in RTX but it makes it unstable and not with improvements in HASH. Other miners like z-enemy still not being optimized, get good numbers

I hope soon an update that takes into account the RTX

Not sure if others are seeing similar behavior, but the phi2 algo on 0.9.2 (all Windows CUDA builds, using 411.63 drivers) seems to bias performance on GPU0 vs. all the other GPUs on two different rigs.

On CD 9.1, I calculated a 5% delta in performance between GPU0 and GPU1 (both identical 1080ti) for phi2. On CD 9.2, this jumps to over 12%. No OC, stock PL.

After benchmarking all 6 x 1080ti in a rig, GPU0 always ends up over 1 MH/s better than all the others but only on phi2. Tests on allium, neoscrypt and lyra2z all show expected deltas in performance between GPUs.

It almost feels like an optimization was applied in code for phi2 specifically, but they forgot to apply it to any GPU past the first.

# Release 0.9.2 - (September 26, 2018)

 * New CryptoNightSaber (BitTube) algorithm
 * Improve Allium and Phi2 for 1070, 1070 Ti, 1080, 1080 Ti
 * Allow to set intensity per GPU
 * Fix failover mechanism
 * Disable miner to auto-start after system restart
 * Add crash reporting mechanism. Check Readme for more information

Are there any speed-up's in 0.9.2?

0.9.2 - phi2 - on 1080 - 7,5-7,8 Mh/s    

I use curve OC, though for each card curve is a bit different cos cards have different OC potential, but not by much. Memory is  set to 5003(-502mhz). And then I adjust power consumption using PL, e.g. for Stris OC's 72%, for Gigabyte OC's its 79%. So you gotta tune your cards individually, my settings might not work for you. If you are mining only lyra2z  algorithm you dont need to limit PL at all, just find proper vcore\clock combination that will make ur card's draw below 200wt. It's just I mine on Nicehash, so I had to make curves for 3 algos for my cards: x16r, lyra2z\rev2, and I also got to use PL hence in lyra2z I'll get more powerdraw (though also higher h\r) if I allow 1080ti's to mine at max clock my curve has. With PL they usually stick to 825-831mV and their respective clocks.

ps Here is example of a curve of 1 of my strix's card:



I get a bit higher h\r - 76-77Mh\s with power draw around 195Wt.

What is it 1060TI?

I get on Lyra2v2  68MH/s on 1080TI with CCminer, and Dredge.0.9.1 I get 74-75MH/s
on 1060TI I get 21.5-22mh/s with CCMiner and Dredge I get 25mh/s

draw more power 30% but you mine faster around 80%

ccminer mine lyra2z around 3.5MH for 1080ti  use cryptodredge it have 6.1MH

what settings if i may ask? thank you

i was trying 75 TPD, 85 - 130 Core and 0 Mem

I get ~5.9MH\s @ 197Wt

yeah but lyra2z was always was drawing about 30 % of the gpu.. not more.. with older ccminers (2017)

1080ti normally it use around 250 watt per card

Just a question. I am getting about 6 MH/s in lyra2z on the 1080 TI.. but the power draw is insane.. about 250 watt
anyone else ?

Periodically I get an error on Lyra2z - "The semaphore timeout period has expired"
With what it can be connected? Other miners do not issue such errors.