Ethereum GPU mining?

[IntelUser2000]

That being said, when Monero wiped out the Bitmain XMR ASICs with an algorithm chain, Bitmain did not strike back, so obviously they weren't prepared there either.

I think its much different in case of the Monero ASIC. The Monero ASIC truly performs like one. It's 100 times faster than the fastest GPU or CPU based miners, at twice the power. The Monero devs had every reason to hard fork.

Also, being that fast means it has to be that specialized. The changes to the hardware may not be trivial at all.

Imagine if Bitcoin did a hard fork to kill current ASICs. The hardware would be so optimized for one task that it would need a total redesign to fit to a new algorithm. That's not the case for Ethereum. Memory is the hard barrier.

 

[DrMrLordX]

I think its much different in case of the Monero ASIC. The Monero ASIC truly performs like one. It's 100 times faster than the fastest GPU or CPU based miners, at twice the power. The Monero devs had every reason to hard fork.

Also, being that fast means it has to be that specialized. The changes to the hardware may not be trivial at all.

Imagine if Bitcoin did a hard fork to kill current ASICs. The hardware would be so optimized for one task that it would need a total redesign to fit to a new algorithm. That's not the case for Ethereum. Memory is the hard barrier.

Well honestly, we don't know what Bitmain used to design their Ethereum ASIC. You do have a point - it is not leaps and bounds better than existing GPU miners (yet), so the incentive for the EF to shut them down with an algorithm change really isn't there anyway.

It still may be possible to render them useless is Bitmain took a true ASIC and strapped it to a powerful memory subsystem to cope with the demands of the Ethereum hashing algorithm. We just don't know.

Personally I would be amused if the Bitcoin core devs did kill the ASICs, but that's not going to happen . . . and the miners would never signal for such a change anyway.

 

[IntelUser2000]

The devs just finished another meeting. I've listened to the end of it. It looks like they finally reached a decision on which EIPs to use. Wisely, it doesn't look like they'll go changing the PoW algorithm to reduce ASICs.

It looks like they'll go about using EIP1234 for Constantinople. Block rewards going down from 3 Eth to 2 Eth.

 

[DrMrLordX]

With the combination of the lower block rewards and the crashing ETH prices, mining is not going to be fun. Might be a fun time to accumulate though.

 

[AdamK47]

I'm so absolutely happy to see ETH tanking.

 

[Elfear]

I'm so absolutely happy to see ETH tanking.

And why did you feel the need to post that in a mining thread? Yay for you I guess?

 

[thilanliyan]

And why did you feel the need to post that in a mining thread? Yay for you I guess?

We're the devils stealing video cards from "gamers"...probably?

 

[ozzy702]

And why did you feel the need to post that in a mining thread? Yay for you I guess?

He's a miserable human being?

 

[DrMrLordX]

I'm so absolutely happy to see ETH tanking.

It isn't falling far enough. I want it at about $20 so I can buy more. Or gimme a flash crash . . .

It'll go back up. By the time it does, PoS will probably be here (2019).

 

[IntelUser2000]

Ethereum price floor is again being interestingly related to mining difficulty.

-End of 2016, price dropped low enough that miners started leaving and difficulty reduced. That provided a floor for bottom price.
-Just a few month ago when Eth dipped temporarily below $400, again difficulty went down.
-In the last few days, difficulty has also been going down. And with it also providing support for Ethereum price.

It won't be so unimaginable to assume lots of miners automatically turn longs for Ethereum.

On ASIC news, Linzhi is saying they'll have an Ethhash miner capable of outputting 1.4GH/s and use 1/8th the power of the Antminer E3. Now, that's a real ASIC miner. They are talking April of next year for possible shipment. The contention between the two camps(one pro anti-ASIC EIP and another against it) are going to really heat up if that happens.

 

[Feld]

Ethereum price floor is again being interestingly related to mining difficulty.

-End of 2016, price dropped low enough that miners started leaving and difficulty reduced. That provided a floor for bottom price.
-Just a few month ago when Eth dipped temporarily below $400, again difficulty went down.
-In the last few days, difficulty has also been going down. And with it also providing support for Ethereum price.

It won't be so unimaginable to assume lots of miners automatically turn longs for Ethereum.

On ASIC news, Linzhi is saying they'll have an Ethhash miner capable of outputting 1.4GH/s and use 1/8th the power of the Antminer E3. Now, that's a real ASIC miner. They are talking April of next year for possible shipment. The contention between the two camps(one pro anti-ASIC EIP and another against it) are going to really heat up if that happens.

1.4GH/s at only 100W is 50x more efficient at hash rate per watt than the most efficient GPU rig setup. If this thing ships in any kind of volume without an anti-ASIC fork first, GPU mining for Ethereum will be completely obsolete.

 

[DrMrLordX]

Yup, I don't see ETH mining as being profitable for anyone with GPUs beyond that point. Not sure how the EF is going to react.

 

[IntelUser2000]

This is why the argument will really heat up. The devs are talking about 8 months for hard fork development now. Possibly the one way they might go about it is:

8 months after Constantinople: Hardfork for ProgPOW for further ASIC-resistance
8 months after that: Casper for PoS
8 months after that: Sharding

 

[DrMrLordX]

Interesting observation. In a way I sort of hope they do go for ASIC-resistance in the short term (though that timeline is ugly; they were supposed to be at sharding by 2019 I thought?). One of the things that has made Ethereum so successful in the past is that votes by the miners have been zero-drama affairs in the near past, with the exception of Ethereum Classic. If ASIC miners dig in and bring a ton of hashpower online, their votes will become as poisonous as those of the BTC ASIC miners (which is why BTC will probably never outgrow ASICs or move to a better consensus model). PoS was going to be a risk anyway, but people using GPU farms at least have the option of selling off their old equipment, and everyone knew it was coming. ASIC investors? Not so much. I wonder who in their right mind would invest in the Linzhi miner if the EF is already in talks to quash it.

 

[beginner99]

I wonder who in their right mind would invest in the Linzhi miner if the EF is already in talks to quash it.

Yeah April 2019 is quiet late but I just read up that POS (Casperg) was delayed (again) indefinitely really since it could be as late as 2021 and then the miner was probably well worth it's money. I fear if they delay even further, the trust will be completely gone. They have been talking about ut already for years.

 

[IntelUser2000]

I fear if they delay even further, the trust will be completely gone. They have been talking about ut already for years.

Rome wasn't built in a day?

Business cycles exist because we, human beings run them and they are a result of our work, hopes, and dreams. Between work we need to sleep, eat, have fun, relax, all in rather periodic timeframes.

Everything takes time, and we can not speed through some things. You can do things to make it look good for the short-term, without any value. Real value takes time to build. If you panic during the depression(whether in economic terms or just mood) and stop, that's the worst decision that can be made. It's painful, but we can't avoid the down phase.

What I am saying is if they are worth even slightly what they claim, the developers will get through the worst times. Looking from a bird's eye view, we'll also see whether this whole thing about "decentralization" was a hopeless grasping for straws or a glimpse of the further future.

 

[DrMrLordX]

Rome wasn't built in a day?

I've said it before, but I'll say it again: blockchain will leave Ethereum behind if they keep lagging. EOS and NEO are attracting enough attention that eventually, people will glom on to a buggy product instead of one riddled with delays.

2021? Please. PoS was supposed to roll out in late 2016 or 2017, at least in some fashion. If the ASIC miners take control of the lion's share of hashpower, Ethereum is over. It'll be another dead legacy coin like Bitcoin, forever at the mercy of insane miners.

 

[AdamK47]

ETH had a good run down to the bottom. Unfortunately it went up a bit. I'm hoping it resumes its downward trajectory.

 

[UsandThem]

I'm so absolutely happy to see ETH tanking.

ETH had a good run down to the bottom. Unfortunately it went up a bit. I'm hoping it resumes its downward trajectory.

Your thoughts on mining was clear after your first
comment on September 9th. At this point, it is trolling.
This is a thread for people to discuss mining, so if
you have nothing to add, move along.

AT Mod Usandthem

 

[DrMrLordX]

ETH had a good run down to the bottom. Unfortunately it went up a bit. I'm hoping it resumes its downward trajectory.

If you really want to bring GPU prices down, go pre-order some of those Linzhi miners. 1.4 GH/s with power usage of ~90W will kill the GPU miners dead, instantly. Of course the GPU miners will just move to XMR for a time, but there isn't enough room for all of them there . . .

 

[thilanliyan]

Anyone running an Eth node or syncing the blockchain?

I was wondering if a NVME drive is absolutely necessary to sync the blockchain and run a full node, or whether I can get away with a regular SATA SSD? I'm assuming writing the blockchain will not be able to saturate the SATA bus, especially if my internet is only 50mbps, but I'm not certain.

I'm trying to decide between 2 drives: Toshiba XG3 512gb NVME (MLC, 1 yr warranty) vs Samsung 860EVO 500GB SATA (V-Nand, 5 yr warranty). The Samsung drive is $12 cheaper. Just trying to decide between NVME (w/ 1 yr warranty) vs SATA (w/ 5 yr warranty).

Thoughts?

 

[Hitman928]

I gave up on running a node a long time ago. There were some tricks to getting it to stay in sync, even on a platter HDD that worked for me, but that was over a year ago. The problem with going with your SSDs is that the blockchain gets big pretty fast. If you are in Eth long term, a 500 GB drive will fill up before too long. There's a way to compress it, but again, it's been a while so I don't remember how I did it anymore but it all just became too much effort for me to deal with having to constantly maintain the size and making sure it was keeping up with the latest updates.

I switched to offline options but am also just in holding mode for now.

 

[thilanliyan]

I gave up on running a node a long time ago. There were some tricks to getting it to stay in sync, even on a platter HDD that worked for me, but that was over a year ago. The problem with going with your SSDs is that the blockchain gets big pretty fast. If you are in Eth long term, a 500 GB drive will fill up before too long. There's a way to compress it, but again, it's been a while so I don't remember how I did it anymore but it all just became too much effort for me to deal with having to constantly maintain the size and making sure it was keeping up with the latest updates.

I switched to offline options but am also just in holding mode for now.

Thanks for the heads up. Yeah, I was also thinking about the size of the db, which is why I was thinking of 500GB+. I really don't know how long it would last but it's just something to tinker with TBH...I was actually going to run it on a NanoPC-T4 running Armbian.

 

[Headfoot]

ETH had a good run down to the bottom. Unfortunately it went up a bit. I'm hoping it resumes its downward trajectory.

Imagine that we all owned horses and I cheered every time the price of gas went up

 

[DrMrLordX]

Anyone running an Eth node or syncing the blockchain?

I was wondering if a NVME drive is absolutely necessary to sync the blockchain and run a full node, or whether I can get away with a regular SATA SSD? I'm assuming writing the blockchain will not be able to saturate the SATA bus, especially if my internet is only 50mbps, but I'm not certain.

I'm trying to decide between 2 drives: Toshiba XG3 512gb NVME (MLC, 1 yr warranty) vs Samsung 860EVO 500GB SATA (V-Nand, 5 yr warranty). The Samsung drive is $12 cheaper. Just trying to decide between NVME (w/ 1 yr warranty) vs SATA (w/ 5 yr warranty).

Thoughts?

I haven't tried syncing the blockchain recently. Supposedly it produces fewer client writes/day now than it used to, but at least using older versions of geth, syncing would beat the hell out of an SSD/NVMe drive. I found that if you're "serious" about running a node, that the way to go is to use a huge mcache setting (10 gigs or larger). Then you can sync to anything you want. Just be advised it'll chew up 12+ GB of RAM, constantly.