Topic: Bitmain E3 Ethash Miner ASIC (Shipping:16-31 July. $800 USD) - page 22. (Read 16395 times)

I manage a bit less than 700 at the wall with 5 x 1070 at right around that same 157 Mhash - but that's on rigs with an AMD FX 8350 CPU (95 watt TDP rating) mining Monero, so NOT well optimized for ETH-specific usage.
A rig using a LOW POWER cpu like a G4600 or a Sempron 145 and not using the CPU for anything else should manage 600 ballpark at the wall for 5 x 1070.

The cards alone are capable of pulling full hashrate at commonly 100-105 watts TDP (nvidia-smi reported, figure 110-115 at the wall per card) depending on the specific card and model, across my stable of 8 or 9 different card models and 30ish 1070 cards.


heavyarms - why do you ASSUME there "haven't been much of GPUs sold since Dec/Jan" when bloody near all GPU models used to mine ETH were TOTALLY SOLD OUT almost all the time in that timeframe?
There is ZERO logic or sense to that assumption.



The F3 is speculation based on ONE video - that didn't show WHAT it was hashing on, what it looked like, what the power consumption is, and even if it exists it might not be for ethash AT ALL.
I concede my "oops" on the 2x hashrate figure - 7.5-8x is the correct number - but that would still need 200,000 OR MORE to match current total ethash network hashrate, NOT going to happen in a few months especially if it's a 14/16nm node ASIC and given it hasn't even been announced for sale (keep in mind that the E3 is not shipping for more than 2 months from now - 15-30 July).

The E3 is information straight from Bitmain's SALE site - it's not speculation and they have stated that while the specs are preliminary they expect final units to have even better performance (take with an ounce of salt, Bitmain has a history of being a bit optimistic on those announcements but are usually at least CLOSE when the units actually reach end users).

There haven't been much of GPUs sold since Dec/Jan (sheer dearth of it) yet the hashrates seems to increase quite a bit?  115 TH to 250+ THs.  That can't be all GPUs right?

Forget the video, talking about everything that was leaked about 2 months ago. There was never an e3 asic.

Is this more speculation or do you have a source for this?

Susquehanna never revealed more details other than confirming Bitmain was working on an ASIC. Zero specs. The video was made by a user on the Ethfans pool and posted to Telegram, not a company. What you're describing is speculation.

The f3 was what people saw on bitmain factories and leaked to the community. It's real and bitmain will never say it exists, you might never hear about it cause if bitmain tries to sell the f3 next year then will be suspicious, so the f3 will never be publicly revealed.

Just want to touch on this because it's a bit misleading.

- Casper testnet is scheduled to be fully operational (currently in "Alpha") by August, no date is set for mainnet launch
- Cost reduction and threat assessment is on-part with devs' wording
- SHA-3 will be considered, a dev asked about the possibility of implementing it but it was left up to further discussion at next month's meeting.
- The 51% attack was deemed to be "difficult to predict" and I believe it was Vitalik who noted that it wouldn't be noticed until it was happening. If this would happen from a governmental or private entity (both were mentioned neither by name), then "emergency measures" would be considered. ETH devs will not be releasing Casper onto mainnet without bug testing, that's a silly assertation and would be dangerous to the coin.

Let's remember that Vitalik brought up a good point; they're not sure how this ASIC functions. Until one is torn apart, it would be foolish to make a move should the unit contain an FPGA as speculated.

There are so many red flags in that video that show it as a fake. Anybody with basic HTML markup knowledge could make the changes to the web GUI in the firmware, the sticker was not the same kind that bitmain uses (less round) and had low-quality ink with smaller stroked accents, plus it was off-center. And we're not going to talk about how "F" is not the same family as "E"? Historically, Bitmain has kept the ASIC models consistent with algorithms.

F3 is pure speculation and spouting nonsense is just upsetting the community over a fake video.

FYI- F3 is 1500 Mh/s in the video. ie 8-10x.
Also if they implement new algo into Metropolis then bitmain is left out in the cold doesn't matter how many they sell.

Even if the F3 exists and meets the stats in that video, it's STILL not a short-term threat - only double the hashrate of the E3, with no information about how much power it draws.
Bitmain would have to sell almost a MILLION of those to match the existing ETH hashrate, and over a million when you toss in the existing hashrate from the other ethash coins like ETC.

NOT HAPPENING in a few months timeframe - though to be fair initial Casper implementation is intented to be the 1%/99% hybrid and still mostly POW.

Let's buy everything guys, let's give power to Jihan

okay so looks like we have a few updates from the Eth dev team after they discussed the relevant eip.

-Casper implementation is scheduled / budgeted around July mid (when these ASICS are due for launch). Progress is good so far and making in roads.
-There is currently a 2.5x cost reduction with e3 but no real gain in hash rate; threat for 51% attack is minimal.
-No emergency measures will need to be taken at the moment; however plan for alternative algo like sha-3 will be integrated into next hard fork.
-If bitmain threatens with higher hashing power and possible 51% attack eth will counter with a cryptonight like hard fork OR they may expedite casper POS launch without bug testing etc. (option 1 seems to be the more widely accepted scenario).
-eth dev will be able to detect any spikes in hashrate etc that signals a high hash rate asic (like F3 eg.) and may choose hard fork earlier.

https://www.youtube.com/watch?v=SoPfoNpqG0k

If this shipped sooner I'd buy it, still you can get 2 eth for what this is worth now

Close - there are probably stlll some "last gen" units like the Innosilicon A2, KNC Titan, and Alcheminer in use since they're still PROFITABLE if you have 4 cent/kwh or less electric cost - but the L3+/A4/L21/A6 have long since become the bulk of Scrypt mining hashrate and likely account for close to 95% of that hashrate.

I am not planning to buy any because of the uncertainty about "when" on the move to Proof of Stake and that I'm currently very short on spare cash, coupled with the relatively LOW reliability of most Bitmain options.
I am not planning to sell a rig to finance buying one (or more likely 2).

I'm on the fence on "would it be a good idea" - if you had sold the rig a month ago when GPU pricing was still crazy-inflated, it would have made sense even WITH the uncertainty of "will an E3 manage to pay for itself" if the E3 was going to ship in the next week or two, and probably even with the "not ship 'till July" issue.
I'm not going to say it's a BAD idea - but it's a bit higher risk than I'm willing to take right now.

If I did buy an ASIC miner in this month, there are other options that I suspect will be more profitable longer than the E3 would be.


There's also the lingering uncertainty about the rumored "F3", if it exists, and what algorithm it's FOR if it does actually exist.
If it turns out the numbers in that one video that caused all of the rumors are correct, and *if* it was mining on Ethereum, that thing WOULD have a noticeable impact sooner than the E3 would - unless the E3 is a 28nm effort and the F3 is a 14/16 nm effort that has to compete for very limited foundry capacity that would reduce it's available numbers to be built A LOT.

Could be not the same chip but something very similar, if you look at the block diagram it resembles very closely GPU without texturing and rasterization units(you dont need these for mining) and pcie bus is moved to an fpga chip. They just have to substitute these "NP" units with something more appropriate for mining. 

I've read your analysis and now i'm curious. Given your arguments, i wouldn't say you are PRO bitmain E3 ASICs but you definitely aren't against them. Are you, jstefanop or QuintLeo, personally buying some of these E3 ASICs? Do you think it would be a wise move to sell a GPU-rig and instead buy one of these now?

You've obviously never been to china.

You are missing my point.  I am saying if ETH went away tomorrow, where would this hashrate switch to exactly?  Like you said, these farms are all using AMD cards, and the only advantage AMD has is with ETH and Cryptonight.  Pretty much every other algo gives the advantage to Nvidia.  I'm guessing a lot of these AMD cards at the farms would simply get sold.  And if there are giant nvidia farms, they are already mining something besides ETH.

Steady rise since December = batch 1 units hit network. Bitmain is rumored to have mined for 3 - 6 months with equipment before refreshing hardware and selling the old stuff. Very likely what's going on here.

Something caused the Ethereum hashrate to Double in less than two months.....I dont think it was this E3.  I dont think it was new GPU's.
It was the hidden Bitmain Eth Asic.  Lets hardfork and watch the hashpower lose at least 35%!!!  Its the only way to know for sure.