Topic: A4 Dominator - Pre-Order Group Buy - 280mh, roughly 1000w, $1800 + shipping - page 18. (Read 122591 times)

Longsnowsm, Poolside hashrate is derived from shares submitted so there is a "luck" element to those numbers that you have to factor in (both good and bad). It's best to look at the longterm average you're getting from all your miners over a 24hr period rather than the 2hr avg (on ProHashing).

Just to prove I'm not delierious here's some shots taken over 20 minutes today. You'll see one miner that's below 250Mh, but it was 290Mh earlier in the day which shows how much they can move around.

I do see a difference between 1200 and 1224 but I'm tracking them over days to see the difference.

After flashing to B3 firmware (the ones in the photos below were originally B1's and B2's), the only variables I can suggest that may be influencing my numbers is that:

• I'm running at 230v and use purpose built 1500w PSUs (designed for Antminers by Rsenda) off a 3 phase feed so have good power, and
• I'm keeping them quite cool most of the time <50 degrees C (Most max out around 55 between Noon and 5pm with a one or two cards reaching 60 on a hot day)

 

HTH








     

on the cgminer GUI, it shows each blade is total chips delivering 69.5MH

so 2 blades per cube is 69.5 x 2 = 139MH

an A4 unit has 2 cubes, so 139 x2 = 278MHs -- thats pretty close to 280MHs

Correct me if I am wrong?

After more than 24 hours seems I get more consistency, I think due to the zpool stratum able to allocate more high diff work for the A4s  ---- however I agree, we need some open source activity here, the A2s survived a nice life with custom firmware thanks to GenTarkin, MarkAZ etc.

With the messy start to the launch of A4s.... the only just consolation for the community is the opensource path.

Until this happens, whatever budget I have allocated for A4s is now redirected to other projects.

I am "proud" owner of 6 x A4s.

Happy New Year 2017.

I get wild spikes and dips in hash rate on zpool.ca.  I just tried it tonight with the higher diff settings.  I do see it roar up to nice high numbers, but then I also see it dip to some extremely low numbers and it just swings back and forth.  I just don't have confidence that I am going to be earning what I presently earn on Prohashing when I see these dips falling down to some completely unreal numbers. 

I have tried zpool.ca several times in the past, and currently do a little cpu mining there, but my performance/earning have never been great there.  Watching the wild swings tonight with the A4's I doubt that it will average out for me as well as what I see on Prohashing... So I just moved them back over to Prohashing for the time being. 

Hopefully we can see the software open sourced and can see some tweaking to get better numbers out of these miners. 

I tried the A4s on several multi-algo pools, and for the first time I see the A4s hitting and above the intended hashrate and staying there long enough. I am mining at a multi-algo pool; zpool.ca

The op asked me to try 65536 x 2 = 131072 as the diff level. I was not able to set  diff level more than 65536 in other pools.

My A4s are B3 with v2.02 firmware - I was lucky to get B3 and only had to upgrade the firmware.

My cubes are setup a bit differently ~ 417MHs per A4.

Anyways, just sharing my findings.

Go and try and see if it works for your A4s. Pays direct to BTC too

Pool: stratum+tcp://scrypt.mine.zpool.ca:3433
Userid: BTC or coin address.
Password: c=BTC d=131072

No registration

one question, one observation...

Q1.  i am in the USA and have decided to buy this product...my bank does not allow my personal account to wire internationally.  what other service allows me to wire from person to HK business?  TIA  (i looked at WU and they said no... most places i found only deal person to person...)

Obsv 1.  did anyone call out innosilicon for magically adding MH?  if you look at the single can they say 135mh, the 280mh is really just two of the single cans, but when i add 135 + 135 it is 270. not 280...

Open source. Is key. 3rd party gen Tarkin saved Titans or the all would be doorstops by now imho.

The next big. Thing you need to fix.  Push etc.

Never any luck with Titans and multi pool. 5% fee 2-3% error rate add no fee
www.litecoinpool.org at NO fee and pays 103% and no errors or down time
In 2 years I never bothered. But if you are getting 20% or say at least 10% more with
above subtracted, I may have to reconsider.

I could only wish that my A4's would pull an avg of 260mhs.  For me the average is in the 240-250 range with occasional blips above that on the pool. The miners at 1224mhz say 281mhs on the miner side, but on the pool after things settle down the hash rate is right back to where it was when set at 1200mhz.   So no real benefit seen on the pool side once the hash rate stabilizes after the restart. 

What the miner says has no bearing on reality for me.  I only wish I was seeing some of the numbers that others here are reporting on the pool side.  What does that translate to?  Well based on the numbers that Vinylwasp just posted it means I am losing on average over $2 a day at today's prices.  So that really stinks.

Also keep in mind if you mine on Prohashing or do conversion to BTC fast and keep BTC. Then those 6$ a day can be more than 10 later next years. (Of course it's some risk to keep BTC.)

Which pool you are using if you only get 230Mhas?

I get about 260 Mh/s poolside on my A4's using 1224Mhz at 814MV on Prohashing with no special settings, and I'd expect a little better if I was mining a single coin like LTC. The coin switching and changing diff per coin has a small negative effect on the poolside hashrate. On the miners, they'll mine at 281 Mhs all day.

But I think this sub is a little off kilter; yes there's been a firmware problem with the A4's, but there's been no "Die failed" issues with A4's, no burnt out 6Pin connectors, the DCDC circuitry isn't running at 105 degrees C, and they don't suffer from long restart penalities like a Titan does so I can mine anywhere I want to without worrying about custom poolside code and special miner settings per pool. Also Inno have been responsive and are working to resolve the firmware problem for everyone. It may be slow but it is happening.

Full credit and kudos to Gentarkin for his work in saving the Titan from all the issues they have but come on folks the A4 is a miles better proposition from a stability/reliability perspective. Also, my recollection is that a Titan uses something like 1300+ W at the wall, even as much as 1500W https://bitcointalk.org/index.php?topic=170332.41700

With respect to ROI, both calculators used above are useless, unless you beleive in a perfectly linear diff and price growth/decline, and as know the diff rate can fall just as fast as if can rise (see LTC diff rate between Dec 2014 and Dec 2015) while the price in truth has been very stable until a week ago.

If you're mining for profit, you should either mine on a multi-pool, or set up your miner so you can switch coins/pools yourself and use http://whattomine.com or coinwarz to help you decide what coins to mine.

Yesterday I was paid (after fees) 0.0000371 BTC per MH on Prohashing so a single A4 would earn you approx $9.35 per day at 260 MH/s.
If your electricity costs $0.12 per Kwh, it would cost you $3.16 per day (@1100w) giving a profit of $6.19.

$2000 (shipped) / 6.19 = 323 days ROI.

Of course, BTC could drop, so could LTC, the diff may fall, or it may rise, who knows, but 0.0000371 BTC per MH is a much better way of looking at ROI than using a single coin calculator, and if want a historical data on BTC per Mh, look here: https://poolpicker.eu/table?algo=scrypt, (just don't trust all the data  )







 

Note: perhaps a tiny bit less with Gen Tarkin 3rd Party Firmware mods. I get 1200/300mh more or less. (may vary according to temps and your equip)
but a bit better then 1250 these days. Also if you do the maxumark (on here) brackets with individual heatsinks and 3000 rpm noctura fan to replace the
knc 1500 rpm..that may have helped this along as well)

but yeah oranges and oranges ..we knc titan owners also feel your pain. (and the all seeing eye of evil asic scrypt manufacturers and their evil ways)
different boats....still its a rowboat

the meme below explains how the asic makers see this (knc/innsilicon/bfl etc)



anyway same issues ...different year...it seems for asic miners

Possible, but you get into things like how do they distribute/queue work to the cores, what is the interrupt method when a core finds a hash, how does it distribute/collect work, etc. I remember when the guy built the chili, he got a 10% hash improvement by totally redesigning how work was sent to the cores.

Stuff like that. Once they open source the software you can fiddle with it and see what happens. Then again you might have to reprogram the MCU to get stuff like that.

Hm.

I think there is still some tweaking needed on the A4's.  We see them very very briefly when first started roar to an impressive hash rate then over time they just keep slowing down. 

I have played with mine some today setting it at the next highest MHZ setting and the miner takes off like a shot, nice numbers very briefly, then the speed gradually slows down to the same hash rate that it was hashing at with the lower MHZ setting. 

So there really isn't any point in doing that exercise I found.  Also interesting is the miner says that it is going roughly 5mhs faster on the miner side, but after the dust settles and it finds the spot it wants to hash you won't see any speed bump at all on the pool side. 

 

Interesting. Titans were built around the 20u die sets, I wonder if the efficiency came from their jamming close to 250 cores on a die. I think they just built the damn things with wizard magic or something.

The pool side hash rate on the A4 (240MH) yields effectively the same power/MH ratio as a Titan.

Titan: 1250/300 = 4.167
A4:  1000/240 = 4.167

You can move these numbers around but these machines compete toe to toe for profitability.

Who mines LTC? I have in my auto-switch SQL logic, but it never rotates through.
Mostly getting GAME, TIPS, EAC and a bunch of those. Typically 120% over LTC

Part of the reason difficulty is going up is due to people with older/less efficient miners getting back into the game for awhile due to the price increase. That should settle out as the GAWs and the A2's drop off again.

I don't know what else there is besides Titans and Inno, and the extra power savings from the A4 doesn't really show a magnitude of improvement over the titan.