Review of the S19pro: Let’s get into it

Just got back from cleaning my gear at Clifton. On my way home, buysolar invited me over and handed me the s19pro. Gonna work on the review today and tomorrow.

spacer it is doing close to spec  at the pool

I have an hourly watt meter. so   723.53 is the reading at 720pm EST and 726.86 is the reading at 820pm EST. so 3.33 kwatts. for 111 on the gui and 121 at the pool and 729.98 is the reading at 920pm EST so 6.45 kwatts/2 = 3.225 kwatts for 110.6  on the gui and 112.1pool and 733.17 is the reading at 1020pm EST so 9.645/3 = 3.215 kwatts for 110.1 on the gui and 109.78pool and 739.64 is the reading at 1220am EST so 16.11/5 = 3.222 kwatts for 111.6 on the gui and 110.9pool and 773.53 is the reading at 10:20 am EST so 50/15 = 3.333 kwatts a 1 hour test is simply not very accurate. a 2 hour test is not long enough. a 3 hour test gets power use closer to accurate a 5 hour test show that same 3.2xx kwatt number. a 15 hour test shows that we rose to 3.333 kwatts. and the most likely reason is the room went from 85f to 96f I will do a few more power hash tests tonight. I will post back on power and hash for a day or 2. Gui has changed bigly. notice first pool is missing one 3   should be 3333 at the end and is 333 cue skips and labels it abnormal notice second pool is rejected mmpool.org has this issue with a lot of mining gear and it is labeled abnormal third pool is good. fan speeds are decent as room is not crazy hot. I suspect room will heat up and fans will get louder. I will play with inspect elemental options. Not much luck so far.

I believe this has to do with the fact the mmpool doesn't support AsicBoost and most gears nowadays come with Asicboost enabled by default. I wouldn't be surprised if they changed the way they hide those settings, I mean look at how much access we had to settings that Bitmain thought they kept hidden using some simple methods, anyway, I heard that Vnish are close to finishing their custom firmware for the S19 series, It would be very interesting to see what this beast is capable of with custom firmware.

It is pretty quiet for a 85f room. Would have been nice to get five of  these a while ago. Looks like it uses 3225 watts for 110 th that is 29.3 watts a th so far the one vnish test miner I had did 25 watts a th running an s17pro at lowest power setting. roughly 990 watts and 39.96 th pretty good. I have to think this gear could approach 22-25 watts if it clocks to 95th it is about 1 inch taller 1 inch wider 1 inch deeper. the psu fans are bigger which cuts down noise.

Thanks for the sneak peek Phil.  Looking forward for the full review.

How ironic does it sound that the S17 pro on a custom firmware is more efficient than the S19 pro on stock? I wonder what does it take for bitmain to start thinking about adding more power modes for its new gears? I personally would love to use the stock firmware with a lot more settings available to me, but I am forced to use custom firmware because Bitmain is too lazy to do anything about their stock firmware, which not only lacks tunning and all the stuff you get on the custom firmware, it lacks basic functions like frequency, heck, even fan speed is disabled on the stock firmware.

Yeah buysolar and I spent the day cleaning 🧹 🧼 the gear. I was driving back to Howell in his tesla and he said I could have the s19pro for a few days to review. I will do a youtube and let people hear how loud it is. I have the dead avalon 1066 I can stack them together to show size. Have to say these are very likely the same chips as the s17pro  just a ton of them per board 114. freq is 525 . I will give the log tomorrow.

interesting !!! so see a fasson for me, to see the thing clearer .... esseyon to identify the chips: S17 pro has 48 chips per board ... so a total of 144. I took at asar a status on google of S17, which shows 57th with an average of 595mhz on the maps ... https://support.bitmain.com/hc/article_attachments/360031053654/2.jpg Now let's see the S19 ... 114 chip x 3 cards, 342 chips, for 110th and an average of 525mhz, are you following me? let's go on !!! the calculation, simulate an S17 at 525mhz: (57000/595) x525 = 50300ghs see the speed per chip, 50300/144 = 349.3ghs per chip so ... Let's compare to the S19 ... 349.26 x 342chips which gives us 119000ghs ... thanks to this little calculation, we have noticed, that in fact, even the S17pro is faster for a given chip speed! not by much ... add the rounded comma, the different miner efficiency (all different), the different models of S17 ... I also believe that Phil is right, the S19, contains the same chips as the S17 ...

that said i am guessing decent firmware rather than bitmain shitware would help the gear. history log. there is a very interesting line in the log 1970-01-01 00:00:10 This is fix-freq version and it did let me set fans to 70% 1970-01-01 00:00:22 fan_eft : 1  fan_pwm : 70 note 1 freq: 1970-01-01 00:03:13 min freq in eeprom = 525 1970-01-01 00:03:13 fixed frequency is 525 I am not sure what dc power is:

ho yes, very good info !!! you don't have hidden menu ? fax fix is hidden ? And, you have "save config" on this miner ? if yes... is the backdoor to crack it

I enjoyed the math, but the S19/S19 pro use BM1398BB chips while the S17/S17pro both use BM1397AD, both look very different and their cost is a lot different, what is your opinion?

AFAIK wasn't the s19 advertised as using the 5nm node chips vs 7nm for the s17's?

well bottom line is either they suck as a new chip design or the firmware is terrible. in 30 minutes I am going to have a 50 38 hour meter reading. And since the garage is now a very warm room over 92f I suspect this will show 3.333 kwatts x 50 hour. The measurement will be  more accurate. Since it is done over 50 hours. I seem to get a steady 110-111 th number. If indeed the chip is new and different and if the firmware is not crippling it that would be very bad news for bitmain as it sucks. When compared to the s17pro chip which did 25 watts a th with vnish firmware. 38 hour readings 723.53 845.65 122.12/38 = 3.213 k- watts  this is decent number and gui says  110.0 real 111.7 avg. pool reads 111.2 for 24 hour read. here is a you tube video https://www.youtube.com/watch?v=pqoQ0jPxs14 note I don't do sponsorships it is not click bait.

you mentioned "firmware rather than bitmain shitware would help the gear." I tried to find a firmware for my S19Pro but can't find anything available yet! while I can see some of the pools are giving me way less than others ( comparing between pools by connecting every 2 in a different pool).

I don't know how you did the test or what pools you tested. Variance is possible. At the moment no one has issued aftermarket firmware.

I run 2 devices on Slushpool and another 2 on f2pool at the moment I got on f2pool 0.0027 - 0.0030 in slushpool we're talking between 0.0025-0.0022 and the hash rate wasn't jumping upve the 110TH per devices one of them was reading less when I fully transferred them f2pool faced another issue " device disconnected ".

Slushpool pays in PPLNS, your payouts will heavily (in fact almost completely) depend on Slushpool's luck, if their luck for that day is 50% your payouts will be 50% less than on F2pool, of course, the opposite is also true and there are a few other factors involved, and since luck doesn't even out in days, sometimes not even in weeks, you should never use the payout as any means of measurement in comparing pools with different payouts. What matters here (the only thing that matters) would be your average ACCEPTED hashrate at each pool, so you allow the miner to run for 24-48 hours, and then look at the average accepted, if they are more or less the same (which is most likely) then there is nothing wrong. Another point is your connection to each of these pools, ping both pools and see if one is over 200ms then avoid it. Where does this error show? did you use the correct worker? confirmed there are no leading/trailing spaces in all 3 fields?

thank you for explaining the pools difference I got the same point on twitter today while asking someone who showed that his 1 85t devices making way more than my 4 S19PRO. it's showed in my user account tried today to reset all the devices and reconnect them and now am facing 2 overheat devices tooked the log and just now get back from the mining area so give me some time I'll upload the log if you allow me to complete the case am facing (cause I feel that I jumped off topic).

Only the admins will have the authority to delete your post, and I suspect that they will since it would be kind of off-topic, so the best thing to do would be to start your own topic in the proper section which is the support board, try to be as detailed as possible, post the kernel log, screenshots of the miner status page as well the pool settings.