Topic: Innosilicon releases A9 Zmaster 50ksol/s Equihash miner at 620W - page 26. (Read 45686 times)

[quote author=Solsichu link=topic=4432819.msg43172941#msg43172941 date=1532978697]
Can anyone please post the number of cores reported when you miner initializes?  After PLL Clock and VID is set it comes up in the log.

or by SSH cgminer-api -o stats | sed 's/,/\n/g' | grep 'cores'

My 28ksol:
Num cores=370
Num cores=373
Num cores=373

My 50ksol:
Num cores=384
Num cores=375
Num cores=377


its not exactly a direct relation in performance to core count but im curious if the lower hashing units always have lower core counts. 

I haven't Bought a new Oscilloscope yet as im still behind from buying my A9's. Dont even have a good multi meter here. But I see very little variation in 'cgminer-api -o stats' Min/Max voltage and nVol per chip. Its incredibly stable. I would suspect its not updating but I did see a very slight difference in reported value over time in performance mode. Think it was 0.001 Volts. lol.  Over and Under volt does nothing significant to performance. I may have gotten another 1ksol avg from overvolting, but i haven't done a comparison with hard data. Hardware Errors vs Valid Nounces seems to be what I need to compare. Whats crazy is underclocking down to 230Mhz didn't effect the error rate. And the errors are randomly spread across chips in a steady ratio.


cgminer-api -o stats | sed 's/,/\n/g' | grep -E 'found|HW'
00 HW errors=140
00 Nonces found=71
01 HW errors=146
01 Nonces found=74
02 HW errors=177
02 Nonces found=31
03 HW errors=89
03 Nonces found=117
04 HW errors=120
04 Nonces found=73
05 HW errors=103
05 Nonces found=112
06 HW errors=137
06 Nonces found=53
07 HW errors=104
07 Nonces found=118
08 HW errors=146
08 Nonces found=67
09 HW errors=77
09 Nonces found=121
10 HW errors=118
10 Nonces found=102
11 HW errors=61
11 Nonces found=136
00 HW errors=106
00 Nonces found=105
01 HW errors=113
01 Nonces found=107
02 HW errors=125
02 Nonces found=78
03 HW errors=54
03 Nonces found=149
04 HW errors=100
04 Nonces found=121
05 HW errors=100
05 Nonces found=122
06 HW errors=197
06 Nonces found=28
07 HW errors=84
07 Nonces found=138
08 HW errors=13
08 Nonces found=198
09 HW errors=156
09 Nonces found=35
10 HW errors=132
10 Nonces found=87
11 HW errors=179
11 Nonces found=22
00 HW errors=67
00 Nonces found=137
01 HW errors=80
01 Nonces found=135
02 HW errors=204
02 Nonces found=7
03 HW errors=55
03 Nonces found=159
04 HW errors=14
04 Nonces found=183
05 HW errors=63
05 Nonces found=139
06 HW errors=24
06 Nonces found=199
07 HW errors=102
07 Nonces found=98
08 HW errors=64
08 Nonces found=184
09 HW errors=42
09 Nonces found=175
10 HW errors=51
10 Nonces found=149
11 HW errors=68
11 Nonces found=155


So I don't know. My gut says these are bad chips and the supporting hardware is fine. But I cant really know that for sure until I get in there and do some measurements.

I was also playing around with 
'bitmine-a1-options 0:0:390:0' to try and limit the spi speed - no idea if it did anything, took the command tho.
'bitmine-a1-options 0:0:0:6' to disable 6 chips in each chain. Workedm I got roughly 1/2 sols. but not exactly as poor hashing chips are randomly placed. Didnt really let it ramp up.

[/quote]

echo "switchpool|1" | nc x.x.x.x 4028

STATUS=E,When=1532879894,Code=45,Msg=Access denied to 'switchpool' command,Description=sgminer 4.4.2

"api-groups": "H:switchpool:enablepool:addpool:disablepool:removepool:poolpriority",
"api-allow": "R:0/0,W:0/0",

"api-allow": "R:0/0,W:*",

echo '{"command":"stats"}' | nc x.x.x.x 4028