|
Agree ver A had few issues as crash after disconnect and also random crashes, B no issue so for
I hope there is more room for optimization, AMD miners is quite competitive on Equihash, 280x 200sol/s and 390x 300sol/s, would be nice if 1070 can at least match 390x with heavy memory OC
|
|
|
|
|
there is some problem with a pool, it keep disconecting, claymore miner working wine
looks like EU server was restarted, shame no failover option to US or..
|
|
|
|
Ok good , now stop .
Work on Nvidia .
Hahahah, Nvidia will wait Spring 2017. The Way It's Meant to be Played | NVIDIA Cya next year  neko ko ce nego balkanska budala  |
|
|
|
clay 8.0 280x 200-210sol 390x 300sol I think its the max from AMD, hope NV 1070 can match 390x soon... |
|
|
|
|
Miner is unstable, when it lose internet or server connection it closes or crash
|
|
|
|
2x exe produce more sol/s and GPU usage 100%, but 0 0 1 1 2 2 does not manage to get GPU at 100%, dev should also look at GPU utilization for more speed! According to our tests, double instances, threads or processes do not result in better performance at all. Really wondering where is the catch that some people are getting more when running two... just try run 2x exe/bat file not two instances in settings but two eqm.exe |
|
|
|
2x exe produce more sol/s and GPU usage 100%, but 0 0 1 1 2 2 does not manage to get GPU at 100%, dev should also look at GPU utilization for more speed! |
|
|
|
Sometimes miner say:
socket.send() raised exception.
duno if that make soma problem
add -v at the end of .bat file for more verbose output. There are few socket exceptions that may cause this. Failed to connect to stratum, auth failed and a few other(I think these two may be the problem) Now I have no error, maybe only few receive job "xx" with no target |
|
|
|
|
As the nicehash zchash pool is closing, what will happen to miner? open source or?
|
|
|
|
|
Sometimes miner say:
socket.send() raised exception.
duno if that make soma problem
|
|
|
|
|
1070, win10
instances=2 gives 99% gpu usage and ~135sol/s
threads=2 does not work on windows 0 sol
|
|
|
|
Still all 0sol but now there is some GPU usage.... I ahd the same with some of the first build when I used bad python |
|
|
|
|
Tested it, win 10, 375.95, 2x 1070 0sol for me
|
|
|
|
Yeah but running Tahiti at 1200mhz engine will require overvolting. Not worth it.
yeah, just for test... tahiti GPU is just weak for blake and zcash to use full memory speed with 384bit bus, maybe only if more optimization can be done to blake on GCN 1.0 |
|
|
|
Claymore V7 seems more Engine clock tuned rather than memory.
Basically all my 7950s are slower than my 7970/280x.
Memory straps do very little now. Just set your core clock as high as you can to get the most performance.
Now memory usage is better and for +160 sol/s you need more GPU power, duno if blace can be optimized more but for ~200sol/s you need +1200Mhz from tahiti 280x and prolly more from 7950 because it has less shaders |
|
|
|
|
rx480 have 8Gbs memory which is fater compared to rx470 so speed is not over theoretical, 480 should be faster then 470 on default
|
|
|
|
|
That is why also on NV 1070 on ETH, memory overclock (samsug) can work +900 (afterburner) and on zcash same card/rig only 600-650, all over is unstable
|
|
|
|
With Claymore V7 some of my GPUs have slowed with the modified bios's, some are faster, some are faster being stock.
Best is the reference R9 290 running at 230 h/s.
Anyone else can report.
Yeah now for full speed you need more GPU power at least on 280x 1000mhz -> 163 1100 -> 180 1200Mhz -> near 200 Also looks like it more now about memory clock and bandwith then about latency |
|
|
|
|
clay 7:
280x modded + OC 180sol, 1200Mhz near 200sol/s
390x - 240sol/s
RX 4xx did not gain any speed
looks like its all about memory speed and some compute for blake ago, so 280x need 1200Mhz to make near 200sol/s
|
|
|
|
Also sliding TDP to 60% and OC-ing you can get undervolt but if miner is low GPU usage it gona "boost" as insane so GPU will crash, I would like to see some new OC tool that can define static freq/voltage to all same GPU in rigs!
Not OC(overclocking). UC(Underclock) the memory of the gpu -500mhz(-1000mhz). Sp-mod #1 is a power saver. Made by a farmer for farmers yeah but for same algo you need OC and less voltage... for instance for lbry ~1970mhz worked on 0.92v fine with memory at -500 stable, was a bit hard to get right values on all rigs, and locking freq/volt you can not save under one profile  |
|
|
|
|
Show Posts
