Miner- SILENTARMY v5

Mining
1031

Even more dramatic CPU usage savings and PCIe bandwidth savings are now committed, thanks to on-device filtering of invalid solutions: filter invalid sols on-device (large PCIe bw and CPU savings) · mbevand/silentarmy@146b8dc · GitHub

PCIe bandwidth usage dropped from ~100 MB/s to 500kB/s per GPU! CPU usage should also now be close to zero. (Well except Nvidia because their OpenCL implementation implement busy waits, but I’ll check in a workaround soon.)

As always, check the changelog which I always update in real-time:

4 Likes
1032

It is good that CPU use is reduced because i had a huge crash on 16.04/SA 3
I got kernel oops, had to reinstall everything, could not even login anymore. tried safe boot and boot-repair. Nothing.

the only thing I did was to connect unpowered monitor to power strip and to the rig.
oops happened. These things are so flaky, you cannot touch anything :unamused:

1033

Can somebody give me his sol/s for a RX 470 4GB with SA ?

1034

I’ve got 36 sols/s, but with unmodded GPU, gen1 risers and weak CPU (G32** series)

1035

I posted mine about 30 posts up from here…

1 Like
1036

fluctuating between 50 and 55 S/s

@mrb: with claymore’s it’s now around 70 S/s, hope you get better soon :slight_smile:

1037

I noticed the new Claymore seems to be Windows-only.

1038

It seems this was very polite way to say ‘f**k you’ to people accusing him of repackaging sa miner (which is for linux) and adding fee.

1039

yeah he’s pissed off with linu users.

1040

I am pretty sure that linux/SA will catch up in speed with claymore’s windows version.
I REALLY don’t want to install more Windows, although i did pay for one more license already.
My single win 10 rig periodically updates itself and crashes.
i thought that i switched the darned update off, but it reanimated itself somehow.
It is such a pain in the neck dealing with this.
i prefer my rigs being unattended: one eth rig did not require anything for more than 30 days on fglrx/14.04

1041

I was not that worried about the speed difference, because I was mining zcl with better profit than zec, but now it’s on whattomine and the net hashrate is skyrocketing >>>>:-(

1042

Some day it will.
The problem is latest silentarmy is slower than claymore 1.1 for linux.
For some cards by a lot.
R9 390 on fglrx gives me 40-42 for silentarmy and 52-55 for claymore 1.1.
Latest claymore is even faster.
So for r9 390 on fglrx it makes no sence to use silentarmy.

1043

And this is bad for miners on Linux only.
Windows miners (Claymore 4.0) gets additional hashrate.
Therefore difficulty raises to compensate.
Since daily coins suply is approxymately constant Linux users will get less coins
than Windows users.

1044

But isn’t there a “dev fee”, so windows users and linux users will probably end up with the same, after you take off the “dev fee”

1045

The hashrate for R9 390 which I can measure right now is:

  1. 40-42H/s for silentarmy
  2. 52-55H/s for claymore 1.1
    Both ubuntu 15.04 with fglrx.
    Latest claymore (windows only) stated to be even faster (by a lot).
    Since “Dev fee” is only 2.5% Linux users does not end up with the same at all.

Maybe it is not so bad for RX 480/470 (have few rigs but too lazy to test)
but for previous generation cards Linux users got screwed.

1046

Claymore v4 using RX480 is doing 72 sol/s per card. You do the math…

1047

That’s an amazing performance jump. What type of sol/s are we looking at for an RX480?

1048

I get 50-52 H/s on a R9 390 ubuntu 14 with fglrx
latest sa miner, compiled with OPTIM_SIMPLIFY_ROUND 1

But it would be nice to see 90 H/s in linux :slight_smile:

1049

And I get 40-42H/s with OPTIM_SIMPLIFY_ROUND 1, current silentarmy git
on headless Ubuntu 15.04 with fglrx.

1050

I’m getting an error:

Mining on 4 devices pipe closed by peer or os.write(pipe, data) raised exception.

Update just had to restart terminal for some reason. Working fine now.