bouc
Jr. Member
Offline
Activity: 44
Merit: 1
|
|
May 25, 2016, 07:13:36 PM |
|
Genoil - I KNOW this is a nitpick - but I noticed as I was going through and rewriting a lot of the Ethash OpenCL... this one just bugs the shit outta me: bool update_share = thread_id == ((a >> 2) & (THREADS_PER_HASH - 1));
Were it performance critical, that would be ouch. As is, on AMD, it just is a little eww, when you could: bool update_share = thread_id == amd_bfe(a, 2U, 3U)
Thanks I'll have a look at it. Up until now it's been highly demotivating to try and optimize the OpenCL kernel, because nothing really made a difference. CUDA has been much more willing to give me a few % improvement over the baseline. That in particular doesn't, but it's nicer. Anyways, the AMD OCL compiler really, REALLY hates this. For the first time, instead of using the AMD OCL compiler, and then modifying the output, I'm considering straight up just rewriting the whole thing and bypassing the OCL compiler entirely. maybe with a custom build of sgminer? (did read claymore's thread... ) would be really nice !
|
|
|
|
bensam1231
Legendary
Offline
Activity: 1750
Merit: 1024
|
|
May 25, 2016, 07:15:15 PM |
|
1.1 pre-release is out: https://github.com/Genoil/cpp-ethereum/tree/110/- no more DAG files (both CUDA/OpenCL) - CUDA Compute 2.0 support is back It looks like it's all working but I'm releasing early so you can help me test. Don't forget to remove your -E and -R params, that is all gone now. CPU util seems down, RAM usage down of course. CPU validation is still in there, using the light cache. no devfee, but do send me some ETH if you like it Dual miner and I'll definitely send you some BTC again. Cool idea with the streaming DAG though.
|
I buy private Nvidia miners. Send information and/or inquiries to my PM box.
|
|
|
|
adaseb
Legendary
Offline
Activity: 3808
Merit: 1723
|
|
May 25, 2016, 11:18:30 PM |
|
So what is exactly the issue here? It crashes sometimes when using stratum mode? That's ok since I still use the old eth-proxy
I want to use this before the DAG file switches over tomorrow.
|
|
|
|
Genoil (OP)
|
|
May 26, 2016, 04:52:10 AM |
|
So what is exactly the issue here? It crashes sometimes when using stratum mode? That's ok since I still use the old eth-proxy
I want to use this before the DAG file switches over tomorrow.
I think I fixed that crash, have ran (and others too) a 12 hr test without deadlock. You can download and run it before the change. If you can't wait, I can also send you a demo that switches DAG every 3 blocks but it isn't very profitable mining with it . im getting this error am i doing something wrong? No, I did. The binary only works in -U mode with GM20x GPUs. Will sort that out with the final release.
|
ETH: 0xeb9310b185455f863f526dab3d245809f6854b4d BTC: 1Nu2fMCEBjmnLzqb8qUJpKgq5RoEWFhNcW
|
|
|
Genoil (OP)
|
|
May 26, 2016, 05:07:28 AM |
|
1.1 pre-release is out: https://github.com/Genoil/cpp-ethereum/tree/110/- no more DAG files (both CUDA/OpenCL) - CUDA Compute 2.0 support is back It looks like it's all working but I'm releasing early so you can help me test. Don't forget to remove your -E and -R params, that is all gone now. CPU util seems down, RAM usage down of course. CPU validation is still in there, using the light cache. no devfee, but do send me some ETH if you like it Dual miner and I'll definitely send you some BTC again. Cool idea with the streaming DAG though. It's not my idea, it's Wolf0's (or sp_'s even if you look at it differently) and Claymore implemented it first . I don't have time for a dual, but I also said that about the streaming DAG
|
ETH: 0xeb9310b185455f863f526dab3d245809f6854b4d BTC: 1Nu2fMCEBjmnLzqb8qUJpKgq5RoEWFhNcW
|
|
|
markoz
Newbie
Offline
Activity: 18
Merit: 0
|
|
May 26, 2016, 06:30:29 AM |
|
No, I did. The binary only works in -U mode with GM20x GPUs. Will sort that out with the final release.
ok will try the final version and yes dual mining would be great, if you find time
|
|
|
|
bensam1231
Legendary
Offline
Activity: 1750
Merit: 1024
|
|
May 26, 2016, 07:56:55 AM |
|
1.1 pre-release is out: https://github.com/Genoil/cpp-ethereum/tree/110/- no more DAG files (both CUDA/OpenCL) - CUDA Compute 2.0 support is back It looks like it's all working but I'm releasing early so you can help me test. Don't forget to remove your -E and -R params, that is all gone now. CPU util seems down, RAM usage down of course. CPU validation is still in there, using the light cache. no devfee, but do send me some ETH if you like it Dual miner and I'll definitely send you some BTC again. Cool idea with the streaming DAG though. It's not my idea, it's Wolf0's (or sp_'s even if you look at it differently) and Claymore implemented it first . I don't have time for a dual, but I also said that about the streaming DAG Episolyn and SP both have open DCR miners, even if they aren't as fast as the private SP miner. Pretty sure you can still charge money for open source miners as well. I linked a article on that in the other thread. It'd be a pretty big boost in profits to Nvidia miners.
|
I buy private Nvidia miners. Send information and/or inquiries to my PM box.
|
|
|
markoz
Newbie
Offline
Activity: 18
Merit: 0
|
|
May 26, 2016, 12:07:50 PM |
|
i have another situation on one of my rigs. Speed is only ~65mhs even when ab shows all 4 gpu being used, rig is 4x380, on other rigs is ~80mhs, and on that rig with clay miner is ~80mhs. https://i.imgur.com/4yP2bPF.jpg
|
|
|
|
Genoil (OP)
|
|
May 26, 2016, 12:21:56 PM |
|
i have another situation on one of my rigs. Speed is only ~65mhs even when ab shows all 4 gpu being used, rig is 4x380, on other rigs is ~80mhs, and on that rig with clay miner is ~80mhs. ...
So on the other rigs you have 1.1 running with ~80MH/s, on this rig with CDM also ~80MH/s, but this rig with 1.1 is ~65MH/s? Or do the other rigs run CDM or some other release of mine?
|
ETH: 0xeb9310b185455f863f526dab3d245809f6854b4d BTC: 1Nu2fMCEBjmnLzqb8qUJpKgq5RoEWFhNcW
|
|
|
markoz
Newbie
Offline
Activity: 18
Merit: 0
|
|
May 26, 2016, 12:59:33 PM |
|
yes 1.1 on all rigs this one on this rig with CDM also ~80MH/s, but this rig with 1.1 is ~65MH/s?
|
|
|
|
Amph
Legendary
Offline
Activity: 3206
Merit: 1069
|
|
May 26, 2016, 03:39:33 PM |
|
hashrate seems a bit lower on the new version, but it could be because of the last drivers, i'm not sure
with two gpu i'm doing 44 but with 3 only 61-62, something is wrong
|
|
|
|
induktor
|
|
May 26, 2016, 05:57:18 PM |
|
Hi Genoil I started using your ethminer (108) about a week ago (windows AMD cards) + (Linux NVIDIA cards) work great but, every now and then i get: Stratum unauthorized workerI know this is most likely a pool issue, is totally random in both rigs. when it happens the program keeps processing (hence using power) but no shares are really being sent to the pool!! can you please add a safety feature that will stop mining and retry every now and then, so it won't keep processing, wasting expensive power (at least stop mining would be enough, we can solve the rest with bash script hehe) I kinda fix it in linux (Ugly, I know) i watch for those words in the syslog and if I see them with a script i kill ethminer process. i have no idea how to do that in windows, so I am stuck with this issue with the AMD cards btw: any chance to include a switch to send the output to the syslog instead to screen?, that would be great for us with LCD displays thank you! indkt.
|
BTC addr: 1vTGnFgaM2WJjswwmbj6N2AQBWcHfimSc
|
|
|
Genoil (OP)
|
|
May 26, 2016, 06:25:44 PM |
|
Hi Genoil I started using your ethminer (108) about a week ago (windows AMD cards) + (Linux NVIDIA cards) work great but, every now and then i get: Stratum unauthorized workerI know this is most likely a pool issue, is totally random in both rigs. when it happens the program keeps processing (hence using power) but no shares are really being sent to the pool!! can you please add a safety feature that will stop mining and retry every now and then, so it won't keep processing, wasting expensive power (at least stop mining would be enough, we can solve the rest with bash script hehe) I kinda fix it in linux (Ugly, I know) i watch for those words in the syslog and if I see them with a script i kill ethminer process. i have no idea how to do that in windows, so I am stuck with this issue with the AMD cards btw: any chance to include a switch to send the output to the syslog instead to screen?, that would be great for us with LCD displays thank you! indkt. Hi thanks for the feedback. I've seen this too. Should be fixable. Can't say when
|
ETH: 0xeb9310b185455f863f526dab3d245809f6854b4d BTC: 1Nu2fMCEBjmnLzqb8qUJpKgq5RoEWFhNcW
|
|
|
voffka05
Newbie
Offline
Activity: 52
Merit: 0
|
|
May 26, 2016, 07:26:18 PM |
|
Hm... Rig with only one 7950 working well 2 days. Rig with 4 x R9 270x toxic (default setting) hangs on mining window, not hashing.Random time from 1 to 5-7minutes... restart again & again,allways same result. Miner v1.1 . Now with CDM v4.4 rig working fine 5 & more days uptime. P.S.: sorry for my bad english.
|
|
|
|
Genoil (OP)
|
|
May 26, 2016, 07:48:21 PM |
|
It's weird that this on-GPU DAG stuff seems to make the actual mining much less stable.
I might just pull the CUDA kernels out and do a ccminer integration. See how that fares..
|
ETH: 0xeb9310b185455f863f526dab3d245809f6854b4d BTC: 1Nu2fMCEBjmnLzqb8qUJpKgq5RoEWFhNcW
|
|
|
bensam1231
Legendary
Offline
Activity: 1750
Merit: 1024
|
|
May 26, 2016, 08:02:48 PM |
|
1.1 pre-release is out: https://github.com/Genoil/cpp-ethereum/tree/110/- no more DAG files (both CUDA/OpenCL) - CUDA Compute 2.0 support is back It looks like it's all working but I'm releasing early so you can help me test. Don't forget to remove your -E and -R params, that is all gone now. CPU util seems down, RAM usage down of course. CPU validation is still in there, using the light cache. no devfee, but do send me some ETH if you like it Dual miner and I'll definitely send you some BTC again. Cool idea with the streaming DAG though. It's not my idea, it's Wolf0's (or sp_'s even if you look at it differently) and Claymore implemented it first . I don't have time for a dual, but I also said that about the streaming DAG Actually, I implemented it first - which is how I knew it worked - in my custom sgminer. If you did, you never talked about it, much like other things that actually help normal miners. Gotta keep it all yourself you know. Pallas or Berthrolobo or whatever his name was talked about it last year some time in the SP thread and I kept poking at it, especially after the Claymore release. Which means, as always, it really doesn't even matter what you're doing.
|
I buy private Nvidia miners. Send information and/or inquiries to my PM box.
|
|
|
|
Genoil (OP)
|
|
May 26, 2016, 08:29:23 PM |
|
1.1 pre-release is out: https://github.com/Genoil/cpp-ethereum/tree/110/- no more DAG files (both CUDA/OpenCL) - CUDA Compute 2.0 support is back It looks like it's all working but I'm releasing early so you can help me test. Don't forget to remove your -E and -R params, that is all gone now. CPU util seems down, RAM usage down of course. CPU validation is still in there, using the light cache. no devfee, but do send me some ETH if you like it Dual miner and I'll definitely send you some BTC again. Cool idea with the streaming DAG though. It's not my idea, it's Wolf0's (or sp_'s even if you look at it differently) and Claymore implemented it first . I don't have time for a dual, but I also said that about the streaming DAG Actually, I implemented it first - which is how I knew it worked - in my custom sgminer. If you did, you never talked about it, much like other things that actually help normal miners. Gotta keep it all yourself you know. Pallas or Berthrolobo or whatever his name was talked about it last year some time in the SP thread and I kept poking at it, especially after the Claymore release. Which means, as always, it really doesn't even matter what you're doing. Christ, you're fucking stupid. Giving Claymore the idea wasn't talking about it? please guys, not in my thread
|
ETH: 0xeb9310b185455f863f526dab3d245809f6854b4d BTC: 1Nu2fMCEBjmnLzqb8qUJpKgq5RoEWFhNcW
|
|
|
|
|