Claymore's Dual Ethereum AMD+NVIDIA GPU Miner v15.0 (Windows/Linux)

[weenkky]

my tests Claymore v14.4
1050 Ti ~ 17.7 strap 5
1060 6Gb ~ 26.5 to 27.6 strap 1 (depends on card)
1070 ~ 33.5 to 34.7 strap 1 (depends on card)
1080 ~ 44.3 strap 2 (without pill)

Any estimate of 1080 ti? I think it should be at about 48 if the basic 1080 is at 44.

I did some quick tests and got 48/49 with 1080TI while Pill gives 52@160W.
1070 arround 34@100W.

I wonder If it will be stable over night.

Are you sure about 34@100W on 1070? I mine Ethereum and one 1070 gives 33.6@110W, another 34.7@114W

[DBIXDBIX]

Teslafunds-TSF(ETHASH) digifinex airdrop

https://bitcointalk.org/index.php?topic=5080832.240

https://bitcointalk.org/index.php?topic=5144043.msg51083552#msg51083552

be ready

[evlo]

I think nvidia performance gains are just troll
12 1060 micron (no strap) @ 635mV (little bit more efficient then pl 70)
ethminer about 275
14.4 clay about 274-devfee
phoenix about 280-devfee, but it drops connections to the pool after about 26 hours and keeps mining, so totally unusuable
reported hashrates - ie. for phoenix it would be about 272 on pool and for ethminer about 269 on pool

clay "-strap 4 -sintensity 1" I need to do more research, but maybe claymore really can change timings on gddr5 nvidias - that would be very very unique and awesome functionality and I don't know of any other publicly available SW that can do that.
Currently I wonder if miner will come back up after restart and still able to mine


Update: so I resterted and got ~284MH/s(-devfee) - I guess maybe in time there will be better straps for nvidias (seems like it is actually applying straps for 1060 and such! WOW). And I also suppose it could get better if cards would not run on 635mV - maybe there is more gain if we let cards bit more juice with better straps?

With 700mV i'm getting 305mH/s per rig @ 900w - which for my power cost is about the same.

But yes, i believe p106 samsung will do 27.5MHs on full power no problem

Update 2: so from the description on first page it seems strap 2 or 3 should be the most intense, now i'm trying 3 and it does not even start minig, so probably I'm on square one and need to lower memory OC etc.
Strap 2 even crashes whole system - nice.

Very interesting and very nice work!

[laik2]

I think nvidia performance gains are just troll
12 1060 micron (no strap) @ 635mV (little bit more efficient then pl 70)
ethminer about 275
14.4 clay about 274-devfee
phoenix about 280-devfee, but it drops connections to the pool after about 26 hours and keeps mining, so totally unusuable
reported hashrates - ie. for phoenix it would be about 272 on pool and for ethminer about 269 on pool

clay "-strap 4 -sintensity 1" I need to do more research, but maybe claymore really can change timings on gddr5 nvidias - that would be very very unique and awesome functionality and I don't know of any other publicly available SW that can do that.
Currently I wonder if miner will come back up after restart and still able to mine


Update: so I resterted and got ~284MH/s(-devfee) - I guess maybe in time there will be better straps for nvidias (seems like it is actually applying straps for 1060 and such! WOW). And I also suppose it could get better if cards would not run on 635mV - maybe there is more gain if we let cards bit more juice with better straps?

With 700mV i'm getting 305mH/s per rig @ 900w - I will need to calculate if it is better then 280 at 800w.

They aren't troll, he really did something. I noted what could happen though in PM

[evlo]

So I'm playing with it and
 -strap 1 -sintensity 100 gives me less hashrate then  -strap 1 -sintensity 80?

[pr3fix]

So I'm playing with it and
 -strap 1 -sintensity 100 gives me less hashrate then  -strap 1 -sintensity 80?

Do DDU uninstall driver, delete the driver. Then, download the latest Nvidia driver, reboot win and try again. My score for now is 42.8 MH/s for P104-100 (-strap 4 -sintensity 95)
Great job Claymore, thank you!  

[evlo]

Much more testing needs to be done, but I will share what I got so far
Micron 1060:
675mV, strap 1,100, mem +600 (on cards that can take it) - 300MHs@850w - so about 25MH/s on a good car, while remaining really restricted in power consumption (lower then PL70!!!).


Tomorrow I will try to find some samsung cards.

So let the finding of new ideal values for each memory type begin yet again.


Yes, seriously awesome work - I mean - it was so hard to believe I thought it was just trolling, I think that speaks for itself

[dema777]

Hi, I have 3 pieces of 1070 and 3 pieces of 1080 in my riga. 1070 work well on strap 1 and 1080 on strap 2, I can’t run all cards on version 14 at the same time, if it is not difficult to show the settings where strap 1 and 2 work simultaneously on different maps. The text is translated by Google. thank 

[br2459]

Ok I have an issue that has been driving me nuts for weeks. It all seemed to start when I upgraded my Desktop/Gaming PC from my Intel i7-3770K to a Ryzen 3 board. The cards are 2 X EVGA 1070 TI's in SLI while gaming and disabled while mining. Before I could run them at +750 Mem clock and they worked great for months without issue. Now I can even go back +600 to get any reliable time out of them. I am constantly having to reset the overclocks. When they crash with either of those two errors below I have to go back to Afterburner and set the Mem clock, power target, and fan. I am hoping someone has had a similar issue and resolved it.


GPU 1, GpuMiner kx failed 1
Set global fail flag, failed GPU1
GPU 1 failed
GPU 1, GpuMiner cu_k1 failed 77, an illegal memory access was encountered
GPU 1, GpuMiner kx failed 1
Set global fail flag, failed GPU1
GPU 1 failed
GPU 0, GpuMiner cu_k1 failed 77, an illegal memory access was encountered
GPU 0, GpuMiner kx failed 1
Set global fail flag, failed GPU0
GPU 0 failed
GPU 0, GpuMiner cu_k1 failed 77, an illegal memory access was encountered
GPU 0, GpuMiner kx failed 1
Set global fail flag, failed GPU0
GPU 0 failed


GPU 1, GpuMiner cu_k1 failed 4, unspecified launch failure
GPU 1, GpuMiner kx failed 1
1Set global fail flag, failed GPU1
GPU 1 failed
GPU 1, GpuMiner cu_k1 failed 4, unspecified launch failure
GPU 1, GpuMiner kx failed 1
Set global fail flag, failed GPU1
GPU 1 failed
GPU 0, GpuMiner cu_k1 failed 4, unspecified launch failure
GPU 0, GpuMiner kx failed 1
Set global fail flag, failed GPU0
GPU 0 failed
GPU 0, GpuMiner cu_k1 failed 4, unspecified launch failure
GPU 0, GpuMiner kx failed 1
Set global fail flag, failed GPU0
GPU 0 failed

[mastercarlox]

Hi.
Thanks Claymore. I have achieved ~56MH with 1080TI's. Seems stable for now.
I used Strap 4 with Pill and Memory 5700Mhz PL 75%-80%

https://i.imgur.com/rJJiuKF.png

[br2459]

I removed the BIOS mod on my RX480, installed AMD drivers 19.4.3 and applied the strap. But I can't get more than 13MH/s out of it. Miner shows it applies the strap just fine but no results. With modified BIOS card runs 30MH/s with blockchain driver. Am I missing something to get back to the 30MH/s without a bios mod?

[fluxy12]

I removed the BIOS mod on my RX480, installed AMD drivers 19.4.3 and applied the strap. But I can't get more than 13MH/s out of it. Miner shows it applies the strap just fine but no results. With modified BIOS card runs 30MH/s with blockchain driver. Am I missing something to get back to the 30MH/s without a bios mod?

Be sure to apply compute mode : -y 1

[br2459]

I removed the BIOS mod on my RX480, installed AMD drivers 19.4.3 and applied the strap. But I can't get more than 13MH/s out of it. Miner shows it applies the strap just fine but no results. With modified BIOS card runs 30MH/s with blockchain driver. Am I missing something to get back to the 30MH/s without a bios mod?

Be sure to apply compute mode : -y 1

Adding compute mode didn't resolve the issue.

Update: Compute mode did solve it. I failed to run it as administrator and then reboot. Thanks for the tip!

Update 2: Well that solved the 13MH/s problem but the temp is out of control now. I dropped the strap to 4 with 100% fan and it's still heating up to 70C in under 5 mins. Not sure what I might be missing now.

[BogdanCo]

Ok I have an issue that has been driving me nuts for weeks. It all seemed to start when I upgraded my Desktop/Gaming PC from my Intel i7-3770K to a Ryzen 3 board. The cards are 2 X EVGA 1070 TI's in SLI while gaming and disabled while mining. Before I could run them at +750 Mem clock and they worked great for months without issue. Now I can even go back +600 to get any reliable time out of them. I am constantly having to reset the overclocks. When they crash with either of those two errors below I have to go back to Afterburner and set the Mem clock, power target, and fan. I am hoping someone has had a similar issue and resolved it.


GPU 1, GpuMiner kx failed 1
Set global fail flag, failed GPU1
GPU 1 failed
GPU 1, GpuMiner cu_k1 failed 77, an illegal memory access was encountered
GPU 1, GpuMiner kx failed 1
Set global fail flag, failed GPU1
GPU 1 failed
GPU 0, GpuMiner cu_k1 failed 77, an illegal memory access was encountered
GPU 0, GpuMiner kx failed 1
Set global fail flag, failed GPU0
GPU 0 failed
GPU 0, GpuMiner cu_k1 failed 77, an illegal memory access was encountered
GPU 0, GpuMiner kx failed 1
Set global fail flag, failed GPU0
GPU 0 failed


GPU 1, GpuMiner cu_k1 failed 4, unspecified launch failure
GPU 1, GpuMiner kx failed 1
1Set global fail flag, failed GPU1
GPU 1 failed
GPU 1, GpuMiner cu_k1 failed 4, unspecified launch failure
GPU 1, GpuMiner kx failed 1
Set global fail flag, failed GPU1
GPU 1 failed
GPU 0, GpuMiner cu_k1 failed 4, unspecified launch failure
GPU 0, GpuMiner kx failed 1
Set global fail flag, failed GPU0
GPU 0 failed
GPU 0, GpuMiner cu_k1 failed 4, unspecified launch failure
GPU 0, GpuMiner kx failed 1
Set global fail flag, failed GPU0
GPU 0 failed

Do you have enough virtual memory?

[UnclWish]

Tried on rig with rx 580 8G hynix memory... All straps up to 4 gives lower speed than my straps with the max 2250MHz MC. Best variant is 2nd strap on 2250MHz. But it still a bit lower than my strap.
Notices:
1. Often if strap set to 0 for use straps from vbios, miner at start can hang the system. But if didn't use -straps option miner always start fine. In vbios flashed eth straps.
2. Often miner hangs system on close if -straps option used.
3. If I use my manual strap (96 symbols) miner can hang on start, and everytime hang on close. But if these strap flashed to vbios and didn't used -straps option, miner works fine.

So, what about AMD hynix MJR 8Gb memory cards problems? And best -dcri settings depends on straps?

[br2459]

Ok I have an issue that has been driving me nuts for weeks. It all seemed to start when I upgraded my Desktop/Gaming PC from my Intel i7-3770K to a Ryzen 3 board. The cards are 2 X EVGA 1070 TI's in SLI while gaming and disabled while mining. Before I could run them at +750 Mem clock and they worked great for months without issue. Now I can even go back +600 to get any reliable time out of them. I am constantly having to reset the overclocks. When they crash with either of those two errors below I have to go back to Afterburner and set the Mem clock, power target, and fan. I am hoping someone has had a similar issue and resolved it.


GPU 1, GpuMiner kx failed 1
Set global fail flag, failed GPU1
GPU 1 failed
GPU 1, GpuMiner cu_k1 failed 77, an illegal memory access was encountered
GPU 1, GpuMiner kx failed 1
Set global fail flag, failed GPU1
GPU 1 failed
GPU 0, GpuMiner cu_k1 failed 77, an illegal memory access was encountered
GPU 0, GpuMiner kx failed 1
Set global fail flag, failed GPU0
GPU 0 failed
GPU 0, GpuMiner cu_k1 failed 77, an illegal memory access was encountered
GPU 0, GpuMiner kx failed 1
Set global fail flag, failed GPU0
GPU 0 failed


GPU 1, GpuMiner cu_k1 failed 4, unspecified launch failure
GPU 1, GpuMiner kx failed 1
1Set global fail flag, failed GPU1
GPU 1 failed
GPU 1, GpuMiner cu_k1 failed 4, unspecified launch failure
GPU 1, GpuMiner kx failed 1
Set global fail flag, failed GPU1
GPU 1 failed
GPU 0, GpuMiner cu_k1 failed 4, unspecified launch failure
GPU 0, GpuMiner kx failed 1
Set global fail flag, failed GPU0
GPU 0 failed
GPU 0, GpuMiner cu_k1 failed 4, unspecified launch failure
GPU 0, GpuMiner kx failed 1
Set global fail flag, failed GPU0
GPU 0 failed

Do you have enough virtual memory?

32GB of Virtual Memory with 8GB of physical mem

[dema777]

Hi, I have 3 pieces of 1070 and 3 pieces of 1080 in my riga. 1070 work well on strap 1 and 1080 on strap 2, I can’t run all cards on version 14 at the same time, if it is not difficult to show the settings where strap 1 and 2 work simultaneously on different maps. The text is translated by Google. thank  

so it was necessary to configure (-strap 1,1,1,2,2,2).
I inadvertently did this, maybe someone will make the same mistake, there is a solution  (-strap 1, 1, 1, 2, 2, 2). wrong

[mozart123]

v14.4 Beta:

- improved mining speed for Nvidia cards by about 1% (depends on card model). Note that dual mining for Nvidia cards suffers from this optimization, use old versions if you use dual mining on Nvidia cards.
- a bit reduced intensity of all Nvidia straps.
- added "-sintensity" option to adjust strap intensity for Nvidia cards, see Readme for details.

кoгдa для Linux бyдeт?

[UnclWish]

v14.4 Beta:

- improved mining speed for Nvidia cards by about 1% (depends on card model). Note that dual mining for Nvidia cards suffers from this optimization, use old versions if you use dual mining on Nvidia cards.
- a bit reduced intensity of all Nvidia straps.
- added "-sintensity" option to adjust strap intensity for Nvidia cards, see Readme for details.

кoгдa для Linux бyдeт?

Teмy читaли, вooбщe? Клeймop пиcaл, чтo нe бyдeт дeлaть для Линyкca, пoтoмy чтo oн в нём нe paзбиpaeтcя. Этo я cвoими cлoвaми вкpaтцe.

[mozart123]

v14.4 Beta:

- improved mining speed for Nvidia cards by about 1% (depends on card model). Note that dual mining for Nvidia cards suffers from this optimization, use old versions if you use dual mining on Nvidia cards.
- a bit reduced intensity of all Nvidia straps.
- added "-sintensity" option to adjust strap intensity for Nvidia cards, see Readme for details.

кoгдa для Linux бyдeт?

Teмy читaли, вooбщe? Клeймop пиcaл, чтo нe бyдeт дeлaть для Линyкca, пoтoмy чтo oн в нём нe paзбиpaeтcя. Этo я cвoими cлoвaми вкpaтцe.

и чтo oпять нa виндy пepexoдить c xaйвa нe peзoннo 120 кapт