ethminer-0.9.41-genoil-1.1

[Genoil]

For added stability during dag generation, add -L sequential to the command line. This is only useful if you have multiple GPUs.

You may also add -SV 2 to the command line for alternative stratum client.

I doubt 1.1 series is really slower than 1.0.8. Personally I feel it is a little bit faster.

What does "-L" do exactly? is it the same as "-T" ? for multiple gpus

It switches the on-GPU dag generation between parallel (all GPU at the same time) and sequential (one GPU at a time)

[Tmdz]

I notice when using ethminer on Windows 7, opencl, AMD 290X that the global overdrive settings for the Fan don't seem to work.
Is this hard coded to override in ethminer?  How can I increase the fan speed (currently on 55%, which isn't bad, but wanting to turn it up)
Also I'm getting 24MH, is that bad?  Can I tweak things to go higher?

Msi afterburner will allow you gain much more control over the gpus so that you can undervolt and change fan speeds as well create custom fan profiles.

24 mh is a bit slow, you should be able to get atleast 25 mh per card or more depending if you over clock.  But from what I read is that can change a bit depending the type of ram the card contains and some other things.  try adding this into the miner bat for starters, --cl-global-work 16384  --cl-local-work 256

[jiggytom]

Anyone hear of a hacked nvidia driver for the GTX1080 so that I can mine at full speed on Windows 10?  I know it requires an older driver, but the older drivers don't support the 1080.  Any help please!

I tried messing with linux install but I hosed something up along the way and haven't had time to try again.

[PeaMine]

I notice when using ethminer on Windows 7, opencl, AMD 290X that the global overdrive settings for the Fan don't seem to work.
Is this hard coded to override in ethminer?  How can I increase the fan speed (currently on 55%, which isn't bad, but wanting to turn it up)
Also I'm getting 24MH, is that bad?  Can I tweak things to go higher?

Msi afterburner will allow you gain much more control over the gpus so that you can undervolt and change fan speeds as well create custom fan profiles.

24 mh is a bit slow, you should be able to get atleast 25 mh per card or more depending if you over clock.  But from what I read is that can change a bit depending the type of ram the card contains and some other things.  try adding this into the miner bat for starters, --cl-global-work 16384  --cl-local-work 256

Thanks for the info! that parameter got me 25mh.
Was there a new DAG just a few days ago?  I remember getting 28-29 not too long ago, or maybe my memory is just bad.
Any recommended core/memory clocks?  I have an R9 Reference card (still going strong) and I believe it's "Hawaii"

[antantti]

I notice when using ethminer on Windows 7, opencl, AMD 290X that the global overdrive settings for the Fan don't seem to work.
Is this hard coded to override in ethminer?  How can I increase the fan speed (currently on 55%, which isn't bad, but wanting to turn it up)
Also I'm getting 24MH, is that bad?  Can I tweak things to go higher?

Msi afterburner will allow you gain much more control over the gpus so that you can undervolt and change fan speeds as well create custom fan profiles.

24 mh is a bit slow, you should be able to get atleast 25 mh per card or more depending if you over clock.  But from what I read is that can change a bit depending the type of ram the card contains and some other things.  try adding this into the miner bat for starters, --cl-global-work 16384  --cl-local-work 256

Thanks for the info! that parameter got me 25mh.
Was there a new DAG just a few days ago?  I remember getting 28-29 not too long ago, or maybe my memory is just bad.
Any recommended core/memory clocks?  I have an R9 Reference card (still going strong) and I believe it's "Hawaii"

Try older 15 series drivers, for some reason these new 16 series give bad performance on windows with genoil's miner.

[lethax]

I'm having issues getting ethminer to work on Windows 7 with GTX 1080 card. I'm using version 1.1.3 It crashes with CUDA error in func 'ethash_cuda_miner::search' at line 303: unknown error. In open cl it crashes with  clEnqueueMapBuffer(-5). Both crash right after the DAG loads before mining starts. the windows nvidia driver crashes and the screen flashes when either of these happen. Any help would be greatly appreciated.

[Tmdz]

Download GPU-Z and take a look at the card. im willing to bet it has Elpida memory which I think is a bit slower than the other one they use.  Hawaii is the name of the gpu processor they use in the 290.

It sounds like we have the same cards with same results, I grabbed 2 of those reference cards from new egg when they went on sale ~ 3 weeks ago.  Over clock really does nothing for it, its the same hash as stock and there is a modified bios thats suppose to improve the timing.  But when I tried that out it put the screen into this checkered pattern and It was nearly impossible to see anything.  So I quit messing around and under volted -100mv in msi and dropped the core clock to 800mhz and left memory at 1250mhz.

[bensam1231]

For added stability during dag generation, add -L sequential to the command line. This is only useful if you have multiple GPUs.

You may also add -SV 2 to the command line for alternative stratum client.

I doubt 1.1 series is really slower than 1.0.8. Personally I feel it is a little bit faster.

What does "-L" do exactly? is it the same as "-T" ? for multiple gpus

It switches the on-GPU dag generation between parallel (all GPU at the same time) and sequential (one GPU at a time)

Still occasionally getting a hang when starting the miner and generating dags even though you can restart it any it seems to work fine.

The problem is when the dag changes when the miner is working it causes the client to hang sometimes for some reason.

1.0.8 runs fine on my machines. I'm not sure what it is. It seems 1.1.3 randomly locks up too while the whole system seems fine. Not sure what's causing it. The epoch change or the miner just having stability issues.

[malekbaba]

I only face problem while DAG becomes invalid. Till date I am mining with almost same pace and getting almost same amount of coins per day with my single gpu

[induktor]

Yup... When a new dag is generated the miner freezes. Was trying to figure out WTF was happening. Quite confusing.

Also it seems like 1.1.3 isn't as stable and hashes about 5% slower compared to 1.0.8.

This happens to me, only with AMD under windows (Nvidia under linux works fine)  with both, version 1.0.8 and 1.1.2
both versions freezes when a new dag is generated.
just close the miner, started it again and it works, but it's downtime until i find out.

[vaulter]

I'm having issues getting ethminer to work on Windows 7 with GTX 1080 card. I'm using version 1.1.3 It crashes with CUDA error in func 'ethash_cuda_miner::search' at line 303: unknown error. In open cl it crashes with  clEnqueueMapBuffer(-5). Both crash right after the DAG loads before mining starts. the windows nvidia driver crashes and the screen flashes when either of these happen. Any help would be greatly appreciated.

Its crashing because currently its super low speed - without the new driver (i cant tell when or if it is coming for win 7/8.1) its pointless to mine on win 7/8.1
For win 10 there will be an update in July and you will be able to mine at ~23mhs
Currently you can set the following parameter for CUDA mode (-U) so it will stop crashing but the speed will be super slow
--cuda-grid-size 2048 --cuda-block-size 128

[bensam1231]

Curious bit here... I overclock my memory on my cards. I wonder if the stability issues and the random hangs when generating the DAG files relate to it becoming corrupt while entering memory and being in memory.

The cards hash fine, but I could see a large chunk of memory becoming corrupted possible causing this to happen. Perhaps some sort of error checking to prevent this from happening? I'm sure I'm not the only one that OCs memory.

[Genoil]

Curious bit here... I overclock my memory on my cards. I wonder if the stability issues and the random hangs when generating the DAG files relate to it becoming corrupt while entering memory and being in memory.

The cards hash fine, but I could see a large chunk of memory becoming corrupted possible causing this to happen. Perhaps some sort of error checking to prevent this from happening? I'm sure I'm not the only one that OCs memory.

Perhaps a separate intensity setting for the dag gen would work.

[carlo_0000]

for the  "-L sequential"  on version ethminer-opencl-0.9.41 it s not recognize

overclock the memory does it really increase performance?

i tryied on r9 380 asus, xfx, sapphire, they don't have same stock memory speed

but i see no difference when increasing memory speed, and 1500mhz is not stable of some of them
so i put them all at 1400 (asus default is 1375, even with 1375 make no difference)

i have only 2gigas's
asus 380's are the fastest  20.8mh for the best one (20.1 for the worst)    best setting --cl-global-work 8192 --cl-local-work 128

sapphire the best 19.3mh ,the worst one 17.6    best setting   --cl-global-work 16384 --cl-local-work 64

xfx did not test yet but it s less than 19mh   current setting --cl-global-work 16384 --cl-local-work 64

for the gpu clock speed asus run at 1010 (default 970), all the other at 1033 (default 1000)

[bensam1231]

Curious bit here... I overclock my memory on my cards. I wonder if the stability issues and the random hangs when generating the DAG files relate to it becoming corrupt while entering memory and being in memory.

The cards hash fine, but I could see a large chunk of memory becoming corrupted possible causing this to happen. Perhaps some sort of error checking to prevent this from happening? I'm sure I'm not the only one that OCs memory.

Perhaps a separate intensity setting for the dag gen would work.

Yeah, not sure if I conveyed that right. My memory OCs are fine when mining, it seems like it might be causing slight corruption somewhere which throws off the files in memory or when they're being generated.  OCs are always stable till they throw a error. The tolerance of a OS and the hardware to handle the errors changes with the severity of the OC. Something like DAG generation or the DAGs that are generated might not remotely tolerate errors or be extremely sensitive to them.

I'm running 1.0.8 again and having no issues, so best guess.

[feriq2]

I've got some rigs running at Win8 with 14.4 whql drivers. Miner 1.0.8 is working just fine (-S -G --farm-recheck 400). However, miner 1.1.3 does not accept/reject any shares.. it just shows [A0+0:R0:F0] all the time. It does show mining speed (pretty similar to 1.0.8 ) in the mining window.. Just not accepting shares.  

[Genoil]

I've got some rigs running at Win8 with 14.4 whql drivers. Miner 1.0.8 is working just fine (-S -G --farm-recheck 400). However, miner 1.1.3 does not accept/reject any shares.. it just shows [A0+0:R0:F0] all the time. It does show mining speed (pretty similar to 1.0.8 ) in the mining window.. Just not accepting shares.  

That's odd, haven't seen that before. What videocards are you using?

[feriq2]

I've got some rigs running at Win8 with 14.4 whql drivers. Miner 1.0.8 is working just fine (-S -G --farm-recheck 400). However, miner 1.1.3 does not accept/reject any shares.. it just shows [A0+0:R0:F0] all the time. It does show mining speed (pretty similar to 1.0.8 ) in the mining window.. Just not accepting shares.  

That's odd, haven't seen that before. What videocards are you using?

These rigs have 5x R9 280X each

[Genoil]

I've got some rigs running at Win8 with 14.4 whql drivers. Miner 1.0.8 is working just fine (-S -G --farm-recheck 400). However, miner 1.1.3 does not accept/reject any shares.. it just shows [A0+0:R0:F0] all the time. It does show mining speed (pretty similar to 1.0.8 ) in the mining window.. Just not accepting shares.  

That's odd, haven't seen that before. What videocards are you using?

These rigs have 5x R9 280X each

Try adding -L sequential to the command line, or alternatively, try using a single GPU to see if that works (--opencl-devices 0). Not sure if it helps, but you could always try. I have 1 7950 (same arch) that works fine with 1.1.3.

[restless]

Or simply copy opencl files from 14.4 into miner directory