Topic: Optiminer/Zero v1.2.0 (AMD+NVIDIA GPU) Linux+Windows (Read 13245 times)

hi i am facing this problem at simpleos/linux
my gpu's are Gtx 1060 6 Gb.

Hello!
My miner reboots every 3-5 minutes. How to fix it?
6x1070, Win-7, RAM 8Gb

this coin could definitely need help from Gtx 1060 6gb cards. since there are a ton of them. i am also waiting for the optiminer for 1060s so that ill mine them. ethereum difficulty is really dead serious.

Please could you release the 6GB version me and 5 other of my friends would love to mine it and we all have 1060 6gbs. it would be awesome!

info?
which cards are you using? the start file included is already good for work you just need to change your user and password of nova and its ready!

Hello all,
I change BAT file with my infos but it did not work. Could someone share a BAT file as an example?
Thanks a lot.

I have nvidia 1070 8gb called device d0 and amd 580 calles device 0 too... How i can make them work separately?

someone is mining with gtx 1060 6gb??  because i cant see them working

Use 1060 GTX 6 gb, too many times get autoreboot miner, because get this message :

[error] Found solution index out of range: 3476816541
[warning] [GPU 0] Share submission failed: job not found

 

hashrate of 1080ti?

I tested version 1.1 on my R9 Nano 4GB (Win 7) with Adrenaline drivers and it does not work. No errors are reported, the command line simply goes to 'pause'. Are 15.12 drivers an absolute requirement?

And I wonder, would there be any additional performance increase from Vega Frontier Edition's 16GB RAM compared to Vega 64's 8GB? Maybe it would make sense to add support for it...

no one can post a work miner for 1060 6gb?
i have 1060 3gb cards but i think its impossible to mine?

I cant use my gtx 970 4Gb cards with the 4GB version

Hello Optiminer,

Version for 6GB cards would be really nice, i would really appreciated it.

Yes, sir! That would be wonderful!
I have 1060 6gb and looking forward to start mining zero on windows.

Does any one know what is the 1080ti OCed performance?

To overcome the issue mentioned in my previous post, I ran each platform (AMD & NVIDIA) on a different optiminer instance running concurrently, using the -c switch.

So regarding the Nvidia platform: I tried to mine zero with three GTX1070 gpus using optiminer zero 1.2, on Windows.
The miner works for about 10-12 min (or even less time) and then it stops mining.
I am using miner-entry pool, if this is relevant.
These are the lines that usually show up:

[2017-11-21 11:13:54.345] [info] [GPU1]  4.8 I/s 11.4 S/s (5s) 5.0 I/s 9.7 S/s (1m)
[2017-11-21 11:13:54.873] [info] [GPU0]  4.8 I/s 9.0 S/s (5s) 4.9 I/s 9.4 S/s (1m)
[2017-11-21 11:13:56.757] [info] [GPU2]  4.8 I/s 8.2 S/s (5s) 4.9 I/s 9.6 S/s (1m)
[2017-11-21 11:13:57.948] [info] [Total]  15.0 I/s 27.8 S/s (5s) 14.8 I/s 28.8 S/s (1m)
[2017-11-21 11:13:58.239] [error] OpenCL error: readBuffer Solutions0 (CL_OUT_OF_RESOURCES)
[2017-11-21 11:13:58.244] [info] Press any key to exit application.

I have tried different gpu settings (default or overclocked core and memory clocks) and different optiminer pci-modes (zero up to three) as well as the experimental-kernel setting without any luck.
The miner either hangs or produces the above message.

On the same rig, I have two AMD RX cards which as explained above, I use to mine on a different instance of the miner, using the -c switch. These run without any problems.

Any ideas? I mean, besides switching to ethos/linux?

I 've used help and I could see that the platform is different, but this still doesn't answer my question how to properly use the -d parameter to select one or a few of the gpus.
For example, how to select only gpu 0 on platform 0 AND gpu 0 on platform 1?

optiminer --list-devices
check platform (-c) 0 or 1

you could have use help

I have a rig with two AMD and three Nvidia cards. Optiminer assigns number 0 and 1 to amd cards and numbers 0, 1 and 2 to the Nvidia cards. How can I distinguish between the cards and choose some of them, if they share the same numbers?

A version for 4 GB GPU,
A version for 8 GB GPU, x2 faster than the 4 GB
...and a version for 11GB GPUs, like nVidia GTX 1080 Ti ?

I get the same when testing GTX 560 Ti for fun with OptiminerEquihash 2.0. Is this a matter of SM version being too low?

yes. the same quastion

I can't make it work with GTX980 4GB (full 4GB) card.

1.1 version tells: "no matching device found (even if it lists it with --list-devices)"

1.2 version tells: "ptxas fatal   : SM version specified by .target is higher than default SM version assumed"

Any help\suggestion?

Vega is not support yet.

I tried mining Zero with 1 X VEGA56, but it will not work. What's the problem? Maybe the miner does not support AMD Vega?

Principally it should work with multiple NVidia CPUs. I suggest to try running it on Linux.

Fixing the high CPU usage will require rewriting the kernel in CUDA which is a major change, so will take time.

The miner only supports stratum mining at the moment.

is this miner support solo mining?

Hey Optiminer, any chance you could fix the Windows miner to allow for multi-GPU configs (nVidia)? Currently it completely saturates the CPU, and ends up crashing the rigs within minutes...

You mean for the 6GB version?

I have gtx 1060 card is there a version which works with my GPU's ?
I hope you will release it soon optiminer

The best thing to get some Zeros then is probably just buying them. The price still is really low as the coin is undervalued.

Yes I understand.. so really the solution is to upgrade my GPUs however I just don't want to do that right now because of cost. I am working with two GTX 970s so unfortunately 4GB would be my only option 

I do appreciate your quick reply!

I could release a 6GB version for nvidia. Would that help? The problem with a 4GB version is that it would be much slower. You can see this in the AMD version, the 4GB version runs about at half the speed.

Hello my friend, i am really hoping that someone will come and say, hey, look, add -BLA BLA on the bat file and that's ok, or the next version working.
I don't know much of linux, you know ? I just used ubuntu for a year because it was simple to use, i don't know if i can format everything and go for linux. haha
thank you for your kindness, i will wait for some answers.
My best regards!

Hey optiminer, do you have any plans to create a 4GB nvidia version? Just curious, not here to badger about it. I'll just buy Zero instead if not 

No panic, it has become even easier, simply ubuntu 16.04 install and the nvidia driver install, everything automatically, you do not have to configure the driver, it is all recognized and you can use directly all graphics cards

I will put it here to see if we can come to something

Sometimes is mining ok, and then stops and give a build status -2
http://imgur.com/cn8wZdA

cl device not found closes my mining after some time
http://imgur.com/Drw7BqP
http://imgur.com/cbbdKAF
http://imgur.com/UN3A8AO

cl out of resources closes too
http://imgur.com/S5msEzE


sometimes closes the mining and give the pause command
http://imgur.com/JSK6uNX

 Undecided 
Im sorry to bother you but i really don't know what else to do.
My virtual memory is 65gb for 7 1070s Huh

What do you guys suggest ?
thank you!

hey dude,

iknow this problem, take linux and this miner in linux and then are all problems solved

regards

what do you mean its not true and closed to a miner ?
:joe
command
goto :joe is like one of the oldest commands and it is true 

It's not true!
Anyway I have found a script which doesn't allow to be closed to a miner =)

ah okay !

well its not wrong i just meant the loop command not the whole thing i just needed device 1 2 3 on platform 0 to work so that worked for me.

thats wrong optiminer -s zero.suprnova.cc:6568 -u (user.wn) (-p wp )-m 8080 (--device 1 -d 2 -d 3 -c 0)

optiminer -s %POOL% -u %USER% -p %PASSWORD% -d 1 -i 1

SET POOL=zero.suprnova.cc:6568
SET USER=user.wn
SET PASSWORD=wp

If you only want to take the first graphic card then  -d 0   or  -d 1 -d 2 etc
If you want to take all you leave the entry free

(for help  only this entry in start.bat    
optiminer --help)

or run benchmark

regards

this makes it loops around if closes by itself.

Rig=4 1070 1 1080.

 also my assumption is the processor utilization,
 But this comes from the interplay (Low CPU - Strong GPU - not optimal programming Mining software.) I had previously 6 x GPU Gtx 1060 with EWBF Miner with 8 GB Ram no problem, now it is 2 GTX 1080 TI with 16 GB Ram and it Runs only under Linux properly

Ha Ha yeah, well, i guess its Nvidia then, i'm aware it takes up much more CPU power for instance. My guess is that the windows geforce drivers aren't playing nice with the internal bus NB/SB. Maybe motherboard BIOS? I don't have an Nvidia rig to test it out though, but I would recommend a "newer" paging Hard drive that was designed after a time workstations only had 4gb of ram.

maybe playing with the sector size could be better for stability and/or even performance in the form of consistency, but again in the wild world of pagefile, everything depends.
*EDIT* tested different cluster sizes, got strange results, but I only have a 512 physical to play with (probably pushing it) and don't feel like formatting my 4kb main drive.

You agree 100%, but under windows 10 Prof. 64 had the increase of the virtual memory no effect,
Also retrofit the physical memory
Or moving the pagefile.sys to an empty HDD was useless
Only the change to Linux brought success.

Shit WINDUMB :-)



regards,
lokko

Good to know, I'm on Radeon/Windows and she's pretty stable, only if i start messing with settings opening and closing the miner multiple times do i get a hangup. Pagefile is important though, as is RAM to a lesser extent. Usually you want to keep pagefile to 1.5 times your RAM at most, and people may have presets limiting the amount of HD space it will use which will crash the miner on startup if you don't have enough allocated. the 1.5 number is just rule of thumb, you can go way beyond that but it comes at a cost of stability, and either way the more pagefile you use say ~32gb, the more you bottleneck your data speeds when doing anything else should you also use it for browsing/entertainment. also having a lot of pagefile will reduce it's lifespan by a bit. all in all having too little = bad. having more than enough = it just works normally.

This is why I recommend having a separate drive handle the pagefile only if you can help it, but ultimately not necessary and in a healthy rig with a healthy amount of ram and a modern HDD, you shouldn't notice a difference. you could always do the USB file transfer test too and see how much it slows down when the miner is running.

and also, don't use an SSD for pagefile, maybe for video games and normal computer stuff it's okay but when you're taxing it with ~30gb and it will not last very long. On the other hand I know Linux handles swapfile much more nicely, but i still wouldn't use that amount on an SSD.

Hello friends, there is only one possible way, under Windows is the Zero Miner not really suitable, I have many days to healthy to solve the problem,

A) Ubuntu Linux 16.04
B) Nvidia driver
C) no more problems

On Windows, it seems to be a driver problem because the virtual memory is practically useless.


best regards

The miner sometimes stops work and is just closed
Can work week, and day can
There is any decision that miner was restarted ?
P.S Rig = 6x480 8 Gb

I think you can have it run on one instance, with the flags  -i 1 -i 1 -d 0 -d 1.

as for virtual memory, you may want to set it a bit higher, at 20gb just to give it breathing room, see if it helps.

same problem here with 2x1080ti.
Use 2 miner instances, one per card.
Still high cpu usage ~60%, but has been stable for 3 days.
win7, 8gb ram, 22gb page file, amd fx 4300 cpu.

Not sure what you have tried... my Zcash miner does not work for NVIDIA cards.

Simple: Because it is the only GPU miner available for ZERO. See first page for a link to more info about this interesting coin

EWBF has a miner for Zero parameters? that would be news to me.
I think you may be confused with Zcash since Opti does have a miner out for that too.

Im really interested in what Vega can do. since it will have 16gb of HBM2, and double speed half-precision floating point operations bringing it to ~25Tflops, I always wondered if mining takes advantage of single or half pass... maybe something special can be drawn up? I would say it's a game changer, but at an MSRP of like 1000 USD and very limited supply many likely going into pre-built systems, there just won't be enough to go around.

this is true, but when shopping around for cards, nothing brings bang for buck like Radeon. Nvidia got better on ETH and it's king of certain algos and definitely best w/h in a lot of cases but they're usually much more expensive, even from a manufacturer refurb, they're still in very high demand too but, obtainable. I guess it's what sort of mining you plan on hedging.

In any case, I have my 390s mining at ~70W 11.1mhs with spikes to the more normal 130W range, one of them doesn't seem to dance much, staying a little higher but it is driving my screen i'm sure if i OCed them a bit and spent a day installing the perfect drivers I could hit +13mh/s each but at the cost of like 140w for 3mh/s. no bad shares either. Even the 390s are going extinct in the wild since they're the next best thing to a 470 at what was usually even under 200$US for a used one.

my 480/580s are nice and cool even with ref coolers in summer heat all under 70c with under 50% fans. going to be treating them with the best paste and pads anyway, without going liquid metal.
I am noticing my GPU1 in afterburner is showing 16% usage and 100% FSB the others read 100% usage and ~40% FSB, it's also the video-out gpu on this rig.

All measurements taken on Gpu-Z/Afterburner, kill-a-watt on its way. so 70W = about ~110W with pcie draw from my understanding. CPU is under 20% (spikes to 28) with some casual browsing, I can even play a couple of games to pass the time with minimal frame penalty so long as i disable that GPU.

well maybe on purpose as they push CUDA, but ppl say there is solution something about GPU waiting on CPU, but I am not a coder so can`t help, not sure there is common fix but more a hard way to find it and make opencl work on NV with no CPU usage

PPS: or time to get used to cuda as AMD HW is out of stock soon we gona have a LOT more on NV hardware + NB mining cards hitting mining space! fu to AMD domination

I fired up the windows miner on my Win 10 Celeron 3900 rig 8 GB of ram with 3 * 1070s and set the virtual memory to min 28 gb to max of 30 gb.  CPU was max'd at 99.0% and would blue screen after about 3 mins.

EBWF mining ZEC puts the CPU at about 20%.  My results confirm what reb0rn21 was seeing.

Looks like I'll have to dual boot Ubuntu to mine this coin...

Great to hear that it works! Yes, the 100% CPU usage is a common issue with opencl and NVIDIA driver.. I do not know why NVIDIA has not fixed that.

2x 1070GTX on i5 cpu win10, works but CPU usage is 100% looks like common opencl NV bug

speed is 10s/s per 1070

its shame RX have same speed

Just put up version 1.2.0 for Windows which should support NVIDIA! Please test and report back. 1.2.0 is also a bit faster for AMD cards, so everyone is suggested to update!

Note that 1.2.0 uses a bit more memory. Make sure your virtual memory size in Windows is large enough! Page size > number of cards * 8GB!

So any word on this coin or any windows miner?  I was running Linux at a time, but went to Windozes to OC and do system tweaks--NVidia fought me in Ubuntu.  I can always make the rig a dual boot and run the Linux miner, but I don't see anyone on the Suprnova mining pool. 

Anyone have a current state of this coin?  Thanks!

Better not. Do you calculation ? Twelve nvidia 1080Ti will demolish this coin. Can not you figure out that this is a coin with a small market capitalization. You'll only make Zero is not profitable. You're just going to bring down a profit the rest of us, and you are leave Zero. This coin is for small miners, king.

Perfect, cant wait for it.
I have 12 1080Ti so I will gladly mine with your miner and contribute some minig fee if you can give us working and profitable Win+nvidia miner

Yes, as soons as I get some time, I will compile a windows version.

It depends on your cards, for the most part yes. 4GB=Maybe   8GB= Most likely. all depending on difficulty and exchange rates of what we're comparing.

*Edit*
Like today, not really worth it.

Are you planning to release any win version for NVidia?

Is Zero more profitable than zcash?

Linux version 1.2.0 released with support for NVIDIA cards! Requires GPUs with at least 8GB of ram.

https://github.com/Optiminer/OptiminerZero

Windows release still pending.

anyone know if it's possible to run this in the panda miner b3

I'm getting these funky messages on my secondary rig. My cards were working fine in my primary rig but I moved them to the other empty one to space it all out
Anyway One card starts to mine, the other shows slight temp bump but nothing happens and I get this message:

OpenCL Error: Readbuffer Solutions1 (CL_Invalid_Event_Wait_List)

edit:
windows 10 build
G1840
H97 anniversary
8GB 1600ddr3
1000w platinum psu

I tried running two separate instances and whichever i start second, regardless which card, will give me the message and fail to mine. Regardless if my main gpu is in the slot proper or on a riser, or which is which. I'm not getting this problem with my other build that i believe only has 16 pcie lanes, although it is a quad.

**EDIT #2**

Had to roll-back entire system, took a few hours to get it running but it works. my guess is it was something that AMD's driver removal missed and so did DDU.

its basically a pool setup that i use to solo mine .... thats all

Good stuff, thanks!

megacrypto, what's the benefit of using znomp?

yes ... it seems that znomp is not ready yet to handle zero !!

Looks like what you want to mine is Zcash not ZERO. Zcash miner is here: https://github.com/Optiminer/OptiminerZcash

im using znomp with my node and using optiminer keeps giving me:


any pointers?!

Any ETA and speculation on Nvidia support?

yea 4gb , was confused cause the profit from mining this then is way lower than the rest right now.

Good to know


on a 4GB? that would be more than decent, on an 8GB that's somewhat lower than I would expect, if you do have an 8GB model stick to v1.0

What is the sol rate on a rx470 ? Anyone tested?

I am getting around 6 sol/s and unsure if that is "decent" or something wrong.

This warning can normally be ignored. It might be a hint that there is a network issue. In this case the miner would disconnect and reestablish the network connection soon after that.

 Concurrent write detected, queuing.

What did optiminer mean by this??

Hopefully this brings some new life to this coin, I might add another GPU to zero.

See https://bitcointalksearch.org/topic/ann-zero-fork-of-zcash-with-harder-mining-params-1796036

Which Zerocoin is this ?
can I have a link to announcement please!
Thanks.

Yeah.. that's expected. With 4gb the memory access pattern are way worse.

Problem with 4gb cards solved for me ... but it is twice as slow ))

Added release for 4GB cards running under linux.

Yeah, I tried -i 2 as well yesterday on a 8gb card. There must be a bug with handling the parallel instances somewhere. But it's rather pointless anyways as for 8gb using the other version is faster. So, I'll probably just remove the code handling parallel solutions and hardcode it to 1 for Zero.

It should work on any card that has at least 4gb of ram. If it doesn't try to reboot the machine and make sure the miner and nothing else runs.

Fantastic work,

Just wanted to point out using the 4gb miner with an 8gb card, I tried using -i 2 but just cranks out a lot of invalid low diff shares. I can sort of figure out why, my lack of expertise can't explain it.

*edit*
any word on r9 280s?

can someone share the speed with 470/480?

so it wont work for r9 290 4gb and what about r9 280 3gb is there anyway to add support to them or not possible thanks

8 mb version works fine for me - r9 390 8gb and rx480 8gb
4 mb version doesn't work - r9 390 8gb and r9 290x 4 gb - it starts and tries to mine but low gpu utilization and low diff shares
windows, 16.9.1 drivers

Thanks

Yes, you can select the cards with '-d N' were N is the number. Use --list-devices to see which device has which number. Then, just run two instances of the miner, one for the 8GB cards and one for the 4GB cards.

Linux version of the 4GB miner will be published soon.

Does the 4gb work on linux as well?

Is it possible to use mixed rigs?
I haven't seen an option to select which card to mine with in the miner...

I am pleased to announce the new version of Optiminer/Zero for the ZERO coin that 8GB AMD+NVIDIA cards.

Support for 4GB cards is limited to AMD for now (it runs quite a bit slower with fewer memory!).

Note there are different binaries for 4GB and 8GB cards! The version for the 8GB cards runs faster but needs more RAM.

Download: https://github.com/Optiminer/OptiminerZero

For more information about ZERO see https://bitcointalksearch.org/topic/ann-zero-fork-of-zcash-with-harder-mining-params-1796036

Note this coin is different from Zcash/Zclassic/Zcoin!

For a Zcash/Zclassic miner see https://bitcointalksearch.org/topic/optiminerzcash-v17-gpu-windowslinux-amd-1686057