fireice-uk / xmr-stak-amd

Monero AMD miner
GNU General Public License v3.0
193 stars 114 forks source link

Error 6x VEGA64 -> CL_MEM_OBJECT_ALLOCATION_FAILURE when calling clEnqueueNDRangeKernel for kernel 0. #174

Open ciclonite opened 6 years ago

ciclonite commented 6 years ago

Hi All, I've problem to run XMR-STAK-AMD on windows 10 with 6GPU AMD VEGA 64... MB : b250 Mining Expert RAM : 16GB OS : Windows 10 pro 64Bit DRIVER : Blockchain driver latest version.

`"gpu_threads_conf" : [ { "index" : 0, "intensity" : 2016, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 0, "intensity" : 1600, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 1, "intensity" : 2016, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 1, "intensity" : 1600, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 2, "intensity" : 2016, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 2, "intensity" : 1600, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 3, "intensity" : 2016, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 3, "intensity" : 1600, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 4, "intensity" : 2016, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 4, "intensity" : 1600, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 5, "intensity" : 2016, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 5, "intensity" : 1600, "worksize" : 8, "affine_to_cpu" : false },

],`

Error : CL_MEM_OBJECT_ALLOCATION_FAILURE when calling clEnqueueNDRangeKernel for kernel 0.

clinfo: https://pastebin.com/cAs2F7ST

I've tried everything but i'm unable to resolve... Any ideas?

UnclWish commented 6 years ago

Try to increase amount of virtual memory.

ciclonite commented 6 years ago

I've setup 32GB of inital virtual Memory without success.. Any ideas? Can i try to put 32GB of RAM?

kschroed commented 6 years ago

Got same issue with same config on Vega 56. 8GB Ram w/ 32GB virtual memory configured.

psychocrypt commented 6 years ago

We released a new version of the miner: https://github.com/fireice-uk/xmr-stak Please check if the issue still exists with the new version. If the problem is not fixed so far please open a new issue in the issue section of the new reposetory

ducho commented 6 years ago

Same issue with 6xVEGA56. App is not able working with more than 4x Vega56 cards.

psychocrypt commented 6 years ago

please use the new release https://github.com/fireice-uk/xmr-stak

Ephrimee commented 6 years ago

I have the same issue with 5 vegas in https://github.com/fireice-uk/xmr-stak/releases

chmarti commented 6 years ago

I'm seeing this when running only 2 Vega 56s. 4GB physical memory, 16 GB virtual memory...

"gpu_threads_conf" : [ { "index" : 0, "intensity" : 1932, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 0, "intensity" : 1932, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 1, "intensity" : 1932, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 1, "intensity" : 1932, "worksize" : 8, "affine_to_cpu" : false }, ],

plashani commented 6 years ago

I have the same issue, currently running 2 vega 64- I know that I can get up to 2kh/s per card so the configuration below is not that intense to the point where it shouldn't work.

when running 2 thread on each gpu I get the error message. I have updated to the newest version of xmr-stak but that didnt help. Increased Virtual memory as well.

"gpu_threads_conf" : [ { "index" : 0, "intensity" : 2000 "worksize" : 8, "affine_to_cpu" : false }, { "index" : 0, "intensity" : 1700, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 1, "intensity" : 2000, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 1, "intensity" : 1700, "worksize" : 8, "affine_to_cpu" : false }, ],

Can some please help.

plashani commented 6 years ago

More important question I forgot to ask is, what causes the error message in general? Is this lack of ram allocation? If I reduce the intensity significantly it gets rid of the error but at the same time, the hash rate drops by 50%. These cards are designed to high intensity, so I'm confused by this error.

chmarti commented 6 years ago

Yes, it would be great if someone could comment on why this error occurs...

nicehet commented 6 years ago

I have latest version of app but still the same problem I have 6x Vega 56. Ethereum mining software work fine with 6 vegas 56. Any new versions coming up with this fix?

seminarian commented 6 years ago

I had the same error with rx580's. Upping page file size to 64000 did it for me.

bex10110 commented 6 years ago

I had the same error with trying only 4 x RX580's. Windows 10 with 4 GB RAM and 12288 MB page file size. I upped the page file size to 24000MB and it is working. I'm adding more cards this week and will see if I have to change that again.

Fonta1n3 commented 6 years ago

I am having same issue running 4 AMD 56, was running 1 fine then added 3 only to get this error. Any suggestions?

Fonta1n3 commented 6 years ago

All fixed, it was a virtual memory issue, upped it to 60000 and it runs fine :)

branislav94 commented 6 years ago

I had the same problem ruining only 3x Vega64 and i had only 20000 free. This solved my problem: LARGE PAGE SIZE: Let's your miner access the memory it needs. Open gpedit.exe

With gpedit open:

I found all this on http://vega.miningguides.com/#WinSetup