Error 381

[Djago]

Hi

I have a nVidia GTX550Ti. It was autodetected as:
[
// gpu: GeForce GTX 550 Ti architecture: 21
// memory: 795/1024 MiB
// smx: 4
{ "index" : 0,
"threads" : 40, "blocks" : 8,
"bfactor" : 8, "bsleep" : 25,
"affine_to_cpu" : true, "sync_mode" : 3,
},
],

I mined a complete day, and suddenly:
[CUDA] Error gpu 0: <C:/xmr-stak/xmr-stak-2.4.5/xmrstak/backend/nvidia/nvcc_code/cuda_extra.cu>:381

If I change threads to, say, 32, I can continue mining. But why? Why it mined a day with 40 and suddenly I can't? What is the correct value?

[ghost]

Top of the head guess, is the card dying? On 10 Jul 2018 19:20, "Djago" <[email protected]> wrote: Hi I have a nVidia GTX550Ti. It was autodetected as: [ // gpu: GeForce GTX 550 Ti architecture: 21 // memory: 795/1024 MiB // smx: 4 { "index" : 0, "threads" : 40, "blocks" : 8, "bfactor" : 8, "bsleep" : 25, "affine_to_cpu" : true, "sync_mode" : 3, }, ], I mined a complete day, and suddenly: [CUDA] Error gpu 0: <C:/xmr-stak/xmr-stak-2.4.5/xmrstak/backend/nvidia/nvcc_code/cuda_extra.cu

:381

If I change threads to, say, 32, I can continue mining. But why? Why it mined a day with 40 and suddenly I can't? What is the correct value? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#234>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AboFkcE0BBMr4C5V4cpL-RVivQtSe6Nbks5uFPBugaJpZM4VJ5QF> .

[Djago]

How can I tell? I'm using it right now and seems to be working fine for everything... except mining...
Now I'm getting the same error on 32 threads...
:P

[ghost]

It just feels like degradation like a downward slope. If it were gaming you'd start to see artefacts and lines etc. But hashing is harder work than rendering so that may take time to show up if my hunch fits. You could listen out for fan imbalance, squealing capacitors or check for excessive heat but unfortunately it may be beyond the senses. You could try running a graphics benchmark tool and see if it crashes. I assume you're up to date with drivers and done all the basic troubleshooting?

…

On Tue, 10 Jul 2018, 20:28 Djago, ***@***.***> wrote: How can I tell? I'm using it right now and seems to be working fine for everything... except mining... Now I'm getting the same error on 32 threads... :P — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#234 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AboFkd6Tb_vh9QRfd9AMhuRO9KAwV3hrks5uFQBUgaJpZM4VJ5QF> .

[psychocrypt]

Reduce threads much more or increase bfactor

[Djago]

It started failing with less and less threads. I've tried with 28, 24, 20, 16 and now it's at 8 since yesterday and working fine. I'll continue like this to check if it's working ok now.

[Djago]

I'm running it with
"threads" : 40, "blocks" : 8,
"bfactor" : 8, "bsleep" : 25,
"affine_to_cpu" : false, "sync_mode" : 3
and it's very stable like this.
Coincidentally these are the same value that XMRIG autodetects!

But there are other problems (see [https://github.com/fireice-uk/xmr-stak/issues/1789]).
So it doesn't seems to be a hardware problem but a CUDA and/or XMR-Stack problem.