Topic: [Cast XMR] high speed XMR/CryptoNight miner for RX Vega GPUs (2 KHash/s) - page 7. (Read 206388 times)

I assume, the difficulty drop was also because a lot of miners/workers did not update the software or incurred in technical problems, therefore there was hashrate drop, which also led to the difficulty drop.

Now is back to the new normal, which means waste electricity for pennies. 

There was a difficulty drop initially, but that's almost vanished now, and the new drivers have wrought havoc with my rigs. Deffo one for a rollback and some kind of CN-Heavy action, methinks.

Uninstall iGPU driver with DDU. Reinstall it properly.

Adrenalin 18.6.1, Windows 10 1803, 5xRX580 3xRX480

I'm currently stuck at "Initializing GPU, loading kernel ..." as long as I got my iGPU enabled, once I turn it off it works right away but I would prefer to keep it enabled.

Is there any way to fix this? Tried "-G 1,2,3,4,5,6,7,8" and ignoring the iGPU with the number 0 but that also doesn't help

Mining Monero seems to be a PITA at the moment.
I cannot make myself bothering to troubleshoot and make this software works properly with the new fork and cannot bother to move to SBR miner or wtf is called.
I was doing a lot of tweaking and testing in my early days of mining, about 10/12 months ago, when the gig was much more profitable than these days.
Also considering the 7% performance hit you take moving to the new fork, IMO mining Monero does not make a lot of sense.
So, hopefully our Dev is gong to make, via an update, the transition to v.8 a little more straight forward than it actually is.

In the mean time, I moved to other coins that are slightly more profitable and liquid enough in exchanges. A couple of them may also have more upside potential than XMR.

I am just curious if other Cast users are doing the same.

Cheers.

Have you tried modding the bios to vega 64? As I know the MSI Airboost based on reference pcb and can be modded. Modded bios gives more voltage to HBM and you can get higher freqs.

Um.. this card crashes at anything above 945 mhz.

It depends on HBM clock. If you can get 1100MHz you should get 1800+ H/s about 200W power consumption.

Hi guys! Is 1550/hs average at 200W okay for vega 56 air boost normal? Or should I try aim for higher. Using wattman for tuning it.

I think fast job option good for rig with hash 8k ++ or maybe even 6k ++

I have not. Gonna try it now. Will report on that later.

Every tried --fastjobswitch option?

Alright been mining crytonight v8 using cast 1.5 now for 2 days straight. So far it's mostly been good. I don't use power play tables or do any bios modding i just use the amd software. My results are definitely lower hash rates all around and more power usage just like everyone else. I am running a few systems of 3x vega 56. All are performing at around 4800-5100 sol/s and consuming around 600-615 watts of power for each system.

The downside here has been my acceptance rating at 95%, where 5% of my shares where rejected due to job change. With mining performance going down all around, this part needs to change. Any ideas of how i can improve that number?

As for now, long as these can mine for a long time without me having to do anything, im happy. I will keep messing around a bit to see if I can get hash rate up without using more power.

some of my cards (1 of first rig and 3 on second one ) have temp problem and fan speeds stock in 100% in the noon
can some of u guys share your new oc/uv or give me an optimized ppt?
im currently on p7 1408/905 p3 1100(1060 on some weaker cards)
ive cheked all cord in hwinfo and every thing seems to be normal on hotter cards but they are just hotter!!!
and another stupid question ,what is the max safe temp for hbm?

In my rig nonce invalid error caused by too high HBM clock. Lower it by 50-100MHz and look for those results.

Maybe this will be useful to someone else:  I used DDU to uninstall the blockchain drivers.  I updated ONT to 0.2.6.  I restarted and applied the current AMD Radeon drivers 18.5.2, release date 5/29/2018 according to driver details.  I created the normal exceptions for virus / anti-malware programs for the new version of Cast.  I restarted.  I applied AMDpixelpathcer 1.4.6, hit OKAY and applied those changes and restarted.  I then installed my same old PowerPlay mods.  Restarted.  In ONT, I went with a conservative P7 state of 1386 / P2 1050 for all 7 cards.  That's the setting that I use for weaker cards, but I wanted to get everything running first.

First couple times, Cast would not launch correctly.  So I went into the Cast 1.5 folder and copied the Run command line and pasted that into my older start file (changing -u value back to my address).  The part at the end was a little different from what I had.

After doing that, Cast started right up.  All 7 Vega 56 (bios flashed to 64) are running fine.  Only 1 error after 40 minutes of up time so far.  Hash rates are in the neighborhood of 1,770 with an occasional spike to 1,800 per card.  I set intensity to -7 in the start file.  I'll play around with ONT settings later on, I would expect to get a few cards up to 1,800 without much effort.  I did not do the usual step of using Regedit to turn off cross-fire (change the value to "0") that I always had to do with the Blockchain drivers.

I do have an integrated Intel graphics driver installed (i7 chip) and I did not uninstall that.

I have not updated the Windows OS in some time.  Probably still on last Fall's Creator Update, maybe a version of two beyond that, but I am using a utility to block other Windows updates.

I have not measured power consumption yet, this was more about seeing if I could get everything to work first.

Maybe lucky or something, but really not a hassle at all for Reference Vega 56s and one Sapphire Pulse Vega 56.  The Pulse Vega hits 1,800 more frequently than the reference cards.

Edit:  6 hours later, no crashes, no drama

https://gyazo.com/fa5f8967ab6f199a4a7006edec69233b

Okay, I solved the problem by installing the driver using :

./amdgpu-pro-install --opencl=legacy,pal --headless

and the miner started ok, but I see a lot of rejected shares :

[16:01:25] GPU0 Found Nonce, submitting...
[16:01:25] !Rejected, Nonce Invalid!
[16:01:25] Shares: 8 Accepted, 7 Errors | Hash Rate Avg: 648.9 H/s | Avg Search Time: 58.4 sec
[16:01:25] GPU0 | 61°C | Fan 847 RPM | 642.3 H/s
[16:01:29] New Job received (CryptoNight-Heavy). Avg Job Time: 37.3 sec
[16:01:31] GPU0 | 61°C | Fan 847 RPM | 675.0 H/s
[16:01:32] GPU0 Found Nonce, submitting...
[16:01:32] Share Accepted -> +1
[16:01:32] Shares: 9 Accepted, 7 Errors | Hash Rate Avg: 649.2 H/s | Avg Search Time: 58.3 sec
[16:01:35] GPU0 | 61°C | Fan 845 RPM | 642.4 H/s
[16:01:36] New Job received (CryptoNight-Heavy). Avg Job Time: 35.3 sec
[16:01:36] Difficulty changed. Now: 6568.
[16:01:40] GPU0 | 61°C | Fan 846 RPM | 674.6 H/s
[16:01:44] GPU0 Found Nonce, submitting...
[16:01:44] !Rejected, Nonce Invalid!
[16:01:44] Shares: 9 Accepted, 8 Errors | Hash Rate Avg: 649.3 H/s | Avg Search Time: 58.3 sec
[16:01:44] GPU0 | 61°C | Fan 843 RPM | 641.7 H/s
[16:01:48] GPU0 | 61°C | Fan 847 RPM | 642.3 H/s
[16:01:51] GPU0 Found Nonce, submitting...
[16:01:51] !Rejected, Nonce Invalid!
[16:01:51] Shares: 9 Accepted, 9 Errors | Hash Rate Avg: 649.2 H/s | Avg Search Time: 58.3 sec
[16:01:53] GPU0 | 61°C | Fan 845 RPM | 642.4 H/s


[Job: #13 | Difficulty: 8183 | Running: 3.0 sec | Avg Job Time: 36.4 sec]

[Hash Rate Avg: 649.7 H/s]

[Shares Found: 11 | Avg Search Time: 73.2 sec]
6 ( 55%) Accepted
0 (  0%) Rejected by pool
5 ( 45%) Invalid result computation failed
0 (  0%) Could not be submitted because of network error
0 (  0%) Outdated because of job change

Why is this happening ? Is this because the Linux version of miner is beta ?

I've tried to run Ubuntu version of the miner : cast_xmr-vega-ubuntu_150.tar.gz

I got this :
[15:31:19] Initializing GPU, loading kernel ...
[15:31:19] Error calling clGetPlatformIDs for number of platforms.

Tried to install the AMD GPU driver in different modes :
./amdgpu-pro-install -y --opencl=legacy
./amdgpu-pro-install -y --opencl=pal
./amdgpu-pro-install -y --opencl=pal --headless

the result is the same : "Error calling clGetPlatformIDs for number of platforms."

GPUs : tried different GPUs : RX 400/500 series (not Vega)
Ubuntu version : 18.04
Driver version : amdgpu-pro-18.30-641594.tar.xz

How to fix this issue ?

vega frontiers are am horribly broken with this version on v8

I got power consumption increased by 13%  using Vega56. Maybe you can tweak a little overclock/undervolt