Asus Strix 480 - problems

[panda86]

Hello,

I'm using 6 GPU Asus Strix 480 8GB with ASRock PRO BTC R2.0 and Claymore 9.3. I have problems with freezing. Bios is modded but only timings (1750 -> 2000) and Max Power Limit set to 130W.

Maximum stable time which I success to archive is 10h and after that system crash. I use different parameters but I cant get stable work.

Does anyone had same problems with Strix 480? Can you help me with setup, do you have tested bios for strix 480 8GB (samsung)?

Everything is powered with one 1200W PSU, and usage is about 1160W when everything work.

Thanks in advance.

[rawnet]

Hi,

I also have the same cards with Samsung memory and after copying 1750 straps to 2000 I've got 27MH/s- whats Yours hashrate? Do You undervolted it?

[Elder III]

It sounds like you are overloading your PSU and it eventually shuts down due to running it so close to the max. Ideally you should only run a PSU at 80% of it's max, as it is now at 1160 you're probably hitting 1200 with spikes. If it's an off brand, low efficiency PSU it's even more likely to not be stable when running it near it's max rated power.

[panda86]

I tried everything, I undervolt for -50mV, -30mV, -10mV, 0mV. I tried GPU clock from 1100MHz to 1170MHz, memory clock from 2000MHz to 2150MHz. I lower powertune -5%.

What setup you use? PSU is gold rated but there is possibility that it makes a problem.

[philipma1957]

I tried everything, I undervolt for -50mV, -30mV, -10mV, 0mV. I tried GPU clock from 1100MHz to 1170MHz, memory clock from 2000MHz to 2150MHz. I lower powertune -5%.

What setup you use? PSU is gold rated but there is possibility that it makes a problem.

pull 1 card

-50mv

gpu clock

1050

memory

1950


see if it runs nonstop  with the five cards and the settings above.

if it does run well  for 30 hours

keep the lower settings  and add the sixth card  if it crashes your psu is too weak.

go back to five cards

[premiero]

It's 99,9% graphic memory realted and heat combined issue. I have 3 rigs based on Asus cards, same mainboards as You have. 2 Rigs 6 x Asus RX480 8GB DUAL and 1 Rig 3xRX480 + 1x Asus R9 290. RM860 / AX1200 / OCZ1250 PSU's. I spent lot of time to diagnose the issue and i can say one thing - avoid to use any of Asus 480 8GB cards. Initially, just like any other i copied 1750 to 2000 memory starps, set all of my 480's to GPU:1010@980mv / MEM:2000@1000mV giving me 28,3 MH/s each while -dcri set to 6 and -ethi set to 16. Than it started. Hangs, freezes, bluescrrens, no signal at all etc. Raising GPU/memory voltage to stock value didnt help at all, neither lowering intensity. So i started to mine with only one card, one by one, to check what's wrong. Out of 15 GPUS: one got RMA, because it even fails at stock bios (no mod at all) and stock clocks (gpu hangs within minutes at PCIEx16 slot, I had to wait almost 3 months for new card!), 4 GPUS cannot handle correctly 2000 MHz memory speed (altough there was no Memory errors at HWINFO), i had to lower it to 1950. Another 2 cards was very temperature sensitive (temps higher than 69 caused hangs - it's very strange cause i tested few other cards and thhey did work well even at 85). One GPU cannot handle even 1950 MHz and i had to set it to 1900 MHz. They are much more stable ATM. GPU temperatures varies from 64 to 73, VRM hit 84. Since then, one rig is rock stable (do not hang at all), other 2 rigs still hang up (and i do not know why, because I have no more "GPU x hangs in OpenCL call,..." i suspect it's network related issue) but now it's like once per 4-7 days, rather than 5-10 times per day. I ended up with stats like below (rejects are due to weak wi-f signal).

115.974 MH/s, 12615/2/1 (0.02%): 28.115+28.110+28.131+31.616 MH/s
164.959 MH/s, 1729/1/0 (0.06%): 27.388+27.393+26.572+27.386+28.089+28.128 MH/s
168.027 MH/s, 2468/0/0 (0.00%): 28.119+28.132+28.128+28.126+28.131+27.388 MH/s

PS1. Claymore 9.1 is more stable for me than 9.3 (single mining)
PS2. There is some corelation between loosing Internet connection and hanging GPU's.

[panda86]

It's 99,9% graphic memory realted and heat combined issue. I have 3 rigs based on Asus cards, same mainboards as You have. 2 Rigs 6 x Asus RX480 8GB DUAL and 1 Rig 3xRX480 + 1x Asus R9 290. RM860 / AX1200 / OCZ1250 PSU's. I spent lot of time to diagnose the issue and i can say one thing - avoid to use any of Asus 480 8GB cards. Initially, just like any other i copied 1750 to 2000 memory starps, set all of my 480's to GPU:1010@980mv / MEM:2000@1000mV giving me 28,3 MH/s each while -dcri set to 6 and -ethi set to 16. Than it started. Hangs, freezes, bluescrrens, no signal at all etc. Raising GPU/memory voltage to stock value didnt help at all, neither lowering intensity. So i started to mine with only one card, one by one, to check what's wrong. Out of 15 GPUS: one got RMA, because it even fails at stock bios (no mod at all) and stock clocks (gpu hangs within minutes at PCIEx16 slot, I had to wait almost 3 months for new card!), 4 GPUS cannot handle correctly 2000 MHz memory speed (altough there was no Memory errors at HWINFO), i had to lower it to 1950. Another 2 cards was very temperature sensitive (temps higher than 69 caused hangs - it's very strange cause i tested few other cards and thhey did work well even at 85). One GPU cannot handle even 1950 MHz and i had to set it to 1900 MHz. They are much more stable ATM. GPU temperatures varies from 64 to 73, VRM hit 84. Since then, one rig is rock stable (do not hang at all), other 2 rigs still hang up (and i do not know why, because I have no more "GPU x hangs in OpenCL call,..." i suspect it's network related issue) but now it's like once per 4-7 days, rather than 5-10 times per day. I ended up with stats like below (rejects are due to weak wi-f signal).

115.974 MH/s, 12615/2/1 (0.02%): 28.115+28.110+28.131+31.616 MH/s
164.959 MH/s, 1729/1/0 (0.06%): 27.388+27.393+26.572+27.386+28.089+28.128 MH/s
168.027 MH/s, 2468/0/0 (0.00%): 28.119+28.132+28.128+28.126+28.131+27.388 MH/s

PS1. Claymore 9.1 is more stable for me than 9.3 (single mining)
PS2. There is some corelation between loosing Internet connection and hanging GPU's.

Yeah, after few modification, I loose GPU x hangs in OpenCL  but it freeze without explanation every few hours. What I found now is UberMix 3.1, memory timing on one GPU bios forum.
UberMix v3.1 memory timing: 777000000000000022CC1C00AD615C41C0590E152ECC8608006007000B031420FA8900A00300000 010122F3FBA354019

After this, my rig works (and still works) flawless without any problem for 5h. I used original bios and change timing 2000 with this one. It is rock stable 175MH/s with sia coin 3160MH/s. I set undervolt -48mV, powertune 0%, core clock 1100MHz and memory clock 2000MHz. First it was set to 2050MHz and it make over 180MH/s.

There I found also another timing:
v2.3 -Less Extreme: 777000000000000022CC1C00AD615B41C0570E152DCB7409006007000B031420FA8900A00300000 010123A46DB354019

I tried it work very good for now.

[premiero]

If You want low temps and lower power draw, set both mem and gpu voltage in WattTool to 900 (don't listen those who write that setting 900 to memory controler does nothing, it's not true, it allows you to lower gpu voltage), GPU core to 1010 MHz, Mem clock to range 1900-2000 MHz. After that my Rigs consume ~800[W] in single ETH mining. Same settings for XMR varies between 700-750[W]. All my Rigs are now rock stable (no crash since last post). Temperatures between 56 to 68 degree at 56% fan speed in closed 20m2 area with poor ventilation.