Ok, I found an old thread where this mobo was discussed in detail. Looks like the clue to get the pcie x1 slots working is to short pins A1-B17 to fake the presence of gpu in the slot. Adding the link to it here for my own reference for later build... https://bitcointalk.org/index.php?topic=36061.140EDIT 20170508 It turned out that the riser I used (old ribbon cable variant) was not working with the pcie-x1 slots on this board. I have now replaced them with the more modern USB riser variant (version 006) and this riser works out of the box on this motherboard  |
|
|
|
|
Hi Phil.
Thanks for the tips. Yeah, I went through a few hours of trial and error. All the ifs you mentioned were positive.
The risers, and GPUs and PSU came from another rig that is working fine.
This ASRock board has 4 pcie x1 and 3 pcie x16 slots.
Back to the problem; I powered off, removed all the GPUs, rebooted and in the BIOS selected Gen 1 for all the PCIe slots.
Then restarted, powered off again, then plugged in one GPU in the first pcie x1 slot (first slot), and powered up.
After reboot the keyboard doesn't work, the display was showing the BIOS boot screen telling me to press F2 to enter the BIOS setup.
I swapped for another keyboard, it, too, doesn't work. I tried all the available USB 2.0 and 3.0 slots - no difference.
With the 1x riser I then connected the single GPU in pcie slot 2 which is the first 16x slot, rebooted.
Now, no display so no telling what went on, but this time I was able to toggle the keyboard caps lock and see the led for that coming on,
so apparently the keyboard was back on.
I powered off, reset the BIOS settings, then repeated all of the above with same result.
Just to check, I powered off, removed the riser and plugged the one GPU into the first 16x slot
After reboot the rig is back up and all is fine.
Can anybody explain why the keyoard doesn't work when I connect the GPU in the first 1x slot?
|
|
|
|
|
I already flashed the mobo with latest bios, 1.70.
I have set the PCIE -16 slots to Gen 1 in BIOS and am able to boot into Win 7 with all 3 cards connected,
however, OS only sees the two cards in the 16-slots, so it will only mine with those two cards.
Any other thing I can try?
|
|
|
|
|
No, I have plugged the gpus back in the rig from where they came. So they are mining away in the old rig.
So the only thing I swapped was the mobo.
But my focus now is to get the Asrock mobo to work..
|
|
|
|
|
I picked up a second hand mobo for a good price - Asrock Fatal1ty Killer Z87 board.
It has 7 pcie slots - 3 x 16 and 4 x 1, so I thought it should be good for gpu-mining.
I got Win 7 installed with drivers and connected 2 x RX 470 cards directly into two of the x16 slots
and claymore eth miner was running fine.
As soon as I install a 3rd card in one of the pcie x 1 slots via powered riser, then upon
reboot the rig hangs with all the gpu fans spinning seemingly at max (It was quite loud).
I know the riser is good as I was using that one in another rig.
In the BIOS there is only option to set the x16 slots to Gen 1, Gen 2 and Auto. I had all
three set to Auto, so that didnt work. There is no option to set the pcie 1 slots to Gen 1 as
they are already Gen 1, right?
I never got so far as to try setting the x16 slots to Gen 1. If that fails what else should I try?
The PSU is a 760Watt Corsair AX760, the kill-a-watt with two cards mining eth showed 230Watts
so power is ok.
|
|
|
|
|
Yes, you would benefit from undervolting core regardless of the coin/algorithm used to mine it.
280W sounds high for mining ZEC. You probably run with a high core clock.
|
|
|
|
Whattomine has a buttom for RX480. You just type in 14 in the cell left of that button  According to Whattomine you could gross $16.77 a day mining ETH. |
|
|
|
|
Yes, it should work on all Hawaii cards, although I have only been able to verify it work on
XFX,Sapphire Nitro and Asus 390-cards
|
|
|
|
|
Some figures for my cards (all mining eth) all running at near stock frequencies
R9 390 180W
R9 380X 140W
RX470 125W
GTX970 140W
Cards mining XMR:
R9 280X/7970 140W
They are all undervolted except the lone GTX970.
|
|
|
|
|
It depends a lot on the hardware and whether you are overclocking the gpus .
I find mining with the older gpus like R9 280x, 290 and 390 tend to be more stable than say
RX470 and RX480. But then it may be because I have overclocked the memory.
Linux tends to be more stable than Win (7 at least) so long as your gpus are not RX-series.
Also, I find it depends on the mobo. I used to like Asus mobos a lot, I have two Z97-As and they are
very finicky.
With R9 gpus on Linux I could take a 3 week vacation and the miners were still running when I came back :-)
|
|
|
|
Operation Sysytem : Win 7 64bit Wallet : My Monero /https://mymonero.com Miner : Claymore CryptoNote CPU Miner v3.5 Beta - POOL / sgminer 5.5.4-gm CPU: Intel i5-3570 / default settings, no overclocking Speed : 190 h/s (3 threads) Pool: https://minexmr.comVGA Card(s) GPU/Mem: Sapphire R9 390 Nitro/Sapphire R9 380X Nitro/ASUS 380X Strix VGA driver : Crimson 15.12 , slight underclock of core, overclock of mem Speed: 1866 h/s |
|
|
|
My statement is regarding the Linux drivers of AMD... if you install Ubuntu 14.04.01 you should still be able to install fglrx via apt. I dont think I can use fglrx drivers with RX 470. For that gpu under Linux, the amdgpu or amdgpu-pro driver is required as far as I understand. |
|
|
|
Trying to mine XMR with sgminer-gm on ubuntu 16.04.1 with three RX470s, but it segfaults immediately. Anybody can give me a hint? Driver is AMDGPU-PRO 16.50. rawintensity/ws/threads: 1008/8/1 Here is the log: [12:27:09] default_profile.name is xmr
[12:27:09] Started sgminer 5.5.4-gm
[12:27:09] * using Jansson 2.7
[12:27:09] Loaded configuration file ./sgminer.conf
[12:27:09] CL Platform vendor: Advanced Micro Devices, Inc.
[12:27:09] CL Platform name: AMD Accelerated Parallel Processing
[12:27:09] CL Platform version: OpenCL 2.0 AMD-APP (2236.5)
[12:27:09] Platform devices: 3
[12:27:09] 0 Ellesmere
[12:27:09] 1 Ellesmere
[12:27:09] 2 Ellesmere
[12:27:09] Default Devices = all
[12:27:09] set_devices(all)
[12:27:09] GPU 0 Worksize set to 8.
[12:27:09] GPU 1 Worksize set to 8.
[12:27:09] GPU 2 Worksize set to 8.
[12:27:09] Loading settings from profile "xmr" for pool 1
[12:27:09] Pool 1 Algorithm set to "cryptonight"
[12:27:09] Pool 1 devices set to "all"
[12:27:09] Pool 1 lookup gap set to "(null)"
[12:27:09] Pool 1 Raw Intensity set to "1008"
[12:27:09] Pool 1 Thread Concurrency set to "(null)"
[12:27:09] Pool 1 GPU Clock set to "1100"
[12:27:09] Pool 1 GPU Memory clock set to "(null)"
[12:27:09] Pool 1 GPU Threads set to "1"
[12:27:09] Pool 1 GPU Fan set to "34,25,25"
[12:27:09] Pool 1 GPU Powertune set to "(null)"
[12:27:09] Pool 1 GPU Vddc set to "(null)"
[12:27:09] Pool 1 Shaders set to "(null)"
[12:27:09] Pool 1 Worksize set to "8"
[12:27:09] Loading settings from profile "xmr" for pool 0
[12:27:09] Pool 0 Algorithm set to "cryptonight"
[12:27:09] Pool 0 devices set to "all"
[12:27:09] Pool 0 lookup gap set to "(null)"
[12:27:09] Pool 0 Raw Intensity set to "1008"
[12:27:09] Pool 0 Thread Concurrency set to "(null)"
[12:27:09] Pool 0 GPU Clock set to "1100"
[12:27:09] Pool 0 GPU Memory clock set to "(null)"
[12:27:09] Pool 0 GPU Threads set to "1"
[12:27:09] Pool 0 GPU Fan set to "34,25,25"
[12:27:09] Pool 0 GPU Powertune set to "(null)"
[12:27:09] Pool 0 GPU Vddc set to "(null)"
[12:27:09] Pool 0 Shaders set to "(null)"
[12:27:09] Pool 0 Worksize set to "8"
[12:27:09] Trying to set current pool...
[12:27:09] Probing for an alive pool
[12:27:09] Testing MineXMR
[12:27:09] Testing XMRDwarfPool
[12:27:10] Succeeded delayed connect
[12:27:10] Succeeded delayed connect
[12:27:10] Socket closed waiting in recv_line
[12:27:10] Closing socket for stratum XMRDwarfPool
[12:27:10] Failed to parse a \n terminated string in recv_line: buffer =
[12:27:10] XMR AuthID: 951616834173910
[12:27:10] parse_notify_cn()
[12:27:10] Stratum authorisation success for MineXMR
[12:27:10] [THR0] gen_stratum_work_cn() - algorithm = cryptonight
[12:27:10] gen_stratum_work_cn() done.
[12:27:10] [THR0] Pushing work from MineXMR to hash queue
[12:27:10] New block: e64fa5012b10d4c58016325c3988a7c1520623a725458f87e9d8a116a3c3052b... diff 24.4M
[12:27:10] Trying to set current pool...
[12:27:10] Switching to MineXMR - first alive pool
[12:27:10] MineXMR alive
[12:27:10] Trying to set current pool...
[12:27:10] Startup GPU initialization... Using settings from pool MineXMR.
[12:27:10] Startup Pool No = 1
[12:27:10] compare_pool_settings()
[12:27:10] set_devices(all)
[12:27:10] Switching to rawintensity: pool = 1008, default = 1008
[12:27:10] rawintensity -> 1008
[12:27:10] GPU 0 Worksize set to 8.
[12:27:10] GPU 1 Worksize set to 8.
[12:27:10] GPU 2 Worksize set to 8.
[12:27:10] Set GPU 0 to cryptonight
[12:27:10] Set GPU 1 to cryptonight
[12:27:10] Set GPU 2 to cryptonight
[12:27:10] Allocate new threads...
[12:27:10] Assign threads for device 0
[12:27:10] Thread 0 set pool = 1 (MineXMR)
[12:27:10] Init GPU thread 0 GPU 0 virtual GPU 0
[12:27:10] CL Platform vendor: Advanced Micro Devices, Inc.
[12:27:10] CL Platform name: AMD Accelerated Parallel Processing
[12:27:10] CL Platform version: OpenCL 2.0 AMD-APP (2236.5)
[12:27:10] Platform devices: 3
[12:27:10] 0 Ellesmere
[12:27:10] 1 Ellesmere
[12:27:10] 2 Ellesmere
[12:27:10] List of devices:
[12:27:10] 0 Ellesmere
[12:27:10] 1 Ellesmere
[12:27:10] 2 Ellesmere
[12:27:10] Selected 0: Ellesmere
EDIT: Trying to clue in on where it fails I attempted to recompile with the -ggdb option, but the compile fails with algorithm/sgminer-blakecoin.o: In function `scanhash_blakecoin': /home/bobben2/sgminer-gm/sgminer/algorithm/blakecoin.c:143: undefined reference to `blakecoinhash' seems like inline functions do not compile properily when -g is used? EDIT2: I managed to find where it breaks: ocl.c: in function create_opencl_command_queue on return from clCreateCommandQueue() status=6 which I think means out of host memory. Then a 2nd call is made to clCreateCommandQueue with command queue propertiy value set to 0 and this is where the segfault occurs. I can boot the rig into Win 8.1 and there sgminer runs ok (but crashes the rig on exit..) I did a fresh install of XUbuntu 14.04 with Amdgpu-pro 16.30 and a single rx470 card and sgminer-gm has the same problem as I posted above. Nothing wrong with my settings rawintensity 512/ws 8/threads 2. So as far as I can see mining XMR with rx470 on Ubuntu is off for now.. From log: [18:58:48] Started sgminer 5.5.4-gm
[18:58:48] * using Jansson 2.7
[18:58:48] Loaded configuration file ./sgminer.conf
[18:58:48] Probing for an alive pool
[18:58:58] Startup GPU initialization... Using settings from pool XMRDwarfPool.
[18:58:58] Startup Pool No = 0
[18:58:58] Success Creating Context. (clCreateContextFromType)Failed in clCreateCommandQueue [-6]
[18:58:58] Error -6: Creating Command Queue. (clCreateCommandQueue)
[18:58:58] Failed to init GPU thread 0, disabling device 0
[18:58:58] Restarting the GPU from the menu will not fix this.
[18:58:58] Re-check your configuration and try restarting.
Press enter to continue:
EDIT: I am wondering if the problem is caused by the driver reporting wrong number of CUs (and possibly other gpu-resources). When I start Claymore's ethminer (which works fine) it reports 14 CUs whereas correct count is 32 in the case of rx470. |
|
|
|
Trying to mine XMR with sgminer-gm on ubuntu 16.04.1 with three RX470s, but it segfaults immediately. Anybody can give me a hint? Driver is AMDGPU-PRO 16.50. rawintensity/ws/threads: 1008/8/1 Here is the log: [12:27:09] default_profile.name is xmr
[12:27:09] Started sgminer 5.5.4-gm
[12:27:09] * using Jansson 2.7
[12:27:09] Loaded configuration file ./sgminer.conf
[12:27:09] CL Platform vendor: Advanced Micro Devices, Inc.
[12:27:09] CL Platform name: AMD Accelerated Parallel Processing
[12:27:09] CL Platform version: OpenCL 2.0 AMD-APP (2236.5)
[12:27:09] Platform devices: 3
[12:27:09] 0 Ellesmere
[12:27:09] 1 Ellesmere
[12:27:09] 2 Ellesmere
[12:27:09] Default Devices = all
[12:27:09] set_devices(all)
[12:27:09] GPU 0 Worksize set to 8.
[12:27:09] GPU 1 Worksize set to 8.
[12:27:09] GPU 2 Worksize set to 8.
[12:27:09] Loading settings from profile "xmr" for pool 1
[12:27:09] Pool 1 Algorithm set to "cryptonight"
[12:27:09] Pool 1 devices set to "all"
[12:27:09] Pool 1 lookup gap set to "(null)"
[12:27:09] Pool 1 Raw Intensity set to "1008"
[12:27:09] Pool 1 Thread Concurrency set to "(null)"
[12:27:09] Pool 1 GPU Clock set to "1100"
[12:27:09] Pool 1 GPU Memory clock set to "(null)"
[12:27:09] Pool 1 GPU Threads set to "1"
[12:27:09] Pool 1 GPU Fan set to "34,25,25"
[12:27:09] Pool 1 GPU Powertune set to "(null)"
[12:27:09] Pool 1 GPU Vddc set to "(null)"
[12:27:09] Pool 1 Shaders set to "(null)"
[12:27:09] Pool 1 Worksize set to "8"
[12:27:09] Loading settings from profile "xmr" for pool 0
[12:27:09] Pool 0 Algorithm set to "cryptonight"
[12:27:09] Pool 0 devices set to "all"
[12:27:09] Pool 0 lookup gap set to "(null)"
[12:27:09] Pool 0 Raw Intensity set to "1008"
[12:27:09] Pool 0 Thread Concurrency set to "(null)"
[12:27:09] Pool 0 GPU Clock set to "1100"
[12:27:09] Pool 0 GPU Memory clock set to "(null)"
[12:27:09] Pool 0 GPU Threads set to "1"
[12:27:09] Pool 0 GPU Fan set to "34,25,25"
[12:27:09] Pool 0 GPU Powertune set to "(null)"
[12:27:09] Pool 0 GPU Vddc set to "(null)"
[12:27:09] Pool 0 Shaders set to "(null)"
[12:27:09] Pool 0 Worksize set to "8"
[12:27:09] Trying to set current pool...
[12:27:09] Probing for an alive pool
[12:27:09] Testing MineXMR
[12:27:09] Testing XMRDwarfPool
[12:27:10] Succeeded delayed connect
[12:27:10] Succeeded delayed connect
[12:27:10] Socket closed waiting in recv_line
[12:27:10] Closing socket for stratum XMRDwarfPool
[12:27:10] Failed to parse a \n terminated string in recv_line: buffer =
[12:27:10] XMR AuthID: 951616834173910
[12:27:10] parse_notify_cn()
[12:27:10] Stratum authorisation success for MineXMR
[12:27:10] [THR0] gen_stratum_work_cn() - algorithm = cryptonight
[12:27:10] gen_stratum_work_cn() done.
[12:27:10] [THR0] Pushing work from MineXMR to hash queue
[12:27:10] New block: e64fa5012b10d4c58016325c3988a7c1520623a725458f87e9d8a116a3c3052b... diff 24.4M
[12:27:10] Trying to set current pool...
[12:27:10] Switching to MineXMR - first alive pool
[12:27:10] MineXMR alive
[12:27:10] Trying to set current pool...
[12:27:10] Startup GPU initialization... Using settings from pool MineXMR.
[12:27:10] Startup Pool No = 1
[12:27:10] compare_pool_settings()
[12:27:10] set_devices(all)
[12:27:10] Switching to rawintensity: pool = 1008, default = 1008
[12:27:10] rawintensity -> 1008
[12:27:10] GPU 0 Worksize set to 8.
[12:27:10] GPU 1 Worksize set to 8.
[12:27:10] GPU 2 Worksize set to 8.
[12:27:10] Set GPU 0 to cryptonight
[12:27:10] Set GPU 1 to cryptonight
[12:27:10] Set GPU 2 to cryptonight
[12:27:10] Allocate new threads...
[12:27:10] Assign threads for device 0
[12:27:10] Thread 0 set pool = 1 (MineXMR)
[12:27:10] Init GPU thread 0 GPU 0 virtual GPU 0
[12:27:10] CL Platform vendor: Advanced Micro Devices, Inc.
[12:27:10] CL Platform name: AMD Accelerated Parallel Processing
[12:27:10] CL Platform version: OpenCL 2.0 AMD-APP (2236.5)
[12:27:10] Platform devices: 3
[12:27:10] 0 Ellesmere
[12:27:10] 1 Ellesmere
[12:27:10] 2 Ellesmere
[12:27:10] List of devices:
[12:27:10] 0 Ellesmere
[12:27:10] 1 Ellesmere
[12:27:10] 2 Ellesmere
[12:27:10] Selected 0: Ellesmere
EDIT: Trying to clue in on where it fails I attempted to recompile with the -ggdb option, but the compile fails with algorithm/sgminer-blakecoin.o: In function `scanhash_blakecoin': /home/bobben2/sgminer-gm/sgminer/algorithm/blakecoin.c:143: undefined reference to `blakecoinhash' seems like inline functions do not compile properily when -g is used? EDIT2: I managed to find where it breaks: ocl.c: in function create_opencl_command_queue on return from clCreateCommandQueue() status=6 which I think means out of host memory. Then a 2nd call is made to clCreateCommandQueue with command queue propertiy value set to 0 and this is where the segfault occurs. I can boot the rig into Win 8.1 and there sgminer runs ok (but crashes the rig on exit..) |
|
|
|
|
Like Golku said - we use calculators now.
whattomine is good.
|
|
|
|
|
You need an updated version of the PolarisEditor that can deal with memory timings for more than memory type.
Do a search for Polaris BIOS Editor Elpida.
The one I have says Polaris BIOS Editor 1.4.1 - Epsylon3
I have used it successfully to alter mem timings on my 470 4Gb cards.
|
|
|
|
Mining XMR with 1 390 and 2 380X sgminer 5.5.4-gm - Started: [2016-12-31 15:15:28] - [0 days 01:34:13]
--------------------------------------------------------------------------------
(5s):2.024K (avg):1.866Kh/s | A:10484018 R:0 HW:0 WU:22869.408/m
ST: 2 SS: 21 NB: 59 LW: 8233 GF: 0 RF: 0
Connected to XMRDwarfPool (stratum) diff as user 47sghzufGhJJDQEbScMCwVBimTuq6L
Block: 56abfd20... Diff:1.09 Started: [16:49:40] Best share: 43.1M 839316aa
--------------------------------------------------------------------------------
[P]ool management [G]PU management [S]ettings [D]isplay options [Q]uit
GPU 0: 55.0C 1572RPM | 731.2/ 740.4h/s | R: 0.0% HW:0 WU:9167.631/m rI:640
GPU 1: 63.0C 1321RPM | 562.7/ 562.5h/s | R: 0.0% HW:0 WU:6892.882/m rI:512
GPU 2: 64.0C 1041RPM | 562.8/ 565.2h/s | R: 0.0% HW:0 WU:6808.895/m rI:512
--------------------------------------------------------------------------------
GPU 0: 0.7 / 0.7 Kh/s | A:140 R:0 HW:0 U:4.21/m I:0 xI:0 rI:640
54.0 C F: 34% (1577 RPM) E: 1000 MHz M: 1500 Mhz V: 0.000V A: 100% P: 0%
Last initialised: [2016-12-31 15:15:27]
Thread 0: 0.4 Kh/s Enabled ALIVE
Thread 1: 0.4 Kh/s Enabled ALIVE
GPU 1: 0.6 / 0.6 Kh/s | A:106 R:0 HW:0 U:3.19/m I:0 xI:0 rI:512
63.0 C F: 40% (1322 RPM) E: 980 MHz M: 1580 Mhz V: 0.000V A: 100% P: 0%
Last initialised: [2016-12-31 15:15:27]
Thread 2: 0.3 Kh/s Enabled ALIVE
Thread 3: 0.3 Kh/s Enabled ALIVE
GPU 2: 0.6 / 0.6 Kh/s | A:99 R:0 HW:0 U:2.98/m I:0 xI:0 rI:512
64.0 C F: 47% (1042 RPM) E: 980 MHz M: 1580 Mhz V: 0.000V A: 100% P: 0%
Last initialised: [2016-12-31 15:15:28]
Thread 4: 0.3 Kh/s Enabled ALIVE
Thread 5: 0.3 Kh/s Enabled ALIVE
[E]nable [D]isable [R]estart GPU [C]hange settings
[I]ntensity E[x]perimental intensity R[a]w Intensity
Or press any other key to continue
Settings: {
"name": "xmr",
"algorithm": "cryptonight",
"rawintensity": "632,512,512",
"worksize": "4",
"gpu-threads": "2",
"gpu-fan": "34,40,47"
}
I suppose I should clock the cores higher for some more hash, but I am having a lazy year.. Win 7 64-bit Catalyst 15.12 drivers |
|
|
|
R9 280X running at 995/1600 gets me 530 h/s with 2 threads: {
"index": 0,
"threads": 2,
"rawintensity": 484,
"worksize": 8
}
R9 390 running at 1000/1500 gets me ~740 h/s with the following config: {
"index": 1,
"threads": 1,
"rawintensity": 984,
"worksize": 8
},
both on Ubuntu 14.04 using fglrx.. |
|
|
|
|
Show Posts
