Cuda initialize crash OR Allocating buffers Crash

[elviture]

Finally got my Dual PSU, AMD&Nvidia rig set up working today, rewired the whole rig to add cards with the extra psus (run in parallel with no crossover power for the cards. i am running the newest version of phoenix, 5.5c (as far as i can tell). right now its a 4x rx570, one 1080 and one 1070... looks like the Nvidia cards are causing the crash. i reinstalled the drivers via geforce experience. also, every single card on this rig has been tested previously so its not a card causing the crash. no blue screen either, so it doesnt seem like an overclocking issue.



Semi new to mining. please advise.

More info on the rig: intel 39?? series cpu, Asrock H110 pro BTC+ mobo, Geskill ageis ddr4 ram, all of the pcie risers are ver07, and i am now running 2x EVGA 850G+ PSUs. (I have not yet modded the bios to boost the RX5xx series boys, i will probably do that eventually going forward once I can make this stable.)



everything has been tested separately and it seems all the hardware is working fine. previously i was running a phanteks revolt pro 1000W but it could only handle 4-5 cards at a time... i think 4 was the max i managed to run stably on that setup.



heres' the phoenix 5.5c notes and crash messages, wallet hidden.





\PhoenixMiner_5.5c_Windows>PhoenixMiner.exe -pool eu1.ethermine.org:4444 -pool2 us1.ethermine.org:4444 -wal (XXXXX) -proto 3

Phoenix Miner 5.5c Windows/msvc - Release build

-----------------------------------------------



CUDA version: 11.0, CUDA runtime: 8.0

OpenCL driver version: 20.11.3/20.12.1

Available GPUs for mining:

GPU1: Radeon RX 570 Series (pcie 1), OpenCL 2.0, 8 GB VRAM, 32 CUs

GPU2: GeForce GTX 1080 (pcie 2), CUDA cap. 6.1, 8 GB VRAM, 20 CUs

GPU3: Radeon RX 570 Series (pcie 6), OpenCL 2.0, 8 GB VRAM, 32 CUs

GPU4: Radeon RX 570 Series (pcie 9), OpenCL 2.0, 8 GB VRAM, 32 CUs

GPU5: Radeon RX 570 Series (pcie 10), OpenCL 2.0, 8 GB VRAM, 32 CUs

GPU6: GeForce GTX 1070 (pcie 11), CUDA cap. 6.1, 8 GB VRAM, 15 CUs

Nvidia driver version: 461.40

Eth: the pool list contains 2 pools (2 from command-line)

Eth: primary pool: eu1.ethermine.org:4444

Starting GPU mining

GPU1: AMD driver 20.11.3

GPU3: AMD driver 20.11.3

GPU4: AMD driver 20.11.3

GPU5: AMD driver 20.11.3

Eth: Connecting to ethash pool eu1.ethermine.org:4444 (proto: QtMiner)

GPU1: 23C 40% 6W, GPU2: 28C 40% 8W, GPU3: 20C 40% 6W, GPU4: 22C 40% 32W, GPU5: 24C 40% 34W, GPU6: 31C 41% 7W

GPUs power: 92.3 W

Eth: Connected to ethash pool eu1.ethermine.org:4444 (172.65.207.106)

Eth: New job #3f303339 from eu1.ethermine.org:4444; diff: 4000MH

GPU1: Starting up... (0)

GPU1: Generating ethash light cache for epoch #394

GPU2: Starting up... (0)

GPU3: Starting up... (0)

GPU4: Starting up... (0)

GPU5: Starting up... (0)

GPU6: Starting up... (0)

Listening for CDM remote manager at port 3333 in read-only mode

Eth: New job #091d0ee4 from eu1.ethermine.org:4444; diff: 4000MH

Light cache generated in 3.5 s (18.5 MB/s)

GPU2: Allocating DAG (4.09) GB; good for epoch up to #396

GPU2: Generating DAG for epoch #394

GPU6: Allocating DAG (4.09) GB; good for epoch up to #396

GPU6: Generating DAG for epoch #394

Eth: New job #5e6e830b from eu1.ethermine.org:4444; diff: 4000MH

GPU1: Free VRAM: 7.948 GB; used: 0.052 GB

GPU1: Allocating DAG (4.09) GB; good for epoch up to #396

GPU1: Generating DAG for epoch #394

Eth speed: 0.000 MH/s, shares: 0/0/0, time: 0:00

GPUs: 1: 0.000 MH/s (0) 2: 0.000 MH/s (0) 3: 0.000 MH/s (0) 4: 0.000 MH/s (0) 5: 0.000 MH/s (0) 6: 0.000 MH/s (0)

GPU3: Free VRAM: 7.948 GB; used: 0.052 GB

GPU3: Allocating DAG (4.09) GB; good for epoch up to #396

GPU2: DAG 9%

GPU3: Generating DAG for epoch #394

GPU6: DAG 11%

GPU1: DAG 15%

GPU4: Free VRAM: 7.948 GB; used: 0.052 GB

GPU4: Allocating DAG (4.09) GB; good for epoch up to #396

GPU2: DAG 19%

GPU4: Generating DAG for epoch #394

GPU6: DAG 23%

GPU3: DAG 13%

GPU1: DAG 27%

GPU5: Free VRAM: 7.948 GB; used: 0.052 GB

GPU5: Allocating DAG (4.09) GB; good for epoch up to #396

GPU5: Allocating buffers failed with: clCreateBuffer (-4).

Eth: New job #96ff5e83 from eu1.ethermine.org:4444; diff: 4000MH

Fatal error detected. Restarting.

GPU6 initMiner error: Unable to initialize CUDA miner

GPU2: DAG 28%

GPU2 initMiner error: Unable to initialize CUDA miner

Eth speed: 0.000 MH/s, shares: 0/0/0, time: 0:00

GPUs: 1: 0.000 MH/s (0) 2: 0.000 MH/s (0) 3: 0.000 MH/s (0) 4: 0.000 MH/s (0) 5: 0.000 MH/s (0) 6: 0.000 MH/s (0)







Eth: New job #c5fb8bae from eu1.ethermine.org:4444; diff: 4000MH







Phoenix Miner 5.5c Windows/msvc - Release build

-----------------------------------------------



Waiting 15 s for previous instance to close

CUDA version: 11.0, CUDA runtime: 8.0

OpenCL driver version: 20.11.3/20.12.1

Available GPUs for mining:

GPU1: Radeon RX 570 Series (pcie 1), OpenCL 2.0, 8 GB VRAM, 32 CUs

GPU2: GeForce GTX 1080 (pcie 2), CUDA cap. 6.1, 8 GB VRAM, 20 CUs

GPU3: Radeon RX 570 Series (pcie 6), OpenCL 2.0, 8 GB VRAM, 32 CUs

GPU4: Radeon RX 570 Series (pcie 9), OpenCL 2.0, 8 GB VRAM, 32 CUs

GPU5: Radeon RX 570 Series (pcie 10), OpenCL 2.0, 8 GB VRAM, 32 CUs

GPU6: GeForce GTX 1070 (pcie 11), CUDA cap. 6.1, 8 GB VRAM, 15 CUs

Nvidia driver version: 461.40

Eth: the pool list contains 2 pools (2 from command-line)

Eth: primary pool: eu1.ethermine.org:4444

Starting GPU mining

GPU1: AMD driver 20.11.3

GPU3: AMD driver 20.11.3

GPU4: AMD driver 20.11.3

GPU5: AMD driver 20.11.3

Eth: Connecting to ethash pool eu1.ethermine.org:4444 (proto: QtMiner)

GPU1: 26C 40% 6W, GPU2: 37C 47% 47W, GPU3: 21C 40% 6W, GPU4: 23C 40% 32W, GPU5: 24C 40% 34W, GPU6: 37C 47% 42W

GPUs power: 167.0 W

Eth: Connected to ethash pool eu1.ethermine.org:4444 (172.65.207.106)

Eth: New job #c221a325 from eu1.ethermine.org:4444; diff: 4000MH

GPU1: Starting up... (0)

GPU1: Generating ethash light cache for epoch #394

GPU2: Starting up... (0)

GPU3: Starting up... (0)

GPU4: Starting up... (0)

GPU5: Starting up... (0)

GPU6: Starting up... (0)

Listening for CDM remote manager at port 3333 in read-only mode

Eth: New job #f65187ec from eu1.ethermine.org:4444; diff: 4000MH

Eth: New job #b73cc90e from eu1.ethermine.org:4444; diff: 4000MH

Light cache generated in 3.5 s (18.6 MB/s)

GPU2: Allocating DAG (4.09) GB; good for epoch up to #396

GPU2: Generating DAG for epoch #394

GPU6: Allocating DAG (4.09) GB; good for epoch up to #396

GPU6: Generating DAG for epoch #394

GPU3: Free VRAM: 7.948 GB; used: 0.052 GB

GPU3: Allocating DAG (4.09) GB; good for epoch up to #396

GPU3: Generating DAG for epoch #394

Eth: New job #9649a9da from eu1.ethermine.org:4444; diff: 4000MH

Eth: New job #0d33116e from eu1.ethermine.org:4444; diff: 4000MH

Eth speed: 0.000 MH/s, shares: 0/0/0, time: 0:00

GPUs: 1: 0.000 MH/s (0) 2: 0.000 MH/s (0) 3: 0.000 MH/s (0) 4: 0.000 MH/s (0) 5: 0.000 MH/s (0) 6: 0.000 MH/s (0)

GPU4: Free VRAM: 7.948 GB; used: 0.052 GB

GPU4: Allocating DAG (4.09) GB; good for epoch up to #396

GPU4: Generating DAG for epoch #394

GPU2: DAG 9%

GPU6: DAG 11%

GPU3: DAG 15%

GPU5: Free VRAM: 7.948 GB; used: 0.052 GB

GPU5: Allocating DAG (4.09) GB; good for epoch up to #396

GPU5: Generating DAG for epoch #394

Eth: New job #db7a404d from eu1.ethermine.org:4444; diff: 4000MH

GPU2: DAG 20%

Eth: New job #c7bbc2c2 from eu1.ethermine.org:4444; diff: 4000MH

GPU4: DAG 12%

GPU6: DAG 23%

GPU1: Free VRAM: 7.948 GB; used: 0.052 GB

GPU1: Allocating DAG (4.09) GB; good for epoch up to #396

GPU1: Allocating buffers failed with: clCreateBuffer (-4).

Fatal error detected. Restarting.

GPU6 initMiner error: Unable to initialize CUDA miner







Eth: New job #b7e7086d from eu1.ethermine.org:4444; diff: 4000MH

Eth: New job #d42a4e81 from eu1.ethermine.org:4444; diff: 4000MH

Eth: New job #a17af0d2 from eu1.ethermine.org:4444; diff: 4000MH

Eth: New job #7759063b from eu1.ethermine.org:4444; diff: 4000MH

Eth: New job #7bca0888 from eu1.ethermine.org:4444; diff: 4000MH

GPU2: DAG 23%

GPU2 initMiner error: Unable to initialize CUDA miner

Eth speed: 0.000 MH/s, shares: 0/0/0, time: 0:00

GPUs: 1: 0.000 MH/s (0) 2: 0.000 MH/s (0) 3: 0.000 MH/s (0) 4: 0.000 MH/s (0) 5: 0.000 MH/s (0) 6: 0.000 MH/s (0)



*** 0:00 *** 2/11 20:06 **************************************

Eth: Mining ETH on eu1.ethermine.org:4444 for 0:00

Available GPUs for mining:

GPU1: Radeon RX 570 Series (pcie 1), OpenCL 2.0, 8 GB VRAM, 32 CUs

GPU2: GeForce GTX 1080 (pcie 2), CUDA cap. 6.1, 8 GB VRAM, 20 CUs

GPU3: Radeon RX 570 Series (pcie 6), OpenCL 2.0, 8 GB VRAM, 32 CUs

GPU4: Radeon RX 570 Series (pcie 9), OpenCL 2.0, 8 GB VRAM, 32 CUs

GPU5: Radeon RX 570 Series (pcie 10), OpenCL 2.0, 8 GB VRAM, 32 CUs

GPU6: GeForce GTX 1070 (pcie 11), CUDA cap. 6.1, 8 GB VRAM, 15 CUs

GPU1: 26C 40% 6W, GPU2: 37C 47% 47W, GPU3: 21C 40% 6W, GPU4: 23C 40% 32W, GPU5: 24C 40% 34W, GPU6: 37C 47% 42W

GPUs power: 167.0 W

Current -gt 15,15,15,15,15,15

Eth: Accepted shares 0 (0 stales), rejected shares 0 (0 stales)

Eth: Incorrect shares 0 (0.00%), est. stales percentage 0.00%

Eth: Average speed (5 min): 0.000 MH/s

[elviture]

So in hopes of finding out whats up...:

1) I first removed the Nvidia cards...
All of the 4x AMD cards are on the main PSU so it was simply a matter of removing the 2nd PSU from the parallel cable and disconnecting the risers. The AMD cards worked perfectly, didn't have any issues during launch. tried several times and worked perfectly several iterations no issues, warnings, errors, crashes etc..

2) I removed the AMD cards and Put back the Nvidia cards...
DCed the AMD cards from pcie slots, moved the Nvidia cards to be powered by the main PSU and reconnected the nvidia card riser usb connections to the pcie slots.... And then this happened, a new error:

2021.02.11:22:15:07.849: GPU1 GPU1: DAG   8%
2021.02.11:22:15:08.245: GPU2 CUDART error in CudaProgram.cu:262 : unknown error (30)
2021.02.11:22:15:08.246: GPU2 GPU2 initMiner error: unknown error
2021.02.11:22:15:08.248: wdog Fatal error detected. Restarting.
2021.02.11:22:15:08.373: GPU1 CUDART error in CudaProgram.cu:262 : unknown error (30)
2021.02.11:22:15:08.373: GPU1 GPU1 initMiner error: unknown error

so interestingly, phoenix did restart and start working again. the cards ran fine for a few minutes and i've stopped the program. I restarted it to see if i could recreate the aforementioned CUDART error but no such luck. seems the GPUs are working fine.

I guess ill try reinstalling all the cards at once again and see if the error has resolved itself.





PS. The only real answer i've found online to address the clCreateBuffer (-4) issue as mentioned in the main post is that i should increase my virtual window memory to 32000, which i had already done previously to this entire issue beginning.

[JayDDee]

Why not try a different miner? Don't mess around with your rig too much until you're sure
that's where the problem is.

[elviture]

i raised my virtual memory to 40000MB and guess what, it worked......


didn't know i could raise the maximum. still don't quite understand what the page memory is even for.... but... I'm mining!



and thats all that matters folks.

[FireBallex]

Raising the memory page file to higher number is very important, for two GPUs I use 9000 MB but you have more, also it's pretty good to separate AMD cards from Nvidia cards, go with two motherboards and have smoother mining experience