Claymore's Dual Ethereum AMD+NVIDIA GPU Miner v15.0 (Windows/Linux)

[dvz35]

21:45:54:383    54900700    Setting DAG epoch #280 for GPU0
21:45:54:383    54900700    Create GPU buffer for GPU0
21:45:54:383    54900700    GPU0 - not enough GPU memory to place DAG, you cannot mine this coin with this GPU
21:45:54:383    54900700    GPU0 - OpenCL error -61 - cannot allocate big buffer for DAG. Check readme.txt for possible solutions.

same trouble now with nicehash

OpenCL error -38 - cannot write buffer for DAG

Switch mining to ETH (epoch 272 for now). I wrote about this few days ago. We need new version of claymore I guess

possible drivers trouble? use 18.6.1
14.6 same trouble
phoenixminer work without this trouble?

add
check again. trouble because ETC coin DAG epoch #281

https://www.youtube.com/watch?v=P8EdQGm9x74&t=1s

[NefariousFool]

Hey Claymore any input or any sort of work going on to get the new rx 5700xt mining on your platform? Thanks!

Yes I'm checking this card, but I need a few days to see what hashrate it can achieve.
[/quote]

Any update friend? We are sitting on the edge of our seats!  Is there anything anybody can contribute to help you out with this process?

Did you try the washer fix for cooling on your card if you have one?

[NefariousFool]

does anyone know what error 1002 means? I'm getting GPU #0 strap "1" failed, error 1002  how do I fix that and what does that error mean? all straps get that error
Using Radeon VII

The straps do not work on Radeon Vii thats why you're getting that error, its not supported. You should try running the intensity tricks on the miner vs straps.

[abg00]

any news about ETC dag 281+ mining? use phoenix now. this miner dont work

[Claymore]

Any update friend? We are sitting on the edge of our seats!  Is there anything anybody can contribute to help you out with this process?
Did you try the washer fix for cooling on your card if you have one?

I'm working on it, just need a couple of days more. AMD even did not release official ISA docs for RDNA yet...

any news about ETC dag 281+ mining? use phoenix now. this miner dont work

What's wrong with dag 281 epoch? It works fine in my tests. If you want to get some support, it's a good idea to post your config.txt file and some hardware info (or log file).

[awarywocki]

What's wrong with dag 281 epoch? It works fine in my tests. If you to get some support, it's a good idea to post your config.txt file and some hardware info (or log file).

I have a miner hang at epoch 280 (both Rx 470 4gb and Rx 480 4gb). After switch to ETH (epoch 272 it still working well). It is not a config problem it is problem with miner. Check phoenix miner topic and there is same epoch problem with older miner versions

C&p
"Read this if you can't mine ETH or ETC with PhoenixMiner: important message for everyone that is running older versions of PhoenixMiner (before 4.2).

IMPORTANT! The versions of PhoenixMiner before 4.2 only support DAG epoch up to 265. Both ETC and ETH has reached epoch 266 already. To ensure uninterrupted operation, please upgrade to 4.2. If you are mining other coins, you can safely disregard this message.

The reason for this limitation is that we need to perform some quite complex calculations for each future DAG epoch in order to increase the hashrate of the optimized kernels. Recently we have added more computational resources and PhoenixMiner 4.2 will work without problems up to DAG epoch 329. We will increase this number significantly (e.g. 400 or more) in the next version."

[UnclWish]

What's wrong with dag 281 epoch? It works fine in my tests. If you to get some support, it's a good idea to post your config.txt file and some hardware info (or log file).

I have a miner hang at epoch 280 (both Rx 470 4gb and Rx 480 4gb). After switch to ETH (epoch 272 it still working well). It is not a config problem it is problem with miner. Check phoenix miner topic and there is same epoch problem with older miner versions

C&p
"Read this if you can't mine ETH or ETC with PhoenixMiner: important message for everyone that is running older versions of PhoenixMiner (before 4.2).

IMPORTANT! The versions of PhoenixMiner before 4.2 only support DAG epoch up to 265. Both ETC and ETH has reached epoch 266 already. To ensure uninterrupted operation, please upgrade to 4.2. If you are mining other coins, you can safely disregard this message.

The reason for this limitation is that we need to perform some quite complex calculations for each future DAG epoch in order to increase the hashrate of the optimized kernels. Recently we have added more computational resources and PhoenixMiner 4.2 will work without problems up to DAG epoch 329. We will increase this number significantly (e.g. 400 or more) in the next version."

Phoenix miner works another way than Claymore's. It use optimized kernels wich is must be calculated with depends of DAG epoch. This is why if miner didn't contains optimized kernels for some DAG it would give error.

Claymore's miner didn't use kernels optimized for each epoch. So it must work on any DAG. You must provide your config, bat file and log to determine what's your problem.

[awarywocki]

Phoenix miner works another way than Claymore's. It use optimized kernels wich is must be calculated with depends of DAG epoch. This is why if miner didn't contains optimized kernels for some DAG it would give error.

Claymore's miner didn't use kernels optimized for each epoch. So it must work on any DAG. You must provide your config, bat file and log to determine what's your problem.

sure. But two of three rigs just hangs at this same moment (Rx 470 & Rx 480), it happens when ETC touch epoch 280. Below is my config.txt content (they nearly just the same on both machines)

[config.txt]
#-epool europe.ethash-hub.miningpoolhub.com:20555 (<-- ETC port, do not work)
-epool europe.ethash-hub.miningpoolhub.com:20535 (<-- ETH port, works well)
-ewal {private, deleted}
-eworker {private, deleted}
-epsw x
-esm 2
-allcoins 1
-allpools 1

-tt 72
-ttdcr 75
-ttli 78
-tstop 82
-fanmin 30
-mport -3333
-ethi 16
-dcri 14
-rxboost 1
-strap 777000000000000022AA1C00315A5B36A0550F15B68C1506004082007C041420CA8980A9020004C 01712262B612B3715

#OC
#core
-cclock 1100
-cvddc 825
#memory
-mclock 2050
-mvddc 825
[/config.txt]

[start.bat]
timeout /t 60
setx GPU_FORCE_64BIT_PTR 0
setx GPU_MAX_HEAP_SIZE 100
setx GPU_USE_SYNC_OBJECTS 1
setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_SINGLE_ALLOC_PERCENT 100

EthDcrMiner64.exe
[/start.bat]

I can paste log too, bu there is strictly NOTHING because miner hangs (sometime with restart, sometime not). It hangs after "setting dag epoch 280" line and before "setting dag epoch 280 done". I tried older version of claymore. It stops hanging but it just creating dag epoch and never finish. Changing port on pool to mining ETH solves the problem... for now

[Claymore]

sure. But two of three rigs just hangs at this same moment (Rx 470 & Rx 480), it happens when ETC touch epoch 280. Below is my config.txt content (they nearly just the same on both machines)

Try "-eres 0".

[UnclWish]

I can paste log too, bu there is strictly NOTHING because miner hangs (sometime with restart, sometime not). It hangs after "setting dag epoch 280" line and before "setting dag epoch 280 done". I tried older version of claymore. It stops hanging but it just creating dag epoch and never finish. Changing port on pool to mining ETH solves the problem... for now

Looks like it's not enough memory to set DAG #280... Do you have enough virtual memory (page file) size?

And another one thing. You use -allcoins 1. This is what Claymore writes in readme:

-allcoins Specify "-allcoins 1" to be able to mine Ethereum forks, in this mode miner will use some default pools for devfee Ethereum mining.
        Note that if devfee mining pools will stop, entire mining will be stopped too.
        Miner has to use two DAGs in this mode - one for Ethereum and one for Ethereum fork, it can cause crashes because DAGs have different sizes.
        Therefore for this mode it is recommended to specify current Ethereum epoch (or a bit larger value),
        for example, "-allcoins 47" means that miner will expect DAG size for epoch #47 and will allocate appropriate GPU buffer at starting, instead of reallocating bigger GPU buffer (may crash) when it starts devfee mining.
        Another option is to specify "-allcoins -1", in this mode miner will start devfee round immediately after start and therefore will get current epoch for Ethereum, after that it will be able to mine Ethereum fork.
        If you mine ETC on some pool that does not accept wallet address but requires Username.Worker instead, the best way is to specify "-allcoins etc", in this mode devfee mining will be on ETC pools and DAG won't be recreated at all.

Try to set "-allcoins etc"

[awarywocki]

sure. But two of three rigs just hangs at this same moment (Rx 470 & Rx 480), it happens when ETC touch epoch 280. Below is my config.txt content (they nearly just the same on both machines)

Try "-eres 0".

hoooooly ff**. You are right! I've checked my config line by line and found that there were "-eres 20" option between my #comment lines at the very bottom of the file. I was sure, that at the bottom are only #comments lines with my tests, thoughts etc (mean et cetera, not ethereum classic ) so I did not paste them. After deleting "-eres 20" my miners can mine ETC again (AFAIR eres=2 by default). So it means miner tried to work with DAG epoch 280 + 20 = 300, and that is why it crashed

THANK YOU!

Looks like it's not enough memory to set DAG #280... Do you have enough virtual memory (page file) size?

And another one thing. You use -allcoins 1. This is what Claymore writes in readme:

-allcoins Specify "-allcoins 1" to be able to mine Ethereum forks, in this mode miner will use some default pools for devfee Ethereum mining.
        Note that if devfee mining pools will stop, entire mining will be stopped too.
        Miner has to use two DAGs in this mode - one for Ethereum and one for Ethereum fork, it can cause crashes because DAGs have different sizes.
        Therefore for this mode it is recommended to specify current Ethereum epoch (or a bit larger value),
        for example, "-allcoins 47" means that miner will expect DAG size for epoch #47 and will allocate appropriate GPU buffer at starting, instead of reallocating bigger GPU buffer (may crash) when it starts devfee mining.
        Another option is to specify "-allcoins -1", in this mode miner will start devfee round immediately after start and therefore will get current epoch for Ethereum, after that it will be able to mine Ethereum fork.
        If you mine ETC on some pool that does not accept wallet address but requires Username.Worker instead, the best way is to specify "-allcoins etc", in this mode devfee mining will be on ETC pools and DAG won't be recreated at all.

Try to set "-allcoins etc"

thanx for response. I tried increase virtual memory before (set it even to 45000MB) and this did not help.
"-allcoins etc" could be solution but miner did not stay alive long enough to change dag epoch because of devfee, so probably this wouldn't help.

Anyway Claymore hit the bull's eye

[abg00]

sure. But two of three rigs just hangs at this same moment (Rx 470 & Rx 480), it happens when ETC touch epoch 280. Below is my config.txt content (they nearly just the same on both machines)

Try "-eres 0".

miner dont work/ test on 8 gb card

EthDcrMiner64.exe -allcoins 1 -eres 0 -mode 1 -asm 2  -esm 3 -rxboost 1

epool
POOL: stratum+tcp://daggerhashimoto.eu.nicehash.com:3353, WALLET: xxx.RIG01, PSW: x, ALLPOOLS: 1

then switch to etc mining DAG 281 have error

pnoenixminer work fine

test on lot of rigs with different card 4\8 gb same error
14.6 14.7 same error
driver 18.6.1 19.7.2 same error

plz fixminer

add test directly on etc pool. same error

[sharmanov]

does anyone know what error 1002 means? I'm getting GPU #0 strap "1" failed, error 1002  how do I fix that and what does that error mean? all straps get that error
Using Radeon VII

The straps do not work on Radeon Vii thats why you're getting that error, its not supported. You should try running the intensity tricks on the miner vs straps.

Intensity trick? could you please specify if it's not too much to ask 🙏

[UnclWish]

sure. But two of three rigs just hangs at this same moment (Rx 470 & Rx 480), it happens when ETC touch epoch 280. Below is my config.txt content (they nearly just the same on both machines)

Try "-eres 0".

miner dont work/ test on 8 gb card

EthDcrMiner64.exe -allcoins 1 -eres 0 -mode 1 -asm 2  -esm 3 -rxboost 1

epool
POOL: stratum+tcp://daggerhashimoto.eu.nicehash.com:3353, WALLET: xxx.RIG01, PSW: x, ALLPOOLS: 1

then switch to etc mining DAG 281 have error

pnoenixminer work fine

test on lot of rigs with different card 4\8 gb same error
14.6 14.7 same error
driver 18.6.1 19.7.2 same error

plz fixminer

add test directly on etc pool. same error

What error you have? Post it here...

[abg00]

21:45:54:383    54900700    Setting DAG epoch #280 for GPU0
21:45:54:383    54900700    Create GPU buffer for GPU0
21:45:54:383    54900700    GPU0 - not enough GPU memory to place DAG, you cannot mine this coin with this GPU
21:45:54:383    54900700    GPU0 - OpenCL error -61 - cannot allocate big buffer for DAG. Check readme.txt for possible solutions.

same trouble now with nicehash

OpenCL error -38 - cannot write buffer for DAG

Switch mining to ETH (epoch 272 for now). I wrote about this few days ago. We need new version of claymore I guess

possible drivers trouble? use 18.6.1
14.6 same trouble
phoenixminer work without this trouble?

add
check again. trouble because ETC coin DAG epoch #281

error already write in this topic. make test again later and write log file. swap 60+ gb. 8 gpu rx570 4 gb

[abg00]

pool nano
coin etc
card 2 x rx570 4gb
swap size 39gb
ram size 4gb

bat file
set GPU_FORCE_64BIT_PTR=0
set GPU_MAX_HEAP_SIZE=100
set GPU_USE_SYNC_OBJECTS=1
set GPU_MAX_ALLOC_PERCENT=100
set GPU_SINGLE_ALLOC_PERCENT=100

EthDcrMiner64.exe -allcoins etc -logfile noappend.log -retrydelay 5 -mport -3336 -ttli 75 -mode 1 -eres 0 -asm 2 -tt 65 -ttli 75 -esm 3 -r 1 -cclock 1100 -cvddc 800,800,870,870,870 -mclock 2020,2020,1950,1950,1950 -mvddc 850,850,900,900,900 -powlim 50 -rxboost 1 -di 01234

config file empty

epool file
POOL: etc-eu1.nanopool.org:19999, WALLET: removed , PSW: x, ESM: 0, ALLPOOLS: 0

error log

11:00:58:848   13cc   Check and remove old log files...
11:00:58:849   13cc   args: -allcoins etc -logfile noappend.log -retrydelay 5 -mport -3336 -ttli 75 -mode 1 -eres 0 -asm 2 -tt 65 -ttli 75 -esm 3 -r 1 -cclock 1100 -cvddc 800,800,870,870,870 -mclock 2020,2020,1950,1950,1950 -mvddc 850,850,900,900,900 -powlim 50 -rxboost 1 -di 01234
11:00:58:849   13cc   
11:00:58:849   13cc   ЙHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH»
11:00:58:849   13cc   є                Claymore's Dual GPU Miner - v14.6               є
11:00:58:849   13cc   є              ETH + DCR/SIA/LBC/PASC/BLAKE2S/KECCAK             є
11:00:58:849   13cc   є                       Supercharged Edition                     є
11:00:58:851   13cc   ИHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHj
11:00:58:852   13cc   
11:00:58:852   13cc   b735
11:00:59:053   13cc   ETH: 6 pools are specified
11:00:59:053   13cc   Main Ethereum pool is etc-eu1.nanopool.org:19999
11:00:59:398   13cc   OpenCL platform: Intel(R) OpenCL
11:00:59:398   13cc   OpenCL platform: AMD Accelerated Parallel Processing
11:00:59:399   13cc   OpenCL initializing...

11:00:59:399   13cc   AMD Cards available: 2
11:00:59:399   13cc   GPU #0: Ellesmere (Radeon RX 570 Series), 4096 MB available, 32 compute units (pci bus 5:0:0)
11:00:59:399   13cc   GPU #0 recognized as Radeon RX 470/570
11:00:59:399   13cc   GPU #1: Ellesmere (Radeon RX 570 Series), 4096 MB available, 32 compute units (pci bus 8:0:0)
11:00:59:399   13cc   GPU #1 recognized as Radeon RX 470/570
11:00:59:399   13cc   POOL/SOLO version
11:00:59:400   13cc   Platform: Windows
11:00:59:436   13cc   start building OpenCL program for GPU 0...
11:00:59:733   13cc   done
11:00:59:937   13cc   start building OpenCL program for GPU 1...
11:01:00:229   13cc   done
11:01:00:469   13cc   GPU #0: algorithm ASM 2
11:01:00:512   13cc   GPU #1: algorithm ASM 2
11:01:00:512   13cc   cudaGetDeviceCount failed (35, CUDA driver version is insufficient for CUDA runtime version), probably no CUDA devices
11:01:00:513   13cc   No NVIDIA CUDA GPUs detected.
11:01:00:513   13cc   Total cards: 2
11:01:00:513   13cc   
Please consider using "-strap" option to increase mining speed, check Readme for details.

11:01:04:117   13cc   
You can use "+" and "-" keys to achieve best ETH speed, see "FINE TUNING" section in Readme for details.

11:01:05:117   13cc   No NVIDIA cards in the list, NVML library will not be used.
11:01:05:119   13cc   Driver version: 0x801
11:01:05:134   13cc   driver init ok
11:01:05:168   13cc   GPU# 0 -rxboost option is applied successfully
11:01:05:201   13cc   GPU# 1 -rxboost option is applied successfully
11:01:06:242   13cc   SSL: Imported 37 certificates from local storage
11:01:06:251   1bc   ETH: Stratum - connecting to 'etc-eu1.nanopool.org' <51.15.89.69> port 19999 (unsecure)
11:01:06:312   1bc   sent: {"worker": "eth1.0", "jsonrpc": "2.0", "params": ["removed", "x"], "id": 2, "method": "eth_submitLogin"}

11:01:06:312   1bc   ETH: Stratum - Connected (etc-eu1.nanopool.org:19999) (unsecure)
11:01:06:343   13cc   ETHEREUM-ONLY MINING MODE ENABLED (-mode 1)

11:01:06:344   13cc   ETH: eth-proxy stratum mode
11:01:06:345   13cc   "-allcoins" option is set, default pools will be used for devfee, check "Readme" file for details.
11:01:06:346   13cc   Watchdog enabled
11:01:06:347   13cc   Remote management (READ-ONLY MODE) is enabled on port 3336
11:01:06:347   13cc   

11:01:06:348   13cc   You did not specify -dcri values directly, so they will be detected automatically
11:01:06:349   13cc   Automatic detection of best -dcri values started, please wait...

11:01:06:389   1bc   buf: {"jsonrpc":"2.0","id":2,"result":true}
{"jsonrpc":"2.0","id":0,"result":["0xc03304e768748f0e1522de45194a56e320af2fbf33d3fcbc79fb37a423ebe8b1","0x706a056e1dad1a5aeabdf3ce09ebccf47a88a8955c0f27181c5b0de3f7ab9ea9","0x000000006df37f675ef6eadf5ab9a2072d44268d97df837e6748956e5c6c2116"]}

11:01:06:389   1bc   ETH: Authorized
11:01:06:389   1bc   sent: {"worker": "", "jsonrpc": "2.0", "params": [], "id": 3, "method": "eth_getWork"}

11:01:06:389   1bc   remove first packet 243
11:01:06:457   1bc   buf: {"jsonrpc":"2.0","id":0,"result":["0xc03304e768748f0e1522de45194a56e320af2fbf33d3fcbc79fb37a423ebe8b1","0x706a056e1dad1a5aeabdf3ce09ebccf47a88a8955c0f27181c5b0de3f7ab9ea9","0x000000006df37f675ef6eadf5ab9a2072d44268d97df837e6748956e5c6c2116"]}

11:01:06:461   1330   Setting DAG epoch #281...
11:01:08:061   6d8   GPU 0 temp = 50, old fan speed = 75, new fan speed = 79

11:01:08:071   6d8   GPU 1 temp = 48, old fan speed = 77, new fan speed = 78

11:01:09:704   1330   Setting DAG epoch #281 for GPU1
11:01:09:706   13b4   Setting DAG epoch #281 for GPU0
11:01:09:707   13b4   Create GPU buffer for GPU0
11:01:09:708   13b4   GPU0 - not enough GPU memory to place DAG, you cannot mine this coin with this GPU
11:01:09:707   1330   Create GPU buffer for GPU1
11:01:09:708   13b4   GPU0 - OpenCL error -61 - cannot allocate big buffer for DAG. Check readme.txt for possible solutions.

11:01:09:709   1330   GPU1 - not enough GPU memory to place DAG, you cannot mine this coin with this GPU
11:01:09:709   1330   GPU1 - OpenCL error -61 - cannot allocate big buffer for DAG. Check readme.txt for possible solutions.

11:01:11:082   6d8   GPU 0 temp = 49, old fan speed = 72, new fan speed = 72

11:01:11:092   6d8   GPU 1 temp = 48, old fan speed = 79, new fan speed = 70

11:01:12:708   13b4   Set global fail flag, failed GPU0
11:01:12:708   13b4   GPU 0 failed
11:01:12:709   c88   Setting DAG epoch #281 for GPU0
11:01:12:710   c88   GPU0, OpenCL error -38 - cannot write buffer for DAG

11:01:12:710   1330   Set global fail flag, failed GPU1
11:01:12:711   1330   GPU 1 failed
11:01:12:712   bb4   Setting DAG epoch #281 for GPU1
11:01:12:712   bb4   GPU1, OpenCL error -38 - cannot write buffer for DAG

11:01:14:103   6d8   GPU 0 temp = 46, old fan speed = 66, new fan speed = 63

11:01:14:113   6d8   GPU 1 temp = 44, old fan speed = 72, new fan speed = 60

11:01:15:712   c88   Set global fail flag, failed GPU0
11:01:15:712   c88   GPU 0 failed
11:01:15:714   bb4   Set global fail flag, failed GPU1
11:01:15:714   bb4   GPU 1 failed
11:01:16:397   1bc   ETH: checking pool connection...
11:01:16:397   1bc   sent: {"worker": "", "jsonrpc": "2.0", "params": [], "id": 3, "method": "eth_getWork"}

11:01:16:459   1bc   buf: {"jsonrpc":"2.0","id":0,"result":["0xc03304e768748f0e1522de45194a56e320af2fbf33d3fcbc79fb37a423ebe8b1","0x706a056e1dad1a5aeabdf3ce09ebccf47a88a8955c0f27181c5b0de3f7ab9ea9","0x000000006df37f675ef6eadf5ab9a2072d44268d97df837e6748956e5c6c2116"]}

11:01:17:124   6d8   GPU 0 temp = 46, old fan speed = 57, new fan speed = 54

11:01:17:135   6d8   GPU 1 temp = 44, old fan speed = 62, new fan speed = 50

11:01:18:174   d20   Quit...
11:01:18:230   c8   GPU 0 temp = 46, old fan speed = 52, new fan speed = 45

11:01:18:233   c8   GPU 1 temp = 44, old fan speed = 54, new fan speed = 40

[abg00]

sure. But two of three rigs just hangs at this same moment (Rx 470 & Rx 480), it happens when ETC touch epoch 280. Below is my config.txt content (they nearly just the same on both machines)

Try "-eres 0".

plz check my answer

[Claymore]

sure. But two of three rigs just hangs at this same moment (Rx 470 & Rx 480), it happens when ETC touch epoch 280. Below is my config.txt content (they nearly just the same on both machines)

Try "-eres 0".

plz check my answer

Error -61 is CL_INVALID_BUFFER_SIZE, you can google about it.
Miner cannot assign enough GPU memory. If some other miner works fine - use it, but it will stop working soon too. The most probable reason (it's related to both nvidia and amd cards):
https://social.technet.microsoft.com/Forums/en-US/15b9654e-5da7-45b7-93de-e8b63faef064/windows-10-does-not-let-cuda-applications-to-use-all-vram-on-especially-secondary-graphics-cards?forum=win10itprohardware
Exact size of reserved GPU memory depends on Windows build and drivers, but afaik the values are similar.

[elzium]

How about #benchmark 300+?

[abg00]

sure. But two of three rigs just hangs at this same moment (Rx 470 & Rx 480), it happens when ETC touch epoch 280. Below is my config.txt content (they nearly just the same on both machines)

Try "-eres 0".

plz check my answer

Error -61 is CL_INVALID_BUFFER_SIZE, you can google about it.
Miner cannot assign enough GPU memory. If some other miner works fine - use it, but it will stop working soon too. The most probable reason (it's related to both nvidia and amd cards):
https://social.technet.microsoft.com/Forums/en-US/15b9654e-5da7-45b7-93de-e8b63faef064/windows-10-does-not-let-cuda-applications-to-use-all-vram-on-especially-secondary-graphics-cards?forum=win10itprohardware
Exact size of reserved GPU memory depends on Windows build and drivers, but afaik the values are similar.

topic about nvidia card

eth dag 278 3.13 GB work fine
etc dag 281 3.2 GB dont work. its realy now. -eres 0 dont work