Hey Claymore, I use several of your miners, they are great and I really appreciate your work!
Right now I'm running into an issue with this cpu miner. I have been mining XMR just fine at 400h/s with my i7-5930k auto-overclocked to 3.9ghz. Today I began manually overclocking it, first by setting voltage to 1.25 and 4.0ghz. However, now your miner says AES-NI is unsupported and I'm getting 115h/s D=
Any idea why bumping my overclock up .1ghz would devastate my hashrate so much?
Edit:
reset voltage to auto and turned MSI's auto-OC back on. Miner is still saying AES-NI is unsupported....
This issue is not related to the miner. It must be some BIOS setting, or OS setting, or hardware problem. |
|
|
|
About private miner - I don't have it so don't know how fast it is, but in my tests I was able to get 200MH/s on LBC for stock 390X. Unfortunately, there is no way to use same code in dual mode, I had to write different code that is slower, it shows only 165MH/s on stock 390X.
Are you going to release LBC only miner? I have only OpenCL-related code, a lot of work is required to create separate miner for LBC, I just don't have time for it. I can put it in dual miner as separate mode but not sure if it worth doing it since there is some private miner with same speed. So I think the answer is no. |
|
|
|
290х get 180 by Wolf
LBC only miner more profitable than dual
Well, use private miner for now if you have it, may be situation will change in future. |
|
|
|
Thanks!
But so less LBC if dcri 500, only 400 mh on my farm that get 935 mh by Wolf privat miner
1. I rescaled -dcri for LBC so you cannot use miner for LBC only. 2. LBC is not so good for dual mode as DCR or SIA. About private miner - I don't have it so don't know how fast it is, but in my tests I was able to get 200MH/s on LBC for stock 390X. Unfortunately, there is no way to use same code in dual mode, I had to write different code that is slower, it shows only 165MH/s on stock 390X. |
|
|
|
|
v7.0 Beta:
- added new coin for dual mode: LBRY(LBC). Now you can mine Ethereum, Ethereum+Decred, Ethereum+Siacoin or Ethereum+Lbry.
- some minor improvements and bug fixes.
|
|
|
|
this is my log file
15:20:47:206 8d4
15:20:47:207 8d4 ÉÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍ»
15:20:47:208 8d4 º Claymore CryptoNote GPU Miner v9.5 Beta º
15:20:47:209 8d4 ÈÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍͼ
15:20:47:690 8d4
Cards available: 1
15:20:47:691 8d4 GPU #0: name: Cedar, 2048 MB available, 2 units
15:20:47:692 8d4 Total cards: 1
15:20:47:696 8d4 Initializing...
Error is CL_INVALID_WORK_ITEM_SIZE You won't be able to start the miner on your videocard. It must be awesome card since all AMD cards I know support 256 local-work size. 2 units? Hmm... Laptop? Or GPU built-in in CPU? |
|
|
|
Dear Claymore when are you going to realase 6.5v ? I have to upgrade some rigs and I wonder if to upgrade to 6.4 or to wait few days..
I'm going to release v7.0 Beta within 24 hours. Doing last tests, may be in a few hours. This morning i found my rig stopped and when i started it win 7 x64, none of the 6 r9 380 card was detected in device manager. I tested 1 card on another pc and it works, but i saw a message saying with red that i never saw before: 1 share rejected gpu 1 (the test one) if u overclocked it too much etc
The log looks suspicious.
It is hardware problem, if you see 511C temp check your risers/clocks. |
|
|
|
Any reasons why Claymore is reporting over 100% fan speed for RX470 and 480? Sometimes up to 192%
Also anyone have the sweet spot OC setting for the EVGA 1070 with Samsung memory? Was told 1070 with Samsung memory performs faster but that's not my experience as at 550 Memory or even OC to 650, it's still slower than the MSI and Asus which have Micron memory.
I've seen the 480 Nitro get to 103%, but I think it's based off the max RPM of the fan and what RPM it's currently running at, hence 103%. But 192%? That's just odd. I was also asking myself why the fan % speed in GPU-Z and claymore miner don't match. It is because ADL returns incorrect data for fan for 4xx cards. For example, I set 1000rpm, then ask current speed, it returns 812. I set 2000 it returns something like 1736. It is senseless and bad for my fan regulation algo. Also I calculate % fan as 100*CurrentFanSpeed/MaxFanSpeed, both values are asked from ADL. But sometimes CurrentFanSpeed can be > MaxFanSpeed which is senseless too. In next update I will limit displayed speed by 100% to hide this issue. AMD created a lot of new ADL data types and functions for new cards for some reason, but they did not document them at all or did it badly. So we have to use what they gave to us, again buggy drivers. |
|
|
|
Hello claymore,
I lately had the issue that the miner stops doing anything and if I press a key it starts saying outdated shares 1000s of them and then stating pool maybe had problems and mines again.
It states then the pool had problem but it happened now 3 times in 1h and minergate techs say no issues :-(
What else could cause the miner to pause or hold ?
Windows 10 if you clicked inside of terminal window, if you see cursor inside of terminal it means that Windows paused application and wait until you select text and press escape. If it is not the case then I have no other ideas. |
|
|
|
Hi  i have about 220h/s XMR minig at Slow mode and 131 h/s "fast mode". How i can enable slow mode again? Core i7 2600 3,4 Specify undocumented option "-del". Miner must show "deletion: 1" message if successfull. Admin rights are required. After it, start miner without admin rights and it won't be able to enable fast mode again. |
|
|
|
AMD - isn't it ?  When Nvidia drivers fail (for example, too much OC, or bug in kernels) - you will see unpleasant things too. After it even if it works, it will work slowly, at least in my tests, and system reboot is necessary anyway. |
|
|
|
@Claymore
Hello, I have a strange problem under Ubuntu 14.04. The miner hangs after bad CL request and watchdog is not able to restart it. When I try to close the miner console the whole rig hangs and one only solution is reboot.
Any help will be appreciated.
Same here, watchdog reports OpenCL hang on card but doesn't restart the rig. Restart (-r) is default. Same in Windows. If OpenCL fails, in many cases you won't be able to restart miner or reboot PC, the only way is to press RESET button. It is not related to miner, it is related to video drivers. |
|
|
|
I had always troubles with AMD cards, they were sensitive to anything sw/hw related.
NVDA is better.
Nvidia is even worse Nvidia was always more expensive but its drivers and stability were always better. And is better now. No flame. Just a fact. But thanks to AMD we can enjoy at least a little competition. Thanks ATI/AMD for trying and thanks their customers for buying Yeah, stability. Stable 3MH/s on ETH on 10xx cards in Windows 10/7 for months until they fixed their drivers finally. 9xx still have this bug in Windows 10. Stable low hashrate on most algos. Also several CUDA versions is a pain for devs, for AMD you can use different OpenCL kernels for different cards and change them in runtime, but for Nvidia you have to make several EXE versions linked to different CUDAs. And yeah, they are more expensive than AMD cards. It is not a bug. It's a hardware limitation (Translation Lookaside Buffer). It's not an issue in anything but random memory accesses across large memory allocations. The Virtual Memory model of Win10 makes the issue surface quicker because of small Page Table Entry size. In Kepler and Maxwell is is limited to 64KB. In Pascal PTE can be set to 1M or 2M with WDDM2.1 As fas as I understand, GCN has a separate IOMMU, no TLB. I did not dig so deeply, may be this is not a bug, but since they fixed it, it can be called so I just try to say that Nvidia is not more easy/powerful/stable than AMD. CUDA is better than OpenCL because it has more device-specific functions, like asm or registers shuffling, I miss it in OpenCL. But for most algos, Nvidia is slower even with CUDA, or there is no good miner for AMD. For example, right now I see that Nvidia is very good on LBRY and AMD is much slower, but it is because there is no fast public miner for AMD, a good miner can be x2 faster and AMD would be on a par with Nvidia. |
|
|
|
I had always troubles with AMD cards, they were sensitive to anything sw/hw related.
NVDA is better.
Nvidia is even worse Nvidia was always more expensive but its drivers and stability were always better. And is better now. No flame. Just a fact. But thanks to AMD we can enjoy at least a little competition. Thanks ATI/AMD for trying and thanks their customers for buying Yeah, stability. Stable 3MH/s on ETH on 10xx cards in Windows 10/7 for months until they fixed their drivers finally. 9xx still have this bug in Windows 10. Stable low hashrate on most algos. Also several CUDA versions is a pain for devs, for AMD you can use different OpenCL kernels for different cards and change them in runtime, but for Nvidia you have to make several EXE versions linked to different CUDAs. And yeah, they are more expensive than AMD cards. |
|
|
|
Why you miner gives me this crappy info? Not enough RAM even 26GB out of 32GB are free on a 64bit System?   Can I hook a graple and shot the joker? API VirtualAlloc returned ERROR_NO_SYSTEM_RESOURCES error. That is all I can say from your screenshot. |
|
|
|
Hello Claymore,
Is it possible to add AEON support to your perfect miner?
Not right now, may be later. They changed PoW algo a bit for some reason, now I will have to create another set of some routines just for this coin (especially for GPU miner). It requires some time, better I'll do something more important/interesting. |
|
|
|
Hi Claymore,
In your FAQ: 2MB per hash is the guide for the -h switch.
So for MSI R9 390 8GB GPU, would it be correct to set the -h as follows:
8GB=8000MB (8192MB to be exact)
h = 8192 / 2 = 4096 hash/s
so -h 4096 ?
Also what is the difference of -a 1 and -a 4 ?
The 390 seems to like -a 4 better than 1
Currently I set the miner to drive the 390 at -h 1376 (same settings for 290x 4GB)
but I am tempted to use 1376 x 2 = 2752 since 390 has twice the amount of memory than 290/290x
Your advise?
Thanks
If GPU has 8GB memory it does not mean that the fastest mode is to use all its memory. It will stuck with memory access. For 290x/390x use "-h 1376", for 290/390 use "-h 1280". Miner does it automatically if you don't specify "-h" option. Select best "-a" by checking hashrates and power consumption (if you necessary) for every "-a" value. |
|
|
|
I had always troubles with AMD cards, they were sensitive to anything sw/hw related.
NVDA is better.
Nvidia is even worse |
|
|
|
claymore, can you assist me... im getting openCL error - cannot alocate big buffer for dag on one 7850 AMD, latest drivers, i know its not problem witg its 2 GB ram, because my other 7850s with same miner, same driver, same OS, are working ok http://prnt.sc/ceecuuThe only thing I can recommend you is to read OP or Readme, "TROUBLESHOOTING" section. If does not help - I cannot help you more because unfortunately I have no idea why sometimes OpenCL fails to allocate memory; if I knew, I would add a solution to Readme. |
|
|
|
see timestamps, it's not 20 minutes!
Miner has two threads: main and for devfee. One of them did not get job for 20 mins. |
|
|
|
|
Show Posts
