|
v9.0 is available:
- Improved speed by 5-15%, depends on card:
About 810 h/s on stock 290X (Hynix memory).
About 710 h/s on stock 290X (Elpida memory).
About 690 h/s on stock 290 (Elpida memory).
About 550 h/s on stock 280X (Hynix memory).
About 440 h/s on stock 270X (Elpida memory).
About 410 h/s on stock 270 (Elpida memory).
- Reduced dev fee, now it is 2.5%.
- Improved "-li" option, Windows freezes even less.
- Best "-a" and "-h" values are set automatically for 290X,290,280X,270X,270,6950_2GB cards.
Note: For cards that are not in this list you need to check different "-a" and "-h" values to get max speed.
|
|
|
|
|
I will release new version with further speedup and reduced dev fee in 2-3 days. I will also release version for Linux, but a bit later.
|
|
|
|
The cards are using ~1GIG less memory than what it shows Claymore has available to it.
It's ok too. |
|
|
|
GPU #0 = Sapphire AMD Radeon R9 280X VAPOR-X 3GB
GPU #1 = RADEON R9 290 4GB POWERCOLOR PCS
When I run Claymore 8, I just noticed this come up
Cards available: 2
GPU #0: name: Tahiti, 3072 MB available, 32 units
GPU #1: name: Hawaii, 3072 MB available, 40 units
Total cards: 2
It's ok, for 4GB cards OpenCL allows 3GB only for using. If v8 is slower than v7 for 290 card, try HashCnt = 1280 for it, for your system with GPU #0 280x and GPU #1 290 it is "-h 896,1280". |
|
|
|
Any update soon for new 14.9 whql drivers? Look like XCN miner dont work so good on them  I'll check it soon. If it is not faster than 14.7 (I mean not this miner but my general tests) I won't optimize miner for it to achieve similar performance as 14.7, it's senseless. |
|
|
|
Tested Sapphire R9 290 Tri-X
7.0 (765/h) core(1100)-memclock(1680) AMD Catalyst 14.7
8.0 (730/h) core(1100)-memclock(1680) AMD Catalyst 14.9
What if specify "-h 1280"? If you use 4 cards then "-h 1280,1280,1280,1280". |
|
|
|
I tried -li 1 option, but this 'low intensity' i still too high for my desktop, is there any way to further adjust it?
You can reduce HashCnt with "-h" option, but it will also reduce hashrate. Anyway, you can try. |
|
|
|
Claymore, any thought on lowering your Dev Fee since Kachur GPU Miner is out and its runs faster than yours with only a 0.7% fee. I like you miner with nice features but I'm getting better performance with Kachur GPU Miner.
It is not faster at least on cards I tested and also I cannot make it work on all cards, but it works in general, so I will reduce dev fee in two times in next update. |
|
|
|
Wow!
r9 280x (hynix) = 521 h/s stock
r9 270x (elpida) = 419 h/s overclocked
r9 290 (hynix) = 790 h/s overclocked
What's the setup for r9 290 ? I have 4x R9 290 Trixx (hynix) 1000/1250 with Win 8.1 + 13.12 Catalyst With 6.1 I get 2900+ i.e. 725 kh/card while with 8.0 I get less than 700 kh/card (~ 695 to 698) However with 280x Asus D2CU (Hynix) @ 1170/1500 I get 550 which was less than 500 with 6.1 So I guess version 8.0 has no added benefit for 290 cards but improves speed by 10% for 280x. Try "-h 1376,1376,1376,1376". In next update I will improve this option so "-h 1376" will mean 1376 for all cards, but right now you have to specify hashrate for every card individually by using comma. Press "m" to see current hashrate. You can also try 1344, 1312, 1280, 1248, 1216 values. |
|
|
|
Time to work on cryptonite m7 pls  support xmg magi & boost perf for amd br Thanks, now I know what to do |
|
|
|
The size of the memory is not playing seems to be no difference between
You need to set HashCnt > 512 to see the difference, but you cannot do it because of 1GB. Therefore, no speedup for 1GB cards. I tried to 2Gb cards all possible values of the parameter h. And compared with 1Gb cards only in the negative (or 0 difference) I don't have such cards here so I cannot investigate it deeply, but if you have 2GB and it does not help - it means that card cannot use entire memory bandwith, too few computing units for that. If so, it's useless to use "-h" parameter, miner will select optimal value automatically. |
|
|
|
The size of the memory is not playing seems to be no difference between
You need to set HashCnt > 512 to see the difference, but you cannot do it because of 1GB. Therefore, no speedup for 1GB cards. |
|
|
|
On 7790(260x) not speedup.   Miner does not set optimal "-h" and "-a" options for many cards. It's because I don't have all cards here. Post first 20-30 log lines here so I can see how many compute units your card has and what's its name. Use "m" key to adjust "-h" value, or try to change "-h" values directly if OpenCL crashes. This version requires some time to find optimal settings. Everyone can post optimal "-h" and "-a" values for your cards here (and also card name as it is displayed and number of compute units), I will use this data to improve miner. Right now it knows only 290x,280x,270x and 6950 cards. Same card, can you enhance its output or it has reached its maximum? Try "-a 4". NOTE: 1GB cards cannot show good results anyway and in most cases you will not see any improvements. |
|
|
|
280x - 485 h/s Hynix
270x - 406 h/s
Very strange that I can't get more on 280x
Start miner for 1-2 minutes and PM me the log file. |
|
|
|
On 7790(260x) not speedup. Miner does not set optimal "-h" and "-a" options for many cards. It's because I don't have all cards here. Post first 20-30 log lines here so I can see how many compute units your card has and what's its name. Use "m" key to adjust "-h" value, or try to change "-h" values directly if OpenCL crashes. This version requires some time to find optimal settings. Everyone can post optimal "-h" and "-a" values for your cards here (and also card name as it is displayed and number of compute units), I will use this data to improve miner. Right now it knows only 290x,280x,270x and 6950 cards. |
|
|
|
|
v8.0 Beta is available:
- Improved speed by 2-20%, depends on card.
- Added "-a 4" mode.
- Added "-h" parameter and tune mode ("m" key).
I get the following results:
About 785 h/s on stock 290X (Hynix memory).
About 690 h/s on stock 290X (Elpida memory).
About 520 h/s on stock 280X (Hynix memory).
About 390 h/s on stock 270X.
About 315 h/s on stock 6950 2GB.
Linux build will be available later.
|
|
|
|
ANN: I will release an update soon, most speedup (about 5%) is for 280x cards and for old cards.
Any news when we can expect new updated miner? Tnx 1-2 days. How many days? I will release new version today, in 1-3 hours. |
|
|
|
Help, often there is an error:
Error in server response
: {"id":1,"jsonrpc":"2.0","error":{"code":-1,"message":"Unauthenticated"}}
What could be the problem?
What pool? You can set "-dbg 2" option and send me the log file. |
|
|
|
How to disable overclocking?
My miner crashes every half hour.
Well, by default cards are not overclocked. So if you want to disable overclocking you must overclock them somehow firstly, and then use same tool to return clocks back |
|
|
|
Hi.
I'm sorry if this was asked before, but is there a way to configure this miner through Tor (Socks) or ISA (http/https) proxy? I have 4 machines at work and the pools cannot resolve.
Thanks in advance.
Current version of the miner does not support proxies. |
|
|
|
|
Show Posts
