|
Gentlemen if you are having issues with claymore 11.x that you didn't have with 10.x but it want to mine keccak or verge do this. Upgrade to 11.5 which was just released. You need to add "-oldkernels 1". You will have your stability back with no incorrect shares and a small speed bump too!
I have one micron card doing 32.2 and 245 keccak. 1215/2200 at 900mv
|
|
|
|
Exactly. With each newer version, you've to find out the best overclocking/etc. values once again. I figured it out and now loving your newer version of miner, with better hash-rate & stability (I just upgraded to your v11.5 from v7, that's an upgrade after a long time).
When you toss in ram straps then it gets really complicated. I'm guessing the guys will have to come up with ones for the new kernel |
|
|
|
I see some people have problems with latest versions because they use hard OC and/or custom bioses created for old versions. They work fine with old GPU kernels, but I modified kernels in v11.2 and you need to readjust your settings.
For those who cannot do it (or don't want to do it) I can revert some changes and release a version with "old-style" binaries. So I will release v11.5 within a few hours and it will be the only change in this version, so don't upgrade if v11.4 works fine for you.
So what would work better on an rx580: - stock bios with 11.4 - modified bios with 11.5 or is it just a matter of testing? Is it no longer worth strapping the ram? The 11.5 should be slightly slower than 11.4 So Claymore optimizing has made ram strap modifications moot? |
|
|
|
Exacly
Seems like if you have 11.4 running stable means that you didn't OCed to the max before.
I have to disagree. My cards are stable up to 1230 and 2225 and beyond on 11.0. It seems like any version above that even 1180/2135 I get 4% incorrect and no amount of underclocking gets rid of them. Anything faster causes instability. For me the faster speeds on 11.0 is a much better deal. I can go a lot faster than 1230/2225 on 11.0 as well. But I do have timing straps done on rx580 micron ram. I have a feeling that is the issue much more than OC speeds. Especially since the Nvidia guys don't seem to have any problems. |
|
|
|
I see some people have problems with latest versions because they use hard OC and/or custom bioses created for old versions. They work fine with old GPU kernels, but I modified kernels in v11.2 and you need to readjust your settings.
For those who cannot do it (or don't want to do it) I can revert some changes and release a version with "old-style" binaries. So I will release v11.5 within a few hours and it will be the only change in this version, so don't upgrade if v11.4 works fine for you.
So what would work better on an rx580: - stock bios with 11.4 - modified bios with 11.5 or is it just a matter of testing? Is it no longer worth strapping the ram? |
|
|
|
|
From testing my msi rx580s with 8gb micron I found they consistently wanted a multiplier of 1.81 between core and memory clocks. This worked from 1900 to 2270 on the memory. Go lower on core and you lose hash. Go higher and you don't gain anything. Plus you're just wasting voltage. I think the multiplier theory will work across all rx580s but it will be a different number for different memory.
Keep this in mind when fine tuning cards. It will help save time.
Another quirk I've found is the higher of the chip and ram voltage is used for both. At least that's what HW reports. It is also about 0.025 lower than what you put in overdriventool. Or is it just my cards?
When they crash they usually reset to the stock bios core and ram speed. Keep that in mind when flashing the BIOS. But they always go to 950 mv on the ram.
|
|
|
|
This is what I've found too with dual mining. 11.3 or 11.4 you have to slow down the cards to get stability but still 4% incorrect shares.
With 11.0 I get slightly slower speeds vs the same card speed but I can turn them up more with stability. I dont have any incorrect shares so far a couple hours in. The effective speed is much higher.
The incorrect shares is due to the speed boost claymore did, claymore increased memory intensity for the boost and now to make it right people need to reduce their memory clock speed. Is it also possible those with modified ram straps are seeing more issues? |
|
|
|
Finally the last version of claymore is stable on my rigs. After some problems, now everything is working just fine. Thanks!
What version did you come from and what changes did you make? Gains? |
|
|
|
nah, save the hassle. reverted back to v11
no matter what i do with downclocking, i always get random incorrect shares. i dont know whats up with this higher hashrates on latest claymores, but its randomly throwing me a lot of incorrect shares.
This is what I've found too with dual mining. 11.3 or 11.4 you have to slow down the cards to get stability but still 4% incorrect shares. With 11.0 I get slightly slower speeds vs the same card speed but I can turn them up more with stability. I dont have any incorrect shares so far a couple hours in. The effective speed is much higher. |
|
|
|
|
Got 4% incorrect over night again even though I turned down the cards further. This is dual mining.
Might just revert to 10.6. The effective rate was much higher and the cards were stable at higher speeds.
|
|
|
|
|
I am stable on 11.4
2 rx580s with micron ram are doing 62.3 combined. Zero rejects, stale, or incorrect.
|
|
|
|
|
Maybe just use nicehash for a while. It will run a benchmark and mine whatever gets you the most profit. You are paid in bitcoin.
|
|
|
|
I do not use overdrewventol. Only AMD Radeon Software. Reset WattMan settings?
Wattman sucks but yes. |
|
|
|
|
Try using overdriventool instead of AB
|
|
|
|
Hi!
Sometimes right after program starts, the fan spins on high speed right away instead of gradually increasing speed to the right cooling temperature. Program settings used are -tt 57.
Actual temperature after the start is, for instance, 48C and fan speed is 93%. Is there a way to correct that?
Before starting go to overdriventool make a small change hit apply then start claymore |
|
|
|
Just finished the test at the WALL
I am sticking with the previous version.
V11.3 does offer a little increase in performance but it come with a Energy demand increase that offset the increase in hashrate.
Overall, I was earning more/card/day a week ago
Ok great now try 11.4 |
|
|
|
So incorrect are just as bad as rejected? Wasted time and power.  (Edit: I did find where someone mentioned incorrect share is local, rejected share is pool and both are equally bad) I'm going to revert to 10.6. I could run the cards faster with almost no incorrect shares.  |
|
|
|
|
In 24 hours I'm getting 0 rejected eth shares and about 22 incorrect per card. 30 rejected decred per card. 0 memory errors. On 11.3. Is this acceptable?
|
|
|
|
|
I had to downclock from 1215/2200 to 1200/1170 to stay stable on claymore 11.3. The faster speeds are stable on 10.6. Maybe ram strappings need to change? Micron ram.
Anyways at the slower speeds and 24 hours
Eth: 30.1, 31.7
Shares: 534, 581
Rejected: 0, 1
incorrect: 28, 18. < ---- about 4%
Decred: 902, 952
rejected: 30, 30 < ---- about 2.5%
Temp: 62c, 63c
gpu memory errors: 0, 0
Should I:
1) slow it down a bit
2) revert to 11.2 or 10.6 I think I could run faster AND more stable
3) try different straps maybe they will work better with 11.3
4) not worry about it? Are the numbers ok?
|
|
|
|
|
I have found with my rx580s when they crash they go back to stock bios settings and 950mv. If ram is set too fast the card can be restarted by restarting claymore or just restart the card in claymore. If that does not work then the gpu core settings are too fast or not enough volts. A computer restart isnneeded. I have found v11.3 contributes to the latter.
It's possible more tweaking on your part is needed or maybe revert back a version. Also dont discount power. Could be your PSUs are at the limit and something is scheduled to come on at that time that causes a blip in voltage to the cards. Could even be power from your utility. Since the card dropping is random I bet it is one of those two!
|
|
|
|
|
Show Posts
