KASPA and Nvidia

Any idea how high the clocks can go?

That varies card to card. Test and see.

If you’re using bzminer try switching to core offsets and power limits. As well as locking the mem clock at 810mhz in the config

I tried the first 3070 using -100 for core offset and the reported core mhz was at 1.8gh, but the power went up to 230 watts. The hashrate did go up to over 600mh but the efficiency was very poor. I switched back to locked core at 1600 and the power went down again to 106. Any particular reason why you suggest offset while using BzMiner?

thank you for your input. We all value it tremendously here

You need to use power limits in addition to the core offset. That’s the most efficient way.

@keaton_hiveon any chance that the devs might allow this as an option in the oc? Last time I tried it, it didn’t cleared when I changed coins. Might be easier if you can do it through hive.

OMG, you are the man! So far I have increased hashrate at least 5% using the same power!

I have seen this issue when you don’t create a Overclock profile for that algo and HiveOs saves the OCs under the default profile. When you chance algo, it will keep the old OC settings.

I hope this is what you are referring to.

no. To lock the memory to 810 you need to use an option in the miner (bz uses oc_lock_memory_clock). When you change miners, the card will remain locked to 810, even if hive has a memory clock value. So your need to remove that lock (bz uses oc_reset_on_exit), but think lolminer doesn’t have an option so you need to reset the rig

I’m not sure why your HiveOs is doing that but my install indeed changes the memory speed.

Mining RavenCoin using T-Rex, nvidia-info shows me memory running at 6800.
When I change flightsheet to mine KAS using bzminer, nvidia-info shows memory running at 7205.
If I switched back from bzminer to t-rex, the memory resets again to the correct 6800

I have an OC profile for each algo with its own OC settings. If you don’t, settings will be saved as the default profile and none of the OC settings will change when you select a different mine or algo.

I was running the memory at 870 (KAS with bzminer) but changed it down to 810 to test your issue but worked ok for me.

bro, please tell me where the command should fix the frequency? on the cards themselves, I tried to put a fixed frequency of 810, it does not work…

Click on this button on the top of your screen after the miner is started and fill in the command

error(

sorry it is just (without the use)

--oc_lock_memory_clock 810

you can separate the values with a comma, if you want to adjust each card (either 0 to not lock the memory or 810)

yes, the team helped! thank you very much!

hi i have try to insert the comment and this is the error code i get

–oc_lock_memory_clock 810 (failed, exitcode=127)

Anyone may help for per GPU command ?

I’d like to try this 810 MCLOCK lock for my 3070 and do not interfer on others (see screenshot)

You can’t lock the other cards mem anyway so it will just fail and move on

1 Like

Ok for this point

As for the 3070, does the command line way is the only way to lock memory at 810 ?
Do not know if I am doing proprely but is this good ?


Your memory aren’t locked at 810, they’re at 7200.

You can lock via nvidia-smi in the shell or in the bzminer config with “oc_lock_memory_clock”: “[810]”

Roger, but sorry your command leads me to this

image