Topic: Claymore's Dual Ethereum AMD+NVIDIA GPU Miner v15.0 (Windows/Linux) - page 95. (Read 6590565 times)

you should not expect to see proper power consumption numbers from software especially on AMD.

Hi community!

Please help me yo understand what am i doing wrong.

I decided to lower powereat of my rig. Before everything CM 14.7 shows something like 90W GPU power consumption. Totoal about 360 on 4 GPUs rig. Okey. Wattmeter show 620 Watt at the wall. 1st WTF.
After a lot of work and experiments i managed to lower core and memory waltages to about 850 , lowered mem clock and etc. Now CM shows 55-57 watt per GPU , it suits info that shows HWinfo about GPU chip power ( so lets say in miner should be "GPU chip power consumption"). BUT...... WTF at the wattmerer i see almost the same 600 Watt , in miner i see "Total GPUs power consumption: 225 Watts"

Where Am i wrong ?

Best Regards.

what?

I'd suggest running @eliovp's AMDMemTweak tool to see if the strap settings are getting the modifications.  Run it without -strap first and note some of the values (tCL, etc) and then run it with -strap to see if indeed they are being applied.  There a thread in Mining (Altcoins) just for this program and the executable is up on GitHub.

Could you share the link of the other thread? I didnt find it, thanks

well first off the fact they are telling you to mod will only work that is out right incorrect with Windows OR Linux.

Another thing is the Rxboost goes from 1 to 30 there is no 50 rxboost.  so setting it to 50 will not do a thing.

For RX570's with Hynix in Windows environment - i suggest running this

-eres 1 -mode 2 -asm 2 -rxboost 21 -POL8H1@2200 -mvddc 875 -cclock 1200 -cvddc 880 -lidag 2

If 2200 doesn't give you a problems try pushing it to 2204 if that's solid try 2215 .. again if that's good try 2235 then 2250.... 2250 is the max you can push a 570 within claymore for the memory speed.
If the settings give you errors take it down to 2190, then try 2150... at worst 2100.
However with a factory bios  you should be able to achieve 2250 with claymore.

If you can get 2250 to hold solid and the cclock to hold 1250 @900 you'll be pushing your card to 33.6 to 34.3 depending on your silicone lottery.  

You may need to bump the mvddc and cvddc to 900 for the extra power on the max settings as well.

Keep in mind, when you have multiple cards int he rigs, I've seen having to push one or 2 cards down to 2100 or even 2000 to keep the rig stable but it'll still hold a 32 to 33 Mhs rate with the other cards. Its got to be an issue with how AMD is handling memory timings.

If you want to try reducing the crashing by slowing down the loading of each dag you can do a -gser 2 - which will load each card 2 seconds apart this also helps with stability.

Now i only use wattman to set the fans, but you can use tt for target temp and fanmax and fanmin to control the fan speed and max temp if you don't want to use wattman or afterburner.

By setting all that in the command string you will NOT need afterburner OR wattman except to control the fan speeds if you want to provide a range of speeds.

Best of luck!!!

@Claymore on the new 14.8 can you possibly add temp levels? I'd really like to get away from having to use a 3rd party tool.  If we can do a temp 1, temp 2 temp 3 ets on temperature levels and fan speeds that would be baller!

I tried all -strap (from 1 to 6) they all applied successfully but no changes.
Rxboost is not an issue (I tried all -rxboost from 1 to 100) they all work fine.
My question is how to get -strap working properly?
I mean I don’t need to bios mod the cards if -strap works fine because it will apply the new timing memory on the fly in Windows.

try other modes of strap 1-6 and rxboost 50 for example

You need mod bios to get more tha 30m/s, with not mod bios is not possible reach thats hash.

Hello guys,

I am experiencing an issue.

I made a new rig with 6 Phantom Gaming D Radeon RX570 8G OC, the memory installed on the cards is Hynix.
They are not bios mod since I wanted to use -strap command directly in Claymore's miner.
I am using Claymore v14.7 miner, with driver 19.4.3 since it is with this driver that Claymore has most tests been performed.

Compute mode ON, with overclocking, I uninstalled and installed the driver in claymore to be able to use -strap command.
And I am mining at 22.5 MHs no matter what -strap I am using.
The strap is applied successfuly : "strap "POL8H1" is applied successfully" on each card.

but the hashrate just refuses to change ! (-rxboost does actually work and allow me to reach 24MHs)

Anyone can help me to get -strap working on my rig and to reach 30+ Mhs per card, maybe by also changing the OC settings?

I will give to the one who provides me with a solution 100$ worth of Ethereum Classic.

The way that the Daggerhashimoto algo is designed it doesn't really matter what software you use, you will most likely get the same speed. Basically from the original ethminer included in the Ethereum github package to the advanced Claymore ETH miner software the actual miner speed is identical. Its just the way they designed the algo from the ground up.

The only difference is that ethminer was extremely buggy, basically from what I recall all my R9 280X rigs crashed every single time there was a DAG change. When claymore was released it fixed that and it also was more stable. And it had more features such as Claymore remote monitoring and Dual mining capability.

So in general the updates are mostly for stability and features instead of actual speed ups for GPUs.

cannot solve its long-standing DAG file problem, so the release is so long

  It doesn't fix the TLB problem with RX5700 (Navi)

I assume @Claymore will let us know if the 19.8.1 driver just released fixes the DAG issue he's mentioned that's holding back the Navi cards.

Hi! Can we expect in v14.8 straps for linux?
Is this even possible?

i begvto argue there. i mean 29to 31 mhs to 33 to 34 mhs i guess isntvTGAT major, but power is reduced for that, plus on vegas going from 37/39 to 42 to 50 depending on lottery isnt technically much? . idunno. i think you're expecting more?

Most of the improved stuff ain't felt (as it has all improved at the back-end), until and unless you actually do use one of the newly introduced features (such as the -showpower command). Plus, there hasn't been a major "hashrate increase" since the v12 anyways.

I have 12 Radeon VII's and switched from v12 to v14.7 and really see no difference.  What am I missing?

could i get some help, a guide on how to use the tool inside the program to overclock the amd graphic card?...

Use a kill-a-watt and convince yourself