Claymore's ZCash/BTG AMD GPU Miner v12.6 (Windows/Linux)

[Marvell1]

Anyones hash rate really low on Nanopool? according to ethmon i'm 3k h/s lower than i should be , usually its 1k diff up or down not 3+ k

Looked like they were getting DDoSed yesterday so I switched to Flypool. Just switched back to Nanopool now. Will let you know what I see.

@fittsy thanks

[Marvell1]

I have few rx470 from xfx, the single fan ones, and with 1500 straps bios mod and clocks: 1200/1000mv 1750/1000mv they are doing 173h/s at 127W from the wall measured with smart plug. This is with V8 and -i 4 setting. This xfx model is the cheapest but it's very silent, running at 70C with 35% fan speed, the room temp is 25C. The nice thing about them is dual bios

Now this is some valuable information, thanks for posting this!

What is this smart plug?

Smart plugs are a nice way to restart machine remotely or monitor how it is runnig. I use tplink hs110.

those a are nice but wont work on a pdu

[m1ner]

how many solve rx 460 does per second ?

[Yaminat]

I'm beginning to believe my motherboard is not capable of handling more than 2 cards ugh! Msi z170a gaming pro carbon either on boot (windows 8 or 10) doesn't recognize cards, afterburner doesn't show temps and card crashes or opencl hangs on miner

I got a Z170 board not knowing issues.  It will see and mine with 4 GPUs no problem (470/480), though if you try 5, it will lock up at bios, even though board has 7 slots. I have http://www.newegg.com/Product/Product.aspx?Item=N82E16813128838 Then I just decided to get H81 BTC pro boards (harder to find).

This MSI Z97 works just as good as an H81 BTC - cheaper in most cases - plus it has 7 pcie slots - I have 2 rigs with this board and 7 RX480's each. Works great.

https://www.amazon.com/gp/product/B00K8KPXUO/ref=oh_aui_detailpage_o07_s01?ie=UTF8&psc=1

anel and cooling (in summer, Georgia), but I can recommend to friends.
[/quote]

For that Motherboard, do you need to short the A1B17 manually or it is automatic like the ASrock boards?

[mettalmag]

after updating to V8 since the release day only three rigs out of 84 has the problem

Code:

Watchdog: GPU 0 hangs in OpenCL call, exit

and after that
Reset control for GPU 0 ...
and miner stops and hangs the PC
in V7 there wasn't a problem like this
any solutions ?

[LACamomile]

v8 -i 5

GPU #0: Ellesmere, 4096 MB available, 36 compute units
GPU #1: Ellesmere, 4096 MB available, 36 compute units
ZEC - Total Speed: 387.347 H/s, Total Shares: 21386, Rejected: 294, Time: 37:56
ZEC: GPU0 193.353 H/s, GPU1 193.995 H/s

MSI RX480 Gaming X 4Gb undervolted
GPU 1250 Vout 0.96V
Mem 1900 (timings from 1500) 0.975V

Both cards in pc box, stable as rock
GPU0 temp/VRM: 68/74 C
GPU1 temp/VRM: 62/66 C

1300/2000 clocks doesnt give too much boost, but will req Gpu 1V/Mem 1.2V, and temps goes to 70-75C.

[adaseb]

after updating to V8 since the release day only three rigs out of 84 has the problem

Code:

Watchdog: GPU 0 hangs in OpenCL call, exit

and after that
Reset control for GPU 0 ...
and miner stops and hangs the PC
in V7 there wasn't a problem like this
any solutions ?

Use -wd 0

When that GPU dies, you need to manually restart computer before re-running the miner.

[mettalmag]

after updating to V8 since the release day only three rigs out of 84 has the problem

Code:

Watchdog: GPU 0 hangs in OpenCL call, exit

and after that
Reset control for GPU 0 ...
and miner stops and hangs the PC
in V7 there wasn't a problem like this
any solutions ?

Use -wd 0

When that GPU dies, you need to manually restart computer before re-running the miner.

So far so good with -wd 0
also I wonder is it mandatory to set 16gb virtual memory ?

[adaseb]

after updating to V8 since the release day only three rigs out of 84 has the problem

Code:

Watchdog: GPU 0 hangs in OpenCL call, exit

and after that
Reset control for GPU 0 ...
and miner stops and hangs the PC
in V7 there wasn't a problem like this
any solutions ?

Use -wd 0

When that GPU dies, you need to manually restart computer before re-running the miner.

So far so good with -wd 0
also I wonder is it mandatory to set 16gb virtual memory ?

No idea but most of us have like 500gb of free hard drive space so why not.

[tanatana]

https://i.imgur.com/ZcZB2S2.png

is that a problem?
after few hours appears incorrect data

this could be due to excessive overclocking or undervolting.

not overcklock, not undervolt, there is change bios
did not claymore 5, 6, 7 only 8

Try reducing the intensity. Use -i 5 or 4 or 3. The V8 stress the GPU more. I had the same problem with V8.

voltage increases on the problematic card
1,09 -> 1.15
problem again
2*7950 card
-------------------------------------
Now I reduced the intensity of 3

thanks. is fine

[mettalmag]

after updating to V8 since the release day only three rigs out of 84 has the problem

Code:

Watchdog: GPU 0 hangs in OpenCL call, exit

and after that
Reset control for GPU 0 ...
and miner stops and hangs the PC
in V7 there wasn't a problem like this
any solutions ?

Use -wd 0

When that GPU dies, you need to manually restart computer before re-running the miner.

So far so good with -wd 0
also I wonder is it mandatory to set 16gb virtual memory ?

No idea but most of us have like 500gb of free hard drive space so why not.

That will be a huge issue for my, most of my SSDs are 16 and 32Gb..
One of three did the same hangs opencl call other two are working fine so far

[adaseb]

after updating to V8 since the release day only three rigs out of 84 has the problem

Code:

Watchdog: GPU 0 hangs in OpenCL call, exit

and after that
Reset control for GPU 0 ...
and miner stops and hangs the PC
in V7 there wasn't a problem like this
any solutions ?

Use -wd 0

When that GPU dies, you need to manually restart computer before re-running the miner.

So far so good with -wd 0
also I wonder is it mandatory to set 16gb virtual memory ?

No idea but most of us have like 500gb of free hard drive space so why not.

That will be a huge issue for my, most of my SSDs are 16 and 32Gb..
One of three did the same hangs opencl call other two are working fine so far

With -i4 each GPU uses like 1.5GB. so you need at least that amount of virtual memory.
If you don't have enough virtual memory, the hashing will fail to start. The GPU 0 hang is something else.

[mettalmag]

after updating to V8 since the release day only three rigs out of 84 has the problem

Code:

Watchdog: GPU 0 hangs in OpenCL call, exit

and after that
Reset control for GPU 0 ...
and miner stops and hangs the PC
in V7 there wasn't a problem like this
any solutions ?

Use -wd 0

When that GPU dies, you need to manually restart computer before re-running the miner.

So far so good with -wd 0
also I wonder is it mandatory to set 16gb virtual memory ?

No idea but most of us have like 500gb of free hard drive space so why not.

That will be a huge issue for my, most of my SSDs are 16 and 32Gb..
One of three did the same hangs opencl call other two are working fine so far

With -i4 each GPU uses like 1.5GB. so you need at least that amount of virtual memory.
If you don't have enough virtual memory, the hashing will fail to start. The GPU 0 hang is something else.

after setting wd to 0 I also did i to 0 too
lets see what will happen

[Flickspeed]

It currently seems that there is much more progress to make for the RX 4xx cards. The v8 miner is not utilizing the full memory speed of the 480s as it seems. Lowering the memory speed from 2200 to 2000 yields same hashrate. While lowering the core clock makes a bigger impact. This should not be the case since the 480 has a faster core than the 390 and the 390 is getting much better results, why? Am I wrong about anything?

[oslak]

It currently seems that there is much more progress to make for the RX 4xx cards. The v8 miner is not utilizing the full memory speed of the 480s as it seems. Lowering the memory speed from 2200 to 2100 yields same hashrate. While lowering the core clock makes a bigger impact. This should not be the case since the 480 has a faster core than the 390. Am I wrong about anything?

You are wrong on everything. LOL

[Flickspeed]

It currently seems that there is much more progress to make for the RX 4xx cards. The v8 miner is not utilizing the full memory speed of the 480s as it seems. Lowering the memory speed from 2200 to 2100 yields same hashrate. While lowering the core clock makes a bigger impact. This should not be the case since the 480 has a faster core than the 390. Am I wrong about anything?

You are wrong on everything. LOL

Do some testing before making such useless posts. Hashrate doesn't even change when lowering the memory clock to 2000. So someone needs to explain why the rx480 is performing so porrly compared to the 390.

[mamedesign]

https://pbs.twimg.com/media/CyQsBvTUUAAGYlv.jpg:large

[deadsix]

It currently seems that there is much more progress to make for the RX 4xx cards. The v8 miner is not utilizing the full memory speed of the 480s as it seems. Lowering the memory speed from 2200 to 2100 yields same hashrate. While lowering the core clock makes a bigger impact. This should not be the case since the 480 has a faster core than the 390. Am I wrong about anything?

You are wrong on everything. LOL

Do some testing before making such useless posts. Hashrate doesn't even change when lowering the memory clock to 2000. So someone needs to explain why the rx480 is performing so porrly compared to the 390.

He doesn't need testing, he (apparently) has spent some time reading about the Algo to know that Equihash is completely compute dependent, and on the current RX cards that's the limiting factor so increasing memory clocks will not help much. Coding for algo's like these is about working on the limiting factors untill you hit their limit or something else becomes the limiting factor. In this case we have hit the limit of the GPU and untill we can surpass it handily memory clocks do not matter beyond a point. It is also the reason much older cards do well on equihash because they have a large number of cores on the GPU. It would do you good to read up about the algo as well, before getting pissed at people.

[Claymore]

I see some posts that V8 reached hardware limits and there is no way to improve speed anymore. It's not true.
I'm working on new version with another good speedup, though it takes more time and efforts.

[Flickspeed]

Thank you for clarifying Claymore. Looking forward to the next release.