Miner- Optiminer/Zcash GPU miner v1.7!

Can someone using nanos please state sol/s and watt at wall please?

I posted those numbers, above, a couple days ago.

1 Like

Sorry my bad think i was drunk at the time lol… 2500 sols at 1325 watts right? Have you overclocked your cards or are they stock? Have you tried undervolting them?

Right. They are not overclocked, but the power level is set to keep them from throttling back from stock clock (1000 MHz). They have a 118 mv core undervolt. Settings were done with BIOS mod, since I can’t change the voltage under linux programmatically.

I should add that a core overclock, with the v1.3 improvements, doesn’t get you anything, although it did in earlier miner versions. I believe opti v1.3 is close to saturating the memory bandwidth/throughput, so upping the core clock doesn’t improve things. I had a ROM with a 550 MHz memory clock mod, but it wasn’t very stable, at all.

1 Like

Thank you for clarification… lets hope we can squeeze some more out of these beasts :wink:

[2017-01-10 11:38:47.321] [error] Found solution index out of range: 2958311445

would be nice to know which GPU produces this error…

I’ve reduce my intensity to -i 3 to avoid GPUs locking up w/ custom bios.

So my MSI RX 470 4G cards are producing ~255 on v1.3.1

[2017-01-10 12:19:51.394] [info] [GPU4] 133.8 I/s 259.0 S/s (5s) 134.4 I/s 255.6 S/s (1m)
[2017-01-10 12:19:52.923] [info] [GPU5] 136.8 I/s 259.2 S/s (5s) 136.9 I/s 257.2 S/s (1m)
[2017-01-10 12:19:53.478] [info] [GPU3] 138.0 I/s 259.4 S/s (5s) 137.6 I/s 260.2 S/s (1m)
[2017-01-10 12:19:54.022] [info] [GPU0] 138.0 I/s 261.0 S/s (5s) 138.1 I/s 251.6 S/s (1m)
[2017-01-10 12:19:54.376] [info] [GPU1] 138.0 I/s 254.8 S/s (5s) 138.1 I/s 257.1 S/s (1m)
[2017-01-10 12:19:54.671] [info] [GPU2] 132.0 I/s 254.6 S/s (5s) 132.2 I/s 248.1 S/s (1m)
[2017-01-10 12:19:55.267] [info] [Total] 818.4 I/s 1568.8 S/s (5s) 817.1 I/s 1531.9 S/s (1m)

Version 1.3.2 released!

This is a bugfix release. No change in speed expected.

[1.3.2] Support older versions of fgrlx again.
[1.3.2] Fix abort when failing to list devices of platform.
[1.3.2] Fix no reconnect after ā€˜No such host’ errors.
[1.3.2] Add --benchmark options.

1 Like

Thank you, now is working fine :slight_smile:

Why the catalyst 15.12 drivers better than all other newer drivers? What the AMD did?

In one of my rigs, I have:
2x SAPPHIRE NITRO+ Radeonā„¢ RX 470 4G D5 OC
1x SAPPHIRE NITRO+ Radeonā„¢ RX 480 4G D5 OC

Using Optiminer I get the best results out of the miners tested, each of those cards provides 250-270 S/s. That’s consistent with what other users are reporting about the RX470, but I would have expected to get a bit more from the RX 480. Currently the 480 behaves exactly like the 470s. Anything I should try to do to improve it? Any tips will be welcome…

I am getting around 285-290S/s on my Sapphire Nitro RX480 with 8GB (stock). I know that the 8GB cards run at a higher memory clock rate. I also got the factory OC at 1342Mhz. But I am only getting ~15 S/s more than you are with a 4GB card. That would be 105 S/s over the whole 7 GPU rig… That is also pulling 1450 Watts to do 2KS/s. I need to figure out if the 8GB cards are worth it in the long run, and or start under volting.

AMD removed some OpenCL extensions from the newer windows drivers.

The 1.3 kernel is memory bound on the RX470/RX480. 4GB RX470 and RX480 have almost the same memory bandwidth. The 8GB 480 version memory bandwith is a bit faster.

I thought I’d mention that the ā€œhost not foundā€ bug fix in v1.3.2 seems to have taken care of a stability issue with v1.3.

Gotcha. Is the ā€œNitroā€ worth it? I can get Sapphire RX470 without Nitro for cheaper, anyone has an insight into those?

Awesome work from Optiminer.

Steady 280sol per GPU (480 4G MXM)

Complete rig 1060 watt :slight_smile:

Modified rom with custom timings…

Where did you buy mxm cards?

There are a few reasons to pick Nitros, but one of the best ones is that they come with dual bios, which means you are significantly less likely to perma-brick a card if you intend bios modding.

1 Like

Nope, i have tested it on 2 machines, 1 has already do it a couple of times