This option will not affect cuda/opencl - relates only to graphics with compute workloads (async compute)
Are you sure vaulter? From the link I posted from the NVIDIA webpage: "This setting is intended to provide additional performance to non-gaming applications that use large CUDA address spaces and large amounts of GPU memory" Yes - you are correct - it relates to virtual memory page size (64kb vs 128kb for Maxwell only) |
|
|
|
@EWBF_ I noticed a new option in the Nvidia control panel when installing the 381.65 drivers (seems to be in there since 378.72 or something) for my GTX 980. It is called: " Optimize for Compute Performance" screenshotAt the moment when I enable that setting the miner (v3.3b) gives an error message: can not run workersMaybe making use of those optimizations could improve efficiency of the miner as it does for some other applications ? This option will not affect cuda/opencl - relates only to graphics with compute workloads (async compute)Correction - this option relates to virtual memory page size (64kb vs 128kb for Maxwell only) |
|
|
|
sp or anyone else. This is from the EWBF thread:
@EWBF_
I noticed a new option in the Nvidia control panel when installing the 381.65 drivers (seems to be in there since 378.72 or something) for my GTX 980.
It is called: "Optimize for Compute Performance"
screenshot
At the moment when I enable that setting the miner (v3.3b) gives an error message: can not run workers
Maybe making use of those optimizations could improve efficiency of the miner as it does for some other applications ?
This could have some impact on our 970/980 cards if someone could get it to work with miners...
This option will not affect cuda/opencl - relates only to graphics with compute workloads (async compute) |
|
|
|
Just wait for the GTX1080 with 2nd Gen GDDR5x.
@Stock the 1080TI is doing 32MH/s on ethash ..with OC I reach 36MH/s
According to NVIDIA the new memory will be deployed along with new GTX1080 too which should make tghen also 32MH/s at stock.
For 1080 11gbps it would be around ~26mh/s - 10% faster then 1080 10gbps version (which is around 24mh/s stock) 1080ti is 36mh/s stock as you are apllying oc to P2 state (which is 500mhz less then P0 game state) = oc'ying for 500mhz results in 5500mhz in afterburner with founders edition |
|
|
|
I think I will buy it for gaming when it is available in my country. We will see how it performs on mining with lower TDP.
Minimum tdp is 125w for founders edition - basically it is almost the same perf per watt as 10xx Makes sense only from density perspective (less mb/cpu /memory/frame material/ventilation - in terms of quantaty of fans/ convinience) 1060oc - we need to see prices first Equilihash can be easily pushed to 700+ but it will require more then 180w - i myself would prefer this if it wouĺd be a little cheaper - i think 1070 would still be optimal (might drop a little in price - i hope) |
|
|
|
Where you have taken the numbers from?
 Consider it titan xp numbers - might be a bit slover on eth (1080ti) - but around 5 % |
|
|
|
|
Its 660 at 180w
Eth is 38 mhs 125w
Pascal is 1650 at 180w
Lbry is 450mh at 180w
Sib at 21mhs at 180w
|
|
|
|
I use one wallet.
what -i and -c param are you using? 1070 rigs? What speed do you have per 1070? Thanks in advance |
|
|
|
I think I can get my 980ti to mine @ 1GHASH as well. But a waste of time.
They tested the opencl bitcoinminer, wich is slower than ccminer.
You can get the performance of the 1080 for free if you just know wich software to run.
And which software as of now is the best to run on the 980ti to get 1080 hashing power? Has this changed in the last 2.5 months that is was posted? The 1080 is the fastest ZEC miner. Well, actually it's the pascal titan-x. Does anyone here have a pascal titan-x? What hashrates is it capable of with ZEC / ETH / XMR? I don't think it can approach the hash/$ of a 1070, but I would still like to know what a 6x titan-x is capable of. 38mh at eth 660sol/s at zec Titan x pascal... |
|
|
|
EU server down.please check
Must be only you, everything working fine here. We are releasing EQM v1.0.4b.
666 Sol/s on Titan Pascal.. Stock or OC? Try with mem overclock (as high as possible to be stable) and then lower the TDP. See how far that brings you with minimal power usage. OCd with +110 Core and +700 Mem (Resulting 5200 - max stable) tdp 60(150W) - 560 s/s tdp 70 (175W) - 620 s/s tdp 80(200w) - 650s/s tdp 90 (225w) - 666s/s tdp 100(250w) - 690 s/s |
|
|
|
We are releasing EQM v1.0.4b.
666 Sol/s on Titan Pascal.. |
|
|
|
Vega uses the same technology as the current Polaris - except for the use of HBM memory instead of GDDR.
It's more the "next generation of Fury" than "the next generation of Polaris".
Think "Fury/FuryX/Nano" vs "R9 390/390x".
The RX 490 (if it actually appears) still looks more likely to be a "dual Polaris" card than anything else.
Vega is new arch (well modified more than Fury or Polaris) Its Single chip (there will be several versions - like Fury and Fury X, Nano? - but I'm not 100% sure) Its likley 8Gb HBM2 but it is announced to work as cash for all memory in the system (it may have onboard SSD?) and we need to see how it will work for mining in terms of latency and other practical aspects of memory (might be similar to furys HBM and GDDR5X) - so it might be good or not so good for mining.. In terms of the TDP - its likely to be 250W card (top card but there might be lower TDP version). It should be expensive and comparable in perf with Titan XP (but that is yet to be seen) The big question is when it is released and quantities (expected to be really poor if it launches feb-march - less than 10000 units world wide) Should OC poorly due to HBM2 |
|
|
|
Some update...
If your GTX 1070 with modified TDP and OC reached 420 with current version, then next version will give you 450 (same settings).
GTX 1060 6G limited to 90W TDP and overclocked memory is hashing at 290. Going to 300 or more with 1060 will be easily possible (with OC of course).
GTX 1080 overclocked to 2100 and memory to 5300 gives 490. But the real power of this card is when limiting TDP. 115W TDP limit with mem oc gives 460 sol/s.
Waiting for the release to post Titan XP numbers - currently its 620 with some OC... |
|
|
|
because you have to pay to breathe
Remember when wolf0 wanted 50BTC on his work on the x11-x15 miner that was sold cheap and stolen? Seems that the wolf is doing pretty good right now. His public adress shows 200 BTC already. That is a $200 000 profit in a couple of years of spare time coding... https://blockchain.info/address/1WoLFumNUvjCgaCyjFzvFrbGfDddYrKNRGood payment for a few lines of code... But worth every penny.. He seems to be in the pocket of genesis mining now, and the opensource competition is only done by students and lamers The xmr hashrates on the cheapest modded bios 470 cards is around 700-800h/s with the wolf0 miner. The unreleased Sp-mod #2 is only doing 550 on the 10603gb. (but only 40 watt in the wall) Hi SP, do you still have 980Ti - what is xmr score on that one? Curious because TitanXP and 1080Ti have 384bit bus and I'm woundering if it has any boost vs 256 on 1070 ( I do have Titan XP but dont have private kernel  |
|
|
|
We have another speedup ready, mostly beneficial to GDDR5X cards (Titan X Pascal will go over 600 for sure). Stock GTX 1070 gets 390 Sol/s.
We have decided to release source of miner and open it up to work on every pool in near future. We will prepare full statement why we have decided that.
Regarding linux: somewhere along the line of optimisations, linux version simply stopped working correctly (is giving incorrect results). We do not have enough man power (+ there are holiday times) to track down the issue, but I am sure community will find it and fix once we release the source.
Weird, interested to hear why that is... I could understand opening up to pools and making it a fee based miner, but open source is something else entirely. Maybe has something to do with the first release by nicehash, which was open source and used work by a few different contributors. Although there is no doubt nicehash has made a great many optimizations to the CUDA code, which will be really interesting to see. This is great news for miners and also very generous by nicehash for sure! I'm actually thinking Nicehash coders have reached a brick wall and they're going open source to get 'input' before going closed source again. BTW Nicehash as I've asked in other threads, consider tackling Cryptonote. The code basically hasn't been touched in years and it's a pretty big economy. Definitely something worth taking a look at. I second that - Nicehash - are you going to look at Cryptonote algo maybe asking the same dev who worked on Zcash djeZo? |
|
|
|
Getting 566 sol/s on my Titan X Pascal at Stock settings.
@vaulter,
What settings you running on the titan?
Didnt tune it much - 90% tdp, +50 core +150 mem Win 10 64 376.48 hotfix driver |
|
|
|
Can you add average hashrate per worker to web stats - it will be eaisier to monitor rigs
Click on the algorithm and then scroll down. There you will see list of your workers and their estimated speed. Yes - but its current speed - it would be nice if you added average 24 hour speed per worker (you have average overall speed but if I could see per worker I could see faulty/glitchy rigs better comparing to each other) |
|
|
|
I've just tried NiceHash 1.7.3.10 and I've got lower results than with Claymore dual ETH/DCR.
It is ok. You can also use other miner to mine on NiceHash, just make sure to enter correct stratum server. Can you add average hashrate per worker to web stats - it will be eaisier to monitor rigs |
|
|
|
|
Titan X Pascal - 590
1080 410
1070 400
titan X 360
980 260
970 240
all with lower tdp and various oc settings
|
|
|
|
There are maybe 10-20% of NVIDIA miners. What you are competing is basically AMD and Claymore. With no CUDA competition, we would be still pushing updates to this miner.
BTW are you going to implement 24h Avg hashrate per worker - it will be easier to monitor per multiple rigs to see if any given rig is acting strange? |
|
|
|
|
Show Posts
