 Show Posts
|
|
Pages: [1]
|
Guys can you explain me this ? https://imgur.com/a/Y7FfW GPU 0-3-4 ABSOLUTLY SAME CARDS 56s... GPU 1-2 Same 64s too... Same clocks.... same cooling... same settings... But different in temperature by 10c  ? How is something like this possible ? Yes, I would bet you have 2 cards that have the filled package (Better) and three that don't have the filled packaging (shit). https://community.amd.com/thread/219528Have you looked at the HBM temperatures that the critical thing? Take the higher ones off and redo the paste, may need a good amount of paste for the non filled HBM cards. |
|
|
|
I have been using CASTxmr for a few weeks and really like it's simplicity and function so thank you, Gandalph!
Had a question...I built 2 rigs:
Rig 1 - 2xRX570 4G and 1xRX580 8G hashing at 2468 h/s avg.
Rig 2 - 2xVega64, 1xVEGA56, 2xRX580 8G hashing at 7680 h/s avg.
I am mining both rigs for ETN at etn.spacepools.org using different workerid's but to the same wallet and have Cast 085, --fastjobswitch activated, AMD Blockchain driver, softPPlay, and OdriveN settings to get those rates on the Vega's and various bios mods for the 570/580 cards from around the web.
I was thinking that Rig2 with the higher total hash rate would find more shares than Rig1, but after starting both rigs at the same time, the shares found for each are about the same with a few more shares actually coming from the rig with the lower hashrate. Probably need to look into the concept of pool mining to figure it out, but was wondering if someone can provide some insight. I was thinking 3x more hashes would at least result in more total shares from one rig vs. the other.
Thanks again.
Check the difficulty on eac miner. I would expect the higher hashing rig to be on a harder difficulty. Pools shift higher hashing miners onto higher difficulty to reduce their loading. You should get paid the same per hash regardless of difficulty. |
|
|
|
Blockchain driver installed miner says it can't find it...  You need to select the right graphics card for a start. Try adding -g 1 to the string. Currently, you are trying to mine on your Intel graphics! |
|
|
|
Boo Yar Got 1300SOC timings running on a Vega 64 BLOCKCHAIN DRIVERS Hashing at 2218H/s.  HBM at 1225 v1.35 Core 1087 v1 Core Temp 37C HBM Temp 42C Stock Bios Win 10. Aug Blockchain. Cast XMR 0.8 5 more on the memory starts to error. Reg Mod for this, Again Vega 64 Powercolor, stock bios NOT water cooled bios. Lowers the p7 so that at -30 hits 1087 on this card. Starts the memory at 1175 so do not even consider using if your card cant do this kind of timings already. The second upload is the bios mod for the same stock 64 bios but at 1200soc. Gives 900mv at -30% core = 980mhz, HBM starts at 1170 on that. Pulls ~2000 with HBM upped to 1190. 1300SOC - slightly lower p7 core https://drive.google.com/file/d/1YVtLP6XUidneMNW7J_XgKsjLELriMM6x/view?usp=sharing1200SOC - lower core speed for p5-p7 and voltage. https://drive.google.com/file/d/1epe-KoXk5wUPn02H7D5qeWAleuLC6DHj/view?usp=sharingBoth cards EK water-cooled. 1300soc is running at 36C and the 1200soc is at 38C. |
|
|
|
Hello I´m trying to get new rig runing. My 3x vega cards on main PSU are woking fine but whenever i try to run the cards which are supplied by second psu it runs for sec and then the second psu and the cards supplied by it shuts down. My main PSu is Evga gold 1000GQ My second PSu is same brand but 700w - Evga gold 700GQ which i use to power 4th and 5th card and their risers(tried to power risers from main psu had the same effect) I can mine on 4th or 5th vega when launched without the cards on main psu
Is it a PSU problem ? i also have just 4gb ram but i have enough paging file memory on ssd
what cables are you using? I experienced a similar problem trying to connect a couple of Vegas to a 750W PSU which only had 3 8 pin PCIe connectors, so I had to use a 2 8Pin PCIe to 1 8Pin PCIe adaptor, and the PSU run for less than a second because that cable in particular created some kind of short circuit, and the solution was to use another conector, a SATA to PCIe adaptor which did the job quite well. It loooks like this https://imgur.com/a/EdcbA after i start cast xmr it works for few seconds then the 2x cards on other psu shuts down im using https://www.miner-shop.cz/prislusenstvi/psu-2-psu/tryed changing gpu powers cables to 2x 8pin in vga and 2 x8pin in psu from 2x 8pin in vga and 1 8pin psu it had same effect I managed to get it work with turning on and off all cards in devices and also turning on and off hbcc on all cards this makes it work but only for 80 mins then same thing happens you don't need to do that. Maybe that cable isn't stable enough and creates some stress on the motherboard. There is a much, much cheaper and energy efficient solution to use 2 PSUs at once --I've seen some miners using a very expensive adaptor to connect 2 PSUs to the same motherboard.  ONE of the PSUs doesn't need to connected to the motherboard at all. Learnt this trick in my first year studying computer science and testing computers, 2 years ago --it was one of the questions in the exam  . The green and black cable of the the PSU can be bridged using a paper clip, this is meant for the PSU to be tested and check if it works or not. If the fan starts spinning then the PSU is fine (so you check for issues on other componentes). The EVGAs include a piece of plastic to make that bridge and test the PSUs. You can use that or just the paper clip trick, a bent needle, etc etc etc. https://aphnetworks.com/tutorials/psu_paperclip_trick my psu´s cables are all black probably an EVGA? EVGAs come with a plastic piece and 2 pins which have the same function as the paper clip trick, checking if the PSU is working well. iirc, it's the 13th and 15th pin you just need to connect, but I am talking from memory here, there should be guides online. both Psu we workin just fine i dont know what was the problem tryed everything several times including diffrent 2x Psu adapter or without adapter using paper clip triick im not amateur when it comes to hardware but ended burning up my second Psu with dat magic paper clip trick , But I managed to get 5 vegas workin on the first psu 1000W evga but undervolting it alot l i get now 9150h but pc crashes from time to time (5times a day least..) So I want to buy second psu to get it workin properly with stable hashing 24h , Problem is the same Evga gold 1000W which i alredy have would take 3 weeks before it gets to me So i want to buy other Psu but dont want to burn it like the last one ? so Should i wait for the same PSU or can i buy diffrent PSu with 1000W power ? Btw thx for answers Since you can get a VEGA 64 hashing at 2100 with under 100W, why are you messing about with dual power supplies, when you have a 1KW one? Rated for what 83Amps on the 12v rails, making 996 watts. With 100W for the board, CPU and stuff, gives ou 150W per vega for 6 of them. OK so it's tight but doable. Another option is a second board/rig. Always handy to have a testing/play rig anyways. Finally, is it just one card that doesn't want to play nice with others? As I have one that will not work in any system with any other card whatsoever. Tried multiple systems and configurations, but just one Vega has to be on its own. Haven't tried since I flashed the bios to copy another card (resulted in better hash rate) as it was one of the very first vegas. |
|
|
|
|
MSI 560 4GB ITX Hynix
Strap 999000000000000022AA9C0073E66C448055131243CFB60D00408400750514204A8900A00200000 014112E34DB324418
Used for 1750 to 2250.
Core -7.5%
Mem 2200
Eth 16Mh
Cryptonight 440H
Cant find who posted the handmade straps, but it works an absolute treat.
|
|
|
|
damn, I would like to see if that is possible SOC 1300 =) Good Luck, Keep updated!
Update from Saturday nights mining rig play time. Applied at 1207 and then a 1300 SOC timings Reg edits. The SOC goes to 1300 on both, according to HWINFO. So far so good, that worked and I could mine at HBM1190 still without any issues. Onto the higher frequencies of HBM, Slowly clocked it up and up, did get to 1250, but after 10 min at that speed, it crashed out. Before it crashed I was pulling +2300H. Keeping the temps at 40C max. HWINFO isnt reporting the memory errors on this rig - no idea why at this time, also one rig the same version of HWINFO doesn't even show the SOC timings. Good job the other rig does so I know the mod works. Put both rigs back on the 1200SOC reg file and left it there for now. Vega 1 is pulling average share find speed of 2050 and vega2 is pulling 2100, (Actual Mining figs are 2100/2300) both on stock Volts, both HBM at 1190 now and both with -30% on the core. The only diff is Vega2 has way better watercooling at 40C max so can be at +50% power and Vega1 is at +10% power and is at 50C. SOO Not sure if its the SOC or the RAM erroring out. Would need to put one on 1300SOC / 1190 HBM and leave it running for a week to see if its stable. Also, Anybody pushed more volts via the HBM? I know everybody wants more efficiency, but I'm free electricity and water-cooled. I want Max performance. Also managed to get two 560s pulling in 440H each so total is now 5k Hashes on Monero, with 2 more Vegas on order  |
|
|
|
|
ETH/Lib 12Mh / 30MH.
XMR 330H
Nicehash, so latest claymore
I have 2 x MIS 560s 4Gb Hynix.
No additional PCIe power required on these, which is often handy but limits the cards.
1500 to 1750 strap. Running at 1750 memory. 1500 timings don't work above 1750, Can get to 2000 mem at 1750 timings, but only just gets back up to the same performance as the 1750mem running at 1500 strap, but with more W and heat.
Core at stock (make no difference) BlockChain drivers, win 10 64,
|
|
|
|
|
My money is on your HBM temperatures being your issue.
If HBM gets above 55C your performance gets hammered. I'm running in CAST with 1300soc timings on two different rigs, as my vegas just would not play nicely together (one card just will not work in any system with any other card in it!!)
I can see on the weaker water cooled rig that if the HBM/Core gets to 55C the performance drops from +2000hs to 17/1800 area.
Do you know what temps your running? Are the higher performance cards cooler by some chance?
Also well done on getting all cards at over 1100 memory timings.
|
|
|
|
why people here like to mine cryptonight algo??? i think it's not profitable
Cause its tweakable as f*k. Cause VEGA (i have non though) can do 2000 h/s while 45 mh/s on Eth? cause with right tweaks, a normal 4gb 580/570 with Hynix can do 925 H/s +++++ and still be power efficient compared to Solo Eth mining (let alone dual mining eth with something else, that overheats cards) As of today though, I think Eth is skyrocketing. but still i rather mine cryptonight, cause CPU also adds to your combined hash. what the hell does tweakable even mean ? even the most profitable cryptonight coin wouldn't make it into the top 10 most profitable coins to mine list, CPU also adds to your combined hash ? it also adds to your wattage so that defeats the purpose, allmost all the Equihash and Neoscrypt coins are more profitable than any Cryptonight, who the hell mines with Vega it's an expensive one trick pony 2 X 3GB 1060 's /570's would outperform it in every single algo and department including PRICE Really? My two vega 64, picked up for £460 each +200 for water blocks onto the existing loops, mine 2100H EACH on Crypto (CASTXMR) at 160W each. Tell me your cards can do 4kh at 320W? Which is pulling £450 per month on Nicehash at the moment. |
|
|
|
|
By your calculations hes fine.
There are three +ve lines in a 6/8 pin PCI-e power line.
10A - 12v = 120W
120W x 3 - 360W
|
|
|
|
UPDATE on my overclocking. Cast 0.8 Blockchain drivers 2 system, 1 card per system, both Powercolor Vega 64s with EK water blocks: Win 10, 16GB ddr3 large pagefiles on SSDs. Vega 64 Watercooled. System 1 holds 50C system 2 holds 35-40C at full mining. Card 1: power +10%, Core -30% (1164): HBM 1150 = 2015Hs average. CARD 2: power +50% Core -30% (1164): HBM 1190 = 2200Hs average. Both cards have accepted SOC timings registry mods to the SOC timings at 1200. This has allowed both cards to hit 1950 memory speeds. However rig1 has a weaker cooling, (single 240rad and 2 week 120 fans) so had the throttle back to keep the core temp below 50C. Above this temp the rig hashes lower (1900) Rig 2 has 3 120rads, 6 120 corsair fans and a massive pump, so is set to 40C and have lots of headroom on that loop. The two cards do not play nicely with each other, else they would both be on Rig2s mega loop. Both at standard voltages at the moment. Next steps, this Saturdays modding session: I have a reg mod for SOC 1300. Which, if the SOC will go that high stable will allow me to see if I can get the HBM past 1190. |
|
|
|
me to 1200 h/s. driver 17.12.2
1200h is about standard before you "restart" the drivers when using Blockchain. Try stopping the miner, open crimson, toggle the "HBCC" switch and apply. Wait for the Crimson to reinitialise and then start Cast. That how you get 2k+ on Blockchain. |
|
|
|
thanks for all help.
looking to buy another vega. tried to buy sapphire on black friday but guess what they OVERSOLD cards and no luck for me.
so now due lack of 56 on sale and way overpriced on ebay I'm looking at vega 64.
Which one to go for? Powercolor? Gigabyte or wait for 3 fan version of powercolor next month for almost 30% extra?
Cant find any comparison of those cards anywhere
I have 2 of the powercolor 64s, both have hit 1100 memory stable, both running at -30% on the core, +50% power and do 2020 to 2050 hashes each. Using 120W core so about 150ish total. Which is nice. You do need watercooling on the 64 to get the best stable results. On Air the hash rate dropped to 1800ish when running for a while due to the HBM getting to warm. |
|
|
|
Finally got single MSI Vega 56 to run stable with xmr-cast-vega. My key was getting the temp right <65 to stay stable and keep MEM < 950 Block chain drivers and reset driver prior to running. Very stable. Freq: -8% Mem: 945 Power +10 Rate: 1980 Fan: ~3400 min Temp: 64 Ave Search 45 secs, no errors or Freq -8% Mem: 935 Power +0 Rate 1950 Fan: ~2500 min Temp: 62 Not bad, both very stable, but I have not been able to get to the fabled 2000  Here's my question. I mined 24 hrs with nicehash Nicehash reports between 1.33 and 2.2 at any given time and is quite variable on their BTC profitability curve through the day. anyone have good/bad NiceHash experience or can explain this phenomenon? Or, suggest a more profitable pool to point to? Thanks You MEM speed is way too low to hit 2k mark, stick it to 1100 and see what you get. Nicehhash speed on the site is calculated from your accepted share rate at the difficulty level submitted. Since there is a large amount of luck in the time between share finds, the reported rate will vary wildly. I get between 0 and 4KH on the site per vega64 running at ~2050each. Its the longer average that's important. So open the Nicehash page and select just Cryptonight and the graph has the average on it and you will see teh average which is usually more stable. |
|
|
|
Hi.
Param "--help" is not working. (-h, /?, -help ... also)
Can anybody put a list of commands in a text file here?
Open CMD and do it from there. Here is what it kicks out Cast XMR 0.6.0 CryptoNight mining optimized for AMD Radeon Vega GPUs Coded by [glph3k] for more info visit http://www.gandalph3000.com Donating 1.5% mining power to developer. Thanks. For current stats press 's' For seamless exit press 'q' ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Generic options: -h [ --help ] show this help message -G [ --gpu ] arg (=0) index of the GPUs to use. Separate by comma to specify multiple GPUs e.g 0,1,2 --opencl arg (=0) index of the OpenCL plattform to use (in case multiple OpenCL verions are installed) Pool Configuration: -S [ --pool ] arg pool server for a cryptonight based currency -u [ --user ] arg user or wallet address for the pool -p [ --password ] arg password for the pool server (if needed) --nicehash enable nicehasher 3 byte nonce Optimizations: -I [ --intensity ] arg (=5) intensity level (0=min power usage, 9=max performance) |
|
|
|
|
I tried this on a powercolor VEGA 64.
Getting a stable 2050Hs for the last 12 hours now, which is more profitable than the 44.2Mh this card does on ETH. Its submitting and being accepted at Nicehash just fine.
Settings
Core -30% (1150mhz) (no difference in hash all the way upto 1800mhz)
Mem 1100
Power limit +50%
Strings used (after the usual site and wallet) -G 1 -I 7
intensity at 7-9 are all the same hash rates so left at 7. I didn't need the --nicehash addon to make this work.
The system has another card as GPU0 which is mining ETH/LCR on the NiceHash Legacy software just fine alongside this.
HWinfo says about 120W core. Which is the same setting and power usage as I use on Claymore for 44.2Mhs on ETH.
The card is water-cooled with a EK full cover and running at 45C.
Now to see how stable this is over the next month or so.
|
|
|
|
|
