Big news from BBT about the Claymore epoch benchmark issue. He reached out to AMD who confirmed they are aware of the issue and are working with Claymore to resolve the issue which DOES NOT affect other miners, such as SGminer. So, it appears it's an issue related to the Claymore miner and NOT the RX platform. https://www.youtube.com/watch?v=l3XOtQtFKSQ&feature=youtu.be&t=978 |
|
|
|
Big news from BBT about the Claymore epoch benchmark issue. He reached out to AMD who confirmed they are aware of the future epoch hashdrop issue and are working with Claymore to resolve the issue which DOES NOT affect other miners, such as SGminer. So, it appears it's an issue related to the Claymore miner and NOT the RX platform. https://www.youtube.com/watch?v=l3XOtQtFKSQ&feature=youtu.be&t=978 |
|
|
|
Big news from BBT about the Claymore epoch benchmark issue. He reached out to AMD who confirmed they are aware of the issue and are working with Claymore to resolve the issue which DOES NOT affect other miners, such as SGminer. So, it appears it's an issue related to the Claymore miner and NOT the RX platform. https://www.youtube.com/watch?v=l3XOtQtFKSQ&feature=youtu.be&t=978 |
|
|
|
|
Not by coin, but by POW algorithm. You can run the benchmark tool included with the NiceHash miner. It will give you the speed of your rig for each of the different algorithms.
|
|
|
|
@rogeriosmorais:
Sata Power only got two 12V lines, each for 1.5A = 36W total
4-Pin Molex are 5A on 12V = 60W
PCIe x16 is specified for 75W and you can find pictures of burnt sata and molex plugs which were used for risers
Sata connectors have 3 12V pins for a total of 54 W. The SATA cable and connector are rated to handle 4.5A over each of the 3.3V, 5V and 12V rails for a total of 91.35 W. http://www.playtool.com/pages/psuconnectors/connectors.html#sataPowered risers draw less than 50 W and are perfectly safe to use with a SATA connector. Burnt connectors are from overloading the cable by using more than two connectors per cable strand MAX or using a Sata connector to power a high draw card. You can use an 8-pin splitter cable into dual 6 pins to power the card and riser on the same cable. https://www.amazon.com/Phobya-PCI-E-Splitter-Cable-Sleeving/dp/B00A5TA7NG |
|
|
|
No, it's based on the fact that a $200 RX 470/570 4 GB card has the best cost/hash performance for just about any algorithm
But try FINDING a 470/570 at $200 for the last month. I also disagree with the "just about any algorithm" part of that statement, though they WERE competitive to superior on the algos used in most or all of the LARGE MARKET CAP coins on both a hahs/$ and a hash/watt basis *IF YOU MODDED THE BIOS*, and fairly close even on non-modded cards. I do suspect they'll sell the things without a problem for a while - just based on there not being enough cards to go around at all - but once the "crash" happens, they're going to have to price them quite a bit less than the consumer version to get any sales at all due to the very very low resale value. Actually I bought four new Powercolor Red Dragon RX 570 4 GB cards for $190 each less than two weeks ago. If you buy the cards right and don't give in to the hype, then I don't think that will change. You also keep saying the RX 470/570 is only good for certain algorithms, first only for ETH based coins and "small-market-cap coins" and now you say the opposite LARGE MARKET CAP coins. Modding the memory timings only affects memory based algorithms. For others like Equihash it has almost no effect. How about you just name an algorithm and a card that supposedly does better than a RX 470/470 4GB card on a cost/hash basis? RX cards are going to be obsolete for mining Ether in a few months...increase in DAG size. That is nonsense based on Claymore future epoch benchmarks, which are only an estimate and in my own testing I found to be overstated. It also assumes that there is nothing AMD can do from a driver or Bios update issue to correct the problem. From a hardware platform standpoint, I don't think there is anything that keeps AMD from being able to address it. This video does a good job explaining why the Dag epoch issue with the RX cards is overstated. Even if the Dag issue is not corrected, there are at least 6 months left where the RX series can mine ETH and dual mine with other coins effectively compared to other alternatives on a cost/hash basis. https://www.youtube.com/watch?v=SQE9WdhupAg |
|
|
|
No, it's based on the fact that a $200 RX 470/570 4 GB card has the best cost/hash performance for just about any algorithm
But try FINDING a 470/570 at $200 for the last month. I also disagree with the "just about any algorithm" part of that statement, though they WERE competitive to superior on the algos used in most or all of the LARGE MARKET CAP coins on both a hahs/$ and a hash/watt basis *IF YOU MODDED THE BIOS*, and fairly close even on non-modded cards. I do suspect they'll sell the things without a problem for a while - just based on there not being enough cards to go around at all - but once the "crash" happens, they're going to have to price them quite a bit less than the consumer version to get any sales at all due to the very very low resale value. Actually I bought four new Powercolor Red Dragon RX 570 4 GB cards for $190 each less than two weeks ago. If you buy the cards right and don't give in to the hype, then I don't think that will change. You also keep saying the RX 470/570 is only good for certain algorithms, first only for ETH based coins and "small-market-cap coins" and now you say the opposite LARGE MARKET CAP coins. Modding the memory timings only affects memory based algorithms. For others like Equihash it has almost no effect. How about you just name an algorithm and a card that supposedly does better than a RX 470/470 4GB card on a cost/hash basis? |
|
|
|
|
Use the forum search box on the upper right. There are at least 10 related threads in the past month.
|
|
|
|
The best risers have a voltage regulator to ensure the voltage to the card doesn't spike. AFAIK, the only risers that have a voltage regulator are the 6-pin risers, which also use four capacitors for a more stable current flow compared to the three used by SATA or 4-pin molex risers. I use a dual PSU setup with all of the risers powered by the same PSU that powers the motherboard. The total draw for the risers and motherboard for five RX 480/580, RX 470 and one Nvidia 1080 is 360 W, so the risers are using between 45 -50 W each for the six card rig.
How is that possible? The RX series have a 8 pin pcie connector which can deliver 150W, how can your risers burn an additional 50W. Are you running the cards at 200W? Btw, if risers would realy burn 50W, that would mean 300W for a 6GPU rig, which would probably burn/kill the wires or even the PSU, as PSU's dont have 300W capability on Sata/Molex rails.... I have a dual PSU setup and use EVGA P2 PSU's that have a single 12V rail and can deliver >99% of the rated power over the 12V line. The best PSU's for multi-psu rigs are dual conversion PSU's that convert the entire energy into a 12V single rail and then uses VRM to generate 5V and 3.3V. The cheaper PSU type converts using different taps in the transformer for each voltage and use the 5V line to regulate the voltage, instead of the 12V line like the dual conversion models. That causes voltage instability from not loading 5V line. The EVGA G2 and P2 series 650W and up are dual conversion models. I use a EVGA P2 1000W with 6 PCI-E connectors for the VGA power inputs only. The total draw for 6 cards is 670 W dual mining. The main PSU that powers the motherboard, risers and SSD is a EVGA P2 750W and uses 360 W. I power the 6 PCI-E risers using four PCI-E connectors and two 6-pin pigtails. |
|
|
|
No data on availability or price available that I've found.
Also, why did they put SLOW MEMORY on a "mining" card?
This thing better be LOW COST if they expect it to sell - though given current "availability" issues, they probably will get away with it even if it's not for a while....
It's based on the RX 470 4GB which has the same memory clock and is already the best cost/hash mining card on the planet. Not at CURRENT pricing on the RX 470 - when you can find them at all - and only ever was true for ETH/ETC/derivatives and possibly a very few "small-market-cap" coins if you limit the list to NEW cards. (My R9 290s cost me $180 from Newegg in a timeframe that the lowest-cost RX 470 there was the SAME price, but outhash pretty much all RX 480s much less RX 470s). Prior to the current "AMD shortage price gouge" the 470 *could* at least argue that point - though a lot of low-cost 480 and 580 cards could argue the point due to higher hashrate and not much higher price. What's odd is that they have this card and the 106 Nvidia varient listed on their "main" website - but they direct US customers by default to www.asus.com/us/ by default (and I have not found a workaround for the main web page to date, just for "further in" ones) which does NOT list these cards at all. The odder part is that it's 470 based instead of 570 based - I wonder if that's why some rumors are talking about "limited run", the GPU chips might be comming from previously rejected chips that were fine EXCEPT for the video-specific part of the chip? No, it's based on the fact that a $200 RX 470/570 4 GB card has the best cost/hash performance for just about any algorithm. The R9 for the same price as a RX 470 is based on that the R9 is a two year old card, uses twice the power and is not as efficient in cost/hash to a RX 470 when the power consumption is factored in. I have all three RX 480/580 8GB and 570 4GB cards and the performance difference for just about any algorithm is not worth the extra ~$50 for the RX 480/580. Nvidia options are nowhere near the RX 470/570 4GB in a cost/hash basis, which why I said the RX 470 4GB is the best mining card for the price on just about any algorithm. |
|
|
|
|
Which version of Windows? Windows 10 should handle up to eight GPU's per driver, so eight AMD cards + eight Nvidia cards.
|
|
|
|
|
You would need to run two instances of the miner with the different wallet addresses and set the cards to use in each with the -di parameter.
|
|
|
|
Hello all,
I'm hoping someone can point me in the right direction. I've built a third mining machine using Windows 10, and currently it has one RX 580 8GB card an an old 7970 3GB card in it. I've installed the latest AMD 17.4.4 drivers, and I copied over my setup from my previous miner including, Claymore miner 9.5 and my batch file from my other miner.
When I run the claymore miner it seems to run OK but I never get the purple line telling me the GPU temperature/fan speed! What am I missing? I don't remember ever having to turn that on before, and I'm using the same batch script I use on my other miner which DOES print out the GPU temperature/fan speed. I copied over the entire folder where I keep the miner and it's config files, and my batch files, etc. so I'm very confused.
Also does anyone know what decred intensity is a good starting point for an RX 580? Thanks!
You need to plug in a monitor or dummy plug for the driver to be able to read and set the temperature and fan speeds. I use -ethi 8 and -dcri 26 dual mining. Thank you! So I'm assuming this is an AMD thing then, yes? Because my Nvidia based rig doesn't have a dummy plug but still prints out the GPU temp/fan speed. Yes, the monitor is required for the AMD drivers. Also a correction, with the latest epoch I get better ETH hash with -dcri 25. |
|
|
|
Hello all,
I'm hoping someone can point me in the right direction. I've built a third mining machine using Windows 10, and currently it has one RX 580 8GB card an an old 7970 3GB card in it. I've installed the latest AMD 17.4.4 drivers, and I copied over my setup from my previous miner including, Claymore miner 9.5 and my batch file from my other miner.
When I run the claymore miner it seems to run OK but I never get the purple line telling me the GPU temperature/fan speed! What am I missing? I don't remember ever having to turn that on before, and I'm using the same batch script I use on my other miner which DOES print out the GPU temperature/fan speed. I copied over the entire folder where I keep the miner and it's config files, and my batch files, etc. so I'm very confused.
Also does anyone know what decred intensity is a good starting point for an RX 580? Thanks!
You need to plug in a monitor or dummy plug for the driver to be able to read and set the temperature and fan speeds. I use -ethi 8 and -dcri 25 dual mining. |
|
|
|
I was having similar issues with Afterburner settings not applying correctly. What fixed it was removing v4.30 and installing the latest Afterburner. Also removing the core and memory clock settings from the miner. I use Afterburner to set the undervolt, core clock, memory clock and fan speeds before starting the miner. https://www.guru3d.com/files-details/msi-afterburner-beta-download.html |
|
|
|
|
The best risers have a voltage regulator to ensure the voltage to the card doesn't spike. AFAIK, the only risers that have a voltage regulator are the 6-pin risers, which also use four capacitors for a more stable current flow compared to the three used by SATA or 4-pin molex risers. I use a dual PSU setup with all of the risers powered by the same PSU that powers the motherboard. The total draw for the risers and motherboard for five RX 480/580, RX 470 and one Nvidia 1080 is 360 W, so the risers are using between 45 -50 W each for the six card rig.
|
|
|
|
Using MSI Afterburner, AMD Wattman, whatever I try, as soon as I apply the new settings they go right back to the previous ones. My -cclock and -mclock parameters in Claymore are having no effect on anything. The weird thing is I have 2 of the exact same card (MSI RX 570 4G) and their settings are different (2000mhz vs 2150mhz memory clock) ... what could have caused this?
How can I regain control over my cards? I have one that is constantly getting incorrect shares so I need to underclock it a bit and I have two or three that could be pushed a little higher.
I've already uninstalled AMD settings so for now it appears Wattman is out of the picture.
Any suggestions? Thanks!
I've had this issue. Using Claymore's to set the clocks is unstable and inconsistent at best. What solved it was removing the -cclock and -mclock parameters from Clayamore, uninstalling Afterburner, reinstalling it and only using Afterburner to undervolt and set the core and memory clocks. |
|
|
|
|
Show Posts
