Be carefull, dual-x don't have vrm temperature sensors, and these get hot. Check the back of the card, if you start seeing some white traces over the pcb, specially near the psu power connection and the caps, it means vrms are overheating. At stock voltage vrms when gpu is 80c should be around 120 C. It looks like you are undervolting in gpu-z, if so you shouldn't have a problem at 80 C. I wouldn't go much further though with dual-x. Stock voltage at 85 C for a long time may fry some caps near the vrm.
I'm using them in summertime atm, ambient temp 35-45 C, open case 80mm apart from each other and 80mm coolers in between them to renew the air. They are all at 85% fan. They never go over 75 C this way, on cool days they are at 65 C. Perhaps you should work a bit on your frame to get it to run cooler.
As for the warranty, if you fry it due to vrm overheating they should replace it. Don't say you used them for mining, tell them you where doing render and after some hours of processing the card blew.
I didnt change anything of voltage settings.Its at stock. My fans are at 75% now.You are running at 85,is it safe? Also I am getting a good usb fan in a few days. I haven't had a fan fail on me yet, and got 5 dual-x running at that speed for almost two months. To undervolt download sapphire trixx, tick force constant voltage in the configuration and drop it to 1175. That is a safe voltage and that plus the extra 10% fan will make them run a few C's lower. If you want to undervolt further do it at small steps and test for stability each time. My 7970 DualX works at core/memory =1000/1500MHz, with core/memory = 1.075/1.5V, the temperature is 70C. I use MSI afterburner. My DualX has VRM monitor. |
|
|
|
How to link a cryptocurrency to a physical or financial asset without centralized control? Decentralization is one of the main features of bitcoin. If people want a centrally-controlled digital currency, they can continue to use digital dollars, euros, etc. Government fiat currencies are not linked to any physical or financial assets either. It seems to me that that doesn't really matter.
Bitcoin the currency is linked to bitcoin the network, and the capabilities of that network are very useful and much better than anything offered by the traditional banking system. That to me seems like more than enough "intrinsic value" to sustain bitcoin.
I agree. Bitcoin is linked to the whole network, internet. So that is the asset backing BTC. |
|
|
|
You have to connect ALL the riser molex to the main PSU.
# 1 - Corsair CX750
* Motherboard 100W
* 1 GPU = 1 * 75 = 75W
* 6 PCIE slot (4+2)*75W = 450W
* Fans 10W
(total = 100+75+450W +10W = 635W 85% of 750W)
# 2 - Fractal R2 750W
* 5 GPU = 5 * 75W = 375W
* (5x 6pin + 0molex)
(total about 50% of 750W)
Like this? The only problem i have then is regarding PSU #2, the Fractal R2. It has 2 rails and limited connectors, im thinking like this: Rail #1 - 30amp*12V=360W 1x pci-e (1*75W) 6x molex -> 3x pci-e 6pin (3*75W=225W) = 300W for 4pcs 6pin Rail #2 - 30amp*12V=360W 1x pci-e (1*75W) -> 2 x 6pin (will make this converter myself) OR1x EA12V -> 6pin? Your picture is right indication of the set up. For fractal R2, you can use the 4*6pin from Rail1 and one from the EA12V->6pin. I do not know the distribution of power between rails. I just assume Rail 1 is the 4*(6+2)pin+peripheral, and Rail 2 is the MB connector +CPU12V. It is better to use two CTX750 if you have not bought the Fractal. |
|
|
|
I have almost the same problem, running 6pcs 270 cards with 2PSU's https://bitcointalk.org/index.php?topic=457287.msg5043351#msg5043351Problem is that my 270's only have one 6pin-connector on the pcb, this allows 75W. The 270 needs about 150W each so the other 75W needed has to come from motherboard/powered riser. That means in my case 6 x 75W = 450W from molex on the risers, and most PSU's only have the molex/sata connectors on 1 rail, and very few PUS's allow 450W on 1 rail. Therefore i need to split the 450W load on 2 PSU, but even 225W is not possible on my PSU (molex on 1 rail, allowing 18Amps, 12V*18Amp = 216W, wich means im 9W short...?) I belive i need to buy a new PSU with a singel 12V rail, like the Corsair CX750, 62AMP on 12V = 744W on 1 rail. The solution is to buy a single rail PSU. You can also undervolt. Yes i belive this is my solution? I will buy a CX750 today. 6 x R9 270 # 1 - Corsair CX750 (1x rail 12V = 744W) # 2 - Fractal R2 750W (2x Rail 12V * 30amp = 360W * 2 = 720W) # 1 - Corsair CX750 * Motherboard 150W * 2 GPU = 4 * 75 = 300W * (2x + 6pin 2molex) * Fans 10W (total about 62% of 750W) # 2 - Fractal R2 750W * 4 GPU = 8 * 75W = 600W * (4x 6pin + 4molex) (total about 80% of 750W) You have to connect ALL the riser molex to the main PSU. # 1 - Corsair CX750 * Motherboard 100W * 1 GPU = 1 * 75 = 75W * 6 PCIE slot (4+2)*75W = 450W * Fans 10W (total = 100+75+450W +10W = 635W 85% of 750W) # 2 - Fractal R2 750W * 5 GPU = 5 * 75W = 375W * (5x 6pin + 0molex) (total about 50% of 750W) |
|
|
|
What about only 3 cards of 6 being connected woth powered molex, is it OK or I need to connect all 6? AFAIK 3 GPUs can sometimes run without powered molex, so is it OK?
If you have enough molex connectors from the PSU, then connect all 6. It is OK to connect only 3. |
|
|
|
I have almost the same problem, running 6pcs 270 cards with 2PSU's https://bitcointalk.org/index.php?topic=457287.msg5043351#msg5043351Problem is that my 270's only have one 6pin-connector on the pcb, this allows 75W. The 270 needs about 150W each so the other 75W needed has to come from motherboard/powered riser. That means in my case 6 x 75W = 450W from molex on the risers, and most PSU's only have the molex/sata connectors on 1 rail, and very few PUS's allow 450W on 1 rail. Therefore i need to split the 450W load on 2 PSU, but even 225W is not possible on my PSU (molex on 1 rail, allowing 18Amps, 12V*18Amp = 216W, wich means im 9W short...?) I belive i need to buy a new PSU with a singel 12V rail, like the Corsair CX750, 62AMP on 12V = 744W on 1 rail. The solution is to buy a single rail PSU. You can also undervolt. |
|
|
|
|
You are right. However you can connect ALL the molex to the main PSU. Never connect the molex to 2nd PSU (if they are not cut).
|
|
|
|
|
You can use MSI afterburner beta to reduce the voltage to ~1.01v. I guess you are running at core/memory 950/1000MHz to achieve 650kH/s. So ~1.01v should be enough.
You can reduce the power/heat so that the other two cards are not throttled. When the core/VRM temperature is too high, the card will reduce the frequency or GPU usage automatically.
|
|
|
|
So I tried middlecoin for 18 hours with 6.3Mh/s and got a pay out of .075... I take it this is higher than normal??
It is, if you only mine for 18 hours and there are some unexchanged coins. I have 4.8Mh/s, mined for many days, today's payout is 0.062. |
|
|
|
A question to all 290X owners here.
Why did you buy theese gpu`s? I can not work out how they can beat a 280X pricewice.
You're right. 1x 290X = 500 EUR 1x 280X = 250 EUR, so 2x280X=500 EUR The final cost is the same. with 2x280X will take more speed than 1x290X true but if you look at it like this. 4xamd 290x = +/- 1800 euro = 3600 - 3700 khash 4xamd 280x = +/- 1040 euro = 2800 - 3000 khash so now thats a diff of about 800 euro's but if you wat to achieve the same hashrate at the 290x you need to buy another mobo, cpu, memory, storage, powersupply and finally a 280x and running 2 computers will have a bigger electricity bill. so in the end you will pay more for 280x to achieve the same khashrate. but thats only if your planning to build a strong rig etc I have to agree with this analysis. For people who just want to build 1 or 2 rigs why should they go for 280X. Yes upfront cost is more but that doesn't matter in longer term. I have 4* XFX 290s set up mining away ( 880 for 3 and 900 for 1) so around 3.5Mhz with one Rig. I got them while they were 440$ at amazon. Unfortunately I cannot find them anymore  at that price and most of them are on back order. I can easily make .08 BTC on that rig / day without lifting a finger.. Now If I build a rig from 280X I can get a max of 2.7Mz I lose around .02BTC each day Cost wise 280X are cheaper by 100$ at max So I spend .5BTC extra on the rig which can be recovered in 1.5 Month ( may be 2 month if earnings keep on falling ) Is there any flaw with this reasoning ? 5X280X have similar hash rate to 4X290X. There are many motherboards supporting 6 or 7 cards. You also need to factor in the cost of electricity. |
|
|
|
h2o is a bad pool operator, very amateur and is stealing. When will you people wakeup? Sounds like you're starting to
Oh i am new here buy i heard this guy is a famous troll  I put him on ignore. Life is much easier after that. |
|
|
|
I am getting alot more rejects then i used to get on hascows.
From 0.17% a few weeks ago, to rejects 3%-4% the last 1,5 weeks.
What happened?
We have higher rejection rate on digitalcoin due to its fast block time. MEOW had 30 seconds block time, so rejection rate was also high. However, MEOW changed its block time to 60 seconds recently. The rejection rate reduces. Other pools use lower difficulty level, so the rejection could be low. |
|
|
|
|
Which Catalyst version do you use to get 6 cards working?
|
|
|
|
the more shares you the faster is the roi
sry but thats wrong. thats completly independent of how many shares you buy. and 200 mil dollar profit(*) from selling products .that will not be done from today to next day.. if the yeven ever archive that hugh profit edit: just read the webpage again. I might have miss read it a bit and they didn't mean to share 5% of the profit only but all the profit and 5% is their calculated profit. This way the numbers looks better (replace the word profit(*) with income) although I very much doubt the 5% profit only. I think your original interpretation was right. If they generate $200M INCOME (revenue) for hardware sale, 5% of the income will be distributed to HW shareholders. That is $10M. Divide that with 0.2M shares, that is $50 per share. I doubt they can sell $200M without causing huge difficulty rise. If we approximate that each chip as 1MH/s, the price is for $500/MH/s, they will sell 400 k chips. That is 400k *1MH/s, or 400GH/s. The net work hash rate is less than 150GH/s at the moment. The difficulty will at least double. There are other ASIC manufacturers as well. |
|
|
|
|
If it hits the market next month, I am happy to pay $0.3/kH/s if I want to have the ROI in one year and have the same ROI as GPU as GPU has resell value. That is because difficulty will sky rocket.
|
|
|
|
|
For the core frequency you are using, there is little difference in performance if you use 1000,1200 or 1500MHz memory frequency. You may save more energy by using 1000MHz.
|
|
|
|
|
Can you change the memory frequency to 1000, 1200 or 1250 MHz? That will reduce the power consumption as well.
For some of my set up, I use 1.0V, core/memory = 950/1000MHz. The power consumption is less than 400W per 7990. The hash rate is 660kH/s for each GPU.
For another setup with 4X7990, I use 0.95V, 850/800MHz. The power consumption for the whole system is 1400W. The hash rate is 4.85MH/s.
|
|
|
|
any advice how to get better numbers? http://www.middlecoin.com/reports/1NZ3JiWLfvwi7J8AFaYZqYQHGSU7wRB9UW.htmlI'm having 5% stales but i see enough people don't having this problem. Running 4x 7950 cards My config = "api-allow" : "W:127.0.0.1", "api-listen" : true, "api-mcast-port" : "4028", "api-port" : "4028", "expiry" : "120", "hotplug" : "5", "kernel-path" : "/usr/local/bin", "log" : "5", "queue" : "1", "scan-time" : "30", "scrypt" : true, "shares" : "0", "gpu-threads" : "1", "gpu-dyninterval" : "7", "gpu-engine" : "1100,1100,1100,1050", "gpu-fan" : "60-85,60-85,60-85,60-85", "gpu-platform" : "0", "gpu-memclock" : "1500,1500,1500,1450", "gpu-memdiff" : "0", "gpu-powertune" : "20,20,20,20", "gpu-vddc" : "1.081", "intensity" : "19,20,20,19", "temp-target" : "78", "temp-overheat" : "83", "temp-cutoff" : "86", "temp-hysteresis" : "3", "vectors" : "1", "worksize" : "256", "lookup-gap" : "2", "shaders" : "0", "thread-concurrency" : "21712", "no-pool-disable" : true Reduce the intensity to 18 and increase TC to 24000. |
|
|
|
|
For the 1000W power supply, it is under strain to supply power to 2X7990. So the efficiency is low, more heat is generated inside the PSU. So the whole system is hotter.
If you take out the PSU from the case, then temperature of the card should drop.
I think you are not using a "Gold" efficiency rated PSU.
If you undervoltage the 7990, power requirement should be lower. Using MSI afterburner Beta 18 or later to reduce the voltage from 1.2V to 1.05V will lower the wattage by 20%.
|
|
|
|
|
I use the GA-Z68AP-D3. It can run 4X7990 or 3X7990+1X7970. If I use 3X7990+2X7970, one of the GPU can be detected by GPUZ and MSI afterburner. But the frequency shows 0MHz for both core and memory.
I need to jump A1B17pin of PCIEX1_2 and 3. The first PCIEX1 slot does not need shorting.
I use powered riser just in case.
|
|
|
|
|
Show Posts
