Why is it that all the phishing/foreign sites always fuck up the punctuation and spaces? It's always like "word , word" or "word,word" when proper use is "word, word"
|
|
|
fine i'll drop it down to 2.05! NEW PRICE!! 2.05BTC!
Just way the equivalent in BTC you want for the cards. BTC prices are so volatile right now that setting a certain BTC price is misleading. How much in dollars do you want for each card (paid in equivalent amount of BTC)
|
|
|
I'm pretty lazy at times, especially when it comes to monotonous repetition. I'm trying to find the optimum memory frequency for a radeon 5870 at certain core speeds, but it's an AWFUL lot of trial and error, and it isn't really instant to test each combination, plus is can be kind of cumbersome. Sooo.. Hoping maybe someone can help me out. I don't really know what to set for a price, because I'm not sure how complicated this will be. Basically, it's kind of like a step 1 > step 2 > step 3 > if this then do this, else do that > step 1. I am using Windows 7 x64 Ultimate, and really quite noobish when it comes to coding.. I don't know any languages except a little bit of LUA (and I do mean little). I want to open my phoenix.exe miner, and after 40 seconds, start recording the hashrate underlined in red here. For the next 20 times it changes (or every 2 seconds for the next 40 seconds, whichever is easier), record the hashrate again. After 20 changes (or 40 seconds, depending on whatever you picked), it would find the average of all these values and save it somewhere, and close phoenix.exe. After 15 seconds, it would need to increase the memory speed by 1 MHz, UNLESS the memory speed is at 210, then instead increase core speed by 5 and reduce memory speed to 140, as well as probably start a new line in wherever this automation is saving the results (Don't care if you use your own proprietary method, or use glakkeclock, or cgminer, or whatever, as long as it WORKS). After this step, wait 5 seconds. Then it would go to back to opening phoenix.exe and repeating the whole process again until core speed is 1025, then it would stop. Of course, all these numbers could possibly be changed to variables, depending if it would be easier for you (or maybe me if I try to use this on other devices) Some source codes that you may need/find useful. Phoenix 2 source code: https://github.com/phoenix2/phoenix Python Glakkeclock source code: https://github.com/Glakke/glakkeclock C++ CGMiner cource code: https://github.com/ckolivas/cgminer/ C
|
|
|
You can keep the newer SDK... if you run diablo.... a 5850 @ 940mhz get's me 380mhash/s
You should be getting a hair above 390 mhash with proper sdk/kernel/settings. 380 is slow Picture below is a 3.7% increase in speed.
|
|
|
Wow, he got each 7970 for less than $320 each.. not counting supporting hardware. I guess lowballers DO get lucky once in a while.
|
|
|
I can help. First step is you sell the power supplies and video cards to me. After you do that, we can go to the next step, which I'll explain after I receive them in the mail.
|
|
|
EVGA nForce 790i SLI Ultra board, Intel Celeron 420, 2x2GB RAM, PC Power & Cooling 1KW-SR, 3 reference 5970, 2 reference 5870. Uses about 1040W AC power. PSU Efficiency is pretty low at around 83%, so about 2.61 MHash/Joule. Win7 x64, Driver 11.12, SDK 2.1. BIOS flashed on each card except 7/8 (new 5970, still breaking it in finding max clocks before I flash). Memory @ 150-160, core speed is specified in the miner window, all voltage is 0.95v except where denoted with a plus (+) sign after the core speed, then it is 1v. I use Phoenix 2 miner, and these arguments are constant between all GPU's. kernel = phatk2, AGGRESSION = 14, VECTORS = true, BFI_INT = true, WORKSIZE = 128. The numbers in the tray in the lower right is the amperage on each volterra programmable voltage regulator. Red is GPU1, orange is GPU2, etc. It helps me notice easier if a GPU is "sick" and adjust frequency to be more stable. And no, I won't use CGMiner. What is the benefit of flashing the card BIOS, do you mean firmware? how many mhash per 5970 and have you tried the memory clock at 1/3 of clock? that increased mhash significantly for me Please make a separate thread if you need anymore info. It's getting offtopic; I just wanted to make a thread about the MSI Z77A-G45 motherboard.
|
|
|
EVGA nForce 790i SLI Ultra board, Intel Celeron 420, 2x2GB RAM, PC Power & Cooling 1KW-SR, 3 reference 5970, 2 reference 5870. Uses about 1040W AC power. PSU Efficiency is pretty low at around 83%, so about 2.61 MHash/Joule. Win7 x64, Driver 11.12, SDK 2.1. BIOS flashed on each card except 7/8 (new 5970, still breaking it in finding max clocks before I flash). Memory @ 150-160, core speed is specified in the miner window, all voltage is 0.95v except where denoted with a plus (+) sign after the core speed, then it is 1v. I use Phoenix 2 miner, and these arguments are constant between all GPU's. kernel = phatk2, AGGRESSION = 14, VECTORS = true, BFI_INT = true, WORKSIZE = 128. The numbers in the tray in the lower right is the amperage on each volterra programmable voltage regulator. Red is GPU1, orange is GPU2, etc. It helps me notice easier if a GPU is "sick" and adjust frequency to be more stable. And no, I won't use CGMiner.
|
|
|
Using single core GPU's right? I've heard it's not possible to have over 7 cores mining at the same time
I'm mining on 8 GPU cores on my EVGA nForce 790i SLI Ultra board right now. 3 PCI-E 16x, 2 PCI-E 1x, all using 1x bandwidth. For the MSI Z77A-G45 I will probably use 1 5970, and the rest will be single core; a mix of 5870 and 5850.
|
|
|
Your problem may be one card doesn't like the ATI driver it has to use. You may get better results with two catalyst drivers, if possible. I believe 5000 series cards prefer ATI Stream 2.6 or less
ATI Stream only goes up to version 2.3. Users of both 2.2 and 2.3 complained about CPU usage bug though. ATI Stream 2.1 is recommended for all 5xxx and other VLIW5 hardware (Up to 6870). AMD APP 2.4 or 2.5 is the lowest compatible SDK for 6930/6950/6970/6990 (VLIW4 hardware), and is slightly slower on lower version cards. AMD APP 2.6/2.7/higher is compatible with 7xxx cards, but takes a big speed hit on 6xxx/5xxx cards. Also, ATI Stream counts as a different OpenCL platform compared to AMD APP. You can have both installed. This means you can mix VLIW5 hardware with VLIW4 or GCN (7xxx series) and not take a hit on speed as long as you configure your miners correctly, but you can't effectively mix VLIW4 and GCN together without having subpar performance on the older hardware, or being entirely incompatible with the newer hardware because AMD APP 2.4/2.5/2.6/2.7 all share the same install path.
|
|
|
Do you have a CPU + heatsink for the mobo? I might be interested then.
Yes, it can include CPU w retail box hsf.. its a sarges 145 single core 45 watt, we have some 3 & 4 core CPUs as well. +$30 for CPU & hsf. I'll leave a 1 GB stick of ram on it too. I broke down and just used my credit account on newegg to get a new MSI Z77A-GD45 + Intel Celeron G440 single core 35W CPU and use some of my existing ramsticks I have laying around. But the info you gave is good for other buyers that may be interested.
|
|
|
Edit: Please stop PMing me for help on how to GPU mine. I will not be able to help you. I stopped GPU mining almost a year ago. Further PMs requiring help for GPU mining will be ignored.So I bit the bullet after having some frustrating problems with one of my miners (only 4 pci-e slots and I think my one pci > pci-e slot adapter is dying, having stability problems) and bought a MSI Z77A-G45 motherboard. I plan on using at LEAST 6 slots, hopefully all 7. I'll be using 1x > 1x risers with al lthe cards and short PCI-E presence pins if necessary. Primary reason why I'm making this thread is I want to document if it is possible for ALL SEVEN slots to be populated and usable at the same time. I haven't encountered any boards yet that completely shut off adjacent PCI-E slots yet with 1x width bandwidth per slot, and I'm hoping it'll be the same for this board too. I'm hoping restricting bandwidth to 1x per slot will keep other slots "on". I -think- my EVGA nForce 790i SLI Ultra board shuts off adjacent slots with larger PCI-E bandwidths (4x/8x/16x), but because I used 1x width risers, it didn't need to shut off any slots. Anyways, in case anyone is wondering what kind of board I'm looking at and doesn't want to click a link, here ya go.
|
|
|
Do you have a CPU + heatsink for the mobo? I might be interested then.
|
|
|
I just recently encountered a similar problem yesterday. I have too many video cards and had to opt for putting one in my main PC for mining. I switched out a 5870 reference with a 5850 reference and moved the 5870 down a few slots, but doing this caused only max resolution 1024x768, no 3d acceleration, memory/core clocks in trixx were 0, temp was -1, obviously everything was fubar. My 5870 wasn't being detected either. In the end, I uninstalled all AMD drivers/SDK, rebooted, used driver sweeper to remove ALL radeon drivers/sdk remnants, reboot, installed 12.1 driver + 2.6 SDK, reboot, installed separate 2.1 SDK (because I like using 2.1 and if dont have 2.4+ installed, my miners dont work), reboot, and voila. All sensors worked and all gpus were detected and installed right.
|
|
|
You don't ship to Canada? Damn!
Not my products. You're free to ask him to ship to canada though.
|
|
|
I probably really should get a per-GPU power draw.. I did one for a 5970 and it was something like 220W full load for both GPU's compared to not being connected or powered at all, but it was also undervolted.
You should have more than enough room on your current PSU. Right now I'm powering 8 (yup, driver limit) Cypress XT (5870/5970) GPU's off of a 1000W PC Power & Cooling 1KW-SR. Somehow, overcurrent protection hasn't kicked in.. I'm drawing 1050W AC and the PSU is about 83% efficient.
|
|
|
Pull requests for cpuminer are still welcomes.
cpuminer is definitely used on testnet and various other alt-coin scenarios.
And botnets
|
|
|
I heard there was free stuff. I want free stuff!
sign me up
|
|
|
|