I'm getting "Failed to connect, retrying..." errors. I have python-twisted installed and my network connection is fine. Any ideas?
Turns out this was only happening with slushs pool, I switched to deepbit and it is working fine. I'm actually having the same problem on one of my rigs. For some reason it will not connect to slush's pool (tried multiple login credentials), but will connect to bitcoinpool and deepbit just fine.
|
|
|
I am completely amazed at how much more power the phoenix miner has been able to unlock in these cards. My rig has gained over 50 mHash/sec in the past 24 hours as new revisions keep rolling out. 380 was pretty much never mentioned on the forums before my post, and now we're seeing 430-440 being very reachable by most 5870 cards with Phoenix.
|
|
|
Another amazing release, went from 421-422 mHash/sec to 423-424 mHash/sec on my 5870s (Aggression 13, Vectors, 128 Worksize, BFI_INT, using 975/300 @ 0.950v).
Just a suggestion, update your original post with the changelog/updates so we can find out what's new without searching through the various replies within the thread.
|
|
|
GPU 1: GigaByte ATI Radeon HD 5870 ($220) GPU 2: MSI Lightning ATI Radeon HD 5870 ($222)
Just hunt on NewEgg & Amazon for 5870s once or twice a day. They tend to pop up cheap (< $230) and sell out within a few hours.
What is the difference between all of the different 5870s? Are they all about the same? I'll be able to tell you on Wednesday how the MSI Lightning compares to the Gigabyte. My main 2 rigs both use GigaByte 5870s. The rig I linked is about to be hooked up, waiting on the power supply to show up. I also just bought another MSI Lightning on NewEgg this morning (bought 1 and they were out of stock before I could buy a 2nd for a complete rig). My understanding is that all the 5870s are BASICALLY the same performance. Slight variations, mostly due to temperatures you can run them out with the slight changes in how each the fans/heat sinks are set up between brands. I do know that you want 1 gig cards over 2 gig cards due to the savings in power consumption.
|
|
|
I'm hoping to push my cards back to 1020/300 once I finish reorganizing my house to accommodate the rigs and can spend some time hooked up to a Kill-A-Watt to find the sweet spot for mHash/watt.
|
|
|
I just switched over to the Phoenix miner 1.1 as well, and now run at 421-423 mHash/sec on all of my 5870s. Clock settings are 975/300, using the 0.950 voltage. They're running cooler, sucking less power, and running significantly faster with Phoenix!
|
|
|
Absolutely amazing work. With the new BFI_INT flag, my 5870s are churning out work at 420-421 mHash/sec each!
|
|
|
1.01 fixed performance issues thanks to adding worksize. It's now performing on-par (hard to say if more/less) with poclbm, with the exception of memory underclocking! Just like bolapara, I was able to drop memory clocks from 975/600 to 975/300 without hurting my MH/s!
UPDATE: Gains of about 3 mHash/sec per 5870 while being able to drop memory from 600 to 300!
|
|
|
79.5 cents/mHash on that rig, and I think you're going to run into big issues cooling 5 GPUs inside that tiny case with poor air flow both from case design and cramming cards in.
My latest rig ended up costing $811 after tax/shipping, 2x 5870s, and the final cost was 93 cents/mHash capacity, while using SIGNIFICANTLY less wattage than what 5x 5770s will run.
Case: HAF 922 ($100) PSU: Cooler Master 750W ($98) Motherboard: MSI AM3 ($70) Processor: Sempron 140 ($38) RAM: 1 gig Kingston ($13) Hard Drive: Kingston 16 GB SSDNow ($50) GPU 1: GigaByte ATI Radeon HD 5870 ($220) GPU 2: MSI Lightning ATI Radeon HD 5870 ($222)
Just hunt on NewEgg & Amazon for 5870s once or twice a day. They tend to pop up cheap (< $230) and sell out within a few hours.
The case is a bit overkill, but I like using cases that are designed for maximum airflow to keep the cards running cooler at higher clock speeds, leading to better mHash/watt ratio, both in terms of power consumption from the GPUs, and the amount of heat that will have to be offset by my air conditioner.
|
|
|
In that case it's slightly slower (~2 mHash/sec) using Aggression 12. It's slightly faster than base poclbm, but slower than the modified poclbm with the H.x, H.y, and H==0 modification.
|
|
|
Comparable to poclbm.py in Linux with my 5870s (using the H.x, H.y, and H == 0 mod) when using aggression 12, compared to poclbm.py with -f1.
Are you using 1024 kHash = 1 mHash, or 1000?
|
|
|
I've got my boxes set with a GUI auto log-in install to make sure all the displays are initialized at the start, but they're all headless. I log into them via SSH and use the following commands to get one SSH window to show multiple miners: screen -S GPU1 ./poclbm.py -d1 (I've modified the python script to auto fill in log in/worksize/vector arguments based on device #) Ctrl a+d screen -S GPU2 ./poclbm.py -d2 Ctrl a+d screen -S Multimon screen -x GPU1 Ctrl+a, Shift+S Ctrl+a, Ctrl+tab Ctrl+a, Ctrl+c screen -x GPU2 The results can be seen here: https://i.imgur.com/b1zg4.gif
|
|
|
That's all you have to do. Open it in notepad/wordpad, change the lines in BitcoinMiner.cl, and see if your speed goes up or down. At this point nobody has been able to pinpoint exactly why it is slower for some and faster for others.
|
|
|
My results (+7 mHash/sec per GPU) were from:
Ubuntu 10.10 (x86_64) ATIStream SDK 2.1 fglrx driver 11.3 HD 5870s
|
|
|
I made the modification and my 5870s went from 363 mHash/sec and 361 mHash/sec to 370 even on both. Ubuntu 10.10.
|
|
|
I'd actually recommend a 16 gig SSD (Kingston). I've put them in my mining rigs, they're only $50 (no tax/shipping) on Amazon. Assuming you're building these as linux boxes, it's more than enough space, almost no power consumption, and the machine goes from off to the Ubuntu desktop in about 8 seconds (less if you use QuickBoot in the BIOS).
|
|
|
Hi slush, Any plans to add some extra data to the JSON API, particularly as it relates to individual worker stats (shares completed, score, last share received, and found blocks)?
I'm working on building an overview page where I can estimate the BTC (and USD at current rate) each rig has generated, offset by cost of electricity and build cost so I can track how long til my multiple rigs pay themselves off. Those 4 stats for each worker added to the API would make it completely automated, and the information is already being tracked on the profile page.
|
|
|
Specs other than GPU would certainly help justify the price. The base system seems far too highly marked up. The price increase between the single system and the dual system is a bit more reasonable ($900 higher than base system for a $720 GPU and a bigger power supply (1200 watt?)).
But $1,800 for 600 mHash/sec? I just built a 740 mHash/sec rig for only $900!
|
|
|
Well, NewEgg just ran out of the 5870 cards I had been buying for $219. I was wondering if anybody had some performance data/comparisons between a 6950 and 5870 card, both in terms of raw mHash/sec, and also mHash/watt. I've found that XFX 6950 cards could be bought for $3 more than I was paying for the 5870s, due to the lack of sales tax from the different vendor.
|
|
|
Have you considered the additional air conditioning load during summer?
My rigs are actually producing very little radiant heat, and I already had the AC running 24/7 last summer, so it's not going to result in any extra hours of AC use.
|
|
|
|