What Software are GPU miners using in Windows nowadays?

[Fuzzy]

I got into mining just under a year ago and have stuck with GUIminer since then in a "ain't broken don't fix" bind. Downtime is money as it goes.

Anything better worth the time to switch to? I'm running win7 and not looking to switch OS, just the software miner.

[rjk]

I got into mining just under a year ago and have stuck with GUIminer since then in a "ain't broken don't fix" bind. Downtime is money as it goes.

Anything better worth the time to switch to? I'm running win7 and not looking to switch OS, just the software miner.

cgminer is the de facto miner for larger installations. It gives you everything in one place, including failover pool support, monitoring, API control, overclocking, and more.

[Mabsark]

I'm a guiminer v2011-07-01 user and for a very good reason.



I use a 6870 and get around 375 MH/s with guiminer compared to around 315 MH/s with cgminer. I've tried all the miners with pretty much every config, and none of them even come close to guiminer/phoenix/phatk2. My hashrate is better than that when doing multiple thing on my desktop including watching video.



I've recently tried the new version of guiminer and noticed that it crippled my internet speed so I went back to v2011-07-01.

[ssateneth]

You'll find a large population uses cgminer, some use DiabloMiner, and not so big population using Phoenix (with or without GUIMiner frontend). Personally I use Phoenix 2, because I get ~2% more than any other miner I've tried for my 57xx/58xx cards.

[grue]

You'll find a large population uses cgminer, some use DiabloMiner, and not so big population using Phoenix (with or without GUIMiner frontend). Personally I use Phoenix 2, because I get ~2% more than any other miner I've tried for my 57xx/58xx cards.

2% is a pretty low margin. are you sure it's not just statistical noise?

[MrTeal]

I'm a guiminer v2011-07-01 user and for a very good reason.

I use a 6870 and get around 375 MH/s with guiminer compared to around 315 MH/s with cgminer. I've tried all the miners with pretty much every config, and none of them even come close to guiminer/phoenix/phatk2. My hashrate is better than that when doing multiple thing on my desktop including watching video.

What SDK and clocks/voltage are you running to get 375MH/s out of a 6870?

[imsaguy]

watching.

[mdude77]

I'm a guiminer v2011-07-01 user and for a very good reason.

I use a 6870 and get around 375 MH/s with guiminer compared to around 315 MH/s with cgminer. I've tried all the miners with pretty much every config, and none of them even come close to guiminer/phoenix/phatk2. My hashrate is better than that when doing multiple thing on my desktop including watching video.

What SDK and clocks/voltage are you running to get 375MH/s out of a 6870?

I got 380MH/s out of my 6870 through poclbm via guiminer in windows 7, SDK 2.1, stock voltage and clock (900 I believe), memory at 300.

[MrTeal]

That almost sounds to me like a bug in the way the hashrate is being reported by GUIminer. There might be improvements to be made with different combinations, but 380@900MHz is more inline with what you'd get out of a 6950 with 25% more shaders than a 6870.

[mdude77]

That almost sounds to me like a bug in the way the hashrate is being reported by GUIminer. There might be improvements to be made with different combinations, but 380@900MHz is more inline with what you'd get out of a 6950 with 25% more shaders than a 6870.

Whoa, looking at this in retrospect, I get 280, not 380, with my 6870.

My fault.

[Mabsark]

What SDK and clocks/voltage are you running to get 375MH/s out of a 6870?

They're in the top right corner of the pic.

• Catalyst 11.9
• 1 GHz core
• 200 MHz mem
• 1.187 V
• 40% fan
• Extra flags: -k phatk2 platform=0 device=0 VECTORS BFI_INT AGGRESSION=6 worksize=128 FASTLOOP=false

The key is to run 2 workers on the one card. With a single worker, even at 13 aggression, the result is similar to the other miners. If you guys get a decent boost in hashrate from this, don't be shy with the donations now, lol.

[ssateneth]

You'll find a large population uses cgminer, some use DiabloMiner, and not so big population using Phoenix (with or without GUIMiner frontend). Personally I use Phoenix 2, because I get ~2% more than any other miner I've tried for my 57xx/58xx cards.

2% is a pretty low margin. are you sure it's not just statistical noise?

My 5870 @ 1035/355 gets 481.84 ±0.17 mhash/sec. It's extremely consistent on dedicated cards that don't have a display assigned to them. I tried Diablominer and got about 473 ± 5 mhash with all sorts of different worksizes/arrays/f numbers/vectors. It's just too inconsistent and low compared to good ol phoenix. The author of course tries to come up with an excuse that phoenix must be inflating mhash somehow, but he's really egotistical and full of himself, so whatever

[ssateneth]

What SDK and clocks/voltage are you running to get 375MH/s out of a 6870?

They're in the top right corner of the pic.

• Catalyst 11.9
• 1 GHz core
• 200 MHz mem
• 1.187 V
• 40% fan
• Extra flags: -k phatk2 platform=0 device=0 VECTORS BFI_INT AGGRESSION=6 worksize=128 FASTLOOP=false

The key is to run 2 workers on the one card. With a single worker, even at 13 aggression, the result is similar to the other miners. If you guys get a decent boost in hashrate from this, don't be shy with the donations now, lol.

Your mhash might hugely be higher, but are your shares/minute higher too? Shares is what counts. Also I realize this might be hypocritical to my above post

[Mabsark]

Your mhash might hugely be higher, but are your shares/minute higher too? Shares is what counts. Also I realize this might be hypocritical to my above post

I'd never considered it before. So, I just stopped everything running on my PC and restarted guiminer to wipe the stats. After 15 minutes, one worker had 43 shares, the other 38.

(43 + 38) / 15 = 5.4 shares per minute.

[mdude77]

I have four 5870s running at 900/300 with phoenix, and I average 425 among them.  Your clock is quite a bit faster, I'd expect more than 481 at 1035.

You'll find a large population uses cgminer, some use DiabloMiner, and not so big population using Phoenix (with or without GUIMiner frontend). Personally I use Phoenix 2, because I get ~2% more than any other miner I've tried for my 57xx/58xx cards.

2% is a pretty low margin. are you sure it's not just statistical noise?

My 5870 @ 1035/355 gets 481.84 ±0.17 mhash/sec. It's extremely consistent on dedicated cards that don't have a display assigned to them. I tried Diablominer and got about 473 ± 5 mhash with all sorts of different worksizes/arrays/f numbers/vectors. It's just too inconsistent and low compared to good ol phoenix. The author of course tries to come up with an excuse that phoenix must be inflating mhash somehow, but he's really egotistical and full of himself, so whatever

[MrTeal]

Your mhash might hugely be higher, but are your shares/minute higher too? Shares is what counts. Also I realize this might be hypocritical to my above post

I'd never considered it before. So, I just stopped everything running on my PC and restarted guiminer to wipe the stats. After 15 minutes, one worker had 43 shares, the other 38.

(43 + 38) / 15 = 5.4 shares per minute.

There's variability in the shares, 15 minutes probably isn't nearly enough to get a good idea of how much you're hashing. Try running for a 24 hour period and see what happens, then try again with a single instance.

[MrTeal]

Your mhash might hugely be higher, but are your shares/minute higher too? Shares is what counts. Also I realize this might be hypocritical to my above post

I'd never considered it before. So, I just stopped everything running on my PC and restarted guiminer to wipe the stats. After 15 minutes, one worker had 43 shares, the other 38.

(43 + 38) / 15 = 5.4 shares per minute.

There's variability in the shares, 15 minutes probably isn't nearly enough to get a good idea of how much you're hashing. Try running for a 24 hour period and see what happens, then try again with a single instance.

Hashing rates reported by miners mean squat.  You should look at: "number of accepted shares"/minute, that is what is making money.  Not what is displayed as "hashing rate" by the miner.  Every miner is doing their own estimate, so the mileage might vary  Pool is doing their own estimation.
Accepted shares per minute does not lie and IMHO, it is the best indicator of the efficiency of your whole mining rig: miner code/PC/network/pool etc.

For me, if the miner is not submitting 1 valid share/minute for every ~70 Mh/s it is reporting, something is wrong.

I would say that shares per minute isn't a large enough sample to get a good estimation. I agree that the number of shares is the most important statistic, but you could get a share after one hash or after billions. It's just about letting the miner run long enough that the law of large numbers comes in on your side.

[Mabsark]

There's variability in the shares, 15 minutes probably isn't nearly enough to get a good idea of how much you're hashing. Try running for a 24 hour period and see what happens, then try again with a single instance.

I'd have to stop using my PC for a day. That's not going to happen. Just out of curiosity, what are you guys getting for shares per minute?

[bmgjet]

Im running  POCLBM in command line on my 6850's

poclbm -d1 --host=xxxxx --port=8332 --user=bmgjet_1 -v 2 -f 2 -w 128
poclbm -d1 --host=xxxxx --port=8332 --user=bmgjet_2 -v 2 -f 2 -w 128

This command gets me 278-283Mhash on each card overclocked to 1ghz core 900mhz memory.
Its using some hacked up kernel that was posted on here last year and modified for best speed on 68XX cards.

Getting 5 shares per min on average sometimes hitting 6 if I dont use the PC.