There may be a 150 difference limit between the engine and memory clocks, just like on the 79x0 cards. Try adding --gpu-memdiff -150
to your cgminer parameters.
to your cgminer parameters.
Does that go before or after my --no-adl ?
Topic: 7770 Mining Stats (Read 3061 times)
Does that go before or after my --no-adl ?
There may be a 150 difference limit between the engine and memory clocks, just like on the 79x0 cards. Try adding --gpu-memdiff -150
to your cgminer parameters.
to your cgminer parameters.
Ah, as far as I know there is no way to get memory down to 300 with a high core still in linux, so I haven't been able to measure it that way.
I just used AB 2.2.1 to lower the mem speed down to 300, and it lowered my system usage down to 149W, which is 66W for the card. This bumps it up from 2.7MHs/w to 3.0MHs/W.
It's currently 67C @ 37% fan.
It's currently 67C @ 37% fan.
Yea, it really is a great value all around ( except for density of course ).
I just got one of these for my brother's new computer, and I decided to start playing with it: http://www.newegg.com/Product/Product.aspx?Item=N82E16814125415
It's mining at 1100,950,1.200, for about 203MH/s. It's factory OC'd, so I didn't touch the core and just set the memory lower manually using CGMiner, along with -k diakgcn -I 6 options on Win7x64 and 12.6(with included SDK).
Idle before card: 83W
Idle with card: 89W
Mining with card: 157W.
157W - 83W = 74W for just the card. That's 2.7MHs/W, which isn't bad.
Note, I bought this card for 140, with a 20 rebate. Its since gone down to 110 with a 20 rebate, so for $90, it's not bad. I'm gonna return mine and buy a new one at the cheaper price.
It's mining at 1100,950,1.200, for about 203MH/s. It's factory OC'd, so I didn't touch the core and just set the memory lower manually using CGMiner, along with -k diakgcn -I 6 options on Win7x64 and 12.6(with included SDK).
Idle before card: 83W
Idle with card: 89W
Mining with card: 157W.
157W - 83W = 74W for just the card. That's 2.7MHs/W, which isn't bad.
Note, I bought this card for 140, with a 20 rebate. Its since gone down to 110 with a 20 rebate, so for $90, it's not bad. I'm gonna return mine and buy a new one at the cheaper price.
Your numbers look a little strange, are you sure they're correct?
For instance, going from 1000 to 1150 is a 15% increase in clock speed and gave you a 12% increase in hash rate, but only a 2.1% increase in power consumption above idle.
For instance, going from 1000 to 1150 is a 15% increase in clock speed and gave you a 12% increase in hash rate, but only a 2.1% increase in power consumption above idle.
Measured everything with my kilawatt... my numbers are as accurate as kilowatt and idle system power stablility are...
I am glad you used a killawatt. Please post the numbers for the entire rig for each clock, that is the only number that really matters anyway
It is 100 watts idle exactly. So just add a 1 in front of those numbers...
Your numbers look a little strange, are you sure they're correct?
For instance, going from 1000 to 1150 is a 15% increase in clock speed and gave you a 12% increase in hash rate, but only a 2.1% increase in power consumption above idle.
For instance, going from 1000 to 1150 is a 15% increase in clock speed and gave you a 12% increase in hash rate, but only a 2.1% increase in power consumption above idle.
Measured everything with my kilawatt... my numbers are as accurate as kilowatt and idle system power stablility are...
I am glad you used a killawatt. Please post the numbers for the entire rig for each clock, that is the only number that really matters anyway
Your numbers look a little strange, are you sure they're correct?
For instance, going from 1000 to 1150 is a 15% increase in clock speed and gave you a 12% increase in hash rate, but only a 2.1% increase in power consumption above idle.
For instance, going from 1000 to 1150 is a 15% increase in clock speed and gave you a 12% increase in hash rate, but only a 2.1% increase in power consumption above idle.
Measured everything with my kilawatt... my numbers are as accurate as kilowatt and idle system power stablility are...
Your numbers look a little strange, are you sure they're correct?
For instance, going from 1000 to 1150 is a 15% increase in clock speed and gave you a 12% increase in hash rate, but only a 2.1% increase in power consumption above idle.
For instance, going from 1000 to 1150 is a 15% increase in clock speed and gave you a 12% increase in hash rate, but only a 2.1% increase in power consumption above idle.
Also prices are falling which is nice to see. I remember the 5770 staying at ~$135 for about 1½ years, even when the 6770 was introduced.
Very nice, thanks for posting.
The 7770 seems to have matured into a decent 5770 replacement.
The 7770 seems to have matured into a decent 5770 replacement.
7770 you mean?
Edit: I see, the topic says 7770 but the body said 7700. That's what was throwing me.
Edit: I see, the topic says 7770 but the body said 7700. That's what was throwing me.
thats pretty damn effecient, wonder how much less it would be in watts if mem was dropped.
I haven't tried much. I did set it with aticonfig but when the clock goes back up, the mem jumps back to the max. It seemed to drop ~4 watts in idle mode when mem was down to 300. I am guessing if I properly setup the profiles I can get it to stay at 300. If not now, then later as the 7xxx is better supported. This card is my main GPU driving 4 monitors, so low memory clock causes screen flickers.. so I have not really looked into it.
thats pretty damn effecient, wonder how much less it would be in watts if mem was dropped.
I got a 7770 today and here are my stats so far. Stock voltage. Watts column is 'watts above idle'. Arch linux, catalyst 12.6-5 beta, cgminer, ambiant temp of 80F
| Intensity | Mh/S Hash | Core | Mem | Watts | Temp (c) | Fan |
| 1 | 90 | 1000 | 1125 | 39 | 59 | 32% |
| 3 | 148 | 1000 | 1125 | 47 | 64 | 37% |
| 3 | 166 | 1150 | 1125 | 48 | 64 | 38% |
| 9 | 186 | 1000 | 1125 | 53 | 68 | 40% |
| 9 | 214 | 1150 | 1125 | 55 | 72 | 44% |
| 9 | 224 | 1200 | 1125 | 58 | 74 | 46% |
Jump to: