pikachuy
|
|
June 07, 2016, 06:16:26 PM |
|
@Claymore can you explain this?
I have a radeon 380 4gb on a windows 7 x64 running 15.12 drivers and running CDM 4.5 that is undervolted -100mv, core clock at 840, memory clock at 1125, and power limit -20% for days using sapphire trixx. And it hashes at a stable 19~ MH/s constantly and never above 20 MH/s.
I then decided to change the memory clock recently to 1500 just to see if it makes any difference on trixx and pressed apply, and let it run for couple of minutes, but the hash rate barely changed or any change at all. So i changed the memory clock back down to 1125 and press apply, and now the hashrate is hashing at a stable 20~ MH/s.
I am clueless of why setting the memory clock from 1125 to 1500 and then back down to 1125 would increase the reported hashrate in CDM. Any thoughts?
|
|
|
|
Claymore (OP)
Donator
Legendary
Offline
Activity: 1610
Merit: 1325
Miners developer
|
|
June 07, 2016, 06:28:59 PM |
|
@Claymore can you explain this? I have a radeon 380 4gb on a windows 7 x64 running 15.12 drivers and running CDM 4.5 that is undervolted -100mv, core clock at 840, memory clock at 1125, and power limit -20% for days using sapphire trixx. And it hashes at a stable 19~ MH/s constantly and never above 20 MH/s. I then decided to change the memory clock recently to 1500 just to see if it makes any difference on trixx and pressed apply, and let it run for couple of minutes, but the hash rate barely changed or any change at all. So i changed the memory clock back down to 1125 and press apply, and now the hashrate is hashing at a stable 20~ MH/s. I am clueless of why setting the memory clock from 1125 to 1500 and then back down to 1125 would increase the reported hashrate in CDM. Any thoughts?
I have no ideas, sorry.
|
|
|
|
restless
Legendary
Offline
Activity: 1151
Merit: 1001
|
|
June 07, 2016, 06:32:32 PM |
|
@Claymore can you explain this?
I have a radeon 380 4gb on a windows 7 x64 running 15.12 drivers and running CDM 4.5 that is undervolted -100mv, core clock at 840, memory clock at 1125, and power limit -20% for days using sapphire trixx. And it hashes at a stable 19~ MH/s constantly and never above 20 MH/s.
I then decided to change the memory clock recently to 1500 just to see if it makes any difference on trixx and pressed apply, and let it run for couple of minutes, but the hash rate barely changed or any change at all. So i changed the memory clock back down to 1125 and press apply, and now the hashrate is hashing at a stable 20~ MH/s.
I am clueless of why setting the memory clock from 1125 to 1500 and then back down to 1125 would increase the reported hashrate in CDM. Any thoughts?
What program are you using to change clocks? What program do you use to verify at what clock is the GPU/mem running? I'd suggest Hwinfo or GPU-Z to verify actual speed&memory speed
|
|
|
|
pikachuy
|
|
June 07, 2016, 06:35:48 PM |
|
@Claymore can you explain this? I have a radeon 380 4gb on a windows 7 x64 running 15.12 drivers and running CDM 4.5 that is undervolted -100mv, core clock at 840, memory clock at 1125, and power limit -20% for days using sapphire trixx. And it hashes at a stable 19~ MH/s constantly and never above 20 MH/s. I then decided to change the memory clock recently to 1500 just to see if it makes any difference on trixx and pressed apply, and let it run for couple of minutes, but the hash rate barely changed or any change at all. So i changed the memory clock back down to 1125 and press apply, and now the hashrate is hashing at a stable 20~ MH/s. I am clueless of why setting the memory clock from 1125 to 1500 and then back down to 1125 would increase the reported hashrate in CDM. Any thoughts?
I have no ideas, sorry. No worries, thought I ask since I can reproduce this result everytime, so now I have to go through this exact pattern everytime I run CDM to get that extra 1 MH/s. Can someone run the same settings to see if they get the same outcome? Wonder if it's CDM related. First, let it run with undervolted -100mv, core clock at 840, memory clock at 1125, and power limit -20%. then once you see the reported hashrate, changed the memory clock to 1500 and apply, then after you see the reported hashrate again, changed the memory clock back down to 1125 and press apply. That's when I then see the increased 1 MH/s everytime. Edit: @restless I am using sapphire trixx
|
|
|
|
adamvp
|
|
June 07, 2016, 06:47:05 PM |
|
do you hava a Saphire card?
|
I am looking for signature campaign pm me
|
|
|
pikachuy
|
|
June 07, 2016, 06:49:57 PM |
|
do you hava a Saphire card?
No, my card is GIGABYTE Radeon R9 380 4gb, but sapphire trixx tool works for it. I see the confirmed changes in GPU-Z.
|
|
|
|
restless
Legendary
Offline
Activity: 1151
Merit: 1001
|
|
June 07, 2016, 06:54:01 PM |
|
First, let it run with undervolted -100mv, core clock at 840, memory clock at 1125, and power limit -20%. then once you see the reported hashrate, changed the memory clock to 1500 and apply, then after you see the reported hashrate again, changed the memory clock back down to 1125 and press apply. That's when I then see the increased 1 MH/s everytime.
Edit: @restless I am using sapphire trixx
Please run also one of hwinfo or gpu-z for confirmation. Trixx and AfterBurner sometimes incorrectly show current clock (both are simply same program, written by Unwinder, with different skins, customized for Saphire and MSI) 20MH/s seems way too good for 380 at 850/1125 PS: Then there could be possibility that somehow going down from 1500, memory is running with better/faster timings Maybe running memory benchmark (Sandra?) will show bigger memcopy speed ?!
|
|
|
|
melloyellow
|
|
June 07, 2016, 06:54:52 PM |
|
@Claymore can you explain this? I have a radeon 380 4gb on a windows 7 x64 running 15.12 drivers and running CDM 4.5 that is undervolted -100mv, core clock at 840, memory clock at 1125, and power limit -20% for days using sapphire trixx. And it hashes at a stable 19~ MH/s constantly and never above 20 MH/s. I then decided to change the memory clock recently to 1500 just to see if it makes any difference on trixx and pressed apply, and let it run for couple of minutes, but the hash rate barely changed or any change at all. So i changed the memory clock back down to 1125 and press apply, and now the hashrate is hashing at a stable 20~ MH/s. I am clueless of why setting the memory clock from 1125 to 1500 and then back down to 1125 would increase the reported hashrate in CDM. Any thoughts?
I have no ideas, sorry. No worries, thought I ask since I can reproduce this result everytime, so now I have to go through this exact pattern everytime I run CDM to get that extra 1 MH/s. Can someone run the same settings to see if they get the same outcome? Wonder if it's CDM related. First, let it run with undervolted -100mv, core clock at 840, memory clock at 1125, and power limit -20%. then once you see the reported hashrate, changed the memory clock to 1500 and apply, then after you see the reported hashrate again, changed the memory clock back down to 1125 and press apply. That's when I then see the increased 1 MH/s everytime. Edit: @restless I am using sapphire trixx It's not CDM related, read here https://forum.ethereum.org/discussion/comment/39962/#Comment_39962 It's a problem with 380's
|
|
|
|
bradleyjay
Newbie
Offline
Activity: 59
Merit: 0
|
|
June 07, 2016, 07:15:21 PM |
|
Dang... thank you for this. I'm not new to mining but I am new to GPU mining and overclocking and I couldn't understand why my 380's hashrate wouldn't budge increasing the core clock but I found a nice little boost by upping the memclock to 1500. After reading the thread you linked to, it makes sense now. I think this card is doing best for me at 980/1500 on W764 with 15.12 drivers and CDM.
|
|
|
|
Xardas2014
|
|
June 07, 2016, 07:19:32 PM |
|
I just tried this on a rig with two Saph 380's and one ASUS 380. It picked up one mh/s per card on ETH and ~10mh/s per card on Decred. Dunno if it is a display anomaly or what.............but I'll leave it anyway.
|
|
|
|
pikachuy
|
|
June 07, 2016, 07:29:50 PM |
|
I just tried this on a rig with two Saph 380's and one ASUS 380. It picked up one mh/s per card on ETH and ~10mh/s per card on Decred. Dunno if it is a display anomaly or what.............but I'll leave it anyway.
Yea, it's quite odd that it does it, but it doesn't bother me, can't complain about increase mh/s. I'm just glad to share what I found and the steps needed to replicate it =) Edit: Can someone else besides a 380 card try the steps i mention to see if it show the same results? First, let it run with undervolted -100mv, core clock at 840, memory clock at 1125, and power limit -20%. then once you see the reported hashrate, changed the memory clock to 1500 and apply, then after you see the reported hashrate again, changed the memory clock back down to 1125 and press apply. That's when I then see the increased 1 MH/s everytime.
Edit: @restless I am using sapphire trixx
Please run also one of hwinfo or gpu-z for confirmation. Trixx and AfterBurner sometimes incorrectly show current clock (both are simply same program, written by Unwinder, with different skins, customized for Saphire and MSI) 20MH/s seems way too good for 380 at 850/1125 PS: Then there could be possibility that somehow going down from 1500, memory is running with better/faster timings Maybe running memory benchmark (Sandra?) will show bigger memcopy speed ?! Here you go restless As you can see, with the undervolt and underclocking and reducing power limit, the power consumption is greatly reduced while still having a stable 20 mh/s. On my screenshot, first 2 cards are powercooler 380x, and the rest are gigabyte 380. If I run it at stock the VDDC power is around 80w, so since power consumption is important to me, lowering about 30w on the VDDC per card is good for me.
|
|
|
|
fredeq
Legendary
Offline
Activity: 1537
Merit: 1005
|
|
June 07, 2016, 07:41:19 PM |
|
@Claymore can you explain this? I have a radeon 380 4gb on a windows 7 x64 running 15.12 drivers and running CDM 4.5 that is undervolted -100mv, core clock at 840, memory clock at 1125, and power limit -20% for days using sapphire trixx. And it hashes at a stable 19~ MH/s constantly and never above 20 MH/s. I then decided to change the memory clock recently to 1500 just to see if it makes any difference on trixx and pressed apply, and let it run for couple of minutes, but the hash rate barely changed or any change at all. So i changed the memory clock back down to 1125 and press apply, and now the hashrate is hashing at a stable 20~ MH/s. I am clueless of why setting the memory clock from 1125 to 1500 and then back down to 1125 would increase the reported hashrate in CDM. Any thoughts?
I have no ideas, sorry. No worries, thought I ask since I can reproduce this result everytime, so now I have to go through this exact pattern everytime I run CDM to get that extra 1 MH/s. Can someone run the same settings to see if they get the same outcome? Wonder if it's CDM related. First, let it run with undervolted -100mv, core clock at 840, memory clock at 1125, and power limit -20%. then once you see the reported hashrate, changed the memory clock to 1500 and apply, then after you see the reported hashrate again, changed the memory clock back down to 1125 and press apply. That's when I then see the increased 1 MH/s everytime. Edit: @restless I am using sapphire trixx It's not CDM related, read here https://forum.ethereum.org/discussion/comment/39962/#Comment_39962 It's a problem with 380's How is that related to what pikachuy is reporting? He says that the hashrate goes up AFTER you downclock again to 1125. Btw I can confirm this as well, pretty weird... got gigabyte 380 on samsung chips @pikachuy dont trust gpu-z readings for core wattage, they are wrong. For me the operation didnt change gpu-z wattage readings (they are still at 60W) and readings from meters suggest around 140W per card.
|
|
|
|
1337leet
|
|
June 07, 2016, 09:48:22 PM |
|
Hey guys,
I'm about to buy a 295x2 but still have some questions.
It's a Sapphire one.
What speed should it get with claymores miner on ETH only Windows 7? Do you think 60 MH/s is realistic?
What core clock, mem clock, voltage should I run it?
Anyone got a rough estimation of the power consumption for ETH only? My rough estimations tells me ~ 350 Watts under full load.
Is the stock hybrid cooler strong enough to hold the card below 75 degrees celsius?
|
|
|
|
Alehandros
Member
Offline
Activity: 72
Merit: 10
|
|
June 07, 2016, 11:07:34 PM |
|
went mode 1 and never looked back!! whoop
|
|
|
|
ExcelThroughKnowledge
|
|
June 08, 2016, 03:09:05 AM |
|
I just tried this on a rig with two Saph 380's and one ASUS 380. It picked up one mh/s per card on ETH and ~10mh/s per card on Decred. Dunno if it is a display anomaly or what.............but I'll leave it anyway.
Yea, it's quite odd that it does it, but it doesn't bother me, can't complain about increase mh/s. I'm just glad to share what I found and the steps needed to replicate it =) Edit: Can someone else besides a 380 card try the steps i mention to see if it show the same results? First, let it run with undervolted -100mv, core clock at 840, memory clock at 1125, and power limit -20%. then once you see the reported hashrate, changed the memory clock to 1500 and apply, then after you see the reported hashrate again, changed the memory clock back down to 1125 and press apply. That's when I then see the increased 1 MH/s everytime.
Edit: @restless I am using sapphire trixx
Please run also one of hwinfo or gpu-z for confirmation. Trixx and AfterBurner sometimes incorrectly show current clock (both are simply same program, written by Unwinder, with different skins, customized for Saphire and MSI) 20MH/s seems way too good for 380 at 850/1125 PS: Then there could be possibility that somehow going down from 1500, memory is running with better/faster timings Maybe running memory benchmark (Sandra?) will show bigger memcopy speed ?! Here you go restless As you can see, with the undervolt and underclocking and reducing power limit, the power consumption is greatly reduced while still having a stable 20 mh/s. On my screenshot, first 2 cards are powercooler 380x, and the rest are gigabyte 380. If I run it at stock the VDDC power is around 80w, so since power consumption is important to me, lowering about 30w on the VDDC per card is good for me. @pikachuy could you teach me how you undervolt and clock your 380x? I'm looking for ways to get stable rates with lesser watts used.. thanks in advance =D
|
|
|
|
wiinston
Newbie
Offline
Activity: 7
Merit: 0
|
|
June 08, 2016, 05:40:21 AM |
|
Can we use antminer S7 or S5 to mine etherum?
never you can lease your rig at mining rig rental then use the BTC earned to buy GPU hashing power. Or you can sell your s7 and start mining with gpus (at least thats what i did)
|
|
|
|
BitJunkie223
Member
Offline
Activity: 88
Merit: 10
Vertcoin stealthy master race.
|
|
June 08, 2016, 02:48:10 PM |
|
Hey guys,
I'm about to buy a 295x2 but still have some questions.
It's a Sapphire one.
What speed should it get with claymores miner on ETH only Windows 7? Do you think 60 MH/s is realistic?
What core clock, mem clock, voltage should I run it?
Anyone got a rough estimation of the power consumption for ETH only? My rough estimations tells me ~ 350 Watts under full load.
Is the stock hybrid cooler strong enough to hold the card below 75 degrees celsius?
WHATEVER you do, don't use Sapphire for mining... It seems like everyone and their brother, sister, and mother, are having issues mining with Sapphire cards... I thought with my R9 270's (2 of them) ~36MH would be my max. I haven't seen it yet. I'm getting about 25MH/s Ethereum, about 380MH/s on Decred. I can get about 26MH/s in Eth only, but it's worth it to mine decred alongside ethereum. Keep in mind, that's considerably overclocked. Stock on mine is 900-925 Core Clock, 1400 Mem clock. I have it at 1050Mhz Core, 1500mhz Memory. My two cards together probably use about 350 watts. I would think a 295x2 would use at LEAST 500 watts. Make sure you have a hefty power supply. As for your expected mining speed... It's hard to really say. Two r9 270's are definitely more powerful than a single 290. So if I'm getting ~26MH/s, I'd imagine you'd get closer to 40. Maybe 45MH/s with a heavy overclock. But again, the problem is that I keep hearing nothing but complaints about Sapphire for mining... Could you perhaps go with two R9 290's, non-Sapphire? I don't want to hate on Sapphire. I used to have a passively-cooled HD5450 that I ran pretty hard. I now use it in my HTPC build. I love the thing, even now. But when you hear nothing but problems about a certain manufacturer in this particular usage... I can't help but recommend against it. I remember a thread on the ethereum forum where a guy had like 6 Sapphire cards for mining. Multiple failed within the first couple weeks. Not a good sign...
|
|
|
|
JuanHungLo
|
|
June 08, 2016, 03:39:08 PM |
|
I cannot disagree more with the above comment on Sapphire.
I have used 8 280x's (Dual-X, 3GB OC, 11221-00-20G) since late 2013 and they still go strong with the exception of changing out some fans.
Of course, YMMV.
|
Bull markets are born on pessimism, grow on skepticism, mature on optimism, and die on euphoria. - John Templeton
|
|
|
CoRpO
|
|
June 08, 2016, 03:41:39 PM |
|
I cannot disagree more with the above comment on Sapphire.
I have used 8 280x's (Dual-X, 3GB OC, 11221-00-20G) since late 2013 and they still go strong with the exception of changing out some fans.
Of course, YMMV.
The bad cards were mainly 7950 with weak power stage design, which overheated and/or burned
|
|
|
|
melloyellow
|
|
June 08, 2016, 03:57:21 PM |
|
Hey guys,
I'm about to buy a 295x2 but still have some questions.
It's a Sapphire one.
What speed should it get with claymores miner on ETH only Windows 7? Do you think 60 MH/s is realistic?
What core clock, mem clock, voltage should I run it?
Anyone got a rough estimation of the power consumption for ETH only? My rough estimations tells me ~ 350 Watts under full load.
Is the stock hybrid cooler strong enough to hold the card below 75 degrees celsius?
WHATEVER you do, don't use Sapphire for mining... It seems like everyone and their brother, sister, and mother, are having issues mining with Sapphire cards... I thought with my R9 270's (2 of them) ~36MH would be my max. I haven't seen it yet. I'm getting about 25MH/s Ethereum, about 380MH/s on Decred. I can get about 26MH/s in Eth only, but it's worth it to mine decred alongside ethereum. Keep in mind, that's considerably overclocked. Stock on mine is 900-925 Core Clock, 1400 Mem clock. I have it at 1050Mhz Core, 1500mhz Memory. My two cards together probably use about 350 watts. I would think a 295x2 would use at LEAST 500 watts. Make sure you have a hefty power supply. As for your expected mining speed... It's hard to really say. Two r9 270's are definitely more powerful than a single 290. So if I'm getting ~26MH/s, I'd imagine you'd get closer to 40. Maybe 45MH/s with a heavy overclock. But again, the problem is that I keep hearing nothing but complaints about Sapphire for mining... Could you perhaps go with two R9 290's, non-Sapphire? I don't want to hate on Sapphire. I used to have a passively-cooled HD5450 that I ran pretty hard. I now use it in my HTPC build. I love the thing, even now. But when you hear nothing but problems about a certain manufacturer in this particular usage... I can't help but recommend against it. I remember a thread on the ethereum forum where a guy had like 6 Sapphire cards for mining. Multiple failed within the first couple weeks. Not a good sign... It's funny 2 years ago, everyone wanted Sapphire cards. I have 2+ year old Sapphire cards and brand new Sapphire cards, only RMA'd one. Maybe 13 Mh/s is about right for a 270 according to this http://www.mininghwcomparison.com/list/index.php?brand=amd There's a 270x mining at 11 mh/s that was added recently but looks like he's not on the best drivers. There is also a 270 at 16mh/s but at a 1200 clock and he may have added that when the dag was smaller. There are others on there more or less in line with your speeds, but nowhere near 18mh/s. My reference 290s do a little less that 27 so one is probably about the same as 2 270s
|
|
|
|
|