A passing "for what it's worth": after some bu**ering about, my Sapphire Nitro RX470s (8GB) seem happiest with memory clocks of 2150MHz, 1750Mhz strap timings and fairly aggressive fan controls (keeping GPU temperatures below 65). With that, I'm getting a steady 27.5Mh/s per card, stable enough (current uptime six and a half days) but this did depend substantially on cooling, down to not having the cards too close together. Prior experience with some 480s suggested that the 1500/1625 strap timings weren't helpful at least under Linux and this seems to carry over to the 470s.
I'm not familiar with the Powercolor cards, but that's what I can sensibly achieve on 8GB Sapphire cards with twin fans. I suspect you'll find odd reports of people getting 28Mh/s out of them - just like you'll see odd reports of 31Mh/s from the 480s - but it might be optimistic to look for a sustained, stable 27-28Mh/s out of 4GB cards. You probably wouldn't be far short, but "is it capable of..." may not answer the question "is it sustainable".
I have been using this driver since it came out . WHQL-Win10-64Bit-Radeon-Software-Crimson-16.9.2-Sep29
i un check the HDMI audio when i install it .
When my rigs hangs it either to high OC on main GPU (the one connected to monitor ) or bad riser Bad drivers either shows less cards or some of the cards under hash .
Is there an original bios for Sapphire rx470 nitro plus 4gb (Elpida) for both silent and overclock (i mean both of the roms) I accidentally lost my backup bios and modded both of them wrongly Now my gpu keeps crashing.
Guess I'm lucky, because my Sapphire Nitro+ RX 470 4 Gb have Hynix memory.
Flashed ROM with PBE (strap 1500, 1000/1900 MHz), got 25.9 MH/s from each GPU. Eth single mode
Temps are around 74 degree, fans speed is below 900 rpm, works pretty stable
Actually you are unlucky. I have 40 cards, 9 of them are elpida and the worst of them hitting 29MH/s with 1150/2000, rest of hynix struggling to get 1130/1920 without errors , some hynix did 1950 and some only 1880
If you raise your core clock to 1130 you can hit 27 MH/s, also increase fan speed and try to play with the vcore
i have 2 elpida 480rx 4gb, cant flash then... but with 470 flashed (also elpida)... no problem, running 28.5 softly - but 480 cant passes 26 and does not accept modded flash, used PBE modified, and a couple of ready bios... no candy for me .. any tips?
i have 2 rigs whit 4 msi rx470 win 7 64x using 16.9.2 drivers i now when i restart claymore constantly geting display driver stopped responding and has recovered error. Any help?
@joca1982 overclocked or undervolted too much most likely.
1100/1900 voltage -36 same settings on both rigs error shows always on just one
Even if the cards look identical, they aren't. Every GPU is different and will overclock differently. Some will overclock savagely high and undervolt well at the same time, yet others will require an overvolt to obtain high overclocks. And then there is variance in VRAM quality which is the typically limiting factor for ethash.
I have 6 x SAPPHIRE Radeon RX 470 DirectX 12 100407-4GOCL 4GB
I first started off with a Gigabyte FX990-GAMING motherboard and could never get it stable with 6 cards, eventually, it wouldn't recognize the 6th card no matter what I did, so I bought an MSI Gaming Z97 GAMING 5 motherboard. I got all 6 cards working and then one of the cards burnt out, it literally caught fire, so I bought a replacement and RMA the burnt out card, this motherboard has 7 pci slots, so when the replacement comes I hope to run it with 7 slots.
I did the bios hack and I can get 28.5 MH/S but not stable, so I have been playing around and this just won't stay up no matter what I do. I have now tried risers of every conceivable flavor, from ebay, newegg, amazon multiple vendors.
It looks like the cards loose "contact" with the motherboard, the rig will be up and working and then randomly claymore reports one ore more GPU's hung and won't restart. Even rebooting the rig doesn't seem to help, sometimes when I reseat all the risers it will magically work again for a while, only to shutdown in the middle of the night.
It looks like 27-28 MH/s is a pipe dream for these cards, I am not trying it at the factory clock speeds and each card seems to be yielding 25.2 - 25.4 MH/S for a total of 154 MH/S
Just when I think its working then I get either GPU 0 or GPU 5: WATCHDOG: GPU 0 hangs in OpenCL call, exit
I am at the end of my wits, I've considered taking my hammer to the fricking thing and call it a loss.
@dbguru got same issues (with rx480 8gb) gpu0 and gpu2 for me , some times 12hours or 24 stable , sometimes reboot after few minutes lot of times ... and get lot of try like you ... it seem there is no real reason, instead gpu are not good as other ... i thought reflash bios help , but to next gpu lost ... and need flash again and again
@dbguru got same issues (with rx480 8gb) gpu0 and gpu2 for me , some times 12hours or 24 stable , sometimes reboot after few minutes lot of times ... and get lot of try like you ... it seem there is no real reason, instead gpu are not good as other ... i thought reflash bios help , but to next gpu lost ... and need flash again and again
Had a similar issue recently on a 6 x RX470 rig, two cards wouldn't play with my normal settings and kept crashing every so often. After some messing around, what seems to have solved it was dropping the memory clock speed down by 25MHz on the two offending cards, leaving the others as they were. It now seems fairly happy at 165Mh/s all up.
For the life of me I can't figure out how windows and claymore numbers the cards. I use sapphire trix to underclock the gpu and overclock memory, I choose the first card in the drop down list and this should be GPU 0, but when I start claymore, GPU 3 is now mining at a higher hash rate, if I overclock the 2nd, now GPU 3, 4 are hashing. how the f** are the cards numbered ?
btw, I have been running all 6 cards at default clock since 2pm (its now 8PM) and they are all still hashing away at 151 MH/S.
With the original bios, these cards get about 22 MH/S, with the modded bios, they get 25.3 MH/s The standard clocks are 1216 and 1750, the supposedly numbers to get 28 MH/S are 1100 and 2000, but I can't get anything stable. So looks like I just have to be contempt with about 151 MH/S
For the life of me I can't figure out how windows and claymore numbers the cards. I use sapphire trix to underclock the gpu and overclock memory, I choose the first card in the drop down list and this should be GPU 0, but when I start claymore, GPU 3 is now mining at a higher hash rate, if I overclock the 2nd, now GPU 3, 4 are hashing. how the f** are the cards numbered ?
btw, I have been running all 6 cards at default clock since 2pm (its now 8PM) and they are all still hashing away at 151 MH/S.
With the original bios, these cards get about 22 MH/S, with the modded bios, they get 25.3 MH/s The standard clocks are 1216 and 1750, the supposedly numbers to get 28 MH/S are 1100 and 2000, but I can't get anything stable. So looks like I just have to be contempt with about 151 MH/S
I'm using the same card and got the same numbers (22.5mh/s stock, 25.5mh/s after modding). Used trixx to push them to 27.5mh/s STABLE (1100/1875-1900). I've been running the rig for 7 days straight (168 hours) without issues, just maintenance because my dog's fur and ambiental dust (every sunday)
What do you do for dist maintenance ? I have a cairn terrier and he sheds like crazy, after the card burnt out I am a little neurotic and try to keep him as far away from the rig as possible, but living in an apartment doesn't give me a lot of options.
I just shut the rig down after it was mining for more than 12 hours at about 160 mh/s only for the miner to refuse starting up again, after changing risers on GPU 1 , I somehow got it working again with all 6 cards at 1100 and 1900 hashing at 161.6 mh/s
But I'm confused about the title of the thread - "GPU mining is out" What does it exactly mean?
If the mining is still possible I want to know something.
1. By how many degrees (C or F) will temperature rise in the room ? 2. How man decibels does the work of the Rig generates ? 3. How can I calculate my Hash power before buying certain GPUs? 4. What is the curent ROI on 5-6 GPU Mining RIG? Let's say I can spend over 1.5k US $ on a Rig.
What do you do for dist maintenance ? I have a cairn terrier and he sheds like crazy, after the card burnt out I am a little neurotic and try to keep him as far away from the rig as possible, but living in an apartment doesn't give me a lot of options.
I just shut the rig down after it was mining for more than 12 hours at about 160 mh/s only for the miner to refuse starting up again, after changing risers on GPU 1 , I somehow got it working again with all 6 cards at 1100 and 1900 hashing at 161.6 mh/s
I have 2 little demons: lhasa apso and a bichon habanero, I need to be very careful about dust, fur, and nasty stuff! I've encased my rig inside a box made of mosquito net and done! The mosquito net keeps out the fur and the rig is always receiving cool air (I live in a tropical country, so my room have an air conditioner and a ceiling fan)
But I'm confused about the title of the thread - "GPU mining is out" What does it exactly mean?
If the mining is still possible I want to know something.
1. By how many degrees (C or F) will temperature rise in the room ? 2. How man decibels does the work of the Rig generates ? 3. How can I calculate my Hash power before buying certain GPUs? 4. What is the curent ROI on 5-6 GPU Mining RIG? Let's say I can spend over 1.5k US $ on a Rig.
ok.... In my personal experience: I'm mining with win7 64, rx470 4gb reference (modded)...
1.- My room (bedroom) is always fresh because I have an air conditioner, no problem with the temp (just one rig - 2 cards) 2.- The noise is very low (I sleep like a dead, so no problems till now) 3.- Look here in the forum... hashrate for Rx470 4gb is about 22mh/s, for 480 is about 25mh/s (without modding) 4.- About ROI: that's very difficult to calculate (because difficulty)... Ex: if you invest 1200$ today you should get about 120$ each 30 days. I suggest you to keep it low: Build a 4 gpu rig but with a mobo with 6 or 7 pci-e slots and buy a good 1200 gold psu. Start mining, check your first month performance and decide if you want to keep going or whatever you want to do...
@Streszcz check out http://whattomine.com its a really awesome site, you can enter the number of GPU's, for example, next to (RX) 470, I entered 6 and clicked the 470 , it turns red and auto fills the hashrates for all the coins. Its very close and then click calculate. It will show the coins to mine in descending order of profitibility.
About ROI: that's very difficult to calculate (because difficulty)... Ex: if you invest 1200$ today you should get about 120$ each 30 days. I suggest you to keep it low: Build a 4 gpu rig but with a mobo with 6 or 7 pci-e slots and buy a good 1200 gold psu. Start mining, check your first month performance and decide if you want to keep going or whatever you want to do...
What do you mean by 4 gpu with a mobo for 6-7 gpus? Means that I should have and option to expand rig from 4 gpus to 6-7 gpus?
and 1200 gold psu is a power supply source, for instance when power blackout will occur?
GPU mining is really out or what it really means? I have heard that Etherum will be going PoS from PoW in 2017, so better to start mining another cryptocurreny?
I told you to keepd it easy with the GPU's because they're the main investment... And @Streszcz is right, ETH will be going PoW this year, so when it happens you could switch to mining ZEC or any other gpu-mineable coin (or just sell your rig and keep one gpu for gaming, IMO)
@charliebridge82 can you provide link to a store with GPUs shiping worldwide?
I always use US Amazon or ebay (lowest prices and my courier doesn't charge too much for sending goods from miami to latin america). If you use Amazon you can claim taxes later (I claimed 13$ for an RX470, not bad)
Comments
What is your PSU model ?
Have you checked for bad riser ?
I'm not familiar with the Powercolor cards, but that's what I can sensibly achieve on 8GB Sapphire cards with twin fans. I suspect you'll find odd reports of people getting 28Mh/s out of them - just like you'll see odd reports of 31Mh/s from the 480s - but it might be optimistic to look for a sustained, stable 27-28Mh/s out of 4GB cards. You probably wouldn't be far short, but "is it capable of..." may not answer the question "is it sustainable".
WHQL-Win10-64Bit-Radeon-Software-Crimson-16.9.2-Sep29
i un check the HDMI audio when i install it .
When my rigs hangs it either to high OC on main GPU (the one connected to monitor ) or bad riser
Bad drivers either shows less cards or some of the cards under hash .
I accidentally lost my backup bios and modded both of them wrongly Now my gpu keeps crashing.
I first started off with a Gigabyte FX990-GAMING motherboard and could never get it stable with 6 cards, eventually, it wouldn't recognize the 6th card no matter what I did, so I bought an MSI Gaming Z97 GAMING 5 motherboard. I got all 6 cards working and then one of the cards burnt out, it literally caught fire, so I bought a replacement and RMA the burnt out card, this motherboard has 7 pci slots, so when the replacement comes I hope to run it with 7 slots.
I did the bios hack and I can get 28.5 MH/S but not stable, so I have been playing around and this just won't stay up no matter what I do. I have now tried risers of every conceivable flavor, from ebay, newegg, amazon multiple vendors.
It looks like the cards loose "contact" with the motherboard, the rig will be up and working and then randomly claymore reports one ore more GPU's hung and won't restart. Even rebooting the rig doesn't seem to help, sometimes when I reseat all the risers it will magically work again for a while, only to shutdown in the middle of the night.
It looks like 27-28 MH/s is a pipe dream for these cards, I am not trying it at the factory clock speeds and each card seems to be yielding 25.2 - 25.4 MH/S for a total of 154 MH/S
Just when I think its working then I get either GPU 0 or GPU 5: WATCHDOG: GPU 0 hangs in OpenCL call, exit
I am at the end of my wits, I've considered taking my hammer to the fricking thing and call it a loss.
btw, I have been running all 6 cards at default clock since 2pm (its now 8PM) and they are all still hashing away at 151 MH/S.
With the original bios, these cards get about 22 MH/S, with the modded bios, they get 25.3 MH/s The standard clocks are 1216 and 1750, the supposedly numbers to get 28 MH/S are 1100 and 2000, but I can't get anything stable. So looks like I just have to be contempt with about 151 MH/S
I just shut the rig down after it was mining for more than 12 hours at about 160 mh/s only for the miner to refuse starting up again, after changing risers on GPU 1 , I somehow got it working again with all 6 cards at 1100 and 1900 hashing at 161.6 mh/s
But I'm confused about the title of the thread - "GPU mining is out" What does it exactly mean?
If the mining is still possible I want to know something.
1. By how many degrees (C or F) will temperature rise in the room ?
2. How man decibels does the work of the Rig generates ?
3. How can I calculate my Hash power before buying certain GPUs?
4. What is the curent ROI on 5-6 GPU Mining RIG? Let's say I can spend over 1.5k US $ on a Rig.
1.- My room (bedroom) is always fresh because I have an air conditioner, no problem with the temp (just one rig - 2 cards)
2.- The noise is very low (I sleep like a dead, so no problems till now)
3.- Look here in the forum... hashrate for Rx470 4gb is about 22mh/s, for 480 is about 25mh/s (without modding)
4.- About ROI: that's very difficult to calculate (because difficulty)... Ex: if you invest 1200$ today you should get about 120$ each 30 days. I suggest you to keep it low: Build a 4 gpu rig but with a mobo with 6 or 7 pci-e slots and buy a good 1200 gold psu. Start mining, check your first month performance and decide if you want to keep going or whatever you want to do...