 Show Posts
|
|
Pages: [1] 2 3 4 »
|
GPU6: Generating DAG for epoch #387
Eth: New job #06735b75 from asia.ethash-hub.miningpoolhub.com:20535; diff: 8589MH
GPU2: DAG 65%
GPU5: DAG 21%
GPU4: DAG 42%
GPU7: Free VRAM: 7.948 GB; used: 0.052 GB
GPU7: Disabling DAG pre-allocation (not enough VRAM)
GPU7: Allocating DAG for epoch #387 (4.02) GB
GPU7: Generating DAG for epoch #387
GPU3: DAG 63%
GPU2: DAG 81%
GPU6: DAG 21%
Eth: New job #7cc46469 from asia.ethash-hub.miningpoolhub.com:20535; diff: 8589MH
GPU5: DAG 41%
GPU1: Free VRAM: 7.948 GB; used: 0.052 GB
GPU1: Disabling DAG pre-allocation (not enough VRAM)
GPU1: Allocating DAG for epoch #387 (4.02) GB
GPU1: Generating DAG for epoch #387
Eth: New job #a67920b6 from asia.ethash-hub.miningpoolhub.com:20535; diff: 8589MH
GPU4: DAG 63%
GPU3: DAG 80%
GPU7: DAG 21%
GPU2: DAG generated in 8.0 s (515.6 MB/s)
GPU2: Using Ethash OCL kernels (Ellesmere; -clkernel 1 -clnew 0)
GPU2: clSetKernelArg (-48)
Fatal error detected. Restarting.
Eth: New job #7482ee9b from asia.ethash-hub.miningpoolhub.com:20535; diff: 8589MH
Eth: New job #9bc3aa8e from asia.ethash-hub.miningpoolhub.com:20535; diff: 8589MH
Eth: New job #b84dedd8 from asia.ethash-hub.miningpoolhub.com:20535; diff: 8589MH
I have that problem, virtual memory in 40.000.
Try to add -gser 2 SAME BRO, same problem you need to find the optimal value for the -gser command for your gpus. Example: PhoenixMiner.exe -pool eth-eu2.nanopool.org:9999 -wal wallet.rig -gser 2 or Perhaps your drivers have been automatically updated. You need to roll back the drivers to the version on which the program works stably. In config.txt add: -rvram -1 And add more virtual memory in windows 16 to 32Go, Me for 6 * RX580 i have 65635Mo virtual memory for no problem Same but i saw this: GPU4: DAG 22% Eth: New job #4b0e9be5 from asia.ethash-hub.miningpoolhub.com:20535; diff: 8589MH GPU3: DAG 65% GPU2: DAG generated in 8.0 s (517.9 MB/s) GPU2: Using Ethash OCL kernels (Ellesmere; -clkernel 1 -clnew 0) GPU2: clSetKernelArg (-48)
Fatal error detected. Restarting.Eth speed: 0.000 MH/s, shares: 0/0/0, time: 0:00 GPUs: 1: 0.000 MH/s (0) 2: 0.000 MH/s (0) 3: 0.000 MH/s (0) 4: 0.000 MH/s (0) 5: 0.000 MH/s (0) 6: 0.000 MH/s (0) 7: 0.000 MH/s (0) GPU5: Free VRAM: 7.948 GB; used: 0.052 GB try to add this: PhoenixMiner.exe -pool eth-eu2.nanopool.org:9999 -wal wallet.rx580 -gser 10 -rvram -1 Same problem, delete all and paste what you write and same vram problem, today was working fine, i did not update nothing and stop mining go back and read few post, I had same problem post #7311 ( https://bitcointalk.org/index.php?topic=2647654.msg56049534#msg56049534 ) Also, windows 20.4 with PM 5.4c and AMD 20.11.2 working nice +merit if solved  Got the same problem today. On 3 miners. 2 miners with Rx580's and a 3rd one with Vegas. All of them showing the same not enough vram and gpu1 error. I updated to the latest edition of phoenix but its still the same.. Could this be a drivers or windows update issue? I had the same issue on my RX580 8GB (clSetKernelArg (-48)). I reinstalled AMD driver software (Adrenalin version 20.8.1) with factory reset option enabled and the issue is gone. I also turned off automatic updates from the Adrenalin settings since I suspect it had auto-updated something. So to fix the issue I suggest you reinstall your drivers to some old version that you know has worked fine before. @PhoenixMiner Could you take a look at what is causing the clSetKernelArg (-48) issue with the auto-updated AMD driver? It seems to repro at least with RX580 8GB cards. |
|
|
|
|
Looks promising. Could you add support for ahashpool? Is there any dev fee?
|
|
|
|
I don't get staking for last 4 days.
I've about 6,400 coins now that usally got atleast 2 -3 stake per day.
Wallet is open 24/7 , 19+ active connection , everythings looks good and works great for a month.
And I already tried resync , No error.
Is there any problem for stake a couple of day?
thanks.
I have 3300 coins and during the past three weeks I've gotten five PoS rewards in total. If you used to get 2-3 rewards per day with 6400 coins then you were lucky. |
|
|
|
|
10 c/kwh for the first 600 kwh per month, then 12 c/kwh. Washington.
|
|
|
|
|
So where do you guys buy replacement fans? It seems to be quite hard to find an exact match for the existing GPU fan.
|
|
|
|
Dont POS mining. I have 615 coins for Sigil-wallets, but in the wallet message: no mature coins found, staking paused. Pool http://sigilmining.com offline?? You need to wait 24 hours before coins have matured for PoS. |
|
|
|
it was 100 coins for staking right? just making sure.
Min 50, however it will take ages to find a block with 50 coins. Better put 500 if you want to get PoS rewards. |
|
|
|
|
Thank you for the bi-weekly update. Wow the new website looks great!
|
|
|
|
Hello guys, I think i have a problem with stacking. I sent 155 sigil coin to my all-24h/7d opened wallet 8 days ago. Now it is 1440 coin days. But I don't get any stacks by now. Is it OK or i did some mistake? Thank you for you help.  You're fine. The difficulty of staking is high currently. I have 2500 coins and staking weight of over 16000 coin days, yet it has been several days since I received my last pos reward. So unfortunately it might take several weeks before you receive your first pos block - unless you get lucky. |
|
|
|
Most HBM usage DOES have the RAM on it's own die.
Reference the AMD "Fury" line, and their current "Vega" line.
I'm pretty sure that's also the case for the Titan V and the Tesla "volta generation" on the Nvidia side.
It's still a bus, it's just a very WIDE bus (4096 bit IIRC?).
Sorry, I ment to say HBM is located on the same silicon. The physical connection between HBM and DDR is quite different so it's not meaningful to compare them. It's like comparing a conveyor belt to a truck - both transport goods from one place to another but that's where the similarities end. Designing a wide bus is not difficult if the physical connection distance is short; Playstation 2's GPU had a 2560 bit bus to its' 4 MB EDRAM memory over 15 years ago. On the other hand with DDR the physical connection is distance is much longer so none of the big silicon giants, Intel, AMD or Nvidia have gone for a very wide DDR bus in their consumer products. To date, HBM used in video cards is NOT on the same silicon, much less in the same package, as the GPU, though that might change when the Bristol Ridge APUs get released. It does tend to be clustered a lot closer TO the GPU though, than ram on a motherboard. HBM and GPU die are connected together via silicon interposer so technically it's all one big silicon. The interposer, HBM and GPU die are manufactured separately before they are integrated together so I'll give you that. That doesn't change the point though; it's much easier to route lots of data lanes when they don't need to though PCB, pins and sockets (like is the case with DDR). |
|
|
|
Most HBM usage DOES have the RAM on it's own die.
Reference the AMD "Fury" line, and their current "Vega" line.
I'm pretty sure that's also the case for the Titan V and the Tesla "volta generation" on the Nvidia side.
It's still a bus, it's just a very WIDE bus (4096 bit IIRC?).
Sorry, I ment to say HBM is located on the same silicon. The physical connection between HBM and DDR is quite different so it's not meaningful to compare them. It's like comparing a conveyor belt to a truck - both transport goods from one place to another but that's where the similarities end. Designing a wide bus is not difficult if the physical connection distance is short; Playstation 2's GPU had a 2560 bit bus to its' 4 MB EDRAM memory over 15 years ago. On the other hand with DDR the physical connection is distance is much longer so none of the big silicon giants, Intel, AMD or Nvidia have gone for a very wide DDR bus in their consumer products. |
|
|
|
Assuming this is a legitimate project, how does your product differ from Baikal's design?
Baikal ASIC works with X11-X15 algos, PASCAL A1 ASIC will support following algos: 1. quark 2. qubit 3. x11 4. x11-gost 5. x13 6. x14 7. x15 8. x17 9. C11 10. GROESTL 11. MYR-GROESTL 12. heavy 13. lyra2re 14. lyra2rev2 15. nist5 16. pascal 17. LBRY 18. KECCAK 19. SKEIN 20. SKEIN2 21. LUFFA 22. BLAKE 23. DECRED 24. BLAKE2S 25. BLAKE2B 26. CRYPTOLIGHT 27. CRYPTONIGHT 28. S3 29. NeoScrypt 30. SK1024 Many of these algorithms have nothing in common. To support all of them you need to build a massively parallel general purpose computing unit. Also known as GPU. So basically you claim you're building a GPU like device and that you're beating nvidia and amd in their own game? With a development team of less than 10 people? I smell bullshit. |
|
|
|
Cryptopia will be better
Check Bulwark bitcointalk page if you like. Couple weeks and there are still major wallet and deposit issues. Right now you can't trade on cryptopia and other altcoins are experiencing similar problems. Exchange also does not give an ETA on fixing problems and everyone just waits. Better off taking your business to another exchange that is trying to grow. No need to dump money to list on an exchange that is already over capacity. Agree! I don't like Cryptopia exchange, which is likely greedy one, always ask high listing fees. The problem with the exchange is its technical staffs seem not proficient enough to deal with growing demands from their customers. At the moment, I think that Coinexchange.io maybe a better option for MUNcoin. Cryptopia has had a lot of technical problems with many coins that work just fine with other exchanges. Sia coin and Signatum among others. It seems they rather focus on adding new coins with high listing fees than fixing their issues. |
|
|
|
Sure, a 128 bit quad channel bus is effectively a 512 bit bus. AMD and Nvidia have used 512 bit buses in the past, however they both have moved away from those due to high cost and high power consumption. If Bitmain's ASIC really does have a separate 512 bit bus for each core (so 18x 512 bit bus in total), then 200+ MH/s is possible. However, in that case the retail cost and the power consumption will be pretty high. I'm sure they wouldn't release anything that's worse than a comparable GPU rig but most likely it's not going to drive GPU rigs out of the market either.
Not entirely true - HBM and HBM2 are both quite a bit wider. True but HBM actually doesn't use a memory bus in the traditional sense since the memory is located on the same physical die. It's more like a huge EDRAM. So it's a completely different technology and not really comparable. It's also rather expensive. But that's not really the point, the point is putting 18 separate 512 bit memory buses on a board is expensive and I wonder if Bitmain has really gone that route and how are they planning to do it cost effectively. |
|
|
|
650mh/s with 72 DDR3 memory? Its Simply Nonsense, and I tell u why: So what we know is the RX 580 has 192Gb/s bandwidth and able to perform ~30mh/s Well if we choose the best DDR3 memory with 1866Mhz (1866000000 Hz) then we can calculate 1 ddr3 memo chips's bandwidth. It's has a 64bit lane so 1866000000*64*8/1024/1024/1024=14.9 GB/s this is 1 DDr3 memo chip's bandwidth. 14.9*72=1074 Gb/s this is the total memory bandwidth with 72 ddr3 chip. So 1074/192*30= ~168 Mh/s this could be the maximum eth mining performance with 72 ddr3 chips. Everything else is bullshit.  Well, you've got the right idea but the wrong numbers. Rx 580 @2Ghz has 256GB/s of bandwidth not 192 (R9 380 @1.5Ghz has 192). ddr3 is available in x8 and x16, and since x16 has twice the bandwidth, that's likely what Bitmain will use. Going with those numbers I come up with 250-300Mh/s. Indeed, you are right with the bandwidth of RX 580, I found that 192gb/s from a shitty site, but your formula is wrong too. I made 2 mistakes, the first is the bad bandwidth, the second is that I used 64bit bitrate for 1 chip but 1 chip has just maximum 16 bit/clock , this is refers to the x16 (or x8 or x4), 64bit is the bitrate of the full DDR3 stick with many memo chips. So we have 3x32 1Gbit chips. Let me recalculate this: 1.866*16*32*3/8/256*30= ~39.3 Mh/s Each core has 32x 1Gbit chips so you should multiply that number by 6 since there are six cores per board. |
|
|
|
I dont think people quite understand the amount of gpus are mining eth. If, and that is a big if this comes out you cant take that amount of gpus and just switch mining to another coin. Yes in theory you can but that will crush diff in a massive amount of those coins.
This works both ways. Massive number of GPUs mining Eth means that adding a few thousands ASICs wont affect the difficulty at all. Even adding a hundred thousands ASICs wont render GPU mining non-profitable. I wonder how many ASICS Bitmain can produce. There's a shortage of DDR3/4 chips so they will be limited by the number of chips they can source. |
|
|
|
Looking at the specs it seems each core has a total of 4 GB of memory. That's going to be an issue in the long run since once Ethereum reaches Epoch 384 the DAG file size will be too large to fit in 4 GB. At the current rate that's going to happen in the beginning of 2021. However, should Ethereum devs decice to speed up Epoch switches then that deadline could be hit much sooner. It seems like an easy way to render the ASIC out of date without resorting to hard forking the algorithm or switching to PoS.
They would NEVER do that because at the same time youd screw over everyone with a 3 or 4 GB GPU.... That means all 1060 3GB's, all 1050ti 4GB's and any RX 470/480/570/580 with 4GB (which is the major one here). Epochs will never be "sped up". Not saying they would or should do it but why would Eth devs care about 3/4GB GPU miners? There's still plenty of 6/8GB GPU miners to keep the network safe and decentralized. |
|
|
|
|
Looking at the specs it seems each core has a total of 4 GB of memory. That's going to be an issue in the long run since once Ethereum reaches Epoch 384 the DAG file size will be too large to fit in 4 GB. At the current rate that's going to happen in the beginning of 2021. However, should Ethereum devs decice to speed up Epoch switches then that deadline could be hit much sooner. It seems like an easy way to render the ASIC out of date without resorting to hard forking the algorithm or switching to PoS.
|
|
|
|
|
Looks interesting. What's the fastest miner for nvidia 10xx cards? Is tpruvot's ccminer still the best?
|
|
|
|
|
