just got http://www.newegg.com/Product/Product.aspx?Item=N82E16814131706
stock setting it does 24 MH ETH and
ZEC 158.848 H
i went to edit the Bios I stopped there because it has one extra memory line I'm not to sure what to do with it goes from 1725 to 1900 then 2000. none of my other cards has the 1900 line. . so i went and down loaded a few of the red devil power color bios sense a review i found said the card is a watered down version of the red devil . it also has Samsung memory and it's temping to pull the cover to see if there is 8gb on board then unlock it . it shows 4gb with any software. it feels nice no cheap feeling .
That's messed up i paid 189 Friday and now it's 169.
settles down @ ZEC - Total Speed: 161.313 H/s, Total Shares: 330, Rejected: 2, Time: 00:52 no over clocking yet using CM with Fee on .
just asked new egg for a 20 bucks back that was point less, they gave me all kinds of exudes why not over 20 bucks lame ...
Topic: SILENTARMY v5: Zcash miner, 115 sol/s on R9 Nano, 70 sol/s on GTX 1070 - page 26. (Read 209263 times)
November 21, 2016, 04:57:59 PM
November 21, 2016, 04:40:19 PM
Is someone working on any improvements for AMD, expecially newer RX cards.
November 21, 2016, 10:44:49 AM
138 sol - already not interested ... ((
Profit below the plinth.
200+ sol on 1070... one might think
I can see there remained some sportsmen altruists)))
Profit below the plinth.
200+ sol on 1070... one might think
I can see there remained some sportsmen altruists)))
That is the reason why this thread is quite now. If you pay $0.25/kWh, it is not profitable to mine ZEC.
November 21, 2016, 10:41:04 AM
138 sol - already not interested ... ((
Profit below the plinth.
200+ sol on 1070... one might think
I can see there remained some sportsmen altruists)))
Profit below the plinth.
200+ sol on 1070... one might think
I can see there remained some sportsmen altruists)))
That is the reason why this thread is quite now. If you pay $0.25/kWh, it is not profitable to mine ZEC.
November 21, 2016, 05:33:28 AM
138 sol - already not interested ... ((
Profit below the plinth.
200+ sol on 1070... one might think
I can see there remained some sportsmen altruists)))
Profit below the plinth.
200+ sol on 1070... one might think
I can see there remained some sportsmen altruists)))
November 21, 2016, 03:53:03 AM
I pushed recent changes to my repo, including my Win32 multithreading mod:
https://github.com/zawawawa/silentarmy
New Windows binaries will be available shortly.
https://github.com/zawawawa/silentarmy
New Windows binaries will be available shortly.
is this version adding the extremal addition codes or any improved hashrate besides fixing the known bugs?
Yes. I haven't uploaded binaries yet, though. I just got new ideas for optimization. Please wait.
November 21, 2016, 03:45:39 AM
I pushed recent changes to my repo, including my Win32 multithreading mod:
https://github.com/zawawawa/silentarmy
New Windows binaries will be available shortly.
https://github.com/zawawawa/silentarmy
New Windows binaries will be available shortly.
is this version adding the extremal addition codes or any improved hashrate besides fixing the known bugs?
November 20, 2016, 07:34:29 PM
@extermal: Nice optimizations I was close to hit the 500sol/s with 2 GTX1080 and 2 GTX1070 :-D
Thanks
Thanks
November 20, 2016, 06:56:31 PM
I pushed recent changes to my repo, including my Win32 multithreading mod:
https://github.com/zawawawa/silentarmy
New Windows binaries will be available shortly.
check also this https://github.com/krnlx/silentarmy-nvmodhttps://github.com/zawawawa/silentarmy
New Windows binaries will be available shortly.
krlnx working on silentarmy too
November 20, 2016, 06:54:20 PM
I am currently getting 100-114 sol/s with RX 480. This is very nice...
102/103 here with 2080 OC.Which card do you have?
Oh, my numbers are with 4 threads per GPU, too. Multithreading seems to be working well so far.
November 20, 2016, 06:51:44 PM
I am currently getting 100-114 sol/s with RX 480. This is very nice...
102/103 here with 2080 OC.Which card do you have?
XFX Black Edition with a modded BIOS.
November 20, 2016, 06:49:10 PM
I pushed recent changes to my repo, including my Win32 multithreading mod:
https://github.com/zawawawa/silentarmy
New Windows binaries will be available shortly.
https://github.com/zawawawa/silentarmy
New Windows binaries will be available shortly.
November 20, 2016, 06:46:59 PM
I am currently getting 100-114 sol/s with RX 480. This is very nice...
102/103 here with 2080 OC.Which card do you have?
November 20, 2016, 06:40:43 PM
I am currently getting 100-114 sol/s with RX 480. This is very nice...
November 20, 2016, 06:34:04 PM
Did you make any progress with AMD?
Last release don't working on AMD, but if I found a reason, it will be same +10-15% on AMD cards. For more speedup, see my previous post.Your last release is working on RX 480 with these modifications. Thanks a bunch!
Code:
// Number of rows and slots is affected by this. 20 offers the best performance
// but occasionally misses ~1% of solutions.
#ifdef cl_nv_pragma_unroll // NVIDIA
#define NR_ROWS_LOG 16
#else
#define NR_ROWS_LOG 18
#endif
// Setting this to 1 might make SILENTARMY faster, see TROUBLESHOOTING.md
#define OPTIM_SIMPLIFY_ROUND 1
// Number of collision items to track, per thread
#ifdef cl_nv_pragma_unroll // NVIDIA
#define THREADS_PER_ROW 32
#define ROWS_PER_WORKGROUP (64/THREADS_PER_ROW)
#define LDS_COLL_SIZE (NR_SLOTS * 24 * (64 / THREADS_PER_ROW))
#else
#define THREADS_PER_ROW 8
#define ROWS_PER_WORKGROUP (64/THREADS_PER_ROW)
#define LDS_COLL_SIZE (NR_SLOTS * 8 * (64 / THREADS_PER_ROW))
#endif
// but occasionally misses ~1% of solutions.
#ifdef cl_nv_pragma_unroll // NVIDIA
#define NR_ROWS_LOG 16
#else
#define NR_ROWS_LOG 18
#endif
// Setting this to 1 might make SILENTARMY faster, see TROUBLESHOOTING.md
#define OPTIM_SIMPLIFY_ROUND 1
// Number of collision items to track, per thread
#ifdef cl_nv_pragma_unroll // NVIDIA
#define THREADS_PER_ROW 32
#define ROWS_PER_WORKGROUP (64/THREADS_PER_ROW)
#define LDS_COLL_SIZE (NR_SLOTS * 24 * (64 / THREADS_PER_ROW))
#else
#define THREADS_PER_ROW 8
#define ROWS_PER_WORKGROUP (64/THREADS_PER_ROW)
#define LDS_COLL_SIZE (NR_SLOTS * 8 * (64 / THREADS_PER_ROW))
#endif
November 20, 2016, 06:19:33 PM
Did you make any progress with AMD?
Last release don't working on AMD, but if I found a reason, it will be same +10-15% on AMD cards. For more speedup, see my previous post. November 20, 2016, 06:13:02 PM
Speedup +10-15% for NVIdia only:
http://coinsforall.io/distr/nvidia/input.cl
http://coinsforall.io/distr/nvidia/param.h
Sorry, but can't work more than 1 hour a day on miner now.
http://coinsforall.io/distr/nvidia/input.cl
http://coinsforall.io/distr/nvidia/param.h
Sorry, but can't work more than 1 hour a day on miner now.
No prob, thank you for your contributions!
November 20, 2016, 06:09:08 PM
Speedup +10-15% for NVIdia only:
http://coinsforall.io/distr/nvidia/input.cl
http://coinsforall.io/distr/nvidia/param.h
Sorry, but can't work more than 1 hour a day on miner now.
http://coinsforall.io/distr/nvidia/input.cl
http://coinsforall.io/distr/nvidia/param.h
Sorry, but can't work more than 1 hour a day on miner now.
No prob, thank you for your contributions!
November 20, 2016, 06:04:06 PM
Speedup +10-15% for NVIdia only:
http://coinsforall.io/distr/nvidia/input.cl
http://coinsforall.io/distr/nvidia/param.h
Sorry, but can't work more than 1 hour a day on miner now.
For other developers, you need:
- Decrease NR_ROWS_LOG to 13 or 12. ht_store function works much faster with low NR_ROWS values and when you decrease NR_ROWS, you also decrease total slots amount, because you can use less values for OVERHEAD constant.
- Optimize equihash round for big NR_SLOTS values. I begin do it in last NVidia release, but need much more work..
http://coinsforall.io/distr/nvidia/input.cl
http://coinsforall.io/distr/nvidia/param.h
Sorry, but can't work more than 1 hour a day on miner now.
For other developers, you need:
- Decrease NR_ROWS_LOG to 13 or 12. ht_store function works much faster with low NR_ROWS values and when you decrease NR_ROWS, you also decrease total slots amount, because you can use less values for OVERHEAD constant.
- Optimize equihash round for big NR_SLOTS values. I begin do it in last NVidia release, but need much more work..
November 20, 2016, 05:57:52 PM
I have been tweaking disassembled GCN codes of SA's kernels, and there seems to be quite a bit of room for performance enhancements, especially by optimizing global memory access by reordering flat_store_dword and s_waitcnt in ht_store(). @eXtremal, how are your next batch of optimizations coming along? If they are almost ready, I will wait for them. Otherwise, I will optimize the OpenCL kernel myself and then tweak the GCN code.
xor_and_store and ht_store must be rewrited, and joined to one function.
unaligned 32 bits reads in xor_and_store -> join in 64bit in half_aligned_long -> 64bit xor in xor_and_store -> on 2,4,6,8 round 256bit shift on xi0xi1xi2xi3 in xor_and_store -> 256bit shift again in ht_store -> split in 32bit, and write in ht_store
must be rewrited to:
unaligned 32 bits reads - > 32 bit xor -> 256bit shift -> 32 or 64 bit, or vector store
or
64 bits reads -> 64 bit xor -> 64 bit 256bit shift -> 64bit or vector store
or
64 and 32 bit reads -> 64 and 32 bit xor -> mixed 256bit shift -> 64bit or 32bit or vector store
depend on round
Excellent suggestions! Let me get to them ASAP.
Jump to: