GPU miner is working.
Need more testing and tweaking still.
Need more testing and tweaking still.
Where can I download GPU miner?
Topic: NSGminer v0.9.4: The Fastest NeoScrypt GPU Miner - page 31. (Read 221777 times)
Where can I download GPU miner?
Please keep us informed on temperatures!
GPU miner is working.
Need more testing and tweaking still.
Need more testing and tweaking still.
First I'm hearing of NeoScrypt. Sounds interesting and could have a future, especially when the Scrypt ASIC miners start to come out.
As far as I know Scrypt ASIC miners are everywhere already...
First generation of ASIC scrypt miners are already obsolete as the latest generation of ASICS have moved the first generator to cost more to run than they make if your electricity costs more than $0.15 to $0.20 a kW, so they are not only everywhere, but the 'arms race' like is happening on bitcoin is now happening with Scrypt
First I'm hearing of NeoScrypt. Sounds interesting and could have a future, especially when the Scrypt ASIC miners start to come out.
As far as I know Scrypt ASIC miners are everywhere already...
This miner is not well optimized, is giving half hash rate for scrypt and sha256d than with Pooler CPU Miner.
This one is pulling only 65 Watts, neoscrypt is worst with 58 Watts. Pooler CPU Miner pulls 77 Watts with scrypt.
This one is pulling only 65 Watts, neoscrypt is worst with 58 Watts. Pooler CPU Miner pulls 77 Watts with scrypt.
First of all its not even live for 1 month, so optimization will follow automaticly, next thing is that CPU isnt the maintarget, GPU should enter this terrain soon.
First I'm hearing of NeoScrypt. Sounds interesting and could have a future, especially when the Scrypt ASIC miners start to come out.
This miner is not well optimized, is giving half hash rate for scrypt and sha256d than with Pooler CPU Miner.
This one is pulling only 65 Watts, neoscrypt is worst with 58 Watts. Pooler CPU Miner pulls 77 Watts with scrypt.
This one is pulling only 65 Watts, neoscrypt is worst with 58 Watts. Pooler CPU Miner pulls 77 Watts with scrypt.
Arithmetic intensity, also known as ALU:TEX ratio... anyhow if it's not for GPUs then I guess this is the future of CPU mining?
I am very confused and I cannot make any sense of what you just wrote.
I am very confused and I cannot make any sense of what you just wrote.
Well, it's still a decent choice for GPUs. Lower memory requirements allow to run more threads in parallel to work around access latency. I cannot tell how good it will be, but I expect NeoScrypt CPU and GPU miners to co-exist for some time. I don't expect ASICs soon because NeoScrypt is unlikely to be implemented as a by-product in SHA-256 or Scrypt devices due to different hardware requirements. SHA-256 ones are very simple high speed ALUs with next to nothing memory requirements. Modify the ALUs and add some eDRAM = universal SHA-256/Scrypt ASIC. I doubt their designers would like to modify the ALUs even more and include fast SRAM to satisfy NeoScrypt.
Quote
My Cuckoo Cycle PoW spends 67% of its time on memory access. In other words, if you leave out all memory accesses, and just do the remaining computations, then runtime reduces to 33% of original.
That's what I would call "memory intensive".
And I'd fully agree with you tromp! By contrast, original scrypt is compute bound and neoscrypt seems to be even more so. That's what I would call "memory intensive".
Arithmetic intensity, also known as ALU:TEX ratio... anyhow if it's not for GPUs then I guess this is the future of CPU mining?
I am very confused and I cannot make any sense of what you just wrote.
I am very confused and I cannot make any sense of what you just wrote.
Why don't you look at the documentation on GPU programming?
Original scrypt was already compute bound. Neoscrypt is even more compute bound. People in GPU programming would go great lengths to have half the arithmetic intensity original scrypt had.
I'll ask you again: what does "memory intensive" mean? Fetching some memory every once in a while isn't an "intensive" operation.
Original scrypt was already compute bound. Neoscrypt is even more compute bound. People in GPU programming would go great lengths to have half the arithmetic intensity original scrypt had.
I'll ask you again: what does "memory intensive" mean? Fetching some memory every once in a while isn't an "intensive" operation.
What am I supposed to do with your documentation on GPU programming? NeoScrypt has no sole purpose of running perfectly on GPUs either existing or to come. Neither Scrypt had it originally. NeoScrypt performs very well on CPUs now and it's going to be even better with the following optimised releases of CPUminer. It takes full advantage of large and fast L2 caches. This is something you don't find in GPUs and ASICs because high clock and low access latency multiported synchronous SRAM is very expensive. GPUs do very limited caching at all. They rely upon GDDR5 memory bandwidth and large thread count because access latency is terrible. Scrypt ASICs must be using eDRAM which is more expensive than GDDR5 and faster due to lower latency and wider internal data bus, but it's DRAM still and doesn't clock very well. Intel Crystalwell has made it to 1.6GHz which is absolutely not impressive if compared to ~4GHz of modern CPUs with their L1 and L2 caches. Even single ported asynchronous SRAM used for L3 caches seems better than eDRAM for performance reasons. Let's see how GDDR5 and eDRAM work against SRAM under NeoScrypt.
What comes to memory intensity, you can calculate how many quarters it takes to produce a single hash for Scrypt and NeoScrypt.
Getting following error when trying to run the miner on PiMP:
[2014-08-19 17:04:10] Stratum requested work restart
Segmentation fault
Thoughts?
[2014-08-19 17:04:10] Stratum requested work restart
Segmentation fault
Thoughts?
Why don't you look at the documentation on GPU programming?
Original scrypt was already compute bound. Neoscrypt is even more compute bound. People in GPU programming would go great lengths to have half the arithmetic intensity original scrypt had.
I'll ask you again: what does "memory intensive" mean? Fetching some memory every once in a while isn't an "intensive" operation.
Original scrypt was already compute bound. Neoscrypt is even more compute bound. People in GPU programming would go great lengths to have half the arithmetic intensity original scrypt had.
I'll ask you again: what does "memory intensive" mean? Fetching some memory every once in a while isn't an "intensive" operation.
My Cuckoo Cycle PoW spends 67% of its time on memory access. In other words, if you leave out all memory accesses, and just do the remaining computations, then runtime reduces to 33% of original.
That's what I would call "memory intensive".
You can do the same test for NeoScypt. Run it with all memory accesses removed, and see how much runtime is reduced.
Why don't you look at the documentation on GPU programming?
Original scrypt was already compute bound. Neoscrypt is even more compute bound. People in GPU programming would go great lengths to have half the arithmetic intensity original scrypt had.
I'll ask you again: what does "memory intensive" mean? Fetching some memory every once in a while isn't an "intensive" operation.
Original scrypt was already compute bound. Neoscrypt is even more compute bound. People in GPU programming would go great lengths to have half the arithmetic intensity original scrypt had.
I'll ask you again: what does "memory intensive" mean? Fetching some memory every once in a while isn't an "intensive" operation.
Considering the massive amount of computation, what does "memory intensive" mean?
Memory read/write operations. Why don't you look at the source code?
Considering the massive amount of computation, what does "memory intensive" mean?
I have a lot of trouble understanding.
Claimed to be harder on memory, takes 20 rounds of salsa (already too much to be memory-bound) adding other 20 rounds of chacha?
Half the memory amount, paper claims to be "1.25 times more memory intensive"... due to cache effects I assume?
Anyway, scrypt was compute bound... I've been looking at this for a few days and I cannot understand what's the whole point.
Claimed to be harder on memory, takes 20 rounds of salsa (already too much to be memory-bound) adding other 20 rounds of chacha?
Half the memory amount, paper claims to be "1.25 times more memory intensive"... due to cache effects I assume?
Anyway, scrypt was compute bound... I've been looking at this for a few days and I cannot understand what's the whole point.
What's your question?
I have a lot of trouble understanding.
Claimed to be harder on memory, takes 20 rounds of salsa (already too much to be memory-bound) adding other 20 rounds of chacha?
Half the memory amount, paper claims to be "1.25 times more memory intensive"... due to cache effects I assume?
Anyway, scrypt was compute bound... I've been looking at this for a few days and I cannot understand what's the whole point.
Claimed to be harder on memory, takes 20 rounds of salsa (already too much to be memory-bound) adding other 20 rounds of chacha?
Half the memory amount, paper claims to be "1.25 times more memory intensive"... due to cache effects I assume?
Anyway, scrypt was compute bound... I've been looking at this for a few days and I cannot understand what's the whole point.
Any pool to mine PXC? bcoz I see all p2p either dead or outdated
MinePool - no neoscrypt support
Multipool - Abondoning PXC
so any pool left to mine?
MinePool - no neoscrypt support
Multipool - Abondoning PXC
so any pool left to mine?
http://pxc.theblocksfactory.com
Jump to: