That caught my attention. I was thinking Ryzen was pushing 2 channel with too many threads
but apparently not.

The only algos I can think of that are memory bottlenecked are srcyptn2 and lyra2z330.

There's no competition from Intel up there. The 3900x and below have recently dropped in price.
3950 still too high.

You haven't addressed the claim there either. Simple denials mean nothing.

Multiple users have reported it so it's not just one.

Now why didn't I think of that?

It is not necessary to reserve a CPU thread for GPU mining as long as you don't raise the
priority of the CPU miner.

When dual mining the GPU miner process can run at a slightly higher priority than the CPU minier
to reduce scheduling delays. But in my experience I rarely see any impact to GPU mining performance
when a CPU miner is also running full throttle. It depends on both the GPU and CPU algos being mined.

It's never a good idea to increase the priority of CPU miners.

Now to get back on topic, I'm looking forward to seeing some result for the 3990x.

Hello Pooler,

I'm not sure you read this thread anymore but you should know your scrypt CPU code is getting some
good use due to Verium which uses N = 1024**2 making it exclusive to CPU mining only.

It's currently optimized up to AVX2 but AVX512 is now available. Unfortunately I don't have
the ASM skills necessary or the understanding of your code to attempt promoting it to AVX512.

There's a big team behind Verium, they may welcome the opprtunity for a more efficient miner.

FYI:

scryptn2 == scrypt:1048576

An FYI for those that noticed the addition of scryptn2 to the pool.

scryptn2 is just scrypt with N = 1024**2 and has been available in cpuminer-opt for a long time:

I just released cpuminer-opt-3.12.2 which fixes xevan skein & skein2 but I overlooked skunk. It will be
in the next release. It also uses skein as the first funtion in the chain so it's the same issue.

If it's a farm it usually means ASICs. ASICs are the biggest risk for several reasons,
including HW quality.

- limited flexibility, usually only one algo supported per device

- short lifespan, ASIC miners become obsolete very fast with no secondary usage (unlike GPUs)

- there are a lot of unethical ASIC peddlers that sell crap, provide no support or are just scammers.

- International borders, most ASICS come from China exacerbating other issues.

- market volatility, 2018, need I say more.

A mixed farm (ASIC, GPU, CPU) would mitigate some of those risks.

Looks very similar to https://github.com/JayDDee/cpuminer-opt/issues/238, which I'm working on as I write.
I'll make sure to test Skein & skein2 as well. I've been doing a lot of testing lately, There was more damage from
the restructuring in the past couple of months than I realized.

I don't test avx-sha because you miss out on 4 way parallel hashing on many algos, including skein.
Going from 4 way AVX2 to 1 way even slows down Ryzen.

Most Lyra2 based algos are ok with AVX because they can still do 4 way (and 16 way AVX512),
but most other algos require AVX2.

You should try both AVX & AVX2 to find the best for each algo.

As far as SHA goes there is a crossover between HW sha256 and parallel SW sha256. With the current situation
HW SHA is used up to and including AVX2, and parallel SW AVX512. When AMD improves AVX2 that may change.

Edit: Yup. Same problem, same fix.

Mining scrypt with a GPU is obsolete, it's all ASICs. I hope your friend realizes that.

Cudaminer is also obsolete, replaced by ccminer (open source) and a few fee based
closed source miners.

As far as pool mining, you get the instructions from the pool, the're all different.

There are a lot of tutorials around, even videos to help you get started.

If you want to help your friend, mining scrypt with a GPU is like filling a swimming pool
with a spoon.

If you want to do some real mining with your 1660s, you need to do some research
to find which algos are suitable for GPU mining. Again there's lot's of info around if you look.

Neoscrypt hasn't had a lot attention in the past couple of years. The last significant update
to neoscrypt for ccminer was for Cuda 5 GTX-9xx. You'll probably have to do your own research about
newer cards.

If you find something interesting you can share it with evryone.

Increasing the displayed hashrate 10% is as simple as adding one line of code.


There are other threads discussing that piece of trash.

Excellent info, very informative. I don't have any Xeons but it was still quite interesting.

Thanks for identifying the 256K L2 dependency, I wasn't previously aware of it.

With improvements to turbo boost, boosting more cores, the benefits of mining with
N/2 threads is increased.

And you're the expert because you tried it for a couple of months and HDDs didn't die?

You lose credibilty when you claim to have knowledge which you fail to share just to try to look smarter than everyone else.
If you're so smart exlain why HDD mining doesn't accelerate aging.

That's crap. Hard drives are mechanical devices subject to physical wear. It will age your HDD faster,
raising the probability of failure. There's a reason HDDs spin down whenever they can, it's to lengthen
their life.

It's a good thing the cache is 256 MB. With only 4 memory channels to serve
128 threads it's best to avoid main memory.

I wonder how Intel will respoond, they currently have nothing to compete with TR.

Was it ever worth it?

I may be biased but I never took it seriously. Doesn't it wear out the drive?
Hard drives and SSDs have a limited lifetime that is shortened by heavy use.

It looks like you are the lucky one, all the other Pi spammers have been shut down.

Who are these PhD's? What are their names? What papers have they published?

Have you caught any fish yet?