Running AWS but 0% CPU usage

Weird. Any other differences, or otherwise identical?

To compile, I just used the instructions in the beta guide. Didn’t change anything.
git clone GitHub - zcash/zcash: Zcash - Internet Money
cd zcash
git checkout v1.0.0-beta1
./zcutil/ -j$(nproc)

only differance is my 3rd system that wont connect is 8core AMD and my other is 6core AMD

oh right. totally beyond my expertise, sorry. I’ll let an expert answer :wink: There are plenty on this forum. I can’t see any obvious reason. Any issues I’d had with compiling have been ultimately due to low RAM

No other differences for me. Total of 6 machines, all i5, 4GB RAM, 80GB hdd. 2 are running properly. A couple are on block 1 or 0, so they might not be connecting. You said that you had an issue with <8GB RAM, whereas I have 4, but it seems to work on 2 of the machines.

Yeah, I’ve had memory-related errors with 4GB RAM, but then I sucessfully compiled with Ubuntu running on my Mac with 4GB. And I’ve heard of others too, so I’ll have to say “I don’t know” :frowning:

Weird. Mine seems to be compiling okay. I even had it running on one machine, recompiled and now it doesn’t work - same machine, nothing else changed. Thanks anyway!

wierd. there is definately some kind of connection issue if im not the only one who has multiple machines and only half of them are working. can we get any devs or experts in here to maybe check the thread out?

let’s start by asking @Shawn because he’s already been involved in this thread, and can advise who else to contact.

just reset all 3 rigs, checked config files, the first 2 booted up and got a connection immediately. the third and same computer just wont get a connection, so i will be trying to re-compile shortly and let you know if i can get a connection after

It sounds like this issue which is inherited from the upstream Bitcoin code

so, all the best benchmarks have a high end i7 processor, but if the algo is mainly memory bandwidth dependant, couldnt you just use an old dual core procesor with good memory and produce near the same results?

The biggest problem with the old PC’s is the memory speed. If you can find an old PC that has a motherboard that will do DDR3 1600, for cheap it may be marginally cost effective. I had an old PC with a core2duo I benchmarked but it was pitifully slow because it had 600mhz memory, it will never have a chance to find a block.

I still don’t understand why greenCube’s i7-6700K 4GHz with DDR4-2400MHz is slower than gareth’s i7-6700K with half as much DDR4, and at 2133MHz. Just chance, I guess. Statistical noise.

so all my amd rigs, which happen to use ddr3 1600, have not found blocks, while my intel has found a block nearly every 3-4 hours with ddr4 2133. so essentially, you need extremely fast memory, and the cheapest possible combo of everything else that can match that to have an optimal rig? looks like AMD may be screwed on this end

after 1400 blocks the intel rig has 12 blocks and amd only 5. i wish i knew why >.<

@shawn, I’ve built a google spreadsheet that automatically accepts data from google’s form - much easier to enter and view benchmark data:

This is the link to fill out the form if you have new data to add:

This is the link the view the data:

Both links are broken, 404, perhaps I need to open them from inside Google spreadsheets?

I had discussed embedding a Google form into the wiki before but when you do it only allows embed of the static viewable table, and if you want to edit it you have to sign in to your Google account. Which is easy if you have a Google account but does not allow for anonymous contributors to the document.

I’ve tried using direct links this time(edited the last message), without google shortner.
hope these last.

I also tried logging in without a google account and it works fine

Ok cool, I’ll see what I can do to embed it into the wiki when I have time. Or if you have time feel free to give it a go.