Topic: Vertcoin - First Scrypt N | First Stealth Address - Privacy without mixer - page 199. (Read 1232754 times)

I stopped to tune GPUs, stopped to play with TC, I, clocking, etc... You can get even 5-10% more hashes but this shit is not stable, you can loose more time trying to figure out why the hell your rig crashed or GPU get sick. If you are hopping to other algos or your rigs are remote, it's better to run it as is. My scrypt-n settings aint special. Nice compromise, you'll almost never get hang or other headache.

I'm sorry, but you DO NOT NEED 8GB of RAM for a Windows install. I've ran Windows 7 SP1 64bit (nothing removed or slipstream) and now I'm running Windows 8.1.. Again nothing removed or slip streamed. I'm using 4 cards ( 1x 280X, 3 R9 270's) and I have maxed out my cards the best of my ability and what my cards will handle (due to poor ASIC quality, and I'm assuming voltage leakage due to GPU load.

There is no solid 100% that you need 8GB of ram, It's all He said She said. I'm telling you, If you are having issues with mining Scrypt-N, please do post in here. I've been messing with this algorithm for 3 months now, It is picky, but it isn't rocket science to calculate and figure out.

Each card, each system is different. You may get lucky and a copy+paste configuration may work for you, or you may crash your drivers, or 1 card is acting a bit off.. Why is that? When it comes to this, during production Elpida, Samsung, and Hynix don't 'cherry pick' memory chips (Which Scrypt-N is very memory intensive as many of you know) so the quality isn't 100% there.

Cherry picking memory is very popular for RAM, benchmarking, specially when D9 chips were the hit with Crucial Tracers 2x1GB sticks for synthetic benchmarking.

Anyways back on topic, what I do for every mining rig, I tell people

- Disable Aero (or set to Max Performance)
- Disable all unnecessary services that don't need to be running on your rig (If you don't know what to disable, check out http://www.blackviper.com/ You can learn a thing or two.

Voltage tolerance, voltage leakage are common issues. Just because you set your cards to 1.112V like I do for my 280X doesn't mean its gonna load at that voltage, thats idle voltage, my 280X loads at iirc 1.048V or somewhere around that area, I forget.

Clean Voltage is the most important part, what I mean is a QUALITY Power Supply. Just because a power Supply from Joe Dirt is 800 W doesn't mean its the same quality as a 800 W from Bruce Willis. (Ya I got bored)

It's the simple things you gotta take into account if your having issues getting the hash rate that you think you should be getting. Ya my 280X can get 350KH/s like others, but due to the cards ASIC quality, and voltage tolerance, I can only get 332-333KH/s. I'm perfectly fine with that.

This post was not intended to bash anyone's experience with mining, I'm just putting information out there that members may not know, and wanna help out fellow Vertans. I've noticed lots of new miners know Nothing of computer rigs, important parts and the internals of everything, they put it together and hope for the best. Gotta do research lads, and you'll be much better off.

In windows the amd libraries compile the opencl file for the cards as one big (memory)image and upload that to the cards. If you don't have enough memory one or more cards won't start. Once the miner is running it uses hardly any memory at all.

Not kidding, it won't work...I run both win and linux mining rigs. (curecoin needs win for now :/)

with win is different you know, you just open 4 chrome page and bam 2.5 gb, also when i'm mining and i reach 4gb+(because i open multiple chrome page or other things) the system tell me there is insufficient memory and i have 8gb total

maybe he reserve 4gb for the gpu

Update: IT'S WORKING 

I've successfully updated the proxypool payout script and sharelogging as well as the frontend in order to handle PLX. You can see it in action by following the link in my signature.

I just tested the transactions by sending 15 PLX to the p2pool account, and the script immediately sent me my due of 7.9 PLX. It also correctly tagged shares as paid. I can't find a block explorer for PLX so at the moment the transaction hash link is a dead-end, but the sum is correct.

Feel free to hop on if you want to triple mine but also support the p2pool. I'll try to get the code available this weekend but I'll be a bit low on time..

Really don't get this memory thing.  I have eight rigs, each with four cards, all running (linux) with 1GB.  Vertminer never uses over 250mb, including caches.  Adding unused ram isn't going to do anything.

If using Windows, get a second 4gb stick...4gb is enough for linux, barely cuts it for win.

It was already off (unchecked).
I guess I never turned it on. When I went in there it first asked me to accept license agreement, which implies I never used it previously.

Right click on desktop => AMD-Catalyst Control Center (CCC) => Performance => AMD Overdrive => remove checkbox "Enable GraphicsOverdrive" => done
Got me 3 brand new grey hairs till i found out what was wrong
Those "OverDrive" tuned the card down and "re-adjusted" my mining settings... *AAARRGGHHHhhhh...*

And as i wrote above: every vendor has its own "sweet spot"... not every time is more = more hashes

insert it into another slot or try a different one

I have a single 4Gb stick.

Looks like RAM issue. How much RAM do you have? You need about 4Gb (up to 3 GPU)
Try to swap RAM modules between slots (disconnect power cable when doing this).

I don't recall the brand offhand, could be PowerColor...
I can check when I get home.
About the AMD-CCC thingy, how/where does one disable this (and re-enable if it doesn't help)?

Changed the TC to 8192 and engine to 1040. Ran about the same, maybe 1% worse than with 1050. I don't think the TC change had any effect.
The -g 2 option killed things totally though, it's terrible with that option. One card basically is crap with this option, the other runs, but slower.

Please disable in AMD-CCC grafic-overdrive.

The GPU-core depends on the vendor of your hardware: i'm using a r9 280x from powercolor(a cheap vendor) and have a GPU-core of 970 with 330kh/s.
On Sapphire cards i'm using a gpu-core of 1050 with 350kh/s.

diff of 20kh/s => at atm price ~ 22$ per yeah diff but costs me 50 bucks less when buying new => 2 years ROI

Tried the config you sent, one GPU is fine but about the same hashrate (215KHs), the other GPU doesn't like it for some reason. Both are the same brand.

Same here.

But it looks like even if you find this pool admin, he do nothing.

I have small amount of VTC there and can not cash out.

If I did not test manual cash out I have there more. It was just before I start my other rigs there.
Of course after I find that I can not cash out, immediately I stop mining there and move my rigs to other pool.

But still I want believe that this bitcrush pool admin is not thief... because I'm not 100% sure about this pool admin nickname here I will not publish it (yet) before more deep investigations.

with my setting i get about 373kh/s, just using this in cgminer

-g 2 --thread-concurrency 8192 -I 13 with that core and mem settings

try TC 8192 instead of 8193 and 1040 core instead of 1050

Try the following, I manage to get 350kh on my 7970s/280x's on sgminer-4.1.271:

"api-allow" : "W:127.0.0.1",
"api-listen" : true,
"expiry" : "5",
"failover-only" : true,
"kernel-path" : "/usr/local/bin",
"no-submit-stale" : true,
"queue" : "0",
"scan-time" : "5",
"gpu-threads" : "2",
"gpu-engine" : "1050",
"gpu-fan" : "60",
"gpu-platform" : "0",
"gpu-memclock" : "1500",
"gpu-memdiff" : "0",
"gpu-reorder" : true,
"intensity" : "13",
"temp-target" : "80",
"temp-overheat" : "85",
"temp-cutoff" : "95",
"vectors" : "1",
"worksize" : "256",
"lookup-gap" : "2",
"shaders" : "0",
"thread-concurrency" : "8192",
"scrypt" : true,
"MinerID" : "0",
"no-pool-disable" : true