The Altcoin Thread

I couldn't stand it any more so I took it offline and put my emergency 7970 spare online instead.

Running the power supply over capacity but in the cooler weather it's probably not working as hard

Now to take apart this powercolor and see WTF it's problem is.
 
well there is a fine coat of dust on everything so that cannot help

and the thermal paste is just nasty, overflowing the sides and just a splotch in the middle

so should as5 be applied as a fine coat across the whole gpu like old cpus or should it be applied as just a dot in the middle like new cpus ?
 
finally got all the nasty paste off, check out that fine fine lapping (not)

EJ0olJn.jpg


So the gpu has a lid that is like glass but the heatsink has nasty grooves. That cannot work well.
 
Another thing, the paste was covering all those transistors (I think?) that surround the gpu without any heatsink contact on them.

That cannot be good for power transfer.

Note most of the paste was stuck to the heatsink side which leads me to believe these were the only points on the gpu actually making contact, or opposite (inverse) contact

Kuwladk.jpg



playah said:
On gpu's I've always used the thin spread technique
Click to expand...

Should I coat the glass-like surface of the gpu, or the uneven heatsink to try to make it even?
 
To my poor eyes those appear to be discrete resistors and capacitors, so no worry there.
My opinion would be to coat the HS to try and get as much contact into the voids as possible.
May not happen if you depend on chip contact alone.
 
Now that is a nice build http://imgur.com/a/IISox

Though if that rack is metal, I would be too nervous to use it.

I think I discovered the ultimate open-air cheap rig holder better than file crates - shoe racks.

If only there was a way to suspend a motherboard from this without it being upside down

http://www.walmart.com/ip/Mainstays-Plastic-Shoe-Rack/16349547

But you could just put it on a table or floor if you had a safe area.

I have to use a crate because of cats.
 
Lol... *I* finally found a block on my pool. As in me. Not someone else.

Of course it was bound to happen, since I'm one of only two people hashing on it right now. :(
 
SunnyD said:
Lol... *I* finally found a block on my pool. As in me. Not someone else.

Of course it was bound to happen, since I'm one of only two people hashing on it right now. :(
Click to expand...

Remind me when I get home tonight after class and I'll move the 2.5Mh area over to it. I jumped over to DOGE last night and haven't had a chance to move away from it.

Anybody on skype? my username is farkedup or google chat [email protected]
 
vibe said:
Now that is a nice build http://imgur.com/a/IISox

Though if that rack is metal, I would be too nervous to use it.

I think I discovered the ultimate open-air cheap rig holder better than file crates - shoe racks.

If only there was a way to suspend a motherboard from this without it being upside down

http://www.walmart.com/ip/Mainstays-Plastic-Shoe-Rack/16349547

But you could just put it on a table or floor if you had a safe area.

I have to use a crate because of cats.
Click to expand...

I'm hanging mine with zip ties but inserting things like this on my rack would probably work for holding the cards a bit cleaner than zip ties because no matter what I do with those ties I ended up really crooked. I ordered Xmas light staples to maybe line the zip ties up better and measure things out but right now I wish I had a bit more heigh between my racks.
 
mkrohn said:
i have an asus 280x where 1 blade randomly flew offf that I noticed yesterday and its creating a vibration. I'd RMA it but do I really wanna lose 750k?
Click to expand...

I've heard of company's just shipping the fan assembly to you rather than having your RMA the whole card.
 
_r2h said:
I've heard of company's just shipping the fan assembly to you rather than having your RMA the whole card.
Click to expand...

PowerColor "offered" to replace the card for $45. This is a brand new card, less than 2 weeks old, but since the fan is broken, "out of warranty".
I have to RMA the card, pay the $45 via cashiers check or money order, ship the card, wait for it to get to them, then wait for them to ship the new one back.
This is what I wrote back:
Raymond,
These terms are ridiculous.
So, let’s see. I send a brand new defective card plus money to you and wait to receive a replacement for who knows how long?
No thanks.
I’ll just determine a way to repair it myself with a decent fan and then be sure I never purchase a PowerColor product again.
And I will be sure to NOT recommend PowerColor to anyone of the many people I know who buy video cards.
Thanks so much,
Click to expand...

Edit: Man. What did I do to make something in quotes look like that!
 
Ya those terms are stupid. I don't buy Powercolor, but because the name is ridiculous though, hah.
 
Core32 said:
PowerColor "offered" to replace the card for $45. This is a brand new card, less than 2 weeks old, but since the fan is broken, "out of warranty".
I have to RMA the card, pay the $45 via cashiers check or money order, ship the card, wait for it to get to them, then wait for them to ship the new one back.
Click to expand...

If you bought that less than 30 days ago from a major retailer like newegg, you might do better negotiating with them. Just make sure you are getting a replacement and not a refund if the price has gone up.

Also on newegg if you leave negative feedback about the manufacturer, they magically show up and make better offers on customer service.
 
Unfortunately I bought these NIB from a seller here.
They were in perfect condition with the box still wrapped in plastic.
I found a fan replacement, same exact one, but it's about $25.
Finding 92mm frameless fans can be difficult...
 
litecoin is about to have a major difficulty correction (increase) sometime tomorrow

last chance to mine some "cheap" litecoin, might never be this low again, unless the price doesn't go over $25 ever again
 
Yep, just got my 3rd 7970 online and gonna let them go on litecoin till tomorrow. Will probably switch back to trademybit once diff goes up.

But I have a feeling it'll come back down a little again. Thing is, I guess I'm one of those still hoping ltc will have another huge spike in price so even with the diff increase, it might still be worth it.
 
vibe said:
litecoin is about to have a major difficulty correction (increase) sometime tomorrow

last chance to mine some "cheap" litecoin, might never be this low again, unless the price doesn't go over $25 ever again
Click to expand...

Who knows, the difficulty may drop again... DOGE has made a bit of an increase in value and if it continues upward, could cause another mass exodus from other coins to DOGE.
 
dogecoin is unproven until after it halves in 15 days and then again 45 days from now

I am curious to see if people will still mine it when they can only make 25% the coins they can now each day

because if the value stays the same at one fourth the coins, it will be not profitable
 
vibe said:
dogecoin is unproven until after it halves in 15 days and then again 45 days from now

I am curious to see if people will still mine it when they can only make 25% the coins they can now each day

because if the value stays the same at one fourth the coins, it will be not profitable
Click to expand...

I'm mining DOGE at least until the first block split, maybe after too. I think as stupid as DOGE is, it's going to stick around for a while. It's actually getting used, tipping etc. plus there's so many of them... and there is something in the human psyche that likes getting 100 of something vs. 0.0002 of something.,

If it's not profitable with the coin reward halving, then a couple things could happen
1) the hash rate will go down = difficulty will go down until it's profitable or break even.
2) price will rise if demand is there.
3) the chain stalls and DOGE dies, not likely since it's a 4 hour re-target and people sleep.
4) a mix of 1+2.
 
canna said:
... and there is something in the human psyche that likes getting 100 of something vs. 0.0002 of something.
Click to expand...
As much as I hate DOGE, I think you're right on this. Which is why it's good for tipping. Both parties feel like they've gotten or given more than they actually have.

The only other thing I like about DOGE is that it's pulled so much having power away from coins that I personally feel will have a much more stable existence in the future.
 
There seems to be a buzz around Dogecoin that is distinctly lacking with Litecoin and other altcoins. I feel like it's going to stick around, even with the silly name.
 
I'm stuck between mining and hoarding a bunch of DOGE before the split or mining the DOGE and dumping it for WDC (more efficient than mining WDC by itself) and holding on to that for the new exchange next month.

There really is a lot to playing these markets that you don't realize at first. Kinda progresses like this:

1) Oh I have a video card in my gaming computer, you mean I can mine some coins with it while I'm not playing BF4? Neat, I'll just do that.
2) Hmm, maybe I can get another video card and stick it in here to get my hash rate up.
3) WHERE CAN I CRAM MORE VIDEO CARDS IN MY HOUSE?!
 
So my new obsession is WU (work utility) which is what your rig is actually achieving vs the fake kh/s which is meaningless

So after much tuning

4 x 7950 @ only 1020mhz

2.566 mh/s = 2420-2440 WU

that is 2420/2566 = 94 percent efficiency

beat that!

See, mhz and raw kh/s doesn't matter. It's WU

let's see what screenshots I can find in this thread to compare... anyone have four 7950 ?
 
Well this is annoying.

Last night for shits and giggles I decided to open up the brand new 7950 that's been sitting in my closet for a month. It required me to upgrade the PSU that was sitting in my ITX box, but whatever. Put in the PSU, but in the 7950, downloaded drivers, installed. Fired up cgminer, it complained because I forgot I was still on the IGP. Oops. Move the monitor over to the 7950, fire up cgminer, it's hashing, but the hash rate is jumping around all over the place and I'm noticing HW errors piling up. Figured I'd tweak the settings a little...

First check AMD Overdrive - it's not even there. Oookay, whatever.
Next, change my conf file:
Change gpu threads from 2 to 1, change concurrency to something reasonable to match like 24000. At this point cgminer says it can't start the gpu and bails. Wut?

I twiddle a few more things and whatnot, still no luck. Worse yet, a couple times I even manage to crash the machine. I give up for the night.

The kicker, I boot the machine this morning, no post. Huh? Move the monitor over to the IGP, looks like Windows failed to boot and wants to know if I want to start normally or safe mode. Joy. I hope I don't have a bad card. I'll have to try it in my other rig to be sure. I wouldn't put it past not liking this ITX machine though, this ITX board is... different... that's for sure.
 
TC is very fickle and will vary drastically from machine to machine. It is very dependant on the amount of RAM you have in a system. Always start your mining at 8192, get it stable and WRITE DOWN your settings then do the threads= and see what the system thinks your TC should be.
 
mkrohn said:
TC is very fickle and will vary drastically from machine to machine. It is very dependant on the amount of RAM you have in a system. Always start your mining at 8192, get it stable and WRITE DOWN your settings then do the threads= and see what the system thinks your TC should be.
Click to expand...

Like I said... -g2 -tc 8192 worked, but I was getting HW errors racking up fairly quickly as the hash rate jumped around erratically. I figured I'd change to -g1 and see what happened, and as soon as I did cgminer "couldn't start the gpu".

Just the fact that I was getting a HW error count at all was disconcerting. Worst case is I'm thinking since it's so cold and dry right now, I could have inadvertently shocked one of the VRAM chips or something while getting everything put together last night. Now THAT would suck.
 
You don't need two threads on a 7950, it is counter productive.

Remember if you change threads. TC settings and such you need to delete the *.bin files in your cgminer directory or the changes won't take effect.

Your 12v may be sagging too much if it is a small psu which will cause HW errors?

Try this to start with

1020mhz gpu
1250mhz ram
TC 24064

You should get 610-630 kh/sec out of that with lower power draw.
 
vibe said:
You don't need two threads on a 7950, it is counter productive.

Remember if you change threads. TC settings and such you need to delete the *.bin files in your cgminer directory or the changes won't take effect.

Your 12v may be sagging too much if it is a small psu which will cause HW errors?

Try this to start with

1020mhz gpu
1250mhz ram
TC 24064

You should get 610-630 kh/sec out of that with lower power draw.
Click to expand...

Which is why I instantly tried going to -g 1. And if I remember right, I did delete the .bin's. No matter what I did though after I changed to 1 gpu thread cgminer refused to run. I tried concurrencies of 28000, 24000, 17920, 16000... it just refused to work.

I have have a barebones conf file, none of the GPU settings for clock speeds, etc are in there. The PSU is a ThermalTake ToughPower 700w psu - it should be fine for the job, though it is a used PSU I picked up locally from a fellow H member.

Hash rates were settling mostly 450-550khash range and occasionally jumping to around 700khash. WU on the other hand was horrible, down in the 110 range.
 
SunnyD said:
Which is why I instantly tried going to -g 1. And if I remember right, I did delete the .bin's. No matter what I did though after I changed to 1 gpu thread cgminer refused to run. I tried concurrencies of 28000, 24000, 17920, 16000... it just refused to work.

I have have a barebones conf file, none of the GPU settings for clock speeds, etc are in there. The PSU is a ThermalTake ToughPower 700w psu - it should be fine for the job, though it is a used PSU I picked up locally from a fellow H member.

Hash rates were settling mostly 450-550khash range and occasionally jumping to around 700khash. WU on the other hand was horrible, down in the 110 range.
Click to expand...

Open a cmd window w/admin privileges then type this in.

setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS 1

I had the same problem with a 270x card I installed last week. I added it to a machine that was already mining with 2x 6970 cards and working fine, but the brand new 270 card did nothing but give me tons of HW errors and wouldn't allow me to use a TC higher than 8192. It drove me crazy for about 2 hours before I stumbled across that on a messageboard somewhere. I really thought I got a lemon card that wasn't working properly, but after I added those commands everything was fine and the card ran like a champ.
 
Pitbully said:
Open a cmd window w/admin privileges then type this in.

setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS 1

I had the same problem with a 270x card I installed last week. I added it to a machine that was already mining with 2x 6970 cards and working fine, but the brand new 270 card did nothing but give me tons of HW errors and wouldn't allow me to use a TC higher than 8192. It drove me crazy for about 2 hours before I stumbled across that on a messageboard somewhere. I really thought I got a lemon card that wasn't working properly, but after I added those commands everything was fine and the card ran like a champ.
Click to expand...

Already done - was one of the first things that I did.

Like I said, the newest symptom as of this morning was the machine crashes as soon as I plug the monitor cable into the card. I'm thinking something isn't happy. Shame I won't get to mess around with it until this evening.
 
SunnyD said:
Already done - was one of the first things that I did.

Like I said, the newest symptom as of this morning was the machine crashes as soon as I plug the monitor cable into the card. I'm thinking something isn't happy. Shame I won't get to mess around with it until this evening.
Click to expand...

BIOS setting? Think there's something about which video to default to onboard or pcie.

Just tossing thoughts out. I've never had an issue I think auto just finds my pcie
 
SunnyD said:
Well this is annoying.

Last night for shits and giggles I decided to open up the brand new 7950 that's been sitting in my closet for a month. It required me to upgrade the PSU that was sitting in my ITX box, but whatever. Put in the PSU, but in the 7950, downloaded drivers, installed. Fired up cgminer, it complained because I forgot I was still on the IGP. Oops. Move the monitor over to the 7950, fire up cgminer, it's hashing, but the hash rate is jumping around all over the place and I'm noticing HW errors piling up. Figured I'd tweak the settings a little...

First check AMD Overdrive - it's not even there. Oookay, whatever.
Next, change my conf file:
Change gpu threads from 2 to 1, change concurrency to something reasonable to match like 24000. At this point cgminer says it can't start the gpu and bails. Wut?

I twiddle a few more things and whatnot, still no luck. Worse yet, a couple times I even manage to crash the machine. I give up for the night.

The kicker, I boot the machine this morning, no post. Huh? Move the monitor over to the IGP, looks like Windows failed to boot and wants to know if I want to start normally or safe mode. Joy. I hope I don't have a bad card. I'll have to try it in my other rig to be sure. I wouldn't put it past not liking this ITX machine though, this ITX board is... different... that's for sure.
Click to expand...

I think I had this problem before. Sometimes setx GPU_MAX_ALLOC_PERCENT 100 doesn't create all the registry settings you need. There are two ways to solve this. One, you can make a new user account and it should work. Two, you can go through that user account's registry settings, find all the instances of GPU_MAX_ALLOC_PERCENT 100 and transfer those registry settings to those of your main user account. That's what I did. When I didn't, it would not let me bump thread concurrency, and without the right thread concurrency you just get hardware errors.
 
allen5055 said:
I think I had this problem before. Sometimes setx GPU_MAX_ALLOC_PERCENT 100 doesn't create all the registry settings you need. There are two ways to solve this. One, you can make a new user account and it should work. Two, you can go through that user account's registry settings, find all the instances of GPU_MAX_ALLOC_PERCENT 100 and transfer those registry settings to those of your main user account. That's what I did. When I didn't, it would not let me bump thread concurrency, and without the right thread concurrency you just get hardware errors.
Click to expand...

You know, I think now that you mention it I may have been doing everything from a non-admin command prompt last night. That would definitely cause setx to fail and not get saved.

But that still doesn't explain why the machine just crashes now when you plug the monitor into the 7950.
 
I've had to uninstall in device manager and then reboot and reinstall drivers when I was getting some wonky output.

Still it's pretty weird about the crashing when plugging in monitor.
 
SunnyD said:
You know, I think now that you mention it I may have been doing everything from a non-admin command prompt last night. That would definitely cause setx to fail and not get saved.

But that still doesn't explain why the machine just crashes now when you plug the monitor into the 7950.
Click to expand...

Yeah, you may want to check that, also reboot when your done to get it to stick. Your problem sounds exactly like what I had with my 270 card. I literately could not get it to do a thing, until I added that from a cmd prompt. All the card would do was throw out HW errors no matter what settings I tried to mine with.
 
Well, it looks like KlondikeCoin prices have gone down far enough where it's time to move on to mining something more profitable (so that I can pay the bills).

Pool will stay open though. I will be diddling around with the server this weekend to see about getting another pool online. I'm tempted to get PotCoin up and running, but I think CASH might be next. Otherwise, any suggestions would be appreciated.
 
You must log in or register to reply here.
Back
Top