Selling FRY for DOGE, LTC, VERT.  PM offers.  Thanks

Hey all!

Well I used to have this working some months ago, though it was glitchy at best, and was using a way older version.

GPU: 1x Sapphire Vapor-X 7950
AMD Driver: 13.9
OS: Windows 7 x64
Command Line: cgminer.exe --scrypt -o stratum+tcp://server:port -u user -p pass -I13
(server:port, user, and pass have all been set in the .bat)

On the current 3.7.2, I can't even get the miner to start as I crash with this error instantly upon starting it:
libusb_init() failed err -99 [2013-11-10 04:57:24] libusb_init() failed

I did try an older 3.0.1 build and it crashes as well, no message in the console but the module that crashed is libusb-1.0.dll.  I also tried 2.7.7 which starts correctly but only will mine at around 5khash/s.

I did some searching on google but so far have come up with nothing as to where to even start troubleshooting the libusb crashing issue.  I am not using any USB FPGA/ASIC miners and the only usb devices attached are a keyboard and mouse.

Any thoughts or advice would be greatly appreciated.

I'm using 13.4 with my old 5870 and getting ~420khash/s with a smallish OC.  Settings for cgminer is --thread-concurrency 8000 -I17 -g1 -w256.  Could probably turn up intensity if you're not using it for displaying video on a monitor.

Hopefully the new board works out better.  I'd ask to have the latest BIOS flashed to it.

You _can_ use 2 different ones on the same card, its just not recommended.

But in reality, this isn't how normal business is conducted.  Microsoft, Sony, HTC, Game Developers, etc. do not raffle pre-orders.  They have a set amount of say the "limited edition", the more abundant "regular" edition, and the super rare "elite" edition, or whatever.  If I was to walk into a gamestop say, and ask to pre-order a game and was told "you have to enter a raffle to even see if you're going to be able to," I would walk out immediately and I wouldn't purchase that game in retaliation to the ridiculousness that is the raffle.

Honestly if you're late to the game, then why should you get the same benefit as those who did their diligence.  I'm all for limiting the number that can be ordered per customer, so one can't buy 1000 units and make everyone down the line wait, but it should be FIFO.

Now for BFL, it was handled poorly, those who did pre-order way in advance obviously are at an advantage, but this is not the fault of the customer, more so it is of the company.  If for instance you have a 3 month pre-order wait time before release date, and you know you can produce 10k units in 3 months, then set the pre-order limit to 10k units.  After that you can either have another pre-order round for the next batch, or just have the item as out of stock until they arrive.

If you announce a pre-order in advance, then how is it not fair to everyone?  Everyone who is keeping up with this will be informed, and can then either purchase a pre-order when available or not.  If the pre-order units run out, that's just what happens, wait for the next time its available.

Edit: Also, if the pre-order isn't a crazy long wait, then auctioning wouldn't be near as big of issue as it is with BFL.

Yea raidmax is definitely bottom of the barrel.  But you could solder that wire back onto the pin instead of crimping it too.

Actually you should be fine with just the single 4 pin CPU power since the CPU wont be doing anything.  Just be sure to tape off that wire so it doesn't short out on anything.

Try each card by itself, starting with the low hashrate one.  Leave it in the current PCI-E slot and see if it gets a low hash rate by itself.  If it still does, move it to the slot the normal hashrate card was in and try again.  If hashrate is still low, it could be that the card is going bad?  Not sure.  I would try stock clocks first and see if there's any improvement.

If it improves by itself in the same slot then its probably an I/O issue with the motherboard, I would try increasing voltages (small increments!) in the BIOS for North Bridge, I/O related chipsets, and PCI-E.  Depends on your board and chipset what you should adjust though.

If the card improves after moving it to the other slot, then it could be that slot it was in is having issues, try another if available.

Hope this helps.

I second the dummy plug suggestion, probably just disabling the card since it thinks there is no display attached.

That's where the 5 came from, thank you! 

Fair enough, was just a thought.

I was apparently up way too long when I read the original post, lol.  Yes adding the power to the riser takes the load away from the motherboard slots if you only connect the card's riser wires to the PSU.  The power draw I still believe is equal from all ports, but I am not 100% sure. 

It is best to have power connections from the same PSU for 1 GPU (in other words, don't have 2 power cables from 2 different PSUs connected to the same card) to avoid weird issues.

You could also try replacing the stock thermal paste with something of higher quality.  Might not help but it's worth a try.

Those temps sound insanely high for stock.  The only thing I can recommend beyond getting a replacement card is to get some high quality thermal paste and replace the crap that is currently on it.

The red light could just mean that the CPU has power from the 8 pin connector.

@ektwr: No onboard speaker from what I could tell on that board.

I'm thinking there might be an issue with the cooler on the card.  What temps do you get at Intensity 20 for stock clocks?

You're better off using the ATI control panel or another tool such as Sapphire TRIXX or MSI Afterburner IMO.  Try setting the Core to 1100 MHz and Mem to 1500MHz, and set the powertune to 20%.  Set thread concurrency to 23777 or 22111.  Also set Intensity to 20.

When a prototype is developed, would it be possible for those of us willing to accept the risk to buy a proto?

I'd recommend testing stability with a 3D benchmark in a loop for a few hours.