Newbie, but flashed my Jalapeno and have a hint for people.....

[lightfoot]

Morning!

Well, we all have to start off as newbies, so here we go:

I got my 5.0gh/s Jally on Friday of last week and ran it through the weekend with BFGminer. Hashed at about 5.0gh/s, just what I paid for. Temps on the unit were a nice cool 38 degrees C. Unit pulled 20 watts with a power factor of .59 (these power supplies are garbage) with a laptop the whole thing pulled 40 watts with the E-meter.

This evening my dragon programmer came from China (ordered Saturday) and of course I had to flash this thing. I've worked on 68HC711K4 type chips and can deal with programmers so I installed studio, flashed the firmware on the dragon, dug out an old 12 pin header cable, tore apart my jally (had to remove heat sink), and got to it.

The heat sink came off with the pads still on it. Chips were super clean. Hooked up the programmer, followed the directions including setting Device=AT32UC3A1128, and formatted the chip. No problem, Jally rebooted to dead.

Then I tried to load the code and failed. Badly. CRUD! So I broke out the hacker helmet, did some poking around and found that my jally has an AT32UC3A1256 chip instead. Very nice of them to give me more memory, but it was close enough to format, not close enough to load.

Loaded the code, put the heat sink back on (snugged the screws down to as snug as they were when I removed them, IE finger tight), fired it up, and it's running at a nice solid 7.5gh as reported in BFG but a temp of *50* C. Pulling a total of 10 watts more as measured by the P3 meter.

So if you're going to do this with a newer Jally, keep in mind the AT chip might be different. Also keep in mind that the temps seem to be higher, and I am not thrilled with this.

Now for the question: Should I put some artic silver or other heat sink compound on the chips before replacing the sink (IE go in there and fix it right)? I don't like hot components, and the heat sink should probably be pulling more of the heat on those chips. But I do not want to tighten the screws, that's the #1 way to break chips.

Hm. I do have that big copper heat sink from my old IBM X366 somewhere....

CZ

[lightfoot]

Update: I shut it down, opened it up, snugged down the bolts on the heat sink plate a bit tighter, and fired it back up. Temps are now at 46c, bit better. 7.5ghs speed, 1.9% error rate (down from 3%).

I'll grab some fujipoly next before screwing with the settings to boost speed.