X6500 Custom FPGA Miner - page 33. | Bitcointalksearch.org

bitcowok

newbie

Activity: 48

Merit: 0

Hi, I've been lurking for ages and only just worked out the 5 post minimum rule... ("wheres the frigging reply button?"

)

Heres a shot of the enclosure i have just built, which will eventually have 30 x6500's in it.

the rest of the images are here if anyone is interested:
http://jabawok.net/gallery/v/random/201201-x6500-30x-enclosure/

thanks to fizzisist for the double stacked fan tower idea. Only difference with my layout is the stack is on its side so the heat will be rising naturally (and sucked out the top with the big fans) Air inlet is through the bottom of the case and lower side vents.

Temp measurements at 166mhz are 39 degrees for the lower heatsinks and 41 degrees for the upper heatsinks. (measured with an IR temp sensor from hobbyking)

All fans are running at 7V at the moment. I plan to put plastic or cardboard sides and top covers to tunnel the air properly, which will hopefully see less temperature difference between the heatsinks.

I'll probably get some power and more temp measurements etc done soon at various hash rates and post back. (I've just changed to 180mhz, 413/6/1 and 433/2/0.. looks ok!)

freshzive

sr. member

Activity: 447

Merit: 250

anyone found an enclosure that will work for these things yet?

ZodiacDragon84

sr. member

Activity: 266

Merit: 250

The king and the pawn go in the same box @ endgame

art sounds like an FPGA ASIC god...lol

njloof

member

Activity: 73

Merit: 10

Quote from: ZodiacDragon84 on January 27, 2012, 03:34:07 PM

Hmmm, wonder how much tweaking in the code could increase efficiency. I figure you guys are already all over that though Tongue

They seem to be doing a pretty good job of optimizing the use of the chip to maximize its throughput given the existing algorithm. ArtForz likes dropping hints about improvements to the structure of the code that they haven't found yet. He was ahead of the FPGA curve by about a year.

ZodiacDragon84

sr. member

Activity: 266

Merit: 250

The king and the pawn go in the same box @ endgame

Quote from: njloof on January 27, 2012, 11:12:20 AM

Quote from: ZodiacDragon84 on January 27, 2012, 02:08:24 AM

So how hard you guys think we can push these things? I would love to put them BFL folks in a competitive place if their hardware exists. With proper cooling, and well regulated power, whats the theoretical limit? I plan on buying one of these and water cooling it when my tax return comes this week, I will offer a bounty if someone can squeeze 500 M/hash out of these. (250 MHz each chip I believe)

You can see more in-depth conversation about trying to squeeze additional performance over at the ztex thread -- my read on it is that 200MHz may be close to the max you'll get from the LX150. There is a little noise about upping the voltage on the chips to try to get a tiny bit more performance (~4%) but it's a good way to blow the chip if you screw up.

Hmmm, wonder how much tweaking in the code could increase efficiency. I figure you guys are already all over that though Tongue

njloof

member

Activity: 73

Merit: 10

they sell usb cables with a right angle connector on them... you can use that to plug into the USB port in the back of the board, then run the cable underneath the board to the side with the power connector.

shad

full member

Activity: 148

Merit: 100

so i mounted my X6500 in a case were i thougt i could put 4 X6500 in it
now i think i can put 2 in the case

my major problem is that the connectors of voltage & usb are not on the same side Sad

if you are working on a new layout i would prefer to have all connectors on the same side, it would be easier for me Cheesy

and i think its the smarter solution anyway

njloof

member

Activity: 73

Merit: 10

Quote from: ZodiacDragon84 on January 27, 2012, 02:08:24 AM

So how hard you guys think we can push these things? I would love to put them BFL folks in a competitive place if their hardware exists. With proper cooling, and well regulated power, whats the theoretical limit? I plan on buying one of these and water cooling it when my tax return comes this week, I will offer a bounty if someone can squeeze 500 M/hash out of these. (250 MHz each chip I believe)

You can see more in-depth conversation about trying to squeeze additional performance over at the ztex thread -- my read on it is that 200MHz may be close to the max you'll get from the LX150. There is a little noise about upping the voltage on the chips to try to get a tiny bit more performance (~4%) but it's a good way to blow the chip if you screw up.

shad

full member

Activity: 148

Merit: 100

the miner is so stable since the last big changes

Quote

314.39 MH/s 0: 20169/167/0 0.8%/0.0% 1: 20276/156/0 0.8%/0.0% 6d2h53m

i am running on 166Mhz

but now i have to shut it down, because i will mount it in a case and put it in the coldest room i have

and try more Mhz

ZodiacDragon84

sr. member

Activity: 266

Merit: 250

The king and the pawn go in the same box @ endgame

So how hard you guys think we can push these things? I would love to put them BFL folks in a competitive place if their hardware exists. With proper cooling, and well regulated power, whats the theoretical limit? I plan on buying one of these and water cooling it when my tax return comes this week, I will offer a bounty if someone can squeeze 500 M/hash out of these. (250 MHz each chip I believe)

tgmarks

donator

Activity: 490

Merit: 500

I'm running at 200mhz off my main desktop in the house with one of these sitting on top of the stock heatsinks:

http://www.newegg.com/Product/Product.aspx?Item=N82E16835233080

and i'm seeing this:

425.22 MH/s | 0: 8756/30/4 0.3%/0.0% | 1: 8925/27/0 0.3% | 2d5h36m

and i'm just warm to the touch.

njloof

member

Activity: 73

Merit: 10

Quote from: ZodiacDragon84 on January 26, 2012, 10:01:57 PM

Quote from: njloof on January 26, 2012, 06:06:23 PM

FWIW, I have the 200MHz bitstreams now running on 4 FPGAs cooled with two of those 80mm USB fans. Granted, they are in a garage in January (~50F most of the day) but the fins and board are just barely warm to the touch.

what are your hash rates?

402.79 MH/s | 0: 7965/127/0 1.6%/0.0% | 1: 7954/97/0 1.2%/0.0% | 2d18m47s 388.65 MH/s | 0: 7821/86/0 1.1%/0.0% | 1: 7715/85/2 1.1%/0.0% | 1d23h10m

ZodiacDragon84

sr. member

Activity: 266

Merit: 250

The king and the pawn go in the same box @ endgame

Quote from: njloof on January 26, 2012, 06:06:23 PM

FWIW, I have the 200MHz bitstreams now running on 4 FPGAs cooled with two of those 80mm USB fans. Granted, they are in a garage in January (~50F most of the day) but the fins and board are just barely warm to the touch.

what are your hash rates?

njloof

member

Activity: 73

Merit: 10

FWIW, I have the 200MHz bitstreams now running on 4 FPGAs cooled with two of those 80mm USB fans. Granted, they are in a garage in January (~50F most of the day) but the fins and board are just barely warm to the touch.

n0ne

hero member

Activity: 2618

Merit: 548

SecureShift.io | Crypto-Exchange

Quote from: spiccioli on January 26, 2012, 08:07:51 AM

Quote from: n0ne on January 25, 2012, 11:23:51 PM

These might work well, cheap too.

http://www.ebay.ca/itm/Low-Profile-Heatsink-Intel-D945GCLF-Atom-Mini-ITX-board-/130303965124?pt=LH_DefaultDomain_0&hash=item1e56b8b7c4

n0ne,

it would be better, IMHO, to have a single heatsink that covers both FPGAs to minimize thermal differences between them.

regards.

spiccioli.

I agree, it's just tough to find rectangle heat sinks that support fans.

Cablesaurus

sr. member

Activity: 302

Merit: 250

We're working on getting more units available as soon as possible.

ZodiacDragon84

sr. member

Activity: 266

Merit: 250

The king and the pawn go in the same box @ endgame

Maybe I need to find a water block that is low profile, and can cover both chips, while having a clamp mechanism that mounts via the standoff hole? I'd like to theoretically mount these puppies over each other in a stack of 5 or so, all water cooled flying balls to the wall out MHz wise. and the water block fittings would have to be mounted on the sides of the block

spiccioli

legendary

Activity: 1378

Merit: 1003

nec sine labore

Quote from: n0ne on January 25, 2012, 11:23:51 PM

These might work well, cheap too.

http://www.ebay.ca/itm/Low-Profile-Heatsink-Intel-D945GCLF-Atom-Mini-ITX-board-/130303965124?pt=LH_DefaultDomain_0&hash=item1e56b8b7c4

n0ne,

it would be better, IMHO, to have a single heatsink that covers both FPGAs to minimize thermal differences between them.

regards.

spiccioli.

li_gangyi

full member

Activity: 157

Merit: 100

You'd need to figure out a way to mount the blocks, as well as still provide some airflow over the board to cool the power supplies etc. We've not released dynamically clocking bitstreams at this point, but it is in the works. So far I've not been able to synthesize anything that'll run over 200Mhz in Xilinx's ISE. You could push it even if the ISE tells you it might not work, but there's no saying if it'll work properly. If that changes I'll post an update. I've yet to run all the possible instances under ISE, I'm only dedicating 2 of my PC's possible 4 cores for running this at the moment because I still game on it.

ZodiacDragon84

sr. member

Activity: 266

Merit: 250

The king and the pawn go in the same box @ endgame

How hard could one push one of these with this water cooling solution?

http://www.frozencpu.com/products/7211/ex-blc-451/EK_ASUS_4_Intel_Northbridge_Southbridge_Chipset_Liquid_Cooling_Block_-_Acetal_EK-NBSB_ASUS_4_Acetal.html?tl=g30c89s147#blank

Topic: X6500 Custom FPGA Miner - page 33. (Read 219954 times)