X6500 Custom FPGA Miner

[shad]

today i tried to get this running on OpenWRT, but there are no ftdi D2xx drivers
any plans to get this running on an non x86 system?

[freshzive]

My reject rate skyrocketed back to 10-20% using the 150mhz and 166mhz bitstreams. This happened on both of the x6500s I tried, so I'm not understanding the 1-2% reject rate I saw on the earlier post. Maybe it's pool dependent?

[Qoheleth]

Just got done finally setting up my "mining rig".

This way they're both in front of the highest-airflow parts of the fan. Of course I'll have to take it apart again when my better heatsinks arrive, but until then it seems to be working

My reject rate skyrocketed back to 10-20% using the 150mhz and 166mhz bitstreams. This happened on both of the x6500s I tried, so I'm not understanding the 1-2% reject rate I saw on the earlier post. Maybe it's pool dependent?

Same. I'm on eligius, for what it's worth.

Edit: Spoke too soon. The reject rate is higher, but not 20% high. More like 6-12%.

[shad]

same here
after 6hours of running ztexmerge_166mhz i have 12% stale, with fpgaminer_133mhz i had 2-3% i dont think this is a pool issue, i am at slushs pool

i will try the 150mhz bitstream

[tgmarks]

with the 166mhz i'm up at 19% reject rate and i'm on btcguild, I was sub 1% at 133mhz

[li_gangyi]

If you're still running the boards passive, it'll probably be a good idea to put on a fan to get adequate airflow over the heatsinks. If the board feels warm under the FPGA, it's probably not cooled enough.

I'm getting about 4% rejects on 190Mhash.

[shad]

so i have the same issue with the 155mhz bitstream, i dont think that i have a heat problem
as more people have that issue with the ztexmerge-bitstream of X6500 i think there is a issue in the bitstream

or has anyone a nomral reject rate with the new bitstreams?

[tf101]

Iv had about a 3 degree increase in temps from mine.. but other than that it seems all good. One of the chips has produced a lot of errors - but most were in the first few hours... so ill just keep an eye on it.

312.86 MH/s | 0: 2099/52/3 2.4% | 1: 2007/60/51 2.9% | 16h30m |

[Qoheleth]

with the 166mhz i'm up at 19% reject rate and i'm on btcguild, I was sub 1% at 133mhz

you've at least got a fan blowing on your heatsinks, right?

[tf101]

..I have an 80mm fan for each board, blowing right ontop (not to the side as others seem to). When I measure the temp, I measure it using an IR thermometer on the back of the board - seems to be where most the heat on my setup...

[tf101]

I can also say they are drawing 1.7A at 12V

[shad]

i have do admit that this could be a cooling problem, i have a 60mm fan blowing at the headsinks, maybe that is not enough

[tgmarks]

I've got an 80mm hdd fan mounted 5mm above the top, heat sinks are seated well, and they don't feel more than warm.  my accepted has risen to 81.5% over the last couple hours, but thats all

Edit: down to 15% reject

[fizzisist]

A customer pointed out to me that he saw high rejects with the faster bitstreams, but most were rejected because they were duplicate submissions. Reducing the getwork interval got rid of this problem.

At 166 MH/s, the entire nonce range should be hashed in 25.9 seconds, so new work should be loaded before that. The default interval in the latest version on Github is 20 seconds, which should be perfect for this (better to always have fresh work then to sometimes have stale work because of a latency problem).

If you're having really high rejects, could you post the getwork interval that you're using?

[Qoheleth]

A customer pointed out to me that he saw high rejects with the faster bitstreams, but most were rejected because they were duplicate submissions. Reducing the getwork interval got rid of this problem.

At 166 MH/s, the entire nonce range should be hashed in 25.9 seconds, so new work should be loaded before that. The default interval in the latest version on Github is 20 seconds, which should be perfect for this (better to always have fresh work then to sometimes have stale work because of a latency problem).

Ah, this is probably my problem. I think I was still running at 30 seconds (the last time I did a git pull was when the submission bugfix came in).

Edit: Yep, back at 2-4%.

[ummas]

@tf101
so You got 310Mhash/21 - impresive

If prize could drop a bit, it would be spectacular 

[tf101]

I cant complain thats for sure... here is the latest from the other FPGA

Run Summary:
-------------
Device: 0
Serial: AH00WOWE
JTAG chain: 2
Number of FPGAs: 2
Running time: 20h15m
Getwork interval: 20 secs
Chain 0:
  Accepted: 2570
  Rejected: 61 (2.32%)
  Invalid: 5 (0.19%)
  Accepted hashrate: 151.32 MH/s
  Hashrate w/ rejects: 154.91 MH/s
  Hashrate w/ invalids: 155.20 MH/s
Chain 1:
  Accepted: 2460
  Rejected: 74 (2.92%)
  Invalid: 63 (2.43%)
  Accepted hashrate: 144.84 MH/s
  Hashrate w/ rejects: 149.20 MH/s
  Hashrate w/ invalids: 152.91 MH/s
Total hashrate for device: 296.16 MH/s / 304.11 MH/s / 308.11 MH/s

I seem to be getting a higher amount of invalids on FPGA1 vs FPGA0 - on both boards, which is odd.... just updating my getwork interval now though as well, ill drop it to 15seconds...

[freshzive]

I was using the default getwork interval (which I thought was 20s?). Dropped to 15s and now my error rates seem to be lower. I'll run it overnight o see what they look like long term.

[freshzive]

Just got done finally setting up my "mining rig".

This way they're both in front of the highest-airflow parts of the fan. Of course I'll have to take it apart again when my better heatsinks arrive, but until then it seems to be working

that is one badass mining rig

[coblee]

Here are my 2 fpga's with the 166 bitstream with default interval:

Code:

Run Summary:
-------------
Device: 0
Serial: AH00WI18
JTAG chain: 2
Number of FPGAs: 2
Running time: 1d19m59s
Getwork interval: 20 secs
Chain 0:
  Accepted: 2766
  Rejected: 72 (2.54%)
  Invalid: 171 (5.68%)
  Accepted hashrate: 135.61 MH/s
  Hashrate w/ rejects: 139.14 MH/s
  Hashrate w/ invalids: 147.52 MH/s
Chain 1:
  Accepted: 2813
  Rejected: 59 (2.05%)
  Invalid: 187 (6.11%)
  Accepted hashrate: 137.92 MH/s
  Hashrate w/ rejects: 140.81 MH/s
  Hashrate w/ invalids: 149.98 MH/s
Total hashrate for device: 273.53 MH/s / 279.95 MH/s / 297.51 MH/s

Code:

Run Summary:
-------------
Device: 1
Serial: AH00WOWD
JTAG chain: 2
Number of FPGAs: 2
Running time: 1d19m54s
Getwork interval: 20 secs
Chain 0:
  Accepted: 2784
  Rejected: 84 (2.93%)
  Invalid: 169 (5.56%)
  Accepted hashrate: 136.50 MH/s
  Hashrate w/ rejects: 140.62 MH/s
  Hashrate w/ invalids: 148.91 MH/s
Chain 1:
  Accepted: 2794
  Rejected: 86 (2.99%)
  Invalid: 146 (4.82%)
  Accepted hashrate: 136.99 MH/s
  Hashrate w/ rejects: 141.21 MH/s
  Hashrate w/ invalids: 148.37 MH/s
Total hashrate for device: 273.50 MH/s / 281.83 MH/s / 297.28 MH/s

I'm not sure why I'm having such high invalids.