Chilling my Jalapenos [ Chill + Flash: 7.2 >> 8.15GH]

peterepeat

full member

Activity: 237

Merit: 100

Quote from: ?? on ??

Quote from: bitpop on July 04, 2013, 06:24:26 AM

What software is that picture?

Quote from: peterepeat on July 03, 2013, 06:16:09 AM

Quote from: philips on July 03, 2013, 05:37:04 AM

Quote from: peterepeat on July 03, 2013, 04:17:10 AM

Just to expand on this, I have now flashed my Jalapenos with the BFL released BitforceSC 1.2.5 firmware and found the 10 minute chill to still be effective.
New speeds: 7.98 improved to 8.18 and 7.62 is now 8.11
so now is 16.29 total hash rate with the pre-chill and new firmware (was around 16.0 without the pre-chill)

You might try the one modified by ckolivas:
https://bitcointalksearch.org/topic/m.2602712
Plus info:
https://bitcointalksearch.org/topic/m.2623529

thanks - I havent tried that one yet. I did try setting
#define __ASIC_FREQUENCY_ACTUAL_INDEX 8
and flashed that, however I was getting 1000s of errors so I went back

http://imgur.com/peKactI

It's cgwatcher, a miner supervisor app.

bitpop

legendary

Activity: 2912

Merit: 1060

What software is that picture?

Quote from: peterepeat on July 03, 2013, 06:16:09 AM

Quote from: philips on July 03, 2013, 05:37:04 AM

Quote from: peterepeat on July 03, 2013, 04:17:10 AM

Just to expand on this, I have now flashed my Jalapenos with the BFL released BitforceSC 1.2.5 firmware and found the 10 minute chill to still be effective.
New speeds: 7.98 improved to 8.18 and 7.62 is now 8.11
so now is 16.29 total hash rate with the pre-chill and new firmware (was around 16.0 without the pre-chill)

You might try the one modified by ckolivas:
https://bitcointalksearch.org/topic/m.2602712
Plus info:
https://bitcointalksearch.org/topic/m.2623529

thanks - I havent tried that one yet. I did try setting
#define __ASIC_FREQUENCY_ACTUAL_INDEX 8
and flashed that, however I was getting 1000s of errors so I went back

http://imgur.com/peKactI

peterepeat

full member

Activity: 237

Merit: 100

Update - the slower unit after 24h is back to 7.9 GH. I tried chill technique and no improvement. Back to firmware tweaking again Embarrassed

Gomeler

hero member

Activity: 697

Merit: 500

Quote from: TheSwede75 on July 03, 2013, 10:21:19 AM

Someone should just try throwing their Jalapeno in a deep-freezer and mining @ -20C. I am serious, I want to see results

If I had one I would. I did all sorts of refrigeration cooling things when I was overclocking CPUs and GPUs with the culmination of a failed 3 stage cascade. If the jalapeno had software voltage control it could really open up some interesting testing. I'd prefer though a single chip PCB like the test PCBs that were used for Bitfury testing. Could be fun seeing how the chips perform down at -100 Celsius

rustyh17

full member

Activity: 213

Merit: 100

Quote from: TheSwede75 on July 03, 2013, 10:21:19 AM

Someone should just try throwing their Jalapeno in a deep-freezer and mining @ -20C. I am serious, I want to see results

Just spray on a little bit of this:

http://www.neverwet.com/videos-news.php#

Click on the CONFORMAL COATINGS video link....

TheSwede75

full member

Activity: 224

Merit: 100

Someone should just try throwing their Jalapeno in a deep-freezer and mining @ -20C. I am serious, I want to see results

peterepeat

full member

Activity: 237

Merit: 100

Quote from: philips on July 03, 2013, 06:25:17 AM

Worth trying.
His modifications consists in relaxing the start-up tests.

Quote

Just comment out these lines in ASIC_Engine.c

Code:

					if (iReadBackNonceCount < 4)
					{
						// This engine is DEAD!
						bWasAnyEngineDecommissioned = TRUE;
						DECOMMISSION_PROCESSOR(cDiagChip, cDiagEngine);
					}

Then it won't decommission any engines regardless. My firmware was slightly more conservative and made sure the engine returned at least one valid nonce.

I will give this a go (just need to get the AVR dragon back - its away for a week or so)

philips

hero member

Activity: 700

Merit: 500

Worth trying.
His modifications consists in relaxing the start-up tests.

Quote

Just comment out these lines in ASIC_Engine.c

Code:

					if (iReadBackNonceCount < 4)
					{
						// This engine is DEAD!
						bWasAnyEngineDecommissioned = TRUE;
						DECOMMISSION_PROCESSOR(cDiagChip, cDiagEngine);
					}

Then it won't decommission any engines regardless. My firmware was slightly more conservative and made sure the engine returned at least one valid nonce.

peterepeat

full member

Activity: 237

Merit: 100

Quote from: philips on July 03, 2013, 05:37:04 AM

Quote from: peterepeat on July 03, 2013, 04:17:10 AM

Just to expand on this, I have now flashed my Jalapenos with the BFL released BitforceSC 1.2.5 firmware and found the 10 minute chill to still be effective.
New speeds: 7.98 improved to 8.18 and 7.62 is now 8.11
so now is 16.29 total hash rate with the pre-chill and new firmware (was around 16.0 without the pre-chill)

You might try the one modified by ckolivas:
https://bitcointalksearch.org/topic/m.2602712
Plus info:
https://bitcointalksearch.org/topic/m.2623529

thanks - I havent tried that one yet. I did try setting
#define __ASIC_FREQUENCY_ACTUAL_INDEX 8
and flashed that, however I was getting 1000s of errors so I went back

http://imgur.com/peKactI

philips

hero member

Activity: 700

Merit: 500

Quote from: peterepeat on July 03, 2013, 04:17:10 AM

Just to expand on this, I have now flashed my Jalapenos with the BFL released BitforceSC 1.2.5 firmware and found the 10 minute chill to still be effective.
New speeds: 7.98 improved to 8.18 and 7.62 is now 8.11
so now is 16.29 total hash rate with the pre-chill and new firmware (was around 16.0 without the pre-chill)

You might try the one modified by ckolivas:
https://bitcointalksearch.org/topic/m.2602712
Plus info:
https://bitcointalksearch.org/topic/m.2623529

peterepeat

full member

Activity: 237

Merit: 100

Quote from: peterepeat on June 25, 2013, 06:32:12 PM

Quote from: PeZ on June 25, 2013, 05:56:19 PM

Aren't you worried about a layer of frozen water condensing on the electronics?

Not really - I only chill it for 10 minutes in the freezer. I guess it gets the PCB to around 5 to 10 °C. Also its winter here at the moment so I will worry about condensation issues later on.

the ideal temperature seems to be between 5 and 10 °C
I tried it again this morning and chilled them for about 10 minutes only in the freezer, then run in ambient temperature of about 18°C
They have been running for about 2 hours now - the 7.75 is actually stable now at 7.98GH/s, the other one is still stable at 7.62GH/s

http://imgur.com/CgQ3YIT

so now its actually a 960MH/s bonus - just like getting 3 ASICMiner USB for free

(disclaimer - this may have other consequences so research first, and consider your ambient environment)

Just to expand on this, I have now flashed my Jalapenos with the BFL released BitforceSC 1.2.5 firmware and found the 10 minute chill to still be effective.
New speeds: 7.98 improved to 8.18 and 7.62 is now 8.11

so now is 16.29 total hash rate with the pre-chill and new firmware (was around 16.0 without the pre-chill)

xgtele

sr. member

Activity: 288

Merit: 250

No Frost fridge is seems fine for your Jallies Cool

GalaxyASIC

full member

Activity: 130

Merit: 100

While working on better cooling solution for my own ASIC, that is still months away, I put together a rig that can keep hashers very cold and keep moisture in the air away from electronics, so condensation is no a problem. This could be applied to BFL miners and I have four of my own 60GH units on order for 4-5 month now, but have to wait till BFL gets up to speed. Buying a 5Gh unit from someone who has it on hand for 15X it's price seems too much since I get my own relatively soon. But I want to test the cooling solution that I have developed.

I have to see if cost of cooling them makes a good business for hosting other people's units, but for now I just want to test how well it works and need a loaner units for about two weeks and I will fully reimburse hashing loses and pay for overnight Fedex shipping both ways. I am in New York city.

You can still mine on your own account from my lab, I just want to see how well it could be cooled. Upside? you get faster hashing unit and make more BTC.

peterepeat

full member

Activity: 237

Merit: 100

Quote from: PeZ on June 25, 2013, 05:56:19 PM

Aren't you worried about a layer of frozen water condensing on the electronics?

Not really - I only chill it for 10 minutes in the freezer. I guess it gets the PCB to around 5 to 10 °C. Also its winter here at the moment so I will worry about condensation issues later on.

the ideal temperature seems to be between 5 and 10 °C
I tried it again this morning and chilled them for about 10 minutes only in the freezer, then run in ambient temperature of about 18°C
They have been running for about 2 hours now - the 7.75 is actually stable now at 7.98GH/s, the other one is still stable at 7.62GH/s

http://imgur.com/CgQ3YIT

so now its actually a 960MH/s bonus - just like getting 3 ASICMiner USB for free

(disclaimer - this may have other consequences so research first, and consider your ambient environment)

PeZ

sr. member

Activity: 297

Merit: 250

Aren't you worried about a layer of frozen water condensing on the electronics?

Hippie Tech

legendary

Activity: 1624

Merit: 1001

All cryptos are FIAT digital currency. Do not use.

There are only 2 possible outcomes. You will.. or you will not find a block. Its always 50/50, Ying Yang, .. ect.

Mr. Nakamoto wrote it into the code. Wink

PaperClip

newbie

Activity: 56

Merit: 0

Quote from: btcusr on June 25, 2013, 06:00:47 AM

Solo mining, no change in total network hash rate, 20 hours a day for 62 days, then you have what percentage of finding a block??

According to this: https://en.bitcoin.it/wiki/Difficulty
time required to find a block = difficulty * 2^48 / 65535 / hashrate

difficulty = 19339258.272387
hashrate = 7.75 GH/s = 7 750 000 000 hash/s
time required to find a block = 19 339 258.272 387 * 2^48 / 65 535 / 7 750 000 000 = 10 717 774 s

20 hours a day for 62 days equals to 4 464 000 seconds
result is 4 464 000 / 10 717 774 = 41.65%

[edit]
Actually, result percentage should be bigger. 10717774s is maximal time to check all possible hashes. In reality miner does not check them all, it stops as soon as correct hash was found.
Its like if somebody has number between 1 and 9 in his mind, and you will guess it by trying all numbers 1,2,3,.. sequently. You do not need to make 9 guesses every time. It will be from 1 to 9; 5 guesses in average.
Soo, in this case result percentage should be 2 times bigger - 83.3%.
Am I correct?

iikun

legendary

Activity: 1062

Merit: 1003

Quote from: rammy2k2 on June 25, 2013, 08:36:11 AM

Quote from: nightyj on June 25, 2013, 08:24:43 AM

This is not a clever experiment to moove object from cooler to hotter place, it can create condensation, which can lead to shorts and damaged unit.

i agree, you might regret the ~5% extra later ...

+1

Extra Mh/s is great but seeing as you've waited a year for the thing it would be a shame to fry it

GigaWave

full member

Activity: 239

Merit: 250

Is this some kind of phenomenon, or was this just something that was shot lived till the device warmed up? I've never heard of the former.

ssateneth

legendary

Activity: 1344

Merit: 1004

Quote from: Rampion on June 25, 2013, 09:09:04 AM

Quote from: Cluster2k on June 25, 2013, 09:05:31 AM

Well, my Jalapenos have shipped (honestly thought I'd never see the day) and was wondering how many people are solo mining with their units. Currently an upgraded Jalapeno will take about half a year for 25 bitcoins, at the current level of difficulty. Realistically it could be years before a single block is hit, and quite possibly never. Risky proposition when one considers the unit may never pay for itself with solo mining.

Solo mining with 5-7GH/s? Seriously guys, what are you smoking?

Topic: Chilling my Jalapenos [ Chill + Flash: 7.2 >> 8.15GH] (Read 5146 times)