Trying to Further Optimize the Jasminer X4 1U ASIC Miner

Trying to Further Optimize the Jasminer X4 1U ASIC Miner


Mining

cryptomining-blog.com

1 h

  

Studying time: ~6 m


We’ve been enjoying round with a Jasminer X4 1U Ethash and ETChash ASIC miner for a few days now and now we have some extra issues to share with you in case you are within the system. We’ve already lined some attention-grabbing findings and attainable tweaks for the Jasminer X4 1U ASIC miner right here, however we continued additional making an attempt to optimize the miner and see if we are able to squeeze a bit of additional efficiency from it. Our subsequent apparent step was to disassemble the hashing board with the ASIC chips…

Eradicating the aluminium cooling heatsink, a single piece for the entire board, revealed the eight Jasminer X4 ASIC chips on the hashing board together with a controller chip and two reminiscence modules. And a ton of thermal grease that must be cleaned and changed earlier than assembling the entire thing again once more. Solely the controller chip isn’t lined with thermal grease, as an alternative it makes use of a thicker thermal pad for contact with the heatsink. The entire chips did have contact with the heatsink, so supposedly no anticipated points with thermal switch and cooling right here. So, our hopes of getting a little bit of thermal enchancment by changing the thermal compound with a greater one would in all probability finish in useless, however since we nonetheless have to interchange the thermal interfacing materials we’ll do it.

Cleansing up the entire thing correctly does require some effort and time as because of the dimension of the crystals on the Jasminer X4 ASIC chips it’s like cleansing 8 CPUs of computer systems. And it’s worthwhile to ensure you are cautious not damaging something, although the cleaned-up board and chips do like fairly good. We’re reusing the unique thermal pad for the controller chip as we don’t consider it’s obligatory to interchange that one with a extra environment friendly one anyway.

The principle controller chip on the hashing board is a Trion T120 FPGA together with two DDR3 reminiscence modules proper subsequent to it (no, that’s not how 5GB of reminiscence seems like!). That is the controller chip for the Jasminer X4 ASIC chips that’s getting used on the hashing boards, the miner does have a second FPGA controller chip with separate RAM chips on the primary management board that hosts the software program and net interface of the miner, there nonetheless we discover a Xilinx Zinq chip.

Here’s a single of the eight Jasminer X4 ASIC chips in a forty five×45 mm package deal (678 sq. millimetres die dimension) with 1TB reminiscence bandwidth, 5GB of reminiscence and a hashrate per chip of 65 MH/s with every of the chips having 384 computing cores and 384 on-chip dies in response to the producer with a 512-bit bus width, an working frequency of 800 to 1 GHz and an influence consumption of 23 Watts.

We’ve used the Arctic Cooling MX4 thermal compound to cowl the chips earlier than reassembling the hashing board with the aluminium heatsink. It’s a decently priced high-performance product that we like to make use of together with Noctua’s NT-H1 resulting from their nice value/efficiency ratio, and there doesn’t appear to be a lot level to attempt to use costlier resolution anyway. When making use of different thermal compound be certain that to place a bit extra on the reminiscence chips as there the hole with the heatsink is a little more in comparison with the one of many ASIC chips.

We’ve in contrast the hashrate and working temperature of the Jasminer X4 1U ASIC miner earlier than altering the thermal compound and after doing so, leaving the system to hash for greater than 10 hours with a view to get common of the working temp and hashrate. Do observe that that is our already silent modified unit with changed followers and firmware, so your system would possibly report a temperature vary and never a single quantity. The earlier than and after outcomes are very related, just about the identical common hashrate and simply 1 diploma Celsius distinction, so there may be actually no have to mess around with the thermal grease because it is not going to provide you with a lot of an enchancment. We additionally did an influence utilization measurement earlier than and after on the wall, the outcomes there weren’t a lot completely different as effectively – 266.1 W earlier than and 265.7 W after the thermal grease substitute.

The following factor we wished to check out right here was to make use of a special energy provide than the one the Jasminer X4 1U comes geared up with. There are three apparent causes to interchange the PSU – to see if the 300W of the inventory one is perhaps a limiting issue, to make use of a silent energy provide and to probably get a decrease energy consumption. We’ve linked a 1000W Corsair HX1000i energy provide (Platinum effectivity ranking) to the ASIC miner instead of the inventory PSU and sadly the outcomes had been not likely that encouraging. You get a silent energy provide because the HX1000i runs passively at sub 300W of energy utilization, however the energy consumption measured on the wall was 263.7 W or simply 2 Watts much less in comparison with the inventory PSU.

The opposite factor we wished to test with the extra highly effective energy provide was if the Jasminer X4 1U would be capable to run at 250 MHz working frequency with the brand new thermal interface materials in place and greater than sufficient energy. Sadly, our miner continued to dislike the 250 MHz working frequency returning a variety of errors making an attempt to hash and thus successfully working at a fraction of the particular efficiency you get at 225 MHz the place every thing runs secure and with out errors. So, the truth that at 250 MHz the mining system makes use of nearly 300 Watts of energy measured on the wall doesn’t imply that the inventory PSU isn’t capable of deal with that. And we don’t suppose that the cooling of the ASIC chips can be a problem right here for them not be capable to run correctly at 250 MHz.

Jasminer X4 does include an excellent thermal grease already utilized, so no want to interchange it. The inventory 300W PSU can be actually good by way of effectivity and isn’t limiting you operating the miner at increased clocks. The one factor is that the inventory energy provide is sort of noisy and also you would possibly wish to modify or exchange it in case you are making your miner right into a silent one like we did with ours. It appears that evidently Sunlune already did very effectively with the {hardware} facet of their Jasminer X4 ASIC miners, each by way of effectivity and efficiency. What they might work a bit extra on nonetheless is the software program facet as their gadgets don’t at present help dual-mining ETH/ETC-based cash together with ZIL, one thing that would simply deliver the miners an additional 30-40% revenue. The dearth of this characteristic signifies that a competitor with dual-mining help already accessible can simply present the identical profitability with a less expensive system that has decrease hashrate and/or isn’t as energy environment friendly because the Jasminer X4.


Source link

Leave a reply

Your email address will not be published. Required fields are marked *

Translator


ArabicChinese (Simplified)DutchEnglishFrenchGermanItalianPortugueseRussianSpanish

Shop Men’s T-Shirt

Shop Hoodies

Shop Women’s T-shirt

  • USD
  • EUR
  • GPB
  • AUD
  • JPY
  • DSLA ProtocolDSLA Protocol(DSLA)
    $0.003681-6.85%
  • lympoLympo(LYM)
    $0.004392-4.43%
  • YAM v2YAM v2(YAMV2)
    $4.70-1.41%
  • PolkaBridgePolkaBridge(PBR)
    $0.439876-7.02%
  • CornichonCornichon(CORN)
    $0.073096-0.86%
  • StacyStacy(STACY)
    $0.0007100.00%
  • RelevantRelevant(REL)
    $0.780.35%
  • Calamari NetworkCalamari Network(KMA)
    $0.0023166.98%
  • TICOEX TokenTICOEX Token(TICO)
    $0.0013640.52%
  • bitcoinBitcoin(BTC)
    $16,468.191.58%