Topic: KnCMiner Jupiter Miner First Impressions - page 5. (Read 65993 times)

Selling Jupiter hashing stable at 525GH/s, to be shipped today: https://bitcointalksearch.org/topic/knc-jupiter-525ghs-stable-in-hand-immediate-delivery-fast-auction-314656

Okay I have no idea what's going on with my Saturn.

Upgraded to 0.96 yesterday (255gh/s, 3850 WU/m) and after a few hours went back to 0.94 (270gh/s, 4050 WU/m). After another few hours, went to 0.93 (270 gh/s, 4050 WU/m) and used enablecores.bin (no difference).

Today I went back to 0.96 and it has the same performance numbers (272gh/s, 4050 WU/m) as 0.93/0.94 with the usual less HW errors.

Here is the previous results from Jupiter 1 on Firmware v0.9.2

Note: that the 1 hour performance vs firmware v0.9.6.

Previous results on 24 hours on firmware v0.9.5 with an ambient room temp of 19°C



24 Hour Results on firmware v0.9.6 with an ambient room temp of 16°C
Note: The first 4 hours the ambient room temp is 19°C I had posted results for each.

Kano thanks again for helping with hosting.

Firmware v0.9.6 Results Update

4Hour Result


13Hour Result


Edit: 13 Hour results updated with proper temps.

Note: The first 4 hours were results in an ambient room temp of 19°C. For the remainder of the time I have lowered the ambient room temp close to the first test in order to see if the 24hour results will be the same as the previous test on firmware v0.9.5

The blue stuff is miner.php I wrote in cgminer ...

Sitarow, from where are you getting the data in the table format (listing all your miners together)?

Here are the results after 3 hours with firmware v0.9.6



Compare it with the 3 hour results with firmware v0.9.5

Unlike before it seemed that now with this new firmware v0.9.6 every time you power the system on, I noticed that within the first 40 minutes performance would take a hit and more hardware errors would occur.

I am still not sure what the problem is.

One possibility could be potential noise on the power such as a motor or fridge compressor. Another possibility could be that this is the result of previously unlocked cores. However I never seen any offline before the update with the exception of  Module #3 on Jupiter #1 that had 4 turned off.

The UPS is my attempt of ruling out the possibility of performance issues due to power line noise and poor power conditioning.

I will swap the systems after my first 1 day results to see if this is indeed the case.

Similar with firmware v0.9.4, given time, both Jupiter Unit #2 and #3 HW error rates do drop considerably.

Here is the results after 2 hours on firmware v0.9.6




Compare with the results after 2 hour on firmware v0.9.5


Oh forgot to mention that on the results with firmware v0.9.5 both Jupiter #2 and Jupiter #4 were on the UPS pure sine wave outlet.
However for the tests with firmware v0.9.6 both Jupiter #1 and Jupiter#4 are on the UPS ups pure sine wave outlet.

Note: That all 4 units are plugged into two separate UPS devices. However one is conditioned outlet and the other is surge protected outlet.

Sitarow, can you explain more what you mean about power sensitive?

Are your Jupiters consistent enough to tell?  If you don't change anything do they stay the same on reboots?

That's one of the ways .96 helps.. I bounced around firmwares again on my troubled miner and came back to that FW

two things .96 can do to help a miner.
First is it can clear up simple core errors and have you hashing with all cores and low HW errors
Second, for troubled miners, you can sometimes get it to just acknowledge troubled cores to stay off so there is less % of HW errors. This is when you do not want to enable cores, and just let it re-enable the simple core issues which are much less

Of course, best to not have a troubled miner, but if you do, try to get it into the state I just described



and the two bad boards

I am testing how well these units handle noise on the power line.

Jupiter #1 and #4 are plugged into a CyberPower CP1500PFCLCD UPS 1500VA 510W PFC compatible Pure sine wave outlet
Jupiter #2 and #3 are on the same outlet however with only a surge suppressor outlet.

Jupiter #1 and #2 are on the same UPS with only #1 using the conditioned power.
Jupiter #4 and #3 are on the same UPS with only #4 using the conditioned power.

CyberPower CP1500PFCLCD UPS 1500VA 510W PFC compatible Pure sine wave
http://www.amazon.com/CyberPower-CP1500PFCLCD-Sinewave-Compatible-Mini-Tower/product-reviews/B00429N19W

Here are the results after 1 hour on the new firmware V0.9.6



At about 45 minutes in both Jupiter #2 and Jupiter #3 hardware errors started to go down. Seems firmware v0.9.6 may be power sensitive much like v0.9.4

Jupiter #1 and #4 hit the target of 550 GH/s only 2 mins from cgminer start.

Compare 1hour results on firmware v0.9.5

Running enablecores.bin helped improve my hashrate on 0.96 from 506 to 518 Gh/s avg on the GUI page and cgminer. WU went up from 7400 to 7700. It shows 520 to 570 Gh/s at the pool (hour averages). Still lots of room for improvement looking at the HW errors and a few select cores being disabled/enabled from time to time, but I'm not too unhappy about its current performance.

One of my miners is a star so far and showing more Gh/s at the pool on avg than on the GUI page
running .96
~565 to 585 at the pool and just a steady 546 on the GUI page.

That helps since my other miner can only creep along around ~420 at the pool while it shows ~495 on the GUI
running .95 since that seems to squeeze the most out of it
(lots of HW errors @ 16% and, drops cores and a whole die off too)

not sure what KNC's RMA policy is but would be nice to exchange the two problem boards/ascis if it is a quick turnaround

My Jupiter is showing 555GHs and the pool is showing 554GHs so I'm pretty happy that what cgminer is showing is what is actually being output. I'm sure there is room for improvement but with 0.96 everything seems ok.

its between 22 and 25max C°, im very happy with this jupiter

I checked but it's ok

Yesterday ASIC#5 work at 38°C and today at 25°C... I think now it don't work...

Did you check if any of your asic modules had the "plastic" taken off the heatsink before being installed?

Some users have reported having that issue. You can read up on this from The Avenger who posted detailed pre-setup steps.
start at Check list item #3
https://bitcointalksearch.org/topic/m.3368753

Here are the results of the units in the previous test after 1 day operating at 19°C -firmware v0.9.5

I had a problem... (at 545W)

http://imgur.com/bofWl3s

Yesterday went at 530GH/s and 580W and the last unit was at 38°C... What's happened?


(room Temp=20°C)