独家优惠奖金 100% 高达 1 BTC + 180 免费旋转

RX2 development report

Attention Luxcore Community!

During the last several weeks the Luxcore team has conducted numerous stress-tests and optimizations for our upcoming RX2 algorithm implementation. For complete transparency, below are the results from beginning until present:

Beginning on October 10th, the first test was conducted using only 8 mining rigs. This was the result:

[2020–10–20 14:05:27] accepted: 160/160 (diff 0.000), latency: 59ms for 8 rig.

The latency and CPU usage were both normal, so we then continued testing with 15 mining rigs:

Despite seeing a significant spike in CPU usage, something which was not a concern at the time, we continued testing with 50 mining rigs:

[2020–10–20 14:21:06] accepted: 149/149 (diff 0.000), latency: 53ms 4431.86 H/s yes!
[2020–10–20 14:21:11] GPU #7: 1873 MHz 5.37 H/W 103W 60C FAN 65%
[2020–10–20 14:21:14] GPU #4: PNY GTX 1070, 552.83 H/s
[2020–10–20 14:21:14] accepted: 150/150 (diff 0.000), latency: 73ms 4431.64 H/s yes!

Both the latency and CPU usage again passed muster. However, we identified an issue with the shares coming in too fast to the pool. This required a difficulty adjustment.

After the adjustment we increased testing to 100 mining rigs:

[2020–10–20 15:26:25] accepted: 69/69 (diff 0.000), latency: 1063ms 3323.53 H/s yes!
[2020–10–20 15:26:26] GPU #2: PNY GTX 1070, 556.72 H/s
[2020–10–20 15:26:27] GPU #4: 1873 MHz 5.43 H/W 100W 53C FAN 65%
[2020–10–20 15:26:27] GPU #4: PNY GTX 1070, 553.46 H/s
[2020–10–20 15:26:27] accepted: 70/70 (diff 0.000), latency: 335ms 3323.81 H/s yes!
[2020–10–20 15:26:28] accepted: 71/71 (diff 0.000), latency: 1388ms 3323.81 H/s yes!

Latency was unsatisfactory with only 100 rigs, so we dove into the reason(s) why and subsequently remedied the cause.

After auditing the pool and miners, and pushing several necessary fixes, we were ready to run more tests to address the latency issue. We:

We then ran tests with 100 rigs for a second time:

[2020–10–22 15:06:24] accepted: 77/77 (diff 0.000), latency: 3688ms 3314.14 H/s yes!
[2020–10–22 15:06:24] accepted: 78/78 (diff 0.000), latency: 3688ms 3314.14 H/s yes!
[2020–10–22 15:06:24] accepted: 79/79 (diff 0.000), latency: 3688ms 3314.14 H/s yes!
[2020–10–22 15:06:25] GPU #3: PNY GTX 1070, 557.08 H/s
[2020–10–22 15:06:27] GPU #1: PNY GTX 1070, 556.30 H/s
[2020–10–22 15:06:28] accepted: 80/80 (diff 0.000), latency: 693ms 3313.72 H/s yes!
[2020–10–22 15:06:29] GPU #2: PNY GTX 1070, 553.67 H/s
[2020–10–22 15:06:32] accepted: 81/81 (diff 0.000), latency: 867ms 3312.88 H/s yes!

During this test, the CPU usage was at 30% per core and we noticed that shares were still piling up too fast, restarting the pool at 10,000 shares.

We tested with 100 rigs once again after adjusting the mining difficulty.

[2020–10–22 15:17:07] randomx block 34598, diff 0.000
[2020–10–22 15:17:09] GPU #0: PNY GTX 1070, 547.62 H/s
[2020–10–22 15:17:10] accepted: 323/323 (diff 0.000), latency: 156ms 3311.83 H/s yes!
[2020–10–22 15:17:11] GPU #3: 1909 MHz 5.26 H/W 105W 49C FAN 64%
[2020–10–22 15:17:18] GPU #0: PNY GTX 1070, 554.25 H/s
[2020–10–22 15:17:18] accepted: 324/324 (diff 0.000), latency: 75ms 3312.23 H/s yes!

This new adjustment yielded far better results for latency. Additionally, cores now had an average of 15% cpu load instead of the prior 30%.

With all signs pointing up, we continued testing with 200 rigs.

[2020–10–22 15:27:44] accepted: 375/375 (diff 0.000), latency: 330ms 3307.19 H/s yes! seed: 3a56fe35ec530efa175d5f259f9588e3ceb6d8f9bb606f24c9c6d0a8c2787c3b
[2020–10–22 15:27:59] randomx block 34634, diff 0.000
[2020–10–22 15:28:07] GPU #3: PNY GTX 1070, 557.15 H/s
[2020–10–22 15:28:07] accepted: 376/376 (diff 0.000), latency: 347ms 3307.20 H/s yes! seed: 3a56fe35ec530efa175d5f259f9588e3ceb6d8f9bb606f24c9c6d0a8c2787c3b
[2020–10–22 15:28:20] randomx block 34636, diff 0.000
[2020–10–22 15:28:32] GPU #3: 1885 MHz 5.10 H/W 108W 59C FAN 65%
[2020–10–22 15:28:33] GPU #5: PNY GTX 1070, 549.18 H/s

Load on the CPU reached out 30% with 200 rigs connecting in at the same time and we saw the latency going up as well, still need more optimization on that part.

We worked diligently and tirelessly over the last several days to identify the reason for increasing latency and implemented several remedial changes.

Below is the result achieved with our new build and 100 rigs.

[2020–10–27 09:53:55] GPU #4: MSI GTX 1070 Ti, 695.31 H/s
[2020–10–27 09:53:55] accepted: 1/1 (diff 0.000), latency: 29ms 695.31 H/s yes!
[2020–10–27 09:53:59] GPU #0: 1882 MHz 0.00 H/W 46W 53C FAN 49%
[2020–10–27 09:54:06] GPU #4: MSI GTX 1070 Ti, 694.48 H/s
[2020–10–27 09:54:06] accepted: 2/2 (diff 0.000), latency: 21ms 694.90 H/s yay!!! solved: 1 seed: 9a55ed078e2f786ec719650140a75597619318a8490a55a6db57bf102a6748d0
[2020–10–27 09:54:07] randomx block 37485, diff 0.000
[2020–10–27 09:54:12] GPU #2: MSI GTX 1070 Ti, 694.54 H/s

CPU load increased slightly but still maintained acceptable support, The latency, however, improved exponentially.

So we added more rigs to the test:

2020–10–27 10:03:47] GPU #3: MSI GTX 1070 Ti, 695.42 H/s
[2020–10–27 10:03:47] accepted: 56/56 (diff 0.000), latency: 22ms 3209.94 H/s yes! seed: 2bbb1286f9018857def60dc49b011bbb66307fc1b75c7021d02ea8129333c595
[2020–10–27 10:03:50] Stratum difficulty set to 0.00192 (0.00001)
[2020–10–27 10:03:50] randomx block 37513, diff 0.000
[2020–10–27 10:03:52] GPU #4: MSI GTX 1070 Ti, 677.54 H/s [
2020–10–27 10:03:52] accepted: 57/57 (diff 0.000), latency: 35ms 3209.94 H/s yes!

We successfully identified the bug causing latency. Everything is running smoothly now.

To push the boundaries of the network even further, we tested 400 rigs and broke 1000k/hs in the pool. Miner latency still continued to do great:

[2020–10–27 10:33:33] GPU #4: MSI GTX 1070 Ti, 689.48 H/s
[2020–10–27 10:33:33] GPU #4: 1848 MHz 5.51 H/W 125W 56C FAN 50%
[2020–10–27 10:33:33] accepted: 190/190 (diff 0.000), latency: 28ms 3201.49 H/s yes! seed: 2bbb1286f9018857def60dc49b011bbb66307fc1b75c7021d02ea8129333c595 [2020–10–27 10:33:42] GPU #2: MSI GTX 1070 Ti, 683.92 H/s
[2020–10–27 10:33:43] accepted: 191/191 (diff 0.000), latency: 43ms 3201.27 H/s yes! seed: 2bbb1286f9018857def60dc49b011bbb66307fc1b75c7021d02ea8129333c595
[2020–10–27 10:34:08] GPU #4: MSI GTX 1070 Ti, 689.23 H/s
[2020–10–27 10:34:08] accepted: 192/192 (diff 0.000), latency: 29ms 3201.14 H/s yes! seed: 2bbb1286f9018857def60dc49b011bbb66307fc1b75c7021d02ea8129333c595

During the above test, we confirmed that everything was now in excellent order with network latency. However, we identified a bug that “crashed” the pool when seeds were changed. We investigated this issue further and fixed it before continuing on with the next phase.

Our new build was ready for a stress test to see if we still had the same “crashing” issue.

[2020–11–07 15:35:16] GPU #5: EVGA GTX 1070, 582.78 H/s
[2020–11–07 15:35:16] accepted: 4/4 (diff 0.000), latency: 23ms 1137.94 H/s yes!
[2020–11–07 15:35:21] GPU #2: EVGA GTX 1070, 579.03 H/s
[2020–11–07 15:35:21] GPU #5: EVGA GTX 1070, 582.62 H/s
[2020–11–07 15:35:21] accepted: 5/5 (diff 0.000), latency: 25ms 1716.90 H/s yes!
[2020–11–07 15:35:21] accepted: 6/6 (diff 0.000), latency: 43ms 1716.90 H/s yes!

The seed change materialized and worked properly, and should continue without anymore crashes.

We have finally reached our end goal for the RX2 algorithm implementation, and are now in the last phase. This last phase includes tweaking the parameters of the algorithm itself to bring up to our taste.

The most complicated part is officially behind us, and we hope that we can tackle the parameter changes in a reasonable amount of time and without any surprises.

The final steps include internally testing the network with the newly configured parameters and then launching the public testnet — until we reach a fork date.

—

Finally, please feel free to leave us your feedback, as we always welcome it!

-The Luxcore Team

Add a comment

What is your view?

Send

Related posts:

5 things your coming soon page needs

You are super excited to get going, you have your domain name and in the process of working with an affordable website designer or maybe you are designing the site yourself. However, you just…

Que nem uma manga verde

Acho que sou uma pessoa reflexiva. Às vezes me pego pensando demais e até ficando mal com as conclusões das minhas viagens (quando, se quer, encontro conclusões). Você já deve ter ouvido falar que é…

Things I Learned About Agile Transformations While Cooking Risotto

Have you ever gone through an Agile transformation where the Development Teams were doing their part, but Product Management and the rest of the Business kept holding to their old ways? Many see…