LBRY Sp-mod #3 vs 1.8.2
Gtx 750ti sp-mod #3 is the fastest
Gtx 950 sp-mod #3 is the fastest
Gtx 960 sp-mod #3 is the fastest
Gtx 980 sp-mod #3 is the fastest
Gtx 980ti sp-mod #3 is the fastest
Gtx 1060 3gb sp-mod #3 is the fastest
Gtx 1060 6gb sp-mod #3 is the fastest
Gtx 1070 sp-mod is the fastest
Gtx 1080 sp-mod is the fastest
Sp-mod #4 vs 1.8.3
Gtx 750ti sp-mod #4 is the fastest
Gtx 950 sp-mod #4 is the fastest
Gtx 960 sp-mod #4 is the fastest
Gtx 980 sp-mod #4 is the fastest
Gtx 980ti sp-mod #4 is the fastest
Gtx 1060 3gb sp-mod #4 is the fastest
Gtx 1060 6gb sp-mod #4 is the fastest
Gtx 1070 sp-mod #4 Same speed
Gtx 1080 sp-mod #4 is the fastest
Salut from paris.
Sp-mod LBRY #4 has been sendt to the LBRY donators.
-Fee reduced to 2.5% (3 minutes for me, 117 for you)
-speedup on the gtx 1060, and small speedups on the rest of the cards. compute 5.0 + only
-Fixes in the donation code.
You can become a donator. Send 0.1 BTC to the adress in my signature and recieve free updates

Just received, thanks
Here's the situation on a Platform with 5 1070:
SP4 (0.1 BTC + 2.5% fee. Hashrate * 0.975 = final value) = 284.505 MH/s on avg
*: I'll need to test to check it this SP4 has the same reconnections issue as the SP3 kernel. I suppose not

Tpruvot (no fee) = 295.8 MH/s on avg

I haven't got a 1060 to test, but it's a long way to go. Any suggestions?
I can confirm that 1.8.3 it's very stable.
SP4 on avg 290MH without the fee. Considering the fee 290*0.975 = 282.75 / 284.73 MH/s