Already upgrade, look coinsforall.io if you need low latency.
Last news about CUDA version, GPU code finished, all benchmarks works. Only miner code remaining, it's not so hard...
- GeForce GTX 1070; Compute capability 6.1
[1] GeForce GTX 750 Ti; Compute capability 5.0
Benchmarking GeForce GTX 1070; 15 compute units
square 320 bits: 36.326ms (3694.812M ops/sec)
multiply 320 bits: 55.180ms (2432.362M ops/sec)
square 352 bits: 44.570ms (3011.392M ops/sec)
multiply 352 bits: 66.563ms (2016.401M ops/sec)
Fermat tests 320 bits: 90.851ms (2.885M ops/sec)
Fermat tests 352 bits: 119.466ms (2.194M ops/sec)
*** hashmod benchmark ***
MHash per second: 737.312
Hash per iteration: 36.969 (0.000441 %)
Average hash multiplier size: 30.779
Hashed with primorial 13 is 12.637%
Hashed with primorial 14 is 68.174%
Hashed with primorial 15 is 19.189%
*** sieve (check) benchmark ***
*
[OK] found candidates by CPU: 1914 by GPU: 1914
* [OK] invalid candidates: 0
* [OK] CPU/GPU candidates difference: 0
*** sieve (performance) benchmark ***
* scan speed: 77.071 G
* iteration time: 10.714ms
* candidates per second: 184692.857
* candidates per iteration: 1978.83 (854.16 320bit, 1124.67 352bit)
* 320bit/352bit ratio: 0.759/1
Benchmarking GeForce GTX 750 Ti; 5 compute units
square 320 bits: 151.284ms (887.191M ops/sec)
multiply 320 bits: 208.753ms (642.950M ops/sec)
square 352 bits: 169.990ms (789.562M ops/sec)
multiply 352 bits: 254.647ms (527.074M ops/sec)
Fermat tests 320 bits: 326.958ms (0.802M ops/sec)
Fermat tests 352 bits: 475.107ms (0.552M ops/sec)
*** hashmod benchmark ***
MHash per second: 133.405
Hash per iteration: 37.844 (0.000451 %)
Average hash multiplier size: 30.626
Hashed with primorial 13 is 13.666%
Hashed with primorial 14 is 68.043%
Hashed with primorial 15 is 18.291%
*** sieve (check) benchmark ***
* [OK] found candidates by CPU: 1398 by GPU: 1399
* [OK] invalid candidates: 0
* [OK] CPU/GPU candidates difference: 0
*** sieve (performance) benchmark ***
* scan speed: 14.236 G
* iteration time: 58.005ms
* candidates per second: 34194.568
* candidates per iteration: 1983.47 (766.39 320bit, 1217.08 352bit)
* 320bit/352bit ratio: 0.630/1