Sure. Setup is
2x Intel(R) Xeon(R) CPU E5-2603 v3
32GB DDR4
3x NVIDIA GeForce GTX 980 Ti
I assume kernel is 4.4.0-31-generic, but since I'm no expert in the Linux world;
uname-a: 4.4.0.-31-linux-generic
cat /proc/version_signature: 4.4.0-31.50-generic 4.4.13
Only used NVIDIA driver 367.35 so far.
I've tried now with both (desktop) 14.04 TLS and (server) 16.04 TLS, and on both versions hashcat v3.0 issues these warnings (and won't retain the temperature specified with the flag**), whereas cudaHashcat v2.01 works just fine without issuing any warnings, which is frustrating since everything else is the same; drivers; installed libraries, etc.
If I set gpu-temp-retain=0, hashcat prints this during the initial phase;
Watchdog: Temperature abort trigger set to 90c
Watchdog: Temperature retain trigger set to 75c
While running under these settings, the temperatures stabilize at about {76c/77c | 83c | 78c/79c} ({device0|device1|device2}).
Some times setting gpu-retain-flag=0 (or a small number) results in no warnings about initial fan speed, but other times the warnings are issued. Kinda confusing.
But just to be clear about it, system works and is stable at the temperatures I've mentioned above (e.g. getting about 18791 MH/s total for all three devices for SHA1 during benchmark and about 17000 MH/s for brute force attack on unsalted SHA1; not sure how that is compared to tuned setups with the same devices, but at least it does the job for my need atm). So in that sense, this isn't a "big" deal, but it'd be nice to find out what causes this and how it can be "fixed". Obviously something isn't as it should be.
2x Intel(R) Xeon(R) CPU E5-2603 v3
32GB DDR4
3x NVIDIA GeForce GTX 980 Ti
I assume kernel is 4.4.0-31-generic, but since I'm no expert in the Linux world;
uname-a: 4.4.0.-31-linux-generic
cat /proc/version_signature: 4.4.0-31.50-generic 4.4.13
Only used NVIDIA driver 367.35 so far.
I've tried now with both (desktop) 14.04 TLS and (server) 16.04 TLS, and on both versions hashcat v3.0 issues these warnings (and won't retain the temperature specified with the flag**), whereas cudaHashcat v2.01 works just fine without issuing any warnings, which is frustrating since everything else is the same; drivers; installed libraries, etc.
If I set gpu-temp-retain=0, hashcat prints this during the initial phase;
Watchdog: Temperature abort trigger set to 90c
Watchdog: Temperature retain trigger set to 75c
While running under these settings, the temperatures stabilize at about {76c/77c | 83c | 78c/79c} ({device0|device1|device2}).
Some times setting gpu-retain-flag=0 (or a small number) results in no warnings about initial fan speed, but other times the warnings are issued. Kinda confusing.
But just to be clear about it, system works and is stable at the temperatures I've mentioned above (e.g. getting about 18791 MH/s total for all three devices for SHA1 during benchmark and about 17000 MH/s for brute force attack on unsalted SHA1; not sure how that is compared to tuned setups with the same devices, but at least it does the job for my need atm). So in that sense, this isn't a "big" deal, but it'd be nice to find out what causes this and how it can be "fixed". Obviously something isn't as it should be.