You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Apr 24, 2022. It is now read-only.
Build: v0.19.0-17-gce52c74021b6fbaaddea3c3c52f64f24e39ea3e9-dirty
Platform: Linux
Compiled with OCL v2.2 on amdgpu and CUDA/Nvidia-OCL v495 on nvdia.
Note: this is for recreational mining on GPU downtime and profitability is not a concern (and please do not dismiss on that ground).
Titan RTX consistently yields a hashrate of around 57Mh, give (for CUDA) or take (for OCL). This is pretty consistent regardless of whether using CUDA or OCL across various settings in both benchmark mode or solo mining mode. Relative to the Radeon TX 580 GTS overclocked at 2GHz, which is 5 ~6 times cheaper than a Titan RTX and produces a maximal stable hashrate of 24.8Mh, the Titan RTX is yielding barely twice the hashrate. This is a rather disappointing outcome. Of course, it's usually hard to compare two architectures based on hardware capacity alone. However, there are a couple indications that this behavior is a bug:
The Titan RTX is underutilized as wattage is consistently 20W on average below maximum at any maximum set via nvidia-smi (the card has its own power supply) and temperature + fan speed don't really go up (temp stays in low 80s with fan speed stays in mid 60s) despite a target temp of 88C, and clock speed stay around 1.8Ghz despite a lock of 2Ghz. A 4K movie or an old game title can work it harder than that. It may be bottlenecked on an unobserved metric, but I can't get the performance up using ethminer settings despite this clear underutilization. The Radeon card is up against all critical/max metrics at maximal stable hashrate.
Just for comparison, using OCL-flavored btx miner nsgminer, I can achieve a maximal stable hashrate of 3.6Ghz on Titan RTX relative to 0.9Ghz on RX 580 GTS. That's 4 times the yield and much better within the expected performance to price ratio, and that's with OCL being suboptimal for nVidia GPUs. Again, no problem with underutilization here, so it can't just be an issue with my nvidia settings throttling the GPU.
Ideas? Thanks.
The text was updated successfully, but these errors were encountered:
Sign up for freeto subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Build: v0.19.0-17-gce52c74021b6fbaaddea3c3c52f64f24e39ea3e9-dirty
Platform: Linux
Compiled with OCL v2.2 on
amdgpu
and CUDA/Nvidia-OCL v495 onnvdia
.Note: this is for recreational mining on GPU downtime and profitability is not a concern (and please do not dismiss on that ground).
Titan RTX consistently yields a hashrate of around 57Mh, give (for CUDA) or take (for OCL). This is pretty consistent regardless of whether using CUDA or OCL across various settings in both benchmark mode or solo mining mode. Relative to the Radeon TX 580 GTS overclocked at 2GHz, which is 5 ~6 times cheaper than a Titan RTX and produces a maximal stable hashrate of 24.8Mh, the Titan RTX is yielding barely twice the hashrate. This is a rather disappointing outcome. Of course, it's usually hard to compare two architectures based on hardware capacity alone. However, there are a couple indications that this behavior is a bug:
The Titan RTX is underutilized as wattage is consistently 20W on average below maximum at any maximum set via
nvidia-smi
(the card has its own power supply) and temperature + fan speed don't really go up (temp stays in low 80s with fan speed stays in mid 60s) despite a target temp of 88C, and clock speed stay around 1.8Ghz despite a lock of 2Ghz. A 4K movie or an old game title can work it harder than that. It may be bottlenecked on an unobserved metric, but I can't get the performance up usingethminer
settings despite this clear underutilization. The Radeon card is up against all critical/max metrics at maximal stable hashrate.Just for comparison, using OCL-flavored btx miner
nsgminer
, I can achieve a maximal stable hashrate of 3.6Ghz on Titan RTX relative to 0.9Ghz on RX 580 GTS. That's 4 times the yield and much better within the expected performance to price ratio, and that's with OCL being suboptimal for nVidia GPUs. Again, no problem with underutilization here, so it can't just be an issue with my nvidia settings throttling the GPU.Ideas? Thanks.
The text was updated successfully, but these errors were encountered: