Page 2 of 2
Re: CUDA invalid value for gpu-architecture
Posted: Wed Jan 06, 2021 7:25 am
by bruce
smurfcorpse wrote:The GTX 1030 couldn't beat my current GTX 580 in overall performance.
I wouldn't expect it to.
You may already know this, but to decode NVidia's model numbering, you're comparing a generation 10* GPU to a generation 5* GPU (which does sound good) but the speed is encoded in the *30 vs. the *80 digits. A *10 is always going to be slower than a *80 GPU, provided they both have the necessary features to run FAH. The first digits do matter, but not as much as the latter.
Re: CUDA invalid value for gpu-architecture
Posted: Thu Jan 07, 2021 12:31 am
by psaam0001
Just to follow up: I did check for GT 1030's for my Dell (for the x8 length slot). EVGA does make one, but they are currently out of stock on it. I'm waiting to get an e-mail from them when/if it is available again. I'll only be able to add one of them (the heat sink would block a slot, so I'll sacrifice the old PCI slot to use that card as a 3rd usable for F@H GPU). So I'd also take the opportunity to replace the 10+ year old SAS HD's & controller with more a current 1TB SATA drive that I can connect to the motherboard.
Paul