Im curious about this,does it matter ? will using pci-e 3.0 card on pci-e 2.0 slot decrase hashrate ?
No, I'm using h81 pro btc with all mine and hashrates are where they're supposed to be
No. But i have the impression that you need PCI x16. Anything lower will damage the hashrate!
NOPE wrong, h81 pro btc has 5 pcie x1 lanes and same hasrate
what OS and miner are you using?
Ive used ethOS and Genoil and Claymore... been same hashrate all across board.
I have 9 rigs with x6 R9 380 each rig so 54 cards total.
whoah ,, how much solve rate you got ? also thanks for info
about 1500 little bit overlocking but each rig is using 300+ watts less than ethereum, which is nice
so do you pay for electric ? also when mining eth what is rigs power draw ?
on a small test rig with three R7 370 vapor x's running win10, amdgpu-pro and genoils0.6 miner I managed to get about 3-5% more Sol/s by taking one card off the riser and plugging it into the 16x slot on the mobo (h81 pro btc). the output of the miner shows that this card also keeps a higher Sol rate then the other two cards. before the change the three cards were doing just shy of 60 Sols and now they are consistently over 60 Sols. Nothing dramatic but a noticeable difference.
I never noticed any difference at all.
Yes I do,
Mining eth I was drawing 1000-1025 watts 8.24 - 8.5 amps
Mining ZEC I am about 700-720 5.9 - 6.2 amps,
I looked into this and am by no means comfortable with the concept but
a) PCIe 2 = 8 Gb/s and PCIe 3 = 16 Gb/s
That is super fast in both cases...do miners benefit from that speed increase? Not sure?
b) PCIe is backward compatible so a GPU with PCIe 3 will run just fine on a PCIe 2 because it uses all available lanes
Other than that I'm not sure.
I think the question is the difference between PCIe 2.0 vs PCIe 3.0 and not whether it is connected to a PCIe x 16 vs x 8 x 4 or x 1
Is your GPU 1 a PCIe 3.0 card on a PCIe 2.0 slot?
not sure, lets assume yes
Mining isn't like playing a video game..
Gpu are just Calculators.
They hash numbers and submit shares
Massive data movement isnt required.
This brings up the big question...what is the prevailing bottleneck?
- For GPUs: RAM vs bus size
- For CPUs: System RAM, Cache sizes, etc
Would love a technical explanation...like Linus Tech Tips
Since GPUs are using less wattage on ZEC vs ETH, it looks like the cores are not being fully utilized. So either there is more algorithm optimization available, or the memory bus is saturated. That's RAM speed times bus size, so it is equally both.
For CPUs, it is more clearly the memory bus because a screaming i7-6700k can't do much better than a new 30 Watt laptop (32 Sol/s) that's using the same memory. However, a i5-3570 (60 W) at the same clock speed as a i7-2600 (100 W) is 25% slower despite the same RAM, so older CPUs have a CPU strength limitation.
edit: there seems to be only a small improvement in going from 4 to 8 GB in GPUs, further indicating memory bus speed saturation.