Topic: Memory errors, do they really matter? (Read 274 times)
avoid memory errors - wasting energy
Long story short, I was getting around 30.9 MH/s ETH + 540 MH/S Sia and I was pretty happy about it.
The reported hashrate on my pool was quite lower (27.5 ish) tho on an average of 6 hours.
I started looking for information and I've read that memory errors can cause it, since memory errors can make many calculations wrong or lose them.
So I launched HWInfo and I've seen that I have TONS of errors, like dozens per second.
I then started underclocking my memory a bit, went from 2230 (strap 1750) to 2190.
I need to drop almost 40 mhz to see a much smaller number of errors, but it also drops the hashrate.
So, what is your position about memory timings?
Lots of memory errors will crash your rig... little bit errors could make bad shares and few errors aren't matter. The reported hashrate on my pool was quite lower (27.5 ish) tho on an average of 6 hours.
I started looking for information and I've read that memory errors can cause it, since memory errors can make many calculations wrong or lose them.
So I launched HWInfo and I've seen that I have TONS of errors, like dozens per second.
I then started underclocking my memory a bit, went from 2230 (strap 1750) to 2190.
I need to drop almost 40 mhz to see a much smaller number of errors, but it also drops the hashrate.
So, what is your position about memory timings?
Long story short, I was getting around 30.9 MH/s ETH + 540 MH/S Sia and I was pretty happy about it.
The reported hashrate on my pool was quite lower (27.5 ish) tho on an average of 6 hours.
I started looking for information and I've read that memory errors can cause it, since memory errors can make many calculations wrong or lose them.
So I launched HWInfo and I've seen that I have TONS of errors, like dozens per second.
I then started underclocking my memory a bit, went from 2230 (strap 1750) to 2190.
I need to drop almost 40 mhz to see a much smaller number of errors, but it also drops the hashrate.
So, what is your position about memory timings?
The reported hashrate on my pool was quite lower (27.5 ish) tho on an average of 6 hours.
I started looking for information and I've read that memory errors can cause it, since memory errors can make many calculations wrong or lose them.
So I launched HWInfo and I've seen that I have TONS of errors, like dozens per second.
I then started underclocking my memory a bit, went from 2230 (strap 1750) to 2190.
I need to drop almost 40 mhz to see a much smaller number of errors, but it also drops the hashrate.
So, what is your position about memory timings?
In my experience memory errors make for unstable rig.
Yours truly,
Nate Higgers
Long story short, I was getting around 30.9 MH/s ETH + 540 MH/S Sia and I was pretty happy about it.
The reported hashrate on my pool was quite lower (27.5 ish) tho on an average of 6 hours.
I started looking for information and I've read that memory errors can cause it, since memory errors can make many calculations wrong or lose them.
So I launched HWInfo and I've seen that I have TONS of errors, like dozens per second.
I then started underclocking my memory a bit, went from 2230 (strap 1750) to 2190.
I need to drop almost 40 mhz to see a much smaller number of errors, but it also drops the hashrate.
So, what is your position about memory timings?
The reported hashrate on my pool was quite lower (27.5 ish) tho on an average of 6 hours.
I started looking for information and I've read that memory errors can cause it, since memory errors can make many calculations wrong or lose them.
So I launched HWInfo and I've seen that I have TONS of errors, like dozens per second.
I then started underclocking my memory a bit, went from 2230 (strap 1750) to 2190.
I need to drop almost 40 mhz to see a much smaller number of errors, but it also drops the hashrate.
So, what is your position about memory timings?
Jump to: