|
Title: Avalon performance Post by: WEB slicer on July 27, 2013, 04:34:34 AM What is considered acceptable HW percentage? All mine run at 350 with HW around 2-3% of diff1shares. Do avalons have HW at stock frequency? Should i downclock to no HW? Is no HW errors worth the lower frequency? I want to try for 360 but i hear dont push past 350. What is unsafe temperatures for avlon?
Title: Re: Avalon performance Post by: crazyates on July 27, 2013, 05:39:12 AM I think you'll have to experiment to see what's more profitable for you. For example, if you OC and get 10% more hashrate, but go from 1% to 3% HW errors, you've still gained ~8%. If you OC by 20% and start getting 30% HW errors, then you come out negative.
Title: Re: Avalon performance Post by: elasticband on July 27, 2013, 05:57:28 AM what firmware are you running?
I have 2 of my batch 1 units overclocked at 355 and they run about 2GH more than my other units..... mine run at 355 with under 1% HW My other machines run at 345 and get around 0.5% HW what temperatures are you displaying? Title: Re: Avalon performance Post by: WEB slicer on July 27, 2013, 06:17:56 AM Most of them have the latest firmware. Temps are 45-50.
Title: Re: Avalon performance Post by: elasticband on July 27, 2013, 06:25:05 AM temps are fine, anything under 55c works
I use the user strombom firmware as it has produce more stable results for me, i really don't mind the fans at 100% though Title: Re: Avalon performance Post by: WEB slicer on July 27, 2013, 07:11:24 AM They all have firmware 20130703. Is this the latest? I set fans to 100%. Some run 3700, 3800, 3900. Why different speeds? Batch 1 doesnt fix fans to 100%. Fans speed up when temps climb.
Title: Re: Avalon performance Post by: -ck on July 27, 2013, 07:43:19 AM Every Avalon will be different just like every GPU will be different so they'll all have different overclock abilities. ASICs like these have hardware errors all the time by design and they're harmless. Fanspeeds at 100% are probably a bad idea because of the poorly designed fan layout in the avalon, it will make the thinner 25mm thick fan blowing to the PSU fail to work because of massive case air pressure and the air will be going backwards. Reported fanspeeds are very inaccurate on the sensors so don't bother trying to compare small differences in their RPMs. With my firmware, hardware errors are not counted as hashrate, so any rise in hashrate reported is a real rise in throughput and should be your ultimate decider in how high you go.
Anything else? ;) Title: Re: Avalon performance Post by: WEB slicer on July 27, 2013, 05:06:18 PM So should i push the chip frequency to highest stable speed under 55c and completely ignore HW ?
Title: Re: Avalon performance Post by: WEB slicer on July 28, 2013, 01:50:48 AM I received a 4 module avalon today. The PSU was not inside the avalon. The seller emailed me and told me not to put the PSU inside the avalon cause it will over heat. This does not seem right. Im assuming if it overheats with the PSU inside there is something wrong with it. What do you guys think?
Title: Re: Avalon performance Post by: elasticband on July 28, 2013, 06:24:31 AM Everything is fine, most people who have 4 modules seem to have their PSU outside in the pictures,
this is due to heat and some PSU(1000w+, which i think is needed for overclock) will not fit into the unit like the smaller PSU All is fine with your unit i am guessing, i think you will jsut need to block the space at the back where the PSU was if this effects airflow inside anyway |