may have got it incorrect an little but on the Single clients alone it should be no worse then 2-5GB per month for 400 single console clients (projects that take 2-3 days would be more like 2gb an month, for short projects that take 3-7hrs more towards 5gb an month) i think if you stick with small packets and adv off and no passkey you get mostly the 72 day projects not sure as i do not monitor the cpu clients at all apart from the 4 i have got here i got adv on and big on but no passkey and i am getting 7-10hr turn around projects with 32day exp on them, i get 100 more PPD with Big and adv on (set in the cfg file and service start) i run all my CPU clients at 82% as that stops the systems from slowing down or been sluggish, as some of the projects work the CPU really hard - do you do any bigadv SMP clients as i work out can 3-5gb an month per client (working off 120mb per packet taking 3-2.5 days as the upload back is like 100mb as well) {not sure how big the bigadv work units are but sure they was around 100mb} GPU clients do not seem to be that big about the size of the CPU clients packet size, but most are done in 2-3 hrs (9800 type cards) so more data would be used so 12 GPUs could use 500mb-1gb an month, what gpus are you using
ok may have it more incorrect then i thought HFM.net has an neat feature that allows you to see the logs of the last 4-7 projects (look at the before decompress size and at the end look at the result writing data size) ok cpu clients not sure if they are compressing the finished data (the GPU clients do state it) so below is not sure at the moment, if it is not compressing then what i have posted is aprox correct but you need to work it out your self (or give me 10-20 completed logs files or the start and end of them with the sizes) as your CPU clients mite be faster then an 2.5 Phenom 9850, not that it matters now due to ISP change my 3g dongle that is running an 9800gx2 and 4 console clients with adv option set and packet size big has reported 35MB today and that's just me remoting in for short times and norton 2010 pulse updates + the folding if the pcs are been used for internet actively then 130gb would not be that hard if the pcs used every day by users
Hey thanks for that, looks like its a little bit tricky to estimate but if your 3g dongle hitting 35mb just with a little use it might be possible to hit 30gb upload with all the pcs and gpus. Its ok though prob best to move isp anyway, strange i have only re-enabled 5 gpus and 2 smps through the alternate isp line (which wont be upgraded until 1/03 but should be ok for just a few fah cores) and im already near 50k. so its entirely possible i should have been hitting much higher points with a few hundred extra clients, gpus etc. It did used to fluctuate and nomatter how many more gpus it wouldent really go over 160k so i might have been hitting some sort of bandwidth limit. thinking about it lots of cores did seem to be always waiting for packets....... could be a busy week getting them all back on and pointing to the new connection. thanks again for going to the effort of calculating it for me
Im just doing smp in windows with no -bigadv, although i have put three of them on the passkey thing hoping they will get the a3 cores? but only one is other two are still doing a1 cores
if they fail to connect to the A3 server you get A1 work units (until they run out of A1 then it keep trying for A3)
