SETI MB CUDA for Linux

[riofl]

interesting.. didnt know freesoft had a prime95 for linux 64. got that too just to have it. could be useful.

[sunu]

if i remember correctly (it has been almost 2 yrs since i did this) i believe i upped the VID to 1.3 v or so.

Please refer to it as vcore. VID is constant something like a characteristic printed on the chip. 1.3 volts is very very good for 3.6Ghz. I also use 1.3 volts but for 3.24 Ghz.

but since i always really wanted a 3.0 ghz machine but at the time could not afford the processor, and this can safely go to that speed i decided to go against my beliefs so that i had the best of both worlds for myself.

That was also the reasoning beside my decision. I settled for 3.24Ghz. I didn't want to go beyond that since my motherboard is very basic.

interesting.. didnt know freesoft had a prime95 for linux 64. got that too just to have it. could be useful.

Yes, though I've never used it. Yes, it could be useful.

[riofl]

i have discovered the following:

in0 = VCore
in1 = DDR2
in2 = 3.3v
in3 = Vcc
in8 = VBatt.

i have changed the labels accordingly in the lmsensors config. the rest are way off calibration and could not possibly be even close.

the temp1 and temp2 that i think are mosfet and southbridge based solely on hair dryer proximity, may be those or they may be what this guy listed. unfortunately he did not have my motherboard, but a close model gigabyte.. he claims they are

temp1 current system temperature
temp2 current cpu casing temperature

i dont know for sure... gigabyte cannot help at all. the guy sounded like i was crazy. they arent too awfully important anyway as the hottest runner, temp2 has never exceeded 55c on peaks and typically runs 45-50c.

so in0 is the Vcore and its voltage matches the  cpu0_vid... which i find out appears to be a static listing. it never moves so maybe it reads something from the cpu? dunno.

[sunu]

The most important temperatures are the four coretemps. If these are good the rest (almost) don't matter
.

so in0 is the Vcore and its voltage matches the  cpu0_vid... which i find out appears to be a static listing. it never moves so maybe it reads something from the cpu? dunno.

That's why I said that cpu0_vid might be the real VID after all. We'll find out if/when you fire up windows and see what Core Temp says.

[riofl]

VERY interesting!

the sensors core0VID is the VID as reported by the chip itself. still there is a difference between linux and windows.
linux says it is 1.219 while coretemp reports it as 1.2500

results of tests:

idle
core0 = 35
core1=35
core2=33
core3=33
cpuz reports coreV = 1.232

under load after 4 iterations of the smal fft test
core0=56
core1=56
core2=51
core3=51
cpuz coreV=1.2500 initially then after a few seconds drops to 1.216

the motherboard health status states the VCore=1.252

i decided to give myself a "newer" computer to celebrate its 2 yr old birthday 

played around at 3.6ghz but since i have changed ram from the last time (old ram was crucial ballistix new ram is OCZ low voltage Blade series) I find that at 3.6ghz the lowest besides 1000 i can go is 1200 which the ram plainly complains even upped from 1.8 to 2.0 volts which is .1 v higher than recommended maximum. so i had no success in stability at 3.6ghz so i dropped it to 3.45ghz along with 1152 on the ram (base is 1066) and VCore set at 1.312v. it appears to be rock stable. I also found that there is a 5c approx difference now between auto and max on the cpu fan so it is set at max and my temps are not much different than at 3.0ghz. maybe 2c higher.. basically nothing... the system passed small fft , large ffp and mixed bag with flying colors. i then booted into memtest86 and ran 2 iterations of its test and all passed just fine.

so i am leaving it at 3.45ghz for now and will run all my apps, test movies, audio, everything.

interersting day

edit addition:
interesting ... the vid cards are running hotter. does that mean maybe my 0.35cpu settings in app_info file and the faster cpu are driving the vid card harder? the 285 is averaging maybe 2c higher than normal but the tesla is running a good 10c hotter! it is now averaging 73-77c at max fan! maybe i jarred it a bit too much in cleaning the system and caused some of the thermalpads to fall or a break in the gpu paste... it shouldnt do that but that thing has been one of those 'jiggle it till it works then dont touch it and hope it stays that way' items.... guess i need to replace it sooner than i wanted to...

20 min later...:   it appears that the problem was with the cpu fan running full bore now (2750rpm approx instead of 2480 approx), it upset the rather delicate airflow balance i had achieved in the case. i had to readjust all fans now the tesla is peaking at 68c which is approx 2c or so warmer than it used to and more than acceptable).

[sunu]

the sensors core0VID is the VID as reported by the chip itself. still there is a difference between linux and windows.
linux says it is 1.219 while coretemp reports it as 1.2500

Why do you say that? We were never sure about core0vid and as I said in another post, 1.219 as a VID, was a strange number anyway. Coretemp clears the things for us, your chip's VID is 1.2500 and that core0VID from lm-sensors is something else.

under load after 4 iterations of the smal fft test
core0=56
core1=56
core2=51
core3=51
cpuz coreV=1.2500 initially then after a few seconds drops to 1.216

1.216 volts under load? At 3Ghz? Might be dangerously low. Maybe not. See also below.

played around at 3.6ghz but since i have changed ram from the last time (old ram was crucial ballistix new ram is OCZ low voltage Blade series) I find that at 3.6ghz the lowest besides 1000 i can go is 1200 which the ram plainly complains even upped from 1.8 to 2.0 volts which is .1 v higher than recommended maximum. so i had no success in stability at 3.6ghz so i dropped it to 3.45ghz along with 1152 on the ram (base is 1066) and VCore set at 1.312v.

What vcore at 3.6GHz? Might not be the memory's fault but too low vcore. You also say "vcore set at 1.312v". Where, in bios? Don't look there. Look at what cpu-z says under 100% load. That's the one to watch. Before I pencil moded my motherboard I had to set the vcore in bios at 1.4750 in order to achieve 1.3000 under 100% load.

the system passed small fft , large ffp and mixed bag with flying colors. i then booted into memtest86 and ran 2 iterations of its test and all passed just fine.

How much time did you let prime95 run? Many people consider a system stable after 12, 18 or even 24 hours of prime95 load.  For my overclock of 3.24Ghz I went over the board and ran prime95 for three days: one day small FFTs, one day large and one day mixed  Also for memtest, I think they recommend to also leave it for many hours to be sure that your ram is good and stable. Whenever I buy new ram the first thing I do is run memtest for 24 hours.

[riofl]

according to several texts the VID is reported by the chip as its nominal operating voltage with stock settings.

i tried VCore all the way to 3.8v .. it ran fine at the lower voltage of 3.68 when i kept the ram set at 1000 with stock timings but i didnt want to bottleneck things so i decided to go down to 3.45ghz which also ran more within my 'comfort range' in cpu temps.

the voltage at 3.0ghz was the 'normal' bios setting and voltages were taken off auto. it ran like that since approximately july 2008 and never locked up.

i agree on the test times .. they should be measured in days, but i cannot be without this machine that long.. i cannot even be without it for more than 2 or 3 hours at most is why i used short tests. if it is going to act up it will do so while i am using it and it is a simple thing to drop it down in minutes if that happens. heck when i switched ram, on customer machines i usually run memtest for a minimum of 24 to 36 hrs, but on mine again due to time constraints, i have found with the way i use the machine if it is gonna mess up it will do it within 2 iterations.

i cant keep  this down long enough to do proper 'burn  in' etc... and i do not have a 2nd machine that is capable of taking its place while being burned in or repaired, so i just have to keep watch on things as i work and if necessary adjust which takes a few min max. due to ship times and the long times it takes to run into town to buy parts, i have spares of every part in the closet so i can do a fast replacement if need be. only spare i dont have is the tesla. i know i could replace it with the spare 285 but that just doesnt sit right with me. a spare is a spare and if i used it i would have to buy yet another one for a spare.

after i build up a newer workstation for myself, i will keep spares of that one, and will use the old spares in another machine to clone the one i have.

then i will either keep them both or sell off one of them. prob. will keep them both since this one proves to be a decent rac bulder compared to the other machines..

[riofl]

forgot.. at 3.45ghz, cpuz reports CoreV to be 1.296. the number i gave in the msg above was the number set in the bios and it also agreed in the bios voltage reading in the 'cpu health' section.

the vcore reading in gkrellm using in0, is 1.31

[sunu]

Very nice "policy" about the spares, but very costly...

[riofl]

thankfully the boss pays for 50% of the cost of spares and he would have a fit if i told him i used a spare as a 2nd card. he has spares for everything in the office and the NOC.. not for each server but since the servers are clones he has one of each  part used in a single computer. doesnt need any more than that really.

i love the spare policy but if i had to foot the bill totally myself i doubt i would have spares of everything, probably just ram and hard drives and fans which are all the most common parts to go.

[riofl]

wish he counted cases as 'spares' as well.. hehe then i wouldnt have to buy anything to build the extra computer once i replace this one.

[riofl]

what do you run on your cpus? multibeam or astropulse? do you find astropulse gives higher credit than the equivalent time with MB would?

[Raistmer]

what do you run on your cpus? multibeam or astropulse? do you find astropulse gives higher credit than the equivalent time with MB would?

There is no AP work now at all, so question highly theoretical one 

[riofl]

ahh .. ok.. guess i will wait to wonder until they bring it back if they do.

thanks

[sunu]

what do you run on your cpus? multibeam or astropulse? do you find astropulse gives higher credit than the equivalent time with MB would?

Well, what raistmer said.
But when astropulse comes back, I don't think there is currently a credit advantage over multibeam. Moreover our Q6600 is not strong enough to support, for example, 4 astropulse workunits plus 1, 2 or more cuda workunits. In that case we would lose RAC.