PrimeGrid
Please visit donation page to help the project cover running costs for this month

Toggle Menu

Join PrimeGrid

Returning Participants

Community

Leader Boards

Results

Other

drummers-lowrise
11) Message boards : Number crunching : Running 2 WU in 1 GPU (GTX560) (Message 41077)
Posted 3643 days ago by uBronan
Well lol when i see the load which is 98-100% i would not advise you to do that since the load is more then enough, running 2 instances for just 2% is kinda a non go for me.
The gpu is constant jumping between 98 and 100% means in my book a 100% load.
If it was under 90% i would try it, the trick is to just use this when the gpu is not fully loaded and enough percentage is not used to gain some.
The lower the load is over time say for instance the einstein app which does only give a gpu load of 40% so 2 instances is fine. Now here you would think 3 instances would work but einstein is a tricky one because it needs between 300 and 450 Mb video ram. So if your card has 1 gb ram running 3 is out of the question :D
The reason why nvidia cards perform better is because cuda is a much older and mature processing tool.
ATI came later on the gpu processing train, so they have a lot to catchup.
Our hope is that opencl will narrow the gap between them, but ofcourse nivida as being the bigger and richer probably spends more money to research and develop.
12) Message boards : Problems and Help : How do I pick up work I do not have, but server says I am running it??? (Message 41075)
Posted 3643 days ago by uBronan
Seems to me the boinc map has some weird configured files if it keeps messing up so much, i agree with mikey to let it run empty and then uninstall boinc
I would then delete the whole boinc data map and start all over with a fresh install
13) Message boards : Project Staging Area : llrCUDA testing (Message 41043)
Posted 3644 days ago by uBronan
As i remember one comment from a gpu guru.
The problem with gpu computing was that normal integer calculations is not the strongest point for these for this function a cpu is still much more effective.
ATI is ofcourse way behing nvidia on the gpu processing simply because they had to start from scratch while nvidia already had made cuda.
But since some good guy found a way to use their gpu was able todo some work lightning fast they stepped on the moving train which we call gpu processing as well.
So the primary language cal was not anything near cuda in terms of usabillity for do the somewhat complexer tasks
Then ati decided that opencl was probably a good bet, which would make gpu processing for all software developers a change to be on the same train.
Do not forget opencl is a standard where also nvidia has a lot to say, as the biggest and richest videocard maker.
But at the moment we see that even though it is becoming better every year still is behind the cuda development. In fact i blame nvidia for not helping much in the opencl development cause they rather want people to adopt the non open cuda standard
Nevertheless i still think that in the near future we probably have nice goodies coming for us with the raw power gpu's has to offer all of us
14) Message boards : Number crunching : The Autumnal Equinox Challenge (Message 40951)
Posted 3645 days ago by uBronan
Sadly Mark it only gives a small increase, but turning programs off which eat unneeded cpu time can pack a punch sometimes
Ofcourse some like firewalls, antivirus and system processes should stay running but for instance almost everyone has some programs running like
Sound managers, Adobe start manager and such those i allways kill
One notorious cpu hogg is the side bar annex the nice gimmicks on your dekstop :D
15) Message boards : Number crunching : The Autumnal Equinox Challenge (Message 40947)
Posted 3645 days ago by uBronan
Speaking of memory bottlenecks,

I am running a Phenom II x6 1055T overclocked to 3.4 ghz. I have four sticks of 2gig DDR2 memory running at 800mhz (stock).

My first six Woodall w/u's took a lot longer than I expected (~246,000 seconds).

Now I am only running four Woodall's to see if my times decrease; I'll post my results in a follow-up 50+ hours from now ;)


Here's the question:

Is there a way to assign each instance of LLR to a specific stick of memory? My gut feeling is that all of my work units are being crunched on a single stick of memory, thus creating a traffic jam, especially so because I'm stuck with DDR2.

If this were possible, would it make a difference?

Neo
Member of Aggie the Pew (AtP)



About your question on memory
There is no need for any change in memory, the only 2 things which do affect the result a bit is overclock or buy 1066 Mhz memory or faster
But to my personal experience its all not worth the trouble
ram is extreme fast so its almost impossible to become a bottleneck

I see you have a very nice overclock on your 6 core sadly my x3 becomes unstable over 3.1 Ghz
But i can give you a few small tips which works well on any machine and for long running units like these woodall make a small difference

First check that your not running too many processes which are not needed
especially check what programs use (too) much cpu time.

Second assign a core to each of the separate threads you can do this by opening the taskmanager and then right click on every and each running project. Make sure not to kill important programs ;)

Another test you could do is to lower the overclock and see if things speed up as i did with my intel, the higher clocked the machine the slower it performed.
I had it running at 3 Ghz but it was finishing jobs slower then when running on its stock speed of 2.1 Ghz
A overclock does not mean it allways makes things faster only by testing you can see if performance is gained
16) Message boards : Number crunching : Strange times (Message 40673)
Posted 3649 days ago by uBronan
Did you compare the instruction sets ?

When you overclock that does not say anything about performance increase also
this is a complete science in itself
You asume that if it gets a higher clock rate the performance will be higher, this in fact is not allways true
For instance i ran 2 q6600 one at 3,2 ghz and the other at 3 ghz
Performance wise the 3 Ghz one beat the faster one in many projects
why i have no clue since everything was the same, mainbord, memory, timings, vga.
But no matter what i did the faster clocked q6600 was allways slower.
So even though it ran 200 mhz faster was not performing better, even switching the cpu's gained me nothing so why does a cpu same model, same brand beat its counter part ?!? if you know that tell me.
The funny thing is when not overclocked and run exactly as intel intended they where exact matched.
Sadly the faster one blow a while ago so i was left with the 3 Ghz one which now runs at 3.2 Ghz with a humongous scyte mugen 2 cooler which hardly fits in the case. I got it so far by lapping the cpu which caused it to run cooler.

With your calculations you count 3 cores and add the results then divide that result by 3 thats another thing i would never do.
I take a complete list of done units on the same machine and divide this by the number done, so if it did 90 units i take the total time / 90 which gives a more acurate average result.
Because simply most units have a different run time and at longer time periods with many more units gets even out over time.

The answer on question one is clear size of caches can have a huge impact, that includes also the other caches, also the type of cache used can have a huge impact. I have here a older opteron cpu which should be much slower then the newer models, but in some cases runs much faster then its newer counter part compared to similar speeds ( i mean by that a 2.2 Ghz opteron gets compared by me by a underclocked x3 set at the same 2.2 ghz not his proper speed 2.8)

Your second question i do not understand when i run 3 units on my machine which means all 3 cores have one unit they actually end in similar times again over time since all units have a more or less different run time.
What has a big influence on performance is that even though boinc gets the major time from the cpu cores other processes running need some time from your cpu as well, this can have a small or bigger effect depending on how much cpu power a process needs.
for instance most people have a firewall, antivirus and other tools running which ofcourse have an impact on your cpu time.
What i am saying with this is that there are so many factors involved that simply compare the same cpu can be a burden even if is exact same machine.
In my case i use the amd fusion tool to almost disable not needed processes, this has a huge effect on time spend on the projects resulting in more efficient use for boinc.
But makes the normal use of the pc a bit handycapped you need to switch profiles to make normal use of the pc when doing this cause everything not needed for boinc gets switched off including several normal system processes.
For some users this is not the solution and i will not recommend people to use it if they do not understand the concept of turning of processes needed.
So in short for me a x2 is most of the time slower since the os needs cpu time as well, when using a x3 or more core cpu's you actually add 1 or more cores which are in fact never used by the os in general
So that can explain a bit why the more core cpu preforms better but i would not dare to claim this as a rule, i just state that this is my observation over the time i use computers.
If you ever tasted a more cpu machine and those where much cheaper then they are now everybody would love to have those. The machines i played with had 2 to 4 cpu's ranging between 4 to 8 cores and where absolute super machines.
But the price of those is also a bit super, go check EVGA the super socket 1366 mainboard for dual cpu machines if i am right still costs over 500 Euro. Thats just the plain mainboard.
17) Message boards : Number crunching : run times with a GTS450 "eco" (Message 40611)
Posted 3650 days ago by uBronan
Einstein is not usefull since there is much difference in runtime between units
Especially since also the settings for cpu load can differ much
I have set ncpu at 0,04 the run times was increase enormous so as a test i now put ncpu at 0,50
And then you will see a faster ending time on gpu but a much bigger load on the cpu as well.
When running 2 units at the gpu that does render a whole cpu core useless for my prime95

Projects like primegrid however run fine next to prime95
Sadly you opened a complete new thread for compare the cards while a older one was already present with some results
18) Message boards : Number crunching : Strange times (Message 40588)
Posted 3650 days ago by uBronan
Now this is a clear case, at first i was not sure what he meant
But when he posted its a X2 all became clear
Amd uses for constructing cpu's a process, they in fact just produce quad cores or 6 cores models

In the process some cpu's fail to the demands for these 4 to 6 cores these failing cpu's will however not be throw away but re used if possible.
They simply get rebranded as long as they can function as a different model such as a 4,3,2 core cpu depending on the failing cores.
At some times the demand for say 2 core cpu's is bigger then the supply of failed quads so they use good ones and turn some of the cores off
What we are here talking about is that by an option on many mainboards you can enable those damaged or turned off cores.
In your case even though the cores work they simply show why they where turned off, they simply did not meet the demands for good function.
So in your case you have a failing quad instead of a quad with 2 cores turned off.
Nevertheless you still may consider yourself lucky that they still function, cause in many cases those cores crash the machines.
That they underperform is not as bad as it seems you still have a usable quad although it is on 2 cores slower
So you can not in any way compare the results of failing cores with the ones
which are intended to run as they should
Its just you gambling being lucky
I am doing the same soon i am going to buy and amd 6950 videocard and going to flash the bios making it think it is a 6970 which if i am lucky enables the disabled core which amd turned off. So the same story if unlucky it will not work and crash badly
Then another thing comparing different cpu's is simply not possible too many factors which can cause speed differences.
In my case my amd x3 - 720 beats a intel Quad 6600 easy on some projects here even though both run almost same mhz
The amd runs 3150 Mhz and the Q6600 3200 Mhz but again to compare them is impossible cause both have different instruction sets.
In my view every cpu has its strong and weak points
So if compare 2 cpu's look at all important factors instead of just state that one is faster then the other
19) Message boards : Problems and Help : What to do about users spamming me? (Message 40564)
Posted 3651 days ago by uBronan
and do not respond in anyway if you do not want it :D
20) Message boards : Problems and Help : No WU for Sophie Germain Prime Search (LLR) via CUDA (Message 40563)
Posted 3651 days ago by uBronan
lol
i have to disappoint you.
even though many boinc users know what they are doing
there also many users who have no clue, simply cause for most of them its a install and run program and only have to choose the project they wish.
this is ofcourse a good thing which have given a huge user base on boinc
i am one person who is allways asked to solve the pc problems of everyone, so i dare to say that about 70% of these people (i figure a few hundred) not even have a clue what pc they have nor do they care, they simply want it to open ie, mail and chat and some even want word or WP9. So for them when i introduce them to my home and see boinc and after i explained it can help solve scientific problems some of them want it installed by me. Now you can laugh at them but many of them not even dare to install programs.

to give you an example a few of them have boinc allways on easy view and allways demand me to put it back from advanced to easy :D
why cause they simply do not care how it runs, but expect it to work > . <
and second they like the moving thingy's they see on the screen :D
the better this thingy looks the more change they pick it as their favorite project ;)
as with all crows when it blinks they want it xD


Next 10 posts
[Return to PrimeGrid main page]
DNS Powered by DNSEXIT.COM
Copyright © 2005 - 2021 Rytis Slatkevičius (contact) and PrimeGrid community. Server load 3.00, 2.41, 2.45
Generated 20 Sep 2021 | 1:40:38 UTC