@Jozef: take a look over there and run 2 concurrent WUs on your GTX980. You might also want to fix the memory clock, if you haven't already done so. Even more so if you run Einstein or SETI.

@The Kitty cruncher: I can understand that you can't change your setup completely. But... there's something I can not understand in your argumentation. It's clear that you want to crunch as fast as possible. And it's also clear that the running cost of your current setup is hurting you financially, up to a point where you don't have enough money left to change parts (you said so yourself).

If this is true, you're running at impressive speed now, but you can't go any further. If I wanted to crunch as much as possible, this would be unbearable for me, seeing how much better a GTX970 is compared to a GTX560 or GTX580 (I'm talking about GPU-Grid, since I'm unfamiliar with performance at SETI).

If I were you I'd scale back a bit now, where it makes the most sense. Save some money for 1 or 2 months, then start replacing parts for more efficient ones. Sell some of the old ones, earn further savings and get the upgrade train rolling. Shoot for a bit lower power consumption than you have now to leave yourself an option for future upgrades. After a year or so you shoudl end up with lower running costs and higher performance. How could one not want this?

Some steps I'd do or consider:
- stop CPU crunching on the Core 2 Quad, if you haven't already
- maybe even stop CPu crunching on the i7's
-> this will hardly impact your throughput, but saves 75 - 100 W per machine

- undervolt the CPUs, although this requires stability testing, as it eats into overclocking headroom (your CPUs have a lot of headroom, though)

- lower the power target of your 600 series GPUs, so that they run at ~1.10 V instead of the default 1.175 V. This improves power efficiency by ~14% at a small core clock drop. I heard SETI is especially demanding on GPU memory throughput (running multiple WUs concurrently for maximum throughput), so the drop in core clock would not even matter that much. This does not impact stability.

- you could OC your GPUs (if not already done so) to make up for the reduced core clock from the previous step.. but this obviously requires testing.

- you could lower the voltage on your 500 series GPUs manually, which increases power efficiency without reducing the clock speed. But this also eats into the overclocking headroom, just as undervolting the CPUs does.

- change the PSUs to 80+ Platinum (e.g. Antec Earthwatts Platinum 650 W for a dual GPU system), this can pay for itself quickly depending on what you're currently using

- software config: are you running optimized apps and multiple concurrent WUs, if this is beneficial to SETI performance?

... optimizing your farm sounds like a fun actvity ;)

MrS

---

Scanning for our furry friends since Jan 2002

Hi Meowsigh

Two things. Ther First: You're a lucky man because you pay for a consumption of 6326 kWh only $ 812.44. Here in Germany the identical consumption would cause costs of € 1770 ($ 2046).
The second: If you don't upgrade your hardware, you remain caught in your own trap. Forever. That's the simple truth.

---

Regards, Josef

@msattler: Man, if you are in debt and crunching costs you 20$ per day I would suggest taking a break / selling machines. At least for me it would not be worth ruining my life prospects for crunching.
Just my personal opinion :) Respect for the dedication though.

For those of you whose crunching rigs serve double duty as space heaters, you might be amused by this company's hardware:

http://www.qarnot-computing.com

(scroll down)

msattler: feel sorry for you - sounds like you have a bit of an addiction problem, all be it a very honorable one. Seriously take these guys advice - put your old machines / cards on ebay whilst they still have some value to gamers. As an incentive think also of the environmental impact of all that energy being wasted.
- My guess the soon to be released 960 will be the bang per buck per watt card to buy, time will tell.

if you are in debt and crunching costs you 20$ per day I would suggest taking a break / selling machines. At least for me it would not be worth ruining my life prospects for crunching.

Believe the man. He's right. The purpose of distributed computing is (imho) not that you burn your entire money. If there is not enough left for the house, the family and your next car, there should be a good reason for that. What would you like to achieve? Do you think that you will be immortalized in the Olympus of the eternal Cruncher? Turn off the tap a little more. Then you're still a highly valued member of GPUGrid. And for that, have a bit more reserves. Think it over.

---

Regards, Josef

I thank you for all of your well-intentioned posts here.
But, don't pity ME, LOL.
I have a good paying job, work some long hours, and things are quite under control. I contribute a bit here at GPUGRID when the crunchers run out of Seti work. My rigs cost a little more than some of yours to keep in electricity, but they crunch hard and I am happy with that.

I am the #1 single home user of Seti on the planet. That is a source of pride and joy to me, even if it won't buy me a cuppa coffee in my retirement age.

For now, it all works. I am advancing on my debts, and when the time comes and the funds allow, I might indulge in some upgrades. Old hardware does present it's problems at times, but I have been adept in dealing with it.

I have built much of my crunching farm with gifts of other cruncher's upgrade pullouts. And that is why I am a generation or two behind some of you.

I don't have the time or inclination to start yanking things apart, investing in new cards, and trying to hawk the old stuff on ebay. It just ain't gonna happen.
I'll pay the slightly increased electricity costs and run what I have until it bites the dust or some kind soul donates some more pulls to me.
For now, my friends, it serves me well.

Meow for now,
Mark

I am the #1 single home user of Seti on the planet. That is a source of pride and joy to me, even if it won't buy me a cuppa coffee in my retirement age.

Congrats on the SETI accomplishment, quite impressive. Maybe you'll get an invite when the first aliens land :-)

I am the #1 single home user of Seti on the planet. That is a source of pride and joy to me, even if it won't buy me a cuppa coffee in my retirement age.

Congrats on the SETI accomplishment, quite impressive. Maybe you'll get an invite when the first aliens land :-)

Thank you.
It 'tis a feather in me cap.

LOL...or not......

I just looked up my past year's electric bills and did the math.
Hang on tight, here is what running my farm costs.....

1/22/15 689.41 5531kwh
12/18/14 473.40 3652kwh
11/17/14 527.82 4267kwh
10/17/14 640.84 4631kwh
9/18/14 532.65 3938kwh
8/19/14 742.42 5256kwh
7/21/14 812.44 6326kwh
6/19/14 768.20 5737kwh
5/20/14 756.71 5502kwh
4/21/14 715.03 5602kwh
3/20/14 644.05 4946kwh
2/19/14 608.29 4689kwh

One year running total?
$7,911.26 and 60,077kwh.

I estimate that 95% of that is supporting the crunchers 24/7/365.

Meow!!!

Hi msattler,

You might look into replacing the 580s with 750tis, which have about the same performance. A fag packet calculation suggests that as you are paying 0.13USD/kWh for power -- and assuming you can sell on the old cards for about 100USD -- you'll recover the initial expense through reduced running costs in 4-6 months. (Saving around 10USD/GPU/month in reduced electricity costs)

Matt

I am just gonna tough it out.
Not that some of the suggestions made here are not valid.
But the time and trouble to implement them are just too much for me to undertake right now.

Besides, the kitties have the crunchers kicking some real ass on Seti at present.
I love it when what I have works well.

I did survey replacing a lower card with a 680 on my top rig just a bit ago.
The problem is with getting 2 six pin connectors on the current card upgraded to 2 eight pin connectors. The PSU has the moxy...but not enough molex connectors to kludge via adapters. The main connectors are already taken by the 3 680s.

Might look for more adapter kludgy thingys tomorrow.

No offensive but man, you have a serious problem.
I have been working for 8 years in support group, you would fit there well.
As I see it you even realised that something is wrong but you are ignoring every suggestion...
BTW "I love it when what I have works well." - that's not true. You are overpaying for wasted electricity. New cards would do the same or better job with less powerdrain and would save you a lot of money.

- Addiction is a state characterized by compulsive engagement in rewarding stimuli, despite adverse consequences.

- The two properties that characterize all addictive stimuli are that they are (positively) reinforcing (i.e., they increase the likelihood that a person will seek repeated exposure to them) and intrinsically rewarding (i.e., they activate the brain's "reward pathways", and are therefore perceived as being something positive or desirable.
- Addictions can include, but are not limited to, exercise addiction, food addiction, drug addiction, computer addiction, sex addiction and gambling addiction.

http://en.m.wikipedia.org/wiki/Addiction
http://en.m.wikipedia.org/wiki/Video_game_addiction

- Denial

1. (Avoidance) Skating Off The Walls: I know I'm using denial when I refuse to directly answer a question and keep trying to change the subject.
2. (Absolute Denial) Saying It Isn't So: I know I'm using denial when I tell people that I don’t have a problem even though I know deep inside that I do.
3. (Minimizing) Saying It Isn’t That Bad: I know I'm using denial when I admit that I have a problem, but try to tell people that it isn't as bad as they think it is.
4. (Rationalizing) Giving Good Reasons: I know I'm using denial when I try to convince people that there are good reasons for me to have the problem and that because there are good reasons I shouldn't be responsible for having to deal with it.

http://www.tgorski.com/clin_mod/dmc/denial_checklist.htm

Guys lay off msattler. If he wants to continue as is then so be it.
What he has is working well (proven by the 928,000+ RAC here not including his seti RAC) even if it isn't as efficient as it could be. Some people (me included) don't like to upgrade when the current hardware is working just fine and producing good results.

Just because there are more efficient vehicles out there did you all run out and buy Prius (or equivalent) cars? I bet not. Older less fuel efficient vehicles are still used all over the world. Whats the difference?

popandbob wrote:

Just because there are more efficient vehicles out there did you all run out and buy Prius (or equivalent) cars? I bet not. Older less fuel efficient vehicles are still used all over the world. Whats the difference?

The difference is simple: anyone who's driving less than 10k km/year can hold on to his old car just fine. The fuel doesn't cost enough to worry him, but replacing the working equipment with something new and shiny would cost a lot.

But for a taxi driver with 100+ km/year the fuel costs amount to a large sum, and a significant portion of the overall cost to run that car. If there's a new vehicle which could cut those costs and quickly pay for the investment through those savings, woudln't this be a smart move? Especially if our "taxi driver" is working on paying pack some debt? It would help him to pay back earlier, or to do whatever else he wants to spend money on. Saying "but the old car still works well" is not helping at all, if things could otherwise be significantly better. We're not living in caves any more, because at some point people said: "it works well, but..."

MrS

---

Scanning for our furry friends since Jan 2002

But for a taxi driver with 100+ km/year the fuel costs amount to a large sum, and a significant portion of the overall cost to run that car. If there's a new vehicle which could cut those costs and quickly pay for the investment through those savings, woudln't this be a smart move? Especially if our "taxi driver" is working on paying pack some debt? It would help him to pay back earlier, or to do whatever else he wants to spend money on. Saying "but the old car still works well" is not helping at all, if things could otherwise be significantly better.

MrS

But that taxi driver would have to spend money to upgrade the taxi and when he's in debt he can't afford to do that. Everyone is saying well stop driving, save the fuel your spending and use that to upgrade but by stopping driving he is now not able to get the desired result (taxi fares in this example).

So as it relates to msattler, everyone is saying saying stop the science being done (this being a project that wants results back very quickly...) and then save up and then buy new stuff that will produce the exact same results. Net result to the science? a loss of 6 months of work done.

People have made suggestions. Msattler doesn't want to do them at the moment. I think we should leave it at that. Personally I feel he deserves a big thanks for supporting science projects. How he decides to do it is his business. If others want to do it differently that's their business. There are many bigger issues in the world to worry about without nitpicking this little one. Let's thank everyone for their contribution to science, and move on.

We should definitely finish this topic. Just one more point:

Mark, I hope it's clear why we posted all those suggestions. It's not to "pick on you" (not that you would have expressed anything like this). I can only speak for myself, but I suppose most will agree.. the point is that your contribution is very welcome, yet we're worried it might not be sustainable.

Popandbob just brought up the point of "if he scales back, there will be science lost or done later". But what you go full throttle now, and maybe for one or two more years, and then it somehow all collapses? If we could help to avoid that by bringing your running costs down while keeping your performance similar, we might loose a bit of speed during the next few months, but could keep you crunching for years afterwards.

How realistic and/or likely is such a scenario? We can't tell, honestly, but from what you wrote many of us would be scared if it were us ;)

MrS

---

Scanning for our furry friends since Jan 2002