KWSN Orbiting Fortress

Pooh Bear 27 · Posted: Thu Jun 21, 2012 12:30 pm Post subject: Rig Lists

Many of us run several rigs and do several projects. I usually try to keep about 4 rigs going, I've had as many as 7, but I do not think I will go to that extreme anytime soon.

Current rigs:
Rig 1:
Core2Quad,Q6600 2.4G, no video (slot died and won't take a card so using on board Intel) - retiring within the next week (Anyone wanna buy it? XP, 4G memory, ~500G HD, DVD R/W, Card reader, 10/100 Eithernet - Dell Brand - Message me)

Rig 2:
I7-920 2.67G, NVidia GT220 video which no longer crunches well so I retired the video.

Rig 3:
I7-920 2.67G, ATI HD5450

Rig 4:
I7-960 2.8, ATI HD5650

Building to replace Rig 1:
I7-3770 3.4/3.9G, NVidia GT 640

No go ahead post what you have. I left off memory, HD size, etc. cause those are not that important)

Sir Cracked of the Mind · Guest

You are a deeply sick man, I have a computer.

Laughing

Sir Papa Smurph · Posted: Thu Jun 21, 2012 3:35 pm Post subject:

I used to have a farm,..

C2duo Lappy, dead & dismembered.

Amd 9850 Black Quad core with, depended on my mood, a 5870 or 2-3 Nvidia 9800's, Won't Boot and I just don't have time to fixit.

My Main Box, Amd 940 Black, with a 5970 dual core Gpu. Still running.

Not to mention the 30-40 hosts I have borged over the years that are all gone now....
_________________
a.k.a. Licentious of Borg.........Resistance Really is Futile.......
and a Really Hoopy Frood who always knows where his Towel is...

dow · Posted: Fri Jun 22, 2012 3:28 am Post subject:

Due to extremely hot weather conditions here in the UK, the shrubbery has gone from 28 to 8 in the last 2 weeks.
some of the shrubs were sorely in need of a prune:

stooper101 · Posted: Fri Jun 22, 2012 9:05 am Post subject:

Main rig: Phenom 2 955 with AMD 5870

Other rig: Phenom 2 B55 (555 unlocked to 4 cores) with AMD 5770

Borged rig: Athlon 4200 with AMD 4770

Very part-time laptop: i5 something.
_________________

>stats<

Putting_things_on_top · Posted: Fri Jun 22, 2012 8:18 pm Post subject:

My current configs are:

Rig 1:
- AMD Phenom II X4 965 BE (3.4GHz), 8192MB DDR3 (4x2GB) [running 4/4 cores]
- ASUS GTX560ti 448 Core (732MHz), 1280MB GDDR5
- ASUS GTX560ti 384 Core (830MHz), 1024MB GDDR5
- Thermaltake 1200W PSU
- Win7 Ultimate

Rig 2:
- AMD FX-8120 8-core (3.1GHz), 8192MB DDR3 (2x4GB) [running 6/8 cores]
- MSI GTX560ti 384 Core (882 MHz), 1024MB GDDR5
- DiabloTek 600W PSU
- Win7 Home Premium

I also have a lowly PS3 running F@H (in 'expert' mode, which just means 'exclusive' mode).
Not much for PPD, but every little bit helps.

So, currently, I have 6 shrubberizers.

By the end July (7/2012), I will bring "ANOTHER SHRUBBERY" online:

Rig 3 (by end-of-month?):
- AMD FX-8150 8-core (3.6GHz), 32768MB DDR3 (4x8GB) [running 6/8 cores]
- EVGA GTX690 3072 Core (915~1019 MHz), 4096MB GDDR5
- OCZ Vertex4 SSD 256GB
- OCZ 1250W PSU
- Win7 Pro

I will be using the most massive case I can find for my new build (click on image for Newegg video review):

That will put the count at 8 shrubberizers.

EDIT (21-Jul-2012):
I was going to get two ASUS GTX 680's (their 'Top' edition), but I got tired/impatient with ASUS (they kept pushing out their ETA). #Mad

So, I opted for a single GTX 690 from EVGA (due for home delivery on 23 July).
I have updated the info (above) to reflect my change.

As budget permits, I will also add another GTX690 to Rig 3, but that will be some months down the road. Sad

I don't know [yet] how much more powerful the GTX 600 series is over the 500 series with regard to F@H.
Some sites claim that F@H performance for 600 series is not too much improved over the 500 series.
But this has more to do with Stanford and the version of 'FAHCore' being used.
It's not a hardware deficiency, it's a matter of programming-efficiency.
Stanford has more work left to do with regard to optimizing the FAHCore_16 module to leverage the NVidia Kepler architecture.
Core-16 is still 'officially' in beta, however the higher-end/newer ATI-Radeon cards seem to be fully capable of using Core-16.

END-EDIT

_________________
Click here for...KWSN F@H team summary at EOC

Or here for...KWSN F@H team overtake at EOC

Killerrabbit · Posted: Fri Jun 22, 2012 11:51 pm Post subject:

I only have one rig a AMD3000.

Ni
_________________

Pooh Bear 27 · Posted: Fri Jul 20, 2012 3:44 pm Post subject:

Now that I've had my 3rd Gen I7 running for a bit, wow does it kick arse on Primegrid. There LLRAVX 64bit app that works on 2nd Gen and newer can haul out work a lot quicker than pre AVX processors.

I've been running CULLEN and PSP units on it. I am doing them about 25% than the 2nd gen chips which have AVX, and more than 50% faster than non AVX I7s.

I have not even played with clocking yet. This is at standard speed. I also have turbo boost which I also have not tried.

Time to start playing with settings this weekend.

Nuadormrac · Prince Joined: 13 Sep 2009 Posts: 506

Pooh Bear 27 · Posted: Fri Jul 20, 2012 7:21 pm Post subject:

Nuadormrac,

They are stock. But because they are now 64bit apps, it runs faster at 4 of 8 cores, which I am doing.

Nuadormrac · Prince Joined: 13 Sep 2009 Posts: 506

Ah, OK... so not an opti-app to make use of instruction set improvements...

If you're using 4 cores on a task, I wonder if the credit return is great enough to make up for the loss of running 3 other tasks concurrently or not... Hmm.... The proc also does have a way of boosting it's clock if it runs on fewer cores (part of the turbo boost 2.0), but in my experience a clock increase wasn't always tit for tat up in output, (aka twice the clock doesn't mean twice the work done) due to other factors.

Was it using this many cores in the past? Or is it crunching them 50% faster, but losing a few cores to crunch other stuff in the process?
_________________

.

Gemjunkie · Posted: Fri Jul 20, 2012 8:59 pm Post subject:

I'd love to see someone do a comparison of i5 vs. i7, same gen., same clock, crunching a project with WUs that take a uniform amount of time, with and without hyperthreading. Cool

_________________

(older, before split CPID)

Grizzly · Posted: Sat Jul 21, 2012 4:30 am Post subject:

Gemjunkie · Posted: Sat Jul 21, 2012 6:10 am Post subject:

Thanks. The i7 costs >50% more than the i5, I'd like to see >50% more performance. Rolling Eyes

Tried the 3rd gen yet?

#ni-1

_________________

(older, before split CPID)

Grizzly · Posted: Sat Jul 21, 2012 9:38 am Post subject:

Gemjunkie · Posted: Sat Jul 21, 2012 10:10 am Post subject:

Nuadormrac · Prince Joined: 13 Sep 2009 Posts: 506

The third gen I imagine would get a bit of an edge on test4theory, given it has optimizations for uses with virtual machines, but that is a seperate project. Test4theory of course runs with virtual box, with scientific Linux running inside.... I didn't delve into exactly how they're improving performance with use of VMs though....

I'm actually fairly impressed thus far, but haven't tested everything out, and what would be the perf dif with multithreading used, vs now is a fair question. Because even on an i7, with 8 "cores", 4 hypter-threaded, one can potentially run 8 tasks at the same time. But if one hyperthreads it, a task will be running on multiple cores (the hyper-threading), but will the credit gain for running a single task on 4 cores, be greater then if one ran 3 other tasks along side it?

In essense for comparison's sake, obviously the single task would run faster with more CPU thrown at it, but would the return (aka credits), and work done (aka tasks completed total) be greater then if one ran 8, in this case prime grad tasks on the CPU in parallel, each being turned in. Running on 4 cores, would mean only running 2 tasks, rather then 8, faster for that task, but would the credit return be > if 8 were running on each individually?

One's now talking about ~30% faster, vs if one ran it on 3 other cores, it would be 400% (or 4x) the number of tasks being run in parallel Hmm #ni-1

With the trade offs, however one approaches it.... And yet, in saying this, I also allow for running 2 tasks on the same core might not be a 2x perf improvement also, for much the same reason 4 cores isn't a straight up 400% perf improvement, there are shared resources one core can end up waiting for....
_________________

.