KWSN Orbiting Fortress

Yankton · Posted: Fri Mar 07, 2014 10:59 pm Post subject:

684575069 512041133 547725 8 Mar 2014, 3:38:51 UTC 8 Mar 2014, 3:50:48 UTC Completed and validated 353.37 5.49 213.76 Milkyway@Home Separation (Modified Fit) v1.28 (opencl_nvidia)
684575068 512041132 547725 8 Mar 2014, 3:38:51 UTC 8 Mar 2014, 3:56:41 UTC Completed and validated 353.42 5.49 213.76 Milkyway@Home Separation (Modified Fit) v1.28 (opencl_nvidia)
684575067 512041131 547725 8 Mar 2014, 3:38:51 UTC 8 Mar 2014, 3:50:48 UTC Completed and validated 353.37 5.65 213.76 Milkyway@Home Separation (Modified Fit) v1.28 (opencl_nvidia)
684575066 512041130 547725 8 Mar 2014, 3:38:51 UTC 8 Mar 2014, 3:56:41 UTC Completed and validated 353.42 5.44 213.76 Milkyway@Home Separation (Modified Fit) v1.28 (opencl_nvidia)
684575065 512041129 547725 8 Mar 2014, 3:38:51 UTC 8 Mar 2014, 3:56:41 UTC Completed and validated 352.42 5.32 213.76 Milkyway@Home Separation (Modified Fit) v1.28 (opencl_nvidia)
684575061 512041125 547725 8 Mar 2014, 3:38:51 UTC 8 Mar 2014, 3:44:55 UTC Completed and validated 353.44 5.73 213.76 Milkyway@Home Separation (Modified Fit) v1.28 (opencl_nvidia)
684575060 512041124 547725 8 Mar 2014, 3:38:51 UTC 8 Mar 2014, 3:44:55 UTC Completed and validated 353.44 5.93 213.76 Milkyway@Home Separation (Modified Fit) v1.28 (opencl_nvidia)
684575059 512041123 547725 8 Mar 2014, 3:38:51 UTC 8 Mar 2014, 3:44:55 UTC Completed and validated 353.44 5.76 213.76 Milkyway@Home Separation (Modified Fit) v1.28 (opencl_nvidia)
684575057 512031725 547725 8 Mar 2014, 3:38:51 UTC 8 Mar 2014, 3:44:55 UTC Completed and validated 353.44 5.62 213.76 Milkyway@Home Separation (Modified Fit) v1.28 (opencl_nvidia)
684575038 512001757 547725 8 Mar 2014, 3:38:51 UTC 8 Mar 2014, 3:50:48 UTC Completed and validated 353.37 5.73 213.76 Milkyway@Home Separation (Modified Fit) v1.28 (opencl_nvidia)
684575037 511981878 547725 8 Mar 2014, 3:38:51 UTC 8 Mar 2014, 3:50:48 UTC Completed and validated 353.37 5.65 213.76 Milkyway@Home Separation (Modified Fit) v1.28 (opencl_nvidia)

No modblpls in this group of shrubs. Running 4 of the shorter ones gpu load was only 70%. Wish you could pick and choose which types of these were sent to you. CPU usage was negligible. Apps showed in the task manager but sat at 0% CPU times above pretty well match that. Very consistent completion times. Last time I waited to post until they all validated. I leave a note tomorrow if something comes up invalid. Hasn't happened yet. All validated.
_________________
Some days are worse than others.

Assimilator1 · Posted: Sat Mar 08, 2014 7:35 am Post subject:

Re cpu useage tag, got ya Smile

Not sure what credit WUs the modbpls are - nm I see they are the long 213.76 WUs & 106.88 (those seem rare atm, only found 3 for my rig over the past day's valid WUs!).

Thx for the results, like you said, very consistent times! & it's a pity you can't choose which WUs you get.
I guess the chances of running 4 modbpls at once are small. If you do happen to get 4 at once LMK what times you get with them Smile

.
_________________
Team AnandTech - SETI, F@H, Muon1, MW@H, Asteroids@H, LHC@H, Skynet, Rosetta@H

Assimilator1 · Posted: Sat Mar 15, 2014 10:53 am Post subject:

Added a HD 5830 to the table.

Still many cards missing from the table e.g ATI/AMD R7s some R9s, HD 6930 (rare 1 that!) o/ced 6900s, 4890, 4700s, NVidia 200s series, some 400s, some 500s, some 600s & 700s.
Not to mention many CPUs!

I would test my own GTX 260 c216 but the 32bit MW mod fit app is bugged (& hence not automatically available).
_________________
Team AnandTech - SETI, F@H, Muon1, MW@H, Asteroids@H, LHC@H, Skynet, Rosetta@H

Yankton · Posted: Sat Mar 15, 2014 11:41 am Post subject:

You have 4770k and 4771 covered already?
_________________
Some days are worse than others.

Assimilator1 · Posted: Sun Mar 16, 2014 7:32 am Post subject:

Err, if you look at the chart you can see they aren't, you know the chart on page 1? Wink

.
Would be good to have those CPU times Smile

, are the 4000 series Ivybridge CPUs?
_________________
Team AnandTech - SETI, F@H, Muon1, MW@H, Asteroids@H, LHC@H, Skynet, Rosetta@H

Yankton · Posted: Sun Mar 16, 2014 5:01 pm Post subject:

Nope, they are both Haswells. Are you just looking cpu times while doing the gpu work? If so you have it for the 4770k already. If not, which other subproject are you looking at?
_________________
Some days are worse than others.

Assimilator1 · Posted: Mon Mar 17, 2014 2:05 pm Post subject:

No I'm after CPU times for the CPU crunching the whole WU itself, same sub project as the GPU. So still the long 213.76 credit WUs.

Ah Haswells, that was my 1st thought Wink

, I'm not upto to speed anymore with CPUs.
I guess the Ivys are the 3000 series then.
What have you o/ced the k to btw?
_________________
Team AnandTech - SETI, F@H, Muon1, MW@H, Asteroids@H, LHC@H, Skynet, Rosetta@H

Yankton · Posted: Tue Mar 18, 2014 7:43 am Post subject:

Ok:

These are from a linux box with a 4771. Like the GPUs there are different types of WUs for the modified fit. I identified 3 but there could be more.

[list=]692640910 518019878 18 Mar 2014, 5:03:31 UTC 18 Mar 2014, 10:52:59 UTC Completed and validated 7,352.16 7,339.11 213.76 Milkyway@Home Separation (Modified Fit) v1.28
692640284 517734927 18 Mar 2014, 5:02:23 UTC 18 Mar 2014, 10:51:51 UTC Completed and validated 7,356.17 7,341.94 213.76 Milkyway@Home Separation (Modified Fit) v1.28
692639638 518018878 18 Mar 2014, 5:01:15 UTC 18 Mar 2014, 12:27:34 UTC Completed and validated 13,100.54 13,076.82 213.76 Milkyway@Home Separation (Modified Fit) v1.28
692638948 518018358 18 Mar 2014, 5:00:07 UTC 18 Mar 2014, 12:27:34 UTC Completed and validated 13,101.64 13,078.48 213.76 Milkyway@Home Separation (Modified Fit) v1.28
692638246 518017799 18 Mar 2014, 4:58:58 UTC 18 Mar 2014, 12:27:34 UTC Completed and validated 13,114.81 13,090.87 213.76 Milkyway@Home Separation (Modified Fit) v1.28
692637870 517968550 18 Mar 2014, 4:57:49 UTC 18 Mar 2014, 12:27:34 UTC Completed and validated 13,110.88 13,084.38 213.76 Milkyway@Home Separation (Modified Fit) v1.28
692637043 518016897 18 Mar 2014, 4:56:39 UTC 18 Mar 2014, 8:48:26 UTC Completed and validated 13,842.25 13,766.73 213.76 Milkyway@Home Separation (Modified Fit) v1.28
692637041 518016895 18 Mar 2014, 4:56:39 UTC 18 Mar 2014, 8:49:35 UTC Completed and validated 13,859.53 13,780.61 213.76 Milkyway@Home Separation (Modified Fit) v1.28
692637040 518016894 18 Mar 2014, 4:56:39 UTC 18 Mar 2014, 8:48:26 UTC Completed and validated 13,782.85 13,703.88 213.76 Milkyway@Home Separation (Modified Fit) v1.28
692637039 518016893 18 Mar 2014, 4:56:39 UTC 18 Mar 2014, 8:49:35 UTC Completed and validated 13,849.42 13,773.36 213.76 Milkyway@Home Separation (Modified Fit) v1.28
692637038 518016892 18 Mar 2014, 4:56:39 UTC 18 Mar 2014, 8:47:16 UTC Completed and validated 13,821.56 13,753.08 213.76 Milkyway@Home Separation (Modified Fit) v1.28
692637036 518016890 18 Mar 2014, 4:56:40 UTC 18 Mar 2014, 8:48:26 UTC Completed and validated 13,795.00 13,713.80 213.76 Milkyway@Home Separation (Modified Fit) v1.28
692637035 518016889 18 Mar 2014, 4:56:40 UTC 18 Mar 2014, 12:26:25 UTC Completed, validation inconclusive 13,125.07 13,092.46 pending Milkyway@Home Separation (Modified Fit) v1.28
692637034 518016888 18 Mar 2014, 4:56:40 UTC 18 Mar 2014, 8:49:35 UTC Completed and validated 13,865.62 13,787.56 213.76 Milkyway@Home Separation (Modified Fit) v1.28 [/list]

These are for a windows machine running a 4770k.
This machine is not currently overclocked. The reason being is if you use a lot of avx (and primegrid does, expect it to appear in more places soon) then anything over 3.7Ghz starts causing temps that are unacceptably high and maxes the output on the voltage regulator which is already high enough that you don't want to mess with it too much on air. Without avx I had acceptable runs at 4.4Ghz. I have a new case for this box and I'm still debating liquid cooling options (primarily so my titan will set at 45 degrees under full load) so I may revisit it but the third problem is that once you are on water and you start overclocking the power draw shoots up really fast. at 4.2Ghz it was drawing 155 watts, double what it does at 3.5. I'm in California so they tend to rake us over on electricity.

Final notes. Most of these units were modbpls so I expect that's why the long run times. A third unit type starting with ps was also present. The gpu ones loaded my card at 28% each. I zonked out again before running checks on the cpu units. I just let these run overnight. Hyperthreading is on and these were allowed to go for the gusto. The windows box allowed 7 simultaneous as I have a logical core set aside for the gpu. Let me know how that jives with what you have! #ni-1

_________________
Some days are worse than others.

Assimilator1 · Posted: Thu Mar 20, 2014 2:42 pm Post subject:

As per my 1st post I just want an average time for the long 213.76 credit WUs, which I identified here (on the 8th) as the bulk of the modbpl WUs. So 213.76 credit mod bpl WUs.

Also as per my 1st post (did you read it? Wink

), I'm after CPU times with HT off, although if you get consistent times with HT on I'll add that time too with a footnote Smile

. Which reminds me can you provide an averaged time for that please?

And yea that is a big jump in power going upto 4.2 GHz! Although it's still a low total, my Q9550 @3.6 GHz draws 175-190w at the wall with just the CPU loaded (draw depends on project, A@H 190w, F@H 175w).
How much is your electricity? Mine's 14p/unit, which is ....... 23c at current exchange rates.

#ni-2

_________________
Team AnandTech - SETI, F@H, Muon1, MW@H, Asteroids@H, LHC@H, Skynet, Rosetta@H

Assimilator1 · Posted: Fri Mar 28, 2014 1:38 pm Post subject:

No times yet?
_________________
Team AnandTech - SETI, F@H, Muon1, MW@H, Asteroids@H, LHC@H, Skynet, Rosetta@H

Yankton · Posted: Sat Mar 29, 2014 11:35 am Post subject:

I had a couple but they fell off the list before the last one finished. My scheduler didn't want to run it. I'll do another batch when I get a chance!
_________________
Some days are worse than others.

Assimilator1 · Posted: Wed Apr 02, 2014 1:06 pm Post subject:

Okey doke

_________________
Team AnandTech - SETI, F@H, Muon1, MW@H, Asteroids@H, LHC@H, Skynet, Rosetta@H

Assimilator1 · Posted: Tue Apr 08, 2014 12:34 pm Post subject:

Added a HD 6990 & a R9 270X.
_________________
Team AnandTech - SETI, F@H, Muon1, MW@H, Asteroids@H, LHC@H, Skynet, Rosetta@H

lvanst · Posted: Sun Apr 13, 2014 7:39 pm Post subject:

Running them now...

Stats in 24 hours Smile

_________________

Assimilator1 · Posted: Mon Apr 14, 2014 12:34 pm Post subject:

Cool

What h/ware?
_________________
Team AnandTech - SETI, F@H, Muon1, MW@H, Asteroids@H, LHC@H, Skynet, Rosetta@H

lvanst · Posted: Mon Apr 14, 2014 10:36 pm Post subject:

My Nvidia GTX 650 Ti consistently runs them in just over 23 minutes.

The card is slightly overclocked...
(GPU 1,110 Mhz, Mem 1,377 Mhz, Core@ 1.087V)

I have an AMD FX-8350 OC'd @ 4.7 Ghz pushing the card...

The CPU WU stats were too inconsistent to post.
_________________

Assimilator1 · Posted: Tue Apr 15, 2014 1:44 pm Post subject:

The table's in seconds, could you give me the 5 (minimum) WU average in seconds please? Smile

(I know 23mins is 1380s but you said just over 23mins Wink

).
Just to confirm a few things, your looking at long 213.76 credit WUs right? The GPU had a dedicated CPU core & GPU utilisation was @ nr 100% for the benchmark run?

Re CPU times, odd, was you running other tasks on the CPU or crunching on the GPU at the same time? Not just seeing the differences between long & short 213.76 credit WUs?
_________________
Team AnandTech - SETI, F@H, Muon1, MW@H, Asteroids@H, LHC@H, Skynet, Rosetta@H

Assimilator1 · Posted: Fri Apr 18, 2014 7:05 am Post subject:

Hmm, looking through my stats last night I noticed that the long 213.76 WUs are now very rare, I think I must of looked through over 100 results & only saw about 1/2 a dozen of the long ones, if that continues then this benchmarking run is coming to an end Sad

, still we've got a good number of results Smile

.

In the meantime hurry up if you want to post some times! #ni-2

_________________
Team AnandTech - SETI, F@H, Muon1, MW@H, Asteroids@H, LHC@H, Skynet, Rosetta@H

Yankton · Posted: Thu Apr 24, 2014 8:07 am Post subject:

4770k on windows at stock speeds with H/T off. Looks like 4660 seconds. Though a few went for about 4910.
_________________
Some days are worse than others.

Assimilator1 · Posted: Thu Apr 24, 2014 2:54 pm Post subject:

Hi, did you read the warning in my previous about the long 213.76 WUs disappearing?

Looking at your results atm I would guess not as you only have 6 results to look at which is too few now, sadly they are probably all the short 213.76s, http://milkyway.cs.rpi.edu/milkyway/results.php?hostid=547725&offset=0&show_names=0&state=0&appid=10

I will see if I can use your times from your previous post & add a foot note about HT.

*********************************************************

Whoops Embarassed

, just noticed that I hadn't clarified which time column should be used, whilst it was obvious for the GPU it wasn't for the CPU!

Going back over my posts in the MW forum I see I chose the run time, but on further investigation it looks like I should of chosen the CPU time for more accurate CPU comparisons (what do you guys reckon?).

Aarrrgg nightmare! I would have to re-calculate mikey's times (fortunately he posted both) & I think for the rest of the CPU times I will just have to put a footnote about it as it will be almost if not totally impossible to get enough long 213.76 times for CPUs now (unless that WU makes a come back which seems unlikely).
_________________
Team AnandTech - SETI, F@H, Muon1, MW@H, Asteroids@H, LHC@H, Skynet, Rosetta@H