Posts by Fred J. Verster


log in
1) Message boards : Number crunching : Seti credit vs Einstein Credit (Message 1329252)
Posted 456 days ago by Profile Fred J. Verster
I would say Einstein offers SLIGHTLY more credits than SETI. Not massively more.

If you just want big numbers to impress your friends with (and if so seriously reconsider your social circle) then try collatz. Honestly I feel I shed 50 IQ points while I crunched it. But my RAC was HUGE.

Ou go to GPUGRID, they "pay" 115k for a single WU! Almos 10x the SETI amount!


But at GPUGrid, it heavily depends on the GPU you use, a FERMI or KEPPLER gives
far more credit, compaired to the 8000/900 & 200 NVidia series.

And for Collatz C. similar, Single Precision AMD/ATI GPUs do very well, especially 5770; 5870 (D.P.).
2) Message boards : Number crunching : Multiple seti issues since updating BOINC Mgr (Message 1326644)
Posted 465 days ago by Profile Fred J. Verster
Anyone wondering, FarmerDon posts with one account, but runs work with another. His computer and tasks are here.

I don't know why the OpenCL app keeps restarting, but apparently it doesn't harm either as the tasks end normally and are validated OK. I just wonder about the credit you managed to get, 701.90 for one AP? Must've been a very special one... ;-)


AstroPulse v6.
ATI host,
in case first link isn't reachable?!

IIRC, this is/was the credit-rate for an AstroPulse WU, with 'normal' output,
i.e. no Infomation Overflow and blanking whithin 2.5%-50%.
\Which has little influence in Credit, but affects CPU-time, higher blanking =
higher CPU-time.

I see 1 host in FarmerDon's account, the setting: Use GPU when idle 10 minutes, can result in lots of shutdowns and restarts.
Just like the :
Stop Computing when non-BOINC load exceeds xx%
, setting.
3) Message boards : Number crunching : Changing GPU - But which will be better? (Message 1326407)
Posted 465 days ago by Profile Fred J. Verster
I like to make some comments.
The 6xx series has a lot more cudas, but more isn't always better.
6xx series is more power efficient,that's a good thing.
But the cuda cores are a lot lot less powerful.

As you can see some 5xx cards outperform the 6xx series.
But I expect there will be some gain in optimizing for the 6xx series


Hasn't that something todo with the Reg_Size, being twice as big in
the 600 series (KEPLER), as the FERMIs with 32K regsize.
Setiathome_CUDA: Found 1 CUDA device(s): Device 1: GeForce GTX 480, 1535 MiB, regsPerBlock 32768 computeCap 2.0, multiProcs 15 clockRate = 1401000


AstroPulse
Two GPU results, 1 rev.1363 (LUNATICs Optimized), 1 stock app. and CPU.

I'm mistaken, wanted to compaire 400 (FERMI) and 600 (KEPLER) and CPU, but
it's a GTX 550 Ti.
And a result overflow :-/.


4) Message boards : Number crunching : Slow WU download speed (Message 1326218)
Posted 466 days ago by Profile Fred J. Verster
I've noticed the prolonged downloads too, but noticed today that it's improved significantly. It's been a few months since I payed much attention to what the clients are doing, so not sure how unusual the slower DL's on my end has been.


Unfortunatly true, crunching is done faster then down- and uploading them .
5) Message boards : Number crunching : Discussion of Invalid Host Messaging (Message 1326061)
Posted 467 days ago by Profile Fred J. Verster
*** I suppose I could just change the Unroll to 2 and see what happens ***

I've completed 6 APs in a row with the Unroll setting at 2. If I would have done that back here ^Something isn't right^ around 5 of those 6 would have ended in a "Computation error".
Here are the first 3 of the 6;
Task 2784334429
Task 2784338147
Task 2784338745

I'm ready to declare Victory.


You could adjust the Fetch and Thread_Block, see that you use 1:3 and
Fetch_Block 2048 and Thread_Block 6144, have you tried other values, f.i. Tread_Block 10240 and Fetch_Block 5120 or 8192 and 4096: (1:2).
Running on device number: 0 DATA_CHUNK_UNROLL at default:2 DATA_CHUNK_UNROLL set to:2 FFA thread block override value:6144 FFA thread fetchblock override value:2048


With UNROLL=2, Tread and Fetch_Block can be bigger!
Which one is the most effective also depends which GPU you're using.
See that you use an 9800GT and a BARTS GPU. Quite different architecture and
thus Regsize.
6) Message boards : Number crunching : Wind tunnel cpu set up (Message 1326051)
Posted 467 days ago by Profile Fred J. Verster
Had this show up on my timeline this morning and thought it was rather "cool". No pun intended. I swear!

Wind Tunnel CPU


This certainly quite another approch to cooling a computer!
I very much like the idea! (Although Water-Cooling may be simpeler
and just as effective).


7) Message boards : Number crunching : Discussion of Invalid Host Messaging (Message 1324587)
Posted 472 days ago by Profile Fred J. Verster
I thought the server was supposed to throttle a client with too many errors, is that not true?

It is true, but effective only when nothing but errors are returned for an extended time. There are delays caused by pending tasks from before the problem, and each error only reduces quota by one. For "SETI@home Enhanced 6.10 windows_intelx86 (cuda_fermi)", Thndr's host is down to "Max tasks per day 2" but for GPUs that is multiplied by a project setting which is probably still 8 here. So for that application version it's down to 16 per day, and that's enough less than the limit of 100 in progress to minimize the damage.
Joe


Its still seems odd that there are users with a thousand inconclusives.

As an example, I've got an inconclusive with this user, who has no tasks in progress, but the application details show he can take 100 WU? http://setiathome.berkeley.edu/show_host_detail.php?hostid=6726942

I don't mean to imply any ill will towards anyone working on sorting out their gpu problem. Maybe the project should throttle the GPUs faster and automatically notify you if you get a lot of errors.



11/14/2012 5:28:45 PM | | suspend work if non-BOINC CPU load exceeds 25 %.



This setting can produce errors, it stops and starts the app..
Better to set it to 0 (zero) and use the (CPU) and GPU setting: run when computer is idle, IMHO

The same applies to the default settings of the AstroPulse stock app.
DATA_CHUNK_UNROLL at default:2 DATA_CHUNK_UNROLL at default:2


Depending of the GPU used, these values should be adjusted accordingly.
F.i. start with UNROLL=8, UNROLL=16 is usually the limit.
Info: BOINC provided device ID used Used GPU device parameters are: Number of compute units: 4 Single buffer allocation size: 256MB max WG size: 1024 FERMI path used: yes

UNROLL=4 should work better, IMHO.
(If it's still possible to change these values?!).
8) Message boards : Number crunching : Merry Christmas (Message 1319869)
Posted 482 days ago by Profile Fred J. Verster
Merry Christmas and a happy

new crunching Year !

9) Message boards : Number crunching : Unexected Logout (Message 1319118)
Posted 484 days ago by Profile Fred J. Verster
Just after one of the recent Windoze updates I found I had been logged out of all sorts of things, including both S@H and S@H-Beta, really frustrating as trying to remember all the various uids, passwords, funny phrases and so on was really frustrating.


Been happening to me for several months. As a test, I disabled Win updates on wife's machine & left mine enabled yet hers remained unaffected.

Dec's patch Tuesday knocked mine out again. Enabled updates on wife's & bingo, had to re log in to many sites from hers.

Thanks M$.


I always have to log in, but only have to enter the first caracter of my
Log-In name. (IIRC you can set this in your browser, remember name
and pw).

Long before the latest security-updates on WIN XP, also WIN 7. (22-12-12).
(And the Cookie-warning).





10) Message boards : Number crunching : Balance CPU and GPUs (Message 1318504)
Posted 485 days ago by Profile Fred J. Verster
You definetly need to free more CPU cores running 4 GPUs with multiple instances.


Very true, also needed with AMD/ATI GPUs, doesn't matter if Cross-Fire
is enabled.
Every GPU has 1 thread of an i7-2600, also running on higher frequncy
by increasing the Base-Clock from 100 to 102 to 104MHz. Since the multiplier is locked, it's the only option and works OK.

But I do have several driver-crashes, specially with SETI work?!

NVidia GPUs also need sometimes a dedicated core, CPU time is higher on NVidia
GPUs.
Almost double compaired to AMD/ATI GPUs.

I've set SETI the same way, 0.5CPU in stead of 0.04 so it uses 1 core/thread
setting this in app_info.xml when doing 2 pper GPU 1 core/thread will
be used, Einstein uses this setting too.

This AMD/ATI host does Docking or Einsteins Gravity Wave search on CPU.
SETI and Einstein only GPU.





1,216.96 runtime

233.94 CPU time

118.06 Credit

SETI@home Enhanced v6.10 (cuda_fermi)



2763176272

6628794

21 Dec 2012 | 17:29:59 UTC

21 Dec 2012 | 21:41:34 UTC

Voltooid en gecontroleerd

1,570.51 runtime

102.46 CPU time. Less as NVidia GPUs.

118.06 Credit.



SETI@home Enhanced
Anoniem platform (ATI GPU)
Both platforms run 2 tasks per GPU.
11) Message boards : Number crunching : SETI/BOINC Milestones [ v2.0 ] - XXV [ CLOSED ] (Message 1318494)
Posted 485 days ago by Profile Fred J. Verster
Well reached over 30 Million, could be more and I keep on CRUNCHING

Last month I got only got Docking or Einsteins Gravity Search, on CPU.
And Binairy Pulsar Search (BRP4CUDA/OpenCL) on 2 NVidia and 2 AMD/ATI 5870GPUs

And since a few hours I gor a lot of SETI work. On all 3 hosts and a Laptop.

A efficient way of crunching, 3 hosts using ~1000 Watt/hour when looking at
the total troughput. > 170 Million credit and 162.000 RAC, for all projects.

And they're running 24/7, ofcoarse, since my electric bill is included in
the rent of this appartment.

In February, i've to do some serious renovating and have to quit for 10 days
or less. :-/
12) Message boards : Number crunching : Balance CPU and GPUs (Message 1318491)
Posted 485 days ago by Profile Fred J. Verster
You definetly need to free more CPU cores running 4 GPUs with multiple instances.


Very true, also needed with AMD/ATI GPUs, doesn't matter if Cross-Fire
is enabled.
Every GPU has 1 thread of an i7-2600, also running on higher frequncy
by increasing the Base-Clock from 100 to 102 to 104MHz. Since the multiplier is locked, it's the only option and works OK.

But I do have several driver-crashes, specially with SETI work?!

NVidia GPUs also need sometimes a dedicated core, CPU time is higher on NVidia
GPUs.
Almost double compaired to AMD/ATI GPUs.

I've set SETI the same way, 0.5CPU in stead of 0.04 so it uses 1 core/thread
setting this in app_info.xml when doing 2 pper GPU 1 core/thread will
be used, Einstein uses this setting too.

This AMD/ATI host does Docking or Einsteins Gravity Wave search on CPU.
SETI and Einstein only GPU.

13) Message boards : Number crunching : Panic Mode On (79) Server Problems? (Message 1318297)
Posted 485 days ago by Profile Fred J. Verster
Having finally received work for main crunching PC
the other day , today I have project down messages...

I lost a few returns last time , hope this is not a major problem again !!
I hate to lose credit because of failure to report :/

Thumbs up for the holiday work , but it has not fixed the problem it seems.
( we do not have that holiday here in the U.K. )

I have NO cuda work again and the cpu work is running out too. :-(

Chris.



Yep, a month later, a lot of work all of a sudden...;-)
(Forgot to post it immediatly).
14) Message boards : News : SETI@home scheduler is down until further notice. (Message 1311809)
Posted 500 days ago by Profile Fred J. Verster
I now have errors in my results, all of which are down to two periods of server downtime.

Because these appear as errors, I take it that my machine has been plodding away for no good reason.


Well, SETI Bêta is up and running,
. UTC= 17:04

Haven't seen much (no) work in the last week, so no errors ;-)

Well you could make a SETI Bêta account, unless you have one already, Less ??1
They're always 'short on testing', too.


15) Message boards : Number crunching : Going from a GTX460 to GTX 600 Series. (Message 1311528)
Posted 501 days ago by Profile Fred J. Verster
---Snipped----

Oh, I thought it meant titanium, and thus stronger and lighter!

----snipped---

Titanium is indeed used, but I don't know for what reason and/or replacing
an other (conducting) material?

I stiil use an GTX470 and 480, cause 470 has twice the CUDA cores as the 460,
in stead off using 2 460s, I rather use 1 470.

But I'm thinking of adding 2 GPUs the 600 series is the better choice,
also a 470 maybe hard to get new ofcoarse.

Although the 600 series has a different architecture, also 64K regsize,
400 series (FERMI) has 32K regsize and most support PCIe 3.0 x16.
And are more efficient compaired too 470 or 480, which (can) run hot, 98C
or the fan can make a lot of RPM c.q. noise.
B.t.w. The 400 or FERMI series already were/are using GDDR5 memory,
quad pumped.

The CUDA app. provides no info about estimated GFLOPS, on 600 series.
Don't know if BOINC 7.0.28 displays an estimate
Atleast the 600 GPU has to have the same or more CUDA cores as
the 470 or 480 NVidia GPUs.
16) Message boards : News : SETI@home scheduler is down until further notice. (Message 1311511)
Posted 501 days ago by Profile Fred J. Verster
I haven't had work for a week, SETI Bêta still sends work, CPU and 0.963CPU+1GPU, v7 MB work.

They urgently need tons of $$$, try to danate as soon as possible, Europe
so the Netherlands too, still are in the 'middle ' of an econimic recess.

But glad, S.E.T.I., is back up and running :-)

17) Message boards : Number crunching : cheap GPU (Message 1311060)
Posted 508 days ago by Profile Fred J. Verster
Is a Galaxy 61TGS4HX2LTX GeForce GT 610 GC worth $25 in terms of Boinc crunching? I'm thinking about adding a GPU to one of my older systems to take the load off its CPU. (It has other duties as well.)


The NVIDIA GeForce 610M is an entry level dedicated laptop graphics card and based on the same chip like the Geforce GT 520M or GeForce GT 520MX. It is based on the GF119 like the GT 520M and can feature clock rates up to 520MX level (900MHz). Using DDR3, the 64 Bit memory bus may become a bottleneck for the graphics card. Furthermore, low clocked versions are also seen in the wild. Our Acer Aspire 5750G review unit for example was clocked only at 672 MHz (lower than the 740MHz of the GT 520M).
GT610.


Not very much, 2 CPU cores I think.
18) Message boards : Number crunching : No new work? (Message 1306915)
Posted 520 days ago by Profile Fred J. Verster
Two hosts, an i7 + 2 ati 5870 GPUs and a C2Q6600+gtx470, have work.
And my LT has had 1 AstroPulse.... taking 40 hours while a ATI 5870 GPU does
it in 40 minutes.

And SETI Bêta ofcoarse, almost always work.

19) Message boards : Number crunching : WASTING MY TIME? (Message 1306867)
Posted 520 days ago by Profile Fred J. Verster
Was everyone told the point is to find ET. Yes.


Using exactly your words...

If the focus is FIND ET, then why try to keep AP-Splitting working and making all this wierd situation worst if is crystal clear the actual configuration can´t handle both project running all full capacity?

Astropulse development was funded in part by an NSF grant. Trying for future NSF grants without continuing AP processing would be silly. Stopping AP temporarily is possible, but the general idea of analyzing all fresh data with both SaH and AP algorithms is necessarily the goal.

In terms of future funding, it might actually be better to skip the SaH processing and do only AP on some of the data.
Joe


I agree to this.
AstroPulse has a better chance of finding 'ET' or something alike.

At least in the same data Einstein has found several Pulsars.

NOT what the majority of the crunchers want to hear, probably, but we do
have to be realistic and funding is dead on necessary for this project!


20) Message boards : Number crunching : Short WU : Good or Evil ? (Message 1305834)
Posted 523 days ago by Profile Fred J. Verster
Well, another thing about 'shorties', the need about the same CPU-time as the
VLARS or ~0.35AR WUs. And more disk I/O, memory I/O, GPU loading, etc.

And do cause a lot of Net Traffic, IMH they're not 'good', compaired to 0.4 ARs.

(If I do compaire this with Einstein, 10x the bandwidth is used and storage, a few GBytes are normal, they've far less users and would experience the same
trouble if having 10x more users).


Next 20

Copyright © 2014 University of California