special astropulse 6.01-build for ATOM-Processor
My long test run with a real astropulse wu ended today
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Quick timetable

WU : #ap_genwis.dat
ap_5.05r409_SSE.exe -verbose :
  Elapsed 104.863 secs
      CPU 101.026 secs
ap_5.05r468_SSE3_ATOM_IXE_MKLS_O3.exe -verbose  :
  Elapsed 8.455 secs, speedup: 91.94%  ratio: 12.40x
      CPU 3.994 secs, speedup: 96.05%  ratio: 25.29x
ap_6.01r548_SSE_331_noAVX.exe -verbose  :
  Elapsed 98.888 secs, speedup: 5.70%  ratio: 1.06x
      CPU 95.785 secs, speedup: 5.19%  ratio: 1.05x
ap_6.01r557_SSE3_ATOM_IXE_MKLS_O3.exe -verbose  :
  Elapsed 6.209 secs, speedup: 94.08%  ratio: 16.89x
      CPU 3.588 secs, speedup: 96.45%  ratio: 28.16x
ap_6.01r557_SSE3_ATOM_IXE_MKLS_O3_libfftwf-3.3.1.exe -verbose  :
  Elapsed 104.192 secs, speedup: 0.64%  ratio: 1.01x
      CPU 100.277 secs, speedup: 0.74%  ratio: 1.01x
ap_6.01r557_SSE3_ATOM_IXE12.1.2.278_MKLS10.3.8_O3.exe -verbose  :
  Elapsed 6.474 secs, speedup: 93.83%  ratio: 16.20x
      CPU 3.869 secs, speedup: 96.17%  ratio: 26.11x

WU : ap_08mr07ag_B4_P1_00025_20100428_07060.wu
ap_5.05r409_SSE.exe -verbose :
  Elapsed 245817.702 secs
      CPU 239508.351 secs
ap_5.05r468_SSE3_ATOM_IXE_MKLS_O3.exe -verbose  :
  Elapsed 175379.443 secs, speedup: 28.65%  ratio: 1.40x
      CPU 173293.118 secs, speedup: 27.65%  ratio: 1.38x
ap_6.01r548_SSE_331_noAVX.exe -verbose  :
  Elapsed 231145.730 secs, speedup: 5.97%  ratio: 1.06x
      CPU 230139.305 secs, speedup: 3.91%  ratio: 1.04x
ap_6.01r557_SSE3_ATOM_IXE_MKLS_O3.exe -verbose  :
  Elapsed 220803.835 secs, speedup: 10.18%  ratio: 1.11x
      CPU 218619.209 secs, speedup: 8.72%  ratio: 1.10x
ap_6.01r557_SSE3_ATOM_IXE_MKLS_O3_libfftwf-3.3.1.exe -verbose  :
  Elapsed 202800.952 secs, speedup: 17.50%  ratio: 1.21x
      CPU 201832.971 secs, speedup: 15.73%  ratio: 1.19x
ap_6.01r557_SSE3_ATOM_IXE12.1.2.278_MKLS10.3.8_O3.exe -verbose  :
  Elapsed 173418.757 secs, speedup: 29.45%  ratio: 1.42x
      CPU 172586.200 secs, speedup: 27.94%  ratio: 1.39x


======================================

Restoring BOINC to pretest state...
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Against ap_5.05r486 we lost no speedup with the new ap_6.01r557
We can be happy with it. 
Full test-result in our beta test area.

_heinz

ATI - driver for HD4670 AGP  12-1_agp-hotfix_xp32_dd_cc

BOINC 7.0.25
30.04.2012 20:19:49 |  | Processor: 1 GenuineIntel               Intel(R) Pentium(R) 4 CPU 2.66GHz [Family 15 Model 2 Stepping 7]
30.04.2012 20:19:49 |  | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pbe
30.04.2012 20:19:49 |  | OS: Microsoft Windows XP: Home x86 Edition, Service Pack 3, (05.01.2600.00)
30.04.2012 20:19:49 |  | Memory: 1023.48 MB physical, 4.90 GB virtual
30.04.2012 20:19:49 |  | Disk: 55.89 GB total, 16.03 GB free
30.04.2012 20:19:49 |  | Local time is UTC +2 hours
30.04.2012 20:19:49 |  | ATI GPU 0: ATI RV730 (CAL version 1.4.1664, 1024MB, 1012MB available, 960 GFLOPS peak)

works fine with primegrid, cpu-usage now down to 7- 10% (tpsieve_1.38)
the machine (P4 Northwood 2.6GHz) shows ~50% cpu usage total while crunching.
This shows: a single CPU can feed a Graphicadapter like HD4670 to crunch

25th april the machine had 2.250.000 pg-credit

_heinz

MSI announces:
N690GTX Dual GPU
N690GTX-P3D4GD5
3072 CUDA-Processors
4GB GDDR5, 6008MHz 
(TDP)  300 Watt
max temp 98 grd C
10 years ultra long lifetime (under full load).

price in germany: 999,00 Euro

happy crunching
_heinz

Ordered now a PSU tester unit to see if PSU or board is dead.
Waiting...

4th of July a great day in Research at CERN,

Higgs within reach

Read the CERN Press Release

Austria-German Press articles
~~~~~~~~~~~~~~~~~~
Wir haben es

Physiker feiern Durchbruch

    
_heinz

Today I cleaned the PSU and had a closer look in it to see something.
PSU_open_cleaned

I found a overheated transformer.
PSU_transformer_E4220_defect
PSU_transformer_E4220_defect_closer_look

After all the cleaning I conneted the PSU-tester, if it shows something.
PSU_tester_connected

And really it shows something now, wow what a effect.
PSU_tester_shows_all_voltages

PSU tester says the PSU is OK.

Now I closed the PSU and set it back into V8-Xeon.
I switched the Power on and the machine runs till to the logon screen, then switched the power automatic off again. Pitty..
A next try to start the machine was not sucessful anymore.
Will order a new PSU now.

_heinz

(a little late, but it is worth to tell and show picture)
As I opened the case I was really surprized about all the dust in it. Last cleaning was sill two a half months ago.
Have a closer look on the picture and you see dust ontop of the grapicadapters too, Its not light, its dust.
v8-xeon_dustbunnies
The most dust is on the wires before the first cpu cooling unit. I had to think about any effective air-filtersystem in the next future.
See you the small space between the graphicadapters, no wounder to get high temperatures, not enough air can come in.
This is not a masterpiece of engeneering, the graphicadapters must be smaller in the part where the fans are to let more air in.
Better I should run still two graphicadapters on this bordlayout.
I could increase the performance using two adapters with dual GPUs. Maybe a mixed configuration, one from NVIDIA the other from ATI.
One of the main-problems are the high temperatures of the FB-DIMM, you know I had already temps over 100 grd Celsius.
On the market I did not found really good freezers for the FB-DIMMs and a lot of memory coolers does not fit.
I got a KINGSTON HyperX RAM-Cooler 2x60mm blue , but it does not fit, it's too big.
I need a 50mm and still 5 mm high to fit under the clamp of the 12cm fan of the left CPU-cooler.
Looks like I must selfcunstruct something.

_heinz
 

10th august 2012, I got  a new milestone with my small i3- laptop gt540m_10Mio_distrtgen 

my comment to the FPGA:
The price is high,  1392 Euro for a PCI-Express-card with XILINX™ VIRTEX-4™ FPGA and 512 MByte SO-DIMM storage-modul
see this entry in my developer-book some years ago
Although if someone has access to all the hard and software it would be nice to have a seti application on a FPGA.
See FPGA thread

_heinz