I have a dual Xeon 3.06GHz with 2GB RAM (an HP xw6000) running FC3 (Fedora Core 3)

all seti workunits are finished in about 10000 seconds (~ 2h 45 min)

I'm using setiathome_SSE2_naparst-r3.4.tar.bz2 and boinc_5.2.14_SSE2.tar.bz2

/proc/cpuinfo contains 4 procs:

processor : 3
vendor_id : GenuineIntel
cpu family : 15
model : 2
model name : Intel(R) Xeon(TM) CPU 3.06GHz
stepping : 9
cpu MHz : 3058.230
cache size : 512 KB
physical id : 3
siblings : 2
core id : 3
cpu cores : 1
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 2
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe cid xtpr
bogomips : 6094.84

this is the last of 4, the others 3 procs look the same

boinc said:
Measured floating point speed: 2310.09 million ops/sec
Measured integer speed: 1960.55 million ops/sec

i've tried:
acpi=ht apm=off in grub.conf
i have resetted the project 2-3 times

top is showing all 4 processes at 99.x%:


PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ #C COMMAND
16173 adib 34 19 38516 9992 2172 S 99.6 0.5 37:16.95 0 albert_4.38_i68
16868 adib 39 19 124m 54m 3068 R 99.6 2.7 37:05.46 1 sulphur_um_4.22
16312 adib 39 19 55224 46m 4016 R 98.5 2.3 119:14.81 2 hadsm3um_4.13_i
16529 adib 39 19 131m 57m 5280 R 98.4 2.8 42:15.51 3 sulphur_um_4.21

(right now seti is not running, but otherwise it says setiathome_SSE2, sometimes on all 4 virtual cpus)

I have other boxes with only 1 HT proc (P4 3.2GHz) which are finishing the wus in ~2000 seconds

if anyone have an idea to increase the speed of my computer, any idea, please answer

thanks

Hi
I have 4 boxes with the same Xeon cpu and also a bunch of other xeons (2.8 - 3.2).
Its always been a puzzle to me why the 306s don't perform as well as the others, especially the 2.8s. One thing is certain though, HT on a dual Xeon throttles them. It seems that it bottlenecks at on-board memory. I have all my Xeons running without HT and I get equal performance to running with HT. Off the top of my head, with HT on each seti WU would take approx 2 hrs ie 4 WUs in 2 hours. With HT off each WU takes 1 hour ie 2 WUs in 1 hour = 4 WUs in 2 hours = HT improves nothing.

So my Xeons run with HT off now

Of all of my Xeon machines (20+) its the 2.8s that run fastest and manage a RAc of round about 1000 each with the 3.06s somewhere between 600-800

heres links to a couple of my xeons 3.06
and a 2.80

---

I forgot to mention:
i tried to stop services: irqbalance, cpuspeed, apmd
no effect :(

and free says:

             total       used       free     shared    buffers     cached
Mem:       2074936    1867716     207220          0     316472     634608
-/+ buffers/cache:     916636    1158300
Swap:      7679060          0    7679060

maybe something to write in /proc/sys/vm?
or in BIOS?

and now seti is running: top:

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  #C COMMAND
17192 adi       39  19 44628  39m 1048 R 99.9  1.9  30:58.27  2 setiathome_SSE2
17191 adi       39  19 44628  39m 1048 R 99.9  1.9  31:07.33  3 setiathome_SSE2
17189 adi       39  19 44632  39m 1048 R 97.0  1.9  30:43.87  0 setiathome_SSE2
17190 adi       39  19 44628  39m 1048 R 97.0  1.9  30:52.41  1 setiathome_SSE2

this is for accuracy, not an answer for the problem

I forgot to mention:
i tried to stop services: irqbalance, cpuspeed, apmd
no effect :(

and free says:

             total       used       free     shared    buffers     cached
Mem:       2074936    1867716     207220          0     316472     634608
-/+ buffers/cache:     916636    1158300
Swap:      7679060          0    7679060

maybe something to write in /proc/sys/vm?
or in BIOS?

and now seti is running: top:

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  #C COMMAND
17192 adi       39  19 44628  39m 1048 R 99.9  1.9  30:58.27  2 setiathome_SSE2
17191 adi       39  19 44628  39m 1048 R 99.9  1.9  31:07.33  3 setiathome_SSE2
17189 adi       39  19 44632  39m 1048 R 97.0  1.9  30:43.87  0 setiathome_SSE2
17190 adi       39  19 44628  39m 1048 R 97.0  1.9  30:52.41  1 setiathome_SSE2

this is for accuracy, not an answer for the problem

The problem is that the xeon 3.06 Ghz has a slow fsb. We're really hitting a bottleneck with harolds version here.

As i ported his app. to windows and got a dual xeon 2,4 GHz myself i now the problem. It's truely realated to the shareing of the fsb.

You could try at least lowering your default settings.
Try to set the max cpu to use at 2.

I've seen that doing this will increase RAC.
But you have to try yourself.

---

Join BOINC United now!

The problem is that the xeon 3.06 Ghz has a slow fsb. We're really hitting a bottleneck with harolds version here.
...
You could try at least lowering your default settings.
Try to set the max cpu to use at 2.

I'm sure that may be the problem

but I can't believe that 2x3.06 HT procs SSE2 (=4 virtual procs) run almost the same as a Celeron 2.4 SSE3 (1 non-HT proc)!
the raport of bogomips is 6/1! (4x6k=24k/4k)!

maybe I should use a slower client, but which runs better in this case?
and, in your opinion, which one?
maybe I'll try them all (latest versions, maybe a few older), to see what happens

The problem is that the xeon 3.06 Ghz has a slow fsb. We're really hitting a bottleneck with harolds version here.
...
You could try at least lowering your default settings.
Try to set the max cpu to use at 2.

I'm sure that may be the problem

but I can't believe that 2x3.06 HT procs SSE2 (=4 virtual procs) run almost the same as a Celeron 2.4 SSE3 (1 non-HT proc)!
the raport of bogomips is 6/1! (4x6k=24k/4k)!

maybe I should use a slower client, but which runs better in this case?
and, in your opinion, which one?
maybe I'll try them all (latest versions, maybe a few older), to see what happens

My Dual Xeon 2.4 GHz runs 4 processes of seti and does a wu in about 2h45min.

I don't think that running a older/slower version will do any better.

I'll guess we have to accept it the way it is.

Anyway things will change when setiathome "enhanced" is relased.

---

Join BOINC United now!

I'll guess we have to accept it the way it is.

maybe yes, maybe no

this night (I'm romanian -> GMT+2) i tried this:
#renice 0 xxxxx
where xxxxx were the 4 PIDs of the setiathome_SSE2
surprise!
WUs were finished in ~7000 secs, all of them!
but, as you can see, this shouldn't happen, the computer is almost all the time idle, else, with nice=19, the seti_SSE2 would't be 99.x%

strange

anyway, probably I'll write a script which runs as root (renice needs root privilege to increase priority) to do this
meanwhile, probably today, I'll try the setiathome_SSE (not 2) variant
and maybe your P3 FFTW version
if this don't work, I'll try some non-static versions (who knows...)
I know this is odd, but who knows?

if I'll get some results, I'll write them here