Post Your Hash/Sec and Hardware

[lachesis]

Hello all! This thread is exactly what the title says: a place to post about the node(s) that you are running, including their performance and hardware.

To find out your performance, you'll need to build from source with laszlo's patch. The performance in hashes / second will be listed in the UI where it used to say "Generating..." and will spam your debug.log.

I'm currently running Bitcoin on three machines (and I tested it on a fourth, but that one is for a business, so I can't keep it running).

My Workstation
Specs:
HP HDX 16t - 16" laptop
Intel Core 2 Duo P8700 (x64) @ 2.53GHz w/ 3072KB cache
4GiB DDR2 RAM
nVidia GeForce GT 130M (OpenCL needs to get its act together!)
Ubuntu 10.04 LTS 64-bit (2.6.32-22)
Performance:
**Perf - thread 3 : 446k iter/sec
**Perf - thread 4 : 451k iter/sec
**Perf - total : 897k iter/sec (2 threads)

My Home Server
Specs:
1400MHz Intel Celeron w/ 512KB L2 Cache
512MiB of (ancient) RAM
Debian testing 32-bit (not sure the kernel version)
Performance:
**Perf - total : 169k iter/sec (1 threads)
Puny!

My Xen VPS
Specs:
Not 100% sure about this, but it's using 4 cores.
256MiB of (fast) RAM
Ubuntu 10.04 LTS 64-bit (not sure the kernel version)
Performance:
**Perf - thread 1 : 436k iter/sec
**Perf - thread 2 : 410k iter/sec
**Perf - thread 3 : 419k iter/sec
**Perf - thread 4 : 429k iter/sec
**Perf - total : 1694k iter/sec (4 threads)

So, anybody else?

[1714133225]

1714133225

[1714133225]

1714133225

[1714133225]

1714133225

[SmokeTooMuch]

someone make a windows binary for noobs/lazy people like me ^^

[laszlo]

Make sure you guys try compiling with -O2 instead of -O0 (check your makefiles).  It improved performance a lot for me.

[laszlo]

E2200 C2D 2Ghz CPU

**Perf - thread 1 : 495k iter/sec
**Perf - thread 2 : 495k iter/sec
**Perf - total : 990k iter/sec (2 threads)

[lachesis]

Make sure you guys try compiling with -O2 instead of -O0 (check your makefiles).  It improved performance a lot for me.

Laszio, I tried that but it failed. I'll see if I can't address the specific error. Line 15 in net.h failed to compile.

someone make a windows binary for noobs/lazy people like me ^^

I can try if nobody else replies and this doesn't get accepted into mainline. Not sure what the Windows dependencies are or if I can cross-compile. Win64 or 32?

[SmokeTooMuch]

Win7 64-Bit here.

[laszlo]

You should compile for 32 bit windows (mingw).. 64 bit would just make it use more memory but you'll have to build all the dependencies that way too, if it's even possible with wxWidgets and stuff..

[lachesis]

I haven't generated a block since 6/3/2010 while running my VPS (up to 2280 khps) 24/7 and my laptop (1200 khps) every night. I'm rather disappointed.

[laszlo]

That seems odd.. are you seeing any instances of 'proof-of-work' if you grep debug.log?

[lachesis]

That seems odd.. are you seeing any instances of 'proof-of-work' if you grep debug.log?

No, I'm not seeing that on either. On my workstation, the debug.log file has for some reason been clipped at the beginning. It's around 4.8 MiB. On my VPS, it's 20MiB and going strong.

[laszlo]

Athlon FX-55 (socket 939) on linux

**Perf - thread 1 : 575k iter/sec
**Perf - total : 575k iter/sec (1 threads)

[laszlo]

The debug.log is truncated when you restart the app, not sure if it does it periodically too.. are you perhaps getting disconnected/reconnected frequently or having really high latency?  A poor connection could prevent you from generating, though I wouldn't think you would be having that issue at the VPS.

[lachesis]

I'm on campus internet. It's pretty reliable. It occasionally asks me to reauthenticate with the master server, but that's fairly rare. The VPS is el-cheepo, but it never drops for more than a second or two. Would it help for me to post or link my debug.log?

[laszlo]

Can you post it on a web server or something?  It sounds pretty large.  I could take a look but it's probably hard to find anything among all the spam.

[lachesis]

Sure. My computers ran all last night without generating again.

http://validity.alloscomp.com/bitcoin/vps.debug.log.bz2 (2.1MiB compressed)
http://validity.alloscomp.com/bitcoin/laptop.debug.log.gz (489KiB compressed)

Thanks.

[laszlo]

Really sorry about this but there was a typo in the patch.. I replaced the original version with a fixed one.  The original patch was actually not incrementing the nonce so it messed up the generation

new patch: http://heliacal.net/~solar/bitcoin/bitcoin-svn-79-perfcounter-2010-06-02.patch

[lachesis]

Really sorry about this but there was a typo in the patch.. I replaced the original version with a fixed one.  The original patch was actually not incrementing the nonce so it messed up the generation

new patch: http://heliacal.net/~solar/bitcoin/bitcoin-svn-79-perfcounter-2010-06-02.patch

Laszlo and I talked about this on IRC. I made a slightly cleaned up version against r80 of the SVN (no differences, but it might be a hair faster). You can get it here:
http://www.alloscomp.com/bitcoin/bitcoin-svn-80-perfcounter-cleaned-2010-06-07.patch

Or you can just get the combined patch that also includes the net.h workaround and some RPC extensions:

http://www.alloscomp.com/bitcoin/bitcoin-svn-80-combined-2010-06-07.patch

[venom]

Intel Core i5-450M (2.4Ghz)
4GBs RAM

*One Core =>  350-450 khash/sec
*Two Cores => 600-850 khash/sec
*Three Cores => 750 - 900khash/sec
*Four Cores => 800-900 khash/sec

Using the regular build, 0.3 on Win7/64bit

[Bitcoiner]

AMD-64 X2 TK-53 1.7Ghz

1 Core: 310 khash/sec
2 Cores: 590 khash/sec (and an overheating laptop)

[ksd5]

Heh. I'm using a HP Pavilion 8670c with 384MB RAM and a 600MHz Pentium III. I'm only getting 50 khash per second.