Testing team?

[Leaps-from-Shadows]

Hello everyone,

I'm Leaps-from-Shadows, but you can call me Leaps.  It was inferred over at the SETI@home forums that y'all need some testers equipped with AMD-based systems.

I've got me a Gateway GT5692, bought in August.  It originally came with a Phenom 8450 (triple core, 2.1GHz), but I upgraded it to a Phenom 9650 (quad core, 2.3GHz) just for running SETI@home.  The 4GB of 667MHz DDR2 RAM and 500GB SATA hard drive have remained the same.  Clock speeds and voltages remain at default because of Gateway's crappy BIOS.  Oh yeah ... it has Vista Home Premium x64 with SP1 for the OS.  I've been using the AK_v8 SSE3 app for a while, and I have been running the SSE3 optimized Astropulse app from the day it was released.  I'm a huge fan of both, and SETI@home is the only project I run.

Anyway ... if it turns out that you do indeed need a tester with my hardware/software configuration, I'd jump at the chance to be involved in development/testing of super-speedy optimized apps to erase the AMD inferiority in SETI@home.

Let me know...

[Raistmer]

Hi
Your config seems unique among our current testers (not because Phenom itself but because its combination with x64 Vista).
Currently development process goes for that we can be restricted our own hosts. But later when pre-release testing will begin your config will be very useful.
Thanks for your proposal, we will use it on pre-release stage.

[Richard Haselgrove]

To clarify (since it was my remark about the limited number of AMD testers that prompted Leaps to volunteer in the first place):

There is an AP app which has the basic upgrades for v5.00 compatibility already in testing. That seems to be stable and reliable so far, and it's probably "good enough" to match the stock v5 release if they spring it on us suddenly.

Raistmer and Jason G are busily working through further optimisations as we speak (I've been watching their conversations this morning from the sidelines - way above my pay-grade to actually join in ): it's those further optimisations which aren't yet ready for live testing.

When testing does begin, it's usually better to test at SETI Beta: not only is it good form to test at a test site, but the tasks you've crunched remain visible on the website indefinitely (no 24-hour purge), which is helpful if you report problems and somebody else needs to look at your output. Oh, and you also get that dinky little 'Volunteer tester' tag on the main site when you register.

Upgrading to pre-release tester status here also requires tweaking by a site administrator, and they are few on the ground. Probably best to start by registering at Beta first, and reading up on some of the history on the Beta message boards, while we're waiting.

[Leaps-from-Shadows]

I had assumed that SETI@home Beta was where they did the 'official' pre-release testing for the optimized Multibeam and Astropulse apps.  Is that not the case?  Does Lunatics do their own independent testing apart from SETI@home Beta?

Pre-release testing at Beta is fine, but I don't really want to test standard apps - unless they get a lot faster on AMD machines.

I was also expecting the forums at SETI@home Beta to be a lot busier than they are.

[Gecko_R7]

I had assumed that SETI@home Beta was where they did the 'official' pre-release testing for the optimized Multibeam and Astropulse apps.  Is that not the case?  Does Lunatics do their own independent testing apart from SETI@home Beta?

Pre-release testing at Beta is fine, but I don't really want to test standard apps - unless they get a lot faster on AMD machines.

I was also expecting the forums at SETI@home Beta to be a lot busier than they are.

Short answer is "It depends".
Testing is done both at Beta & here.  Richard's point re: ability to monitor output is spot-on.  It's a better "control" environment to test functionality in a live mode.  Some of the current work being done also has relevance to stock Berkeley development as well.  Optimized development aps are tested here @ Lunatics in the MANY development builds that eventually coalesce into an agreed release candidate build.  A release candidate may need multiple compiles/flavors to suit different archs, i.e.  ssse3 & sse4.1 for Intel, sse2, sse3 for AMD etc.  as well as 32bit & 64 bit .  At this point, more people and platforms are needed to test all these alphas sufficiently.  Much of this goes on here.  Once the gremlins are sorted & confidence is high in accuracy and stability, some limited testing may be done @ Beta prior to "public release". 

The one common element is that testing groups here are tight control groups & developers COUNT on their active participation & quick response.  Each person generally is contributing a certain combination of CPU + OS flavor that helps complete a whole picture.  If called-upon, it's not casual testing & can in fact involve a considerable amount of work & time.
The developers pretty much determine the scope of testing and what specific resources are needed & when.
Thanks for your interest. 

[Leaps-from-Shadows]

Thanks for the added detail in your answer, Gecko_R7 - sounds like the kind of operation I'd like to be involved in.  I'm a night owl that sleeps an average of four hours per night.  I have a retail job, so my hours are a bit variable, but they tend to be fairly short shifts (~6 hours) so I've got lots of time on my hands.

My system:
AMD Phenom X4 9650 2.3GHz quad core
4GB 667MHz DDR2 RAM
500GB SATA HD
Stock clocks and voltages
Vista Home Premium x64 SP1
BOINC 6.2.19 32-bit client
AK_v8_win_SSE3 32-bit optimized app
ap_4.35rev24b54_SSE3 32-bit optimized app
Runs SETI@home (no other BOINC projects) 24/7 with other programs (mainly Firefox 3.0.4, as I'm an internet junkie)

Since the unique thing about my machine is the Phenom CPU plus x64 OS combo, I would also be willing to switch to the 64-bit BOINC manager and apps if necessary.