I now have two machines both AMD64 3200's one with 1 gig and the other with two gig.... Running Win x64 and they are ripping through the data sets. Is anyone else running x64?

---

Gigabyte K8NS Ultra 939, AMD64 3200, 2.0 Gigabytes PC3200, Windows XP x64

I hate to disappoint you, but IMO those times are nothing special.
My XP64 3000+ is returning results in ~5,000 seconds, yours are ~9,000 secs.
Take a look at the optimized versions here: http://www.marisan.nl/seti/

---

Isn't there potentially a big difference in the size of the Seti WUs a machine can request ?

I'm using a mix of AMD Opteron 248s and 250s with 4GB or 8GB or RAM and I've seen completition times ranging from ~3,000 seconds up to ~6,800 seconds (averaging ~5,800 seconds) on a pre-compiled Solaris 10 x64 client.

Benchmark Results (248/250)

Memory 4030.96 MB / 8126.96 MB
Cache 976.56 KB
Swap space 50 MB
Measured floating point speed 2109.97 million ops/sec / 2302.11 million ops/sec
Measured integer speed 5757.14 million ops/sec / 6230.04 million ops/sec


When a machine requests work and asks for xx,xxx seconds of work is that the expected time to completion or the amount of telescope data ?

---

Isn't there potentially a big difference in the size of the Seti WUs a machine can request ??

There currently isn't a large difference between WUs (except for ones that contain alot of noise and those terminate pretty quickly). The Beta group is testing a new seti application that doubles the sensitivity, and these take longer to process.

When a machine requests work and asks for xx,xxx seconds of work is that the expected time to completion or the amount of telescope data ?

It is the "Estimated" time to completion. It's currently over estimating the total time, this difference changes betweeen hosts.

does this help?

tony

When a machine requests work and asks for xx,xxx seconds of work is that the expected time to completion or the amount of telescope data ?

It is the "Estimated" time to completion. It's currently over estimating the total time, this difference changes betweeen hosts.

does this help?

tony

yeah thanks, I'd noticed HUGE differences in the requested work, yet very small differences in the time to completion (which was always significantly smaller than the 'requested work' time) which was confusing me.
But if it is over estimating and some WUs are full of noise then I guess that explains it.

---

Isn't there potentially a big difference in the size of the Seti WUs a machine can request ?

No, all SETI WUs are the same size - at least in terms of actual size (~340Kb) and recording time - so many seconds of recorded date.

The content of the data can vary though, from nothing interesting at all through to solid earth generated radio signals.

I'm using a mix of AMD Opteron 248s and 250s with 4GB or 8GB or RAM and I've seen completition times ranging from ~3,000 seconds up to ~6,800 seconds (averaging ~5,800 seconds) on a pre-compiled Solaris 10 x64 client.

Benchmark Results (248/250)

Memory 4030.96 MB / 8126.96 MB
Cache 976.56 KB
Swap space 50 MB
Measured floating point speed 2109.97 million ops/sec / 2302.11 million ops/sec
Measured integer speed 5757.14 million ops/sec / 6230.04 million ops/sec


When a machine requests work and asks for xx,xxx seconds of work is that the expected time to completion or the amount of telescope data ?

Neither really. It is the difference in seconds between how much you have in your cache and how much you want in your cache. If you connect every 24hrs, you want 86,400 seconds of work cached. If you have .5 days in your cache, you only have 43,200 seconds, so you will request another 43,200.

---

Neither really. It is the difference in seconds between how much you have in your cache and how much you want in your cache. If you connect every 24hrs, you want 86,400 seconds of work cached. If you have .5 days in your cache, you only have 43,200 seconds, so you will request another 43,200.

That makes more sense. Thanks.

---

Yep..

AMD64 Winchester +3500, 1 Gig ram, completes in about 3500 - 3600 seconds.
Running Win XP PRO 64bit

Bukken

Keep Crunching

---

I think there are a lot of us AMD 64 users.

This one is mine and it's nothing special at 3800-4300sec/wu.
But I am afraid to try the MS WinXP 64 OS.
CPU type AuthenticAMD
AMD Athlon(tm) 64 Processor 4000+
Number of CPUs 1
Operating System Microsoft Windows XP
Professional Edition, Service Pack 2, (05.01.2600.00)
Memory 2048 MB
Measured floating point speed 2824.27 million ops/sec
Measured integer speed 4366.45 million ops/sec

HAVE FUN!!!

I have three all running XP64 Pro. An AMD 3000, An AMD 3400 and one with dual 275 Opterons. I didn't see much if any increase in speed between XP32 and XP64. XP64 is more responsive however.

---

I have three all running XP64 Pro. An AMD 3000, An AMD 3400 and one with dual 275 Opterons. I didn't see much if any increase in speed between XP32 and XP64. XP64 is more responsive however.

Besides the responsivness are there any compatibility issues?
Drivers and the like? Cuz there is a trade-in at MS for users to upgrade to XP64 pro and I'm trying to figure out if I should take advantage of it.

Thanks,
Daemon