 |
 |
Is the client taking advantage of Altivec?
|
 |
|
 |
|
Juan Font
|
|
After initial testing on both my PB G3 250 WallStreet and my G4 400 PCI Video, I believe that the client is not using the Altivec on the G4. The time difference between calculations on both computers roughly corresponds to the processor speed difference. If this is so, then using the Altivec should give us a substantial lead on the G4 respect tot he other platforms. I would appreciate someone to give some information on this. Thanks.
|
|
|
| |
|
|
|
 |
|
 |
|
Dedicated MacNNer
Join Date: Mar 2000
Location: Toledo, Ohio
Status:
Offline
|
|
Hello Juan,
You're right about Seti@Home not taking advantage of Altivec. I guess they have no plans on making an Altivec cause the users are already more than they anticipated. There is no need for them to speed up people's processes out there. Matter of fact, v.3.0 will be coming out soon and I believe that it will run test on another part of the data that has been already collected.
So you were right, and for the most part, the main different is your processor speeds. It would be nice to have an Altivec enhanced S@H client to show off the great arcitecture of the G4.
Welcome to the team all new members!!
brock
|
|
|
| |
|
|
|
 |
|
 |
|
Fresh-Faced Recruit
Join Date: Jun 2000
Location: Mesa, AZ
Status:
Offline
|
|
SETI Team,
My question concerning Alti-Vec support comes from my PC use (my only sin in life  ). If they can do a "DOS" based version for the PC using command line use, then how come an Alti-Vec enabled version using some sort of non-graphical based isn't possible? Even just a non-graphical version would be kick-butt because of testing I've done to show it is about 50% faster or more without graphics! I think we should start a petition to get support for Alti-Vec. It should be viewed as a seperate platform in of itself because of it's differing nature in calculations. If BSD and BeOS can get support, then what's holding back this?
-jeffdano
|
|
|
| |
|
|
|
 |
|
 |
|
Professional Poster
Join Date: Mar 1999
Status:
Offline
|
|
seems to me there was mention of an altivec-enhanced version of seti that was in the works... maybe i am having a false remembrance... search the other fora
i admit to not being very concerned about an altivec version as i only own G3 Macs

|
|
|
| |
|
|
|
 |
|
 |
|
Dedicated MacNNer
Join Date: Jan 2000
Status:
Offline
|
|
Actually, I believe they have committed to AltiVec. Just down the road, not immediately.
And as someone who is newly concerned about AltiVec (i have a bouncing baby G4/400), I will keep you all abreast of the information.
|
|
|
| |
|
|
|
 |
|
 |
|
Mac Elite
Join Date: Jun 2000
Location: Pasadena
Status:
Offline
|
|
no, Seti@home is NOT committed to Alti-vec, at least not "Officially." I know this because about 2 months ago, I emailed the lead developer of the Seti project, regarding the issue, and his reply was something along the lines of "G4, K7, and many other CPUs have their special set of instructions to optimize performances, and we at Seti do not have the resources to optimize our client software for specific CPUs..."
Unfortunately, they won't release the code for developers to do the optimization, fearing of people tampering with the code to mess around with results etc.
I don't think I kept that email, but I'll look for it and if I find it, I'll post it here.
|
|
G4/450, T-bird 1.05GHz, iBook 500, iBook 233...4 different machines, 4 different OSes...(9, 2k, X.1, YDL2.2 respectively) PiA to maintain...
|
| |
|
|
|
 |
|
 |
|
Mac Elite
Join Date: Apr 2000
Location: Minneapolis, MN USA
Status:
Offline
|
|
I recently read an article that at the rollout of the G4 computers (the initial introduction) an Altivec optimized client was shown. However, it wasn't available to the public and it was probably a v1.1 client.
(1) no altivec client is expected for 3.x.
(2) seti wanted 200,000 people in this, they got 2 million plus
(3) they would prefer we slow down rather than speed up
(4) at this rate, they will run out of data with any more speed/users added.
(5) it stinks, but thems the breaks. they'd need to restructure the whole enterprise to do it differently.
|
|
|
| |
|
|
|
 |
|
 |
|
Mac Elite
Join Date: Apr 2000
Location: Minneapolis, MN USA
Status:
Offline
|
|
Very interesting: I'm reading that some of the G4 users in the beta program are actually getting shorter block times. Stay tuned, this may be very interesting.
|
|
|
| |
|
|
|
 |
|
 |
|
Dedicated MacNNer
Join Date: Mar 2000
Location: Toledo, Ohio
Status:
Offline
|
|
Todd you are right. I have two other G4 friends that are averaging about 1 hour less with the beta 3.
G4 Friend 1: 6:30 -> 5:25
G4 Friend 2: 7:08 -> 6:11
It looks like it's opimized for something.
Here's the proof and plus you can see how it is killing the Winblows machines. Win/Seti is about 4-5x longer with beta 3.
Here is where one of my friends posted his times.
http://www.kwsn.net/seti/
Go to the link right under the top picture (NEW New Charts On Processing Times NEW), and then 1/2 way down the page you'll see the beta results for Mac/Win machines.
brock
|
|
|
| |
|
|
|
 |
 |
|
 |
|
|
|
|
|

|
|
 |
Forum Rules
|
 |
 |
|
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
|
HTML code is Off
|
|
|
|
|
|
 |
 |
 |
 |
|
 |
|