A1 units are brutal

J

jebo_4jc

[H]ard|DCer of the Month - April 2011
Joined
Apr 8, 2005
Messages
14,567
It's hard to watch a Q6600's ppd drop from ~6500 to 1700

Why are these units still floating around anyway?
 
Because they need to be completed. We all have dealt with them at some point.
 
Used to be all we got back in the day. I remember running two VM's on a q6600 just so I would maximize the productivity with those old A1 units, back before the GPU clients came out. While I know there are still A1's out there that need to be processed, it wouldn't hurt my feelings at all if they were to get rid of them and set them up for the A2 or A3 platform. ;)
 
My unlocked 720BE goes from just over 6 minutes per frame at about 6k ppd to almost 11 minutes per frame and 2000ppd.

I bet they're still coming to us because most people ditch them when they see it come up as an a1.
 
OvrrDrive said:
I bet they're still coming to us because most people ditch them when they see it come up as an a1.
Click to expand...

This is very likely the case. I don't monitor mine that frequently so when it comes, it comes. The number of A1's given the number of clients I am running is pretty small.
 
There are plenty of A3's available, but the A1's have to get finished and every once and a while you get lucky :)
 
OvrrDrive said:
I bet they're still coming to us because most people ditch them when they see it come up as an a1.
Click to expand...
Yep, pretty much what happens. They will be completed but if people delete them it will take that much longer. There are still months left and we will likely see them come and go for most of the remainder of the year.
 
They suck but what else can you do except "take one for the team". ;)
Process them and move on so someone else doesn't get stuck doing them. :) Karma will repay you.
 
I know the pain of A1s all too well. I've gotten three of them in the last day or so. Going from about 7500PPD to about 2500PPD is not a pleasant sight. Obviously, I just let them process but it is rather painful when I see two or more of them crunching at the same time.

 
If you are running VM, or have VMware player installed what you can do is just run a second instance of the client so you can run the A1, and then another A2 simultaneously so your machine isnt just idling.

A1 only use 2 cores, so if you have an i7920 and pull an A1 your rig is 75% unutilized.
 
nitrobass24 said:
If you are running VM, or have VMware player installed what you can do is just run a second instance of the client so you can run the A1, and then another A2 simultaneously so your machine isnt just idling.

A1 only use 2 cores, so if you have an i7920 and pull an A1 your rig is 75% unutilized.
Click to expand...
For those with i7s or other machines that have many cores, just start a few single core clients with the -oneunit flag to prevent all those cores idling for potentially days without work. Some of these A1 WUs like the P5100s take an excruciating long time, better make the most of it. ;)
 
Why don't they just leave the A1s for people running the standard windows client that only uses one core/thread anyways? Seems most people running these clients wouldn't care what they got as long as it's running since I'm sure they wouldn't know an A1 from a VM, hah.
 
T4rd said:
Why don't they just leave the A1s for people running the standard windows client that only uses one core/thread anyways?
Click to expand...
One core\thread is not supported since it's a SMP WU. there's a minimum of two cores required to run them.
 
APOLLO said:
One core\thread is not supported since it's a SMP WU. there's a minimum of two cores required to run them.
Click to expand...

Ahh, my bad. For some reason I remembered them only using one of my threads. I don't think I've seen one in a couple months, hah. But I've only been running bigadv until just now where I've been recruiting a few quads at work for A3 duty. I'm sure I just jinxed myself now :(.
 
T4rd said:
Ahh, my bad. For some reason I remembered them only using one of my threads..
Click to expand...
The big problem with the A1 Wus is essentially their inefficiency in core utilization. They cannot max each core and hence perform less effectively than the A2/A3 WUs. They were the first SMP WUs released by Stanford years ago, and the sooner we complete them, the sooner Stanford will EOL them, so to speak.
 
That's pretty much how I've been treating the A1s. They have to get done eventually plus the extra PPD I get over my old notfred setup evens out the drop from the occasional A1.

As long as the A1 units still provide good results for Standford it's all good.
 
I set up 4 xp VM's with the old SMP client so i can crunch 4 of these at a time.
Lets get em gone!
 
+1 cookie to Nitro for being the man
Posted via [H] Mobile Device
 
Vaulter98c said:
+1 cookie to Nitro for being the man
Posted via [H] Mobile Device
Click to expand...
+1

In my wildest dreams:

Since the problem with A1s is core utilization, they should be saved for dual cores, which, in my experience, produce about the same PPD with A1 units as they do A3s
 
You must log in or register to reply here.
Back
Top