![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Yeah, I noticed a lot of 2653 is giving less PPD on many of my dual-core boxes. E6300 at stock usually average 1150ppd with those WU but now, it's down to 995ppd. Even my T7200 laptops which used to average 1400 ppd now get just 1000ppd. I guess they are getting more complicated...
Well, I don't know about temperature and different WUs, but these new WUs definitely decrease PPD as has been mentioned here and other threads. So far they seem to be in the minority, though. Some of the new WUs do provide similar PPD to the P2653 like the P3065. The P3064 however are much lower. YMMV.dang i picked up my first 3065 and it sucks! I went from about 16mins a frame to about 25mins a frame! AND my 2140 runs hotter now by about 3C!!!
It's the slower WU not your machine. I have a 3065 on one of my machines worth 2144 points, yet the PPD is approximately 700PPD less than that of a P2653.
3065 blows.
It's slow and has one of the tightest preferred deadlines I've seen. If they keep sending them, I'll have to rethink my dual-SMP setup. This is the first work unit I've seen that I can't hit the preferred deadlines with my split quads.
3065 blows.
It's slow and has one of the tightest preferred deadlines I've seen. If they keep sending them, I'll have to rethink my dual-SMP setup. This is the first work unit I've seen that I can't hit the preferred deadlines with my split quads.
Here's the funny thing: Cores 0 and 1 in Windows SMP are working on 3065 and are getting 1660 PPD. Cores 2 and 3 are working on 2653 are getting 2372 PPD, which is normal. You'd think if 3065 needed a ton of memory bandwidth or some other resource, my 2653 unit would be impacted in some way. That's what used to happen with 2610s. Like others have noticed, it looks to me like a ton of steps but crappy points.
Whenever I receive one of the new slower WUs such as a P3065, my other SMP instances speed up for some reason. Meaning, their PPD increase above normal. However, the P3065 WU processes extremely slowly. It's as if my systems don't want to process the new WUs or are lazy about it and appropriate a greater portion of their processors' capabilities to the P2653s. It's very strange, but there's something inefficient about these new WUs.
I have a P3065 WU on one of my AMD machines. It will barely make the final deadline. Forget about AMD hardware with these WUs. A Phenom will definitely be better but by how much is anyone's guess.
Whenever I receive one of the new slower WUs such as a P3065, my other SMP instances speed up for some reason. Meaning, their PPD increase above normal. However, the P3065 WU processes extremely slowly. It's as if my systems don't want to process the new WUs or are lazy about it and appropriate a greater portion of their processors' capabilities to the P2653s. It's very strange, but there's something inefficient about these new WUs.
No, it would just slow everything else down and some of the very slow WUs like the P3065 might not make the deadline on older systems.
I wish Stanford would change some of the other WUs to more accurately reflect work/time processed.
Yeah, I had to shut down the second client in my dual Opteron machine or it might not have made the deadline...
Whatever it is it's killing my dual Opteron with 4 cores...I've got:
5x 2653 = 1760pts
1x 3065 = 2144pts
1x 3062 = 1732pts
WTF is a 3065
Whatever it is it's killing my dual Opteron with 4 cores...
I had to drop the second client when the system grabbed one of these. Even with only one client it takes 29-30 minutes per frame with little else going on except some surfing and a TV app playing. The system is running at only 2.8GHz, though. Too bad I can't clock it higher.My headless Q6600 has 2 clients, but it's making 24 min / 1%
no to bad..... your poor opties though.....
You guys will just love the 3065s.
