Project 5102 PPD to low?

[Karamiekos]

Does anyone know if there is a way not to get this WU? I tried setting WU packet size from large to small, but that didn't seems to work. I know of course it is all about the science, but I just went from 1200ppd to 600... and due to the competitive nature of folding this make me pretty sad in the pants.

I deleted the queue and work folder once to get something else, and someone explained to me how that would hurt the science since that work unit would not be issued until the time expired. So I won't do that again, but its hard for me to stay competitive like this... So if there was a way I could change a setting to not get them....

If not oh well, and such is life, but thank you for your time.


Updated thread title. -7im

[Baowoulf]

Is there an actual problem with the WU as to why you wouldn't want to get a WU from this project? If it's just for points I don't think there is anything you can do about it. Otherwise you'd have people doing it all the time to get the WU's for the most points instead of just for the science of it.

[Foxery]

Hardware specs? Log file? Recent changes or other unusual activity on the machine? You haven't given us any way to determine if there's a problem...

[Karamiekos]

There is no problem other than it takes over 5.5 days to complete one of these guys. It's huge! I was told points were also linked to how much science being done. If that's the case my Comp isn't giving the best science/time output.

Absolutely no problems though. Must just be the new A2 core is Really that much better at utilization. Been folding since January and just never had a WU take this long.

5102's are a beast!!!! lol

[toTOW]

Could you give more details about your machine if you want us to tell whether your PPD is normal or not

Note that deadline is 8 days for this WU.

[Karamiekos]

Could you give more details about your machine if you want us to tell whether your PPD is normal or not

Note that deadline is 8 days for this WU.

I'm pretty sure it's completely normal. IIRC, the 5102 is core A1 based and thus locked to 4 threads and optimized for only quad cores really. The A2 Core for project 2662, which I love BTW, Is better optimized for either quad or dual.

I do get higher CPU utilization with the A2 core, I even notice my temp creep up a degree or two. Here's the rundown:
HP Laptop with Turion X2 @ 1.8GHz with 2 Gigs o' RAM
5102 nets 630 ppd
2106 nets 830 ppd
2662 nets 1200 ppd

All is normal, I may have been getting greedy about points. I just finished a 5102 early this morn, I think it said I had 30-some % of time remaining. I apologize, its all about the science, but points are nice too! I bought my new graphics card banking on the fact ATI cards will someday soon get a little boost. Sad huh....had this 4850 for a week and haven't even gamed on it for an hour....been folding non stop....lol gets slightly better ppd than my CPU though in this DeskStar. 1200-1600 on my x2 3800 @ 2.5GHz versus 1700-2100 ppd on the GPU depending on which unit I get.

[RipD]

I'm running a 5102 on dual core virtual machine (VMWare) with 700M of memory allocated. My PPD is about 1/2 of what I typically get from this VM.

[Trivolve]

whichhh leads us to question why the speed of folding on the benchmark machine is not scaling properly, or whether there was an error in benchmarking.

[Flathead74]

I'm running a 5102 on dual core virtual machine (VMWare) with 700M of memory allocated. My PPD is about 1/2 of what I typically get from this VM.

Welcome to p5102 lovin'.

[Xilikon]

whichhh leads us to question why the speed of folding on the benchmark machine is not scaling properly, or whether there was an error in benchmarking.

The main problem with the benchmark machine is that it's not a true quad core, it's 2 dual-core processors. It's not the first time we saw discrepancies between the benchmark machine and the actual production.

[7im]

whichhh leads us to question why the speed of folding on the benchmark machine is not scaling properly, or whether there was an error in benchmarking.

The main problem with the benchmark machine is that it's not a true quad core, it's 2 dual-core processors. It's not the first time we saw discrepancies between the benchmark machine and the actual production.

That's not the cause. It's the same questions debated to death in the P2665 thread, and A1 core vs A2 core.

Please also note that getting half the PPD on one work unit vs. another work unit is not considered to be a problem unless more details can be provided. Some work units get a bonus, so they will always score higher. If you should happen to be running bonus work units, and then get a regular work unit, your PPD is expected to drop in half.

Not saying that is the issue here, but we DO need more info to make determinations like that. Don't just assume that because the PPD dropped in half that it is actually a problem. It may not be a problem at all...

Karamiekos, feel free to PM the researcher for this project and ask.

[RipD]

Don't just assume that because the PPD dropped in half that it is actually a problem. It may not be a problem at all...

Fair point. I'm just stating what I'm seeing - didn't mean to imply it was wrong.

btw, after letting this run longer I appear to getting about 70% of my previous average rather than 50%. I also happened to get above average scoring WUs just before this one (2662s? Can't recall the exact number). So while this WU is slower, it's still generating over 1K PPD for me. Some are slower, some are faster - my average stays the same.