TPU's WCG/BOINC Team

[twilyth]

Happy b-day bow.

 

[GREASEMONKEY]

Happy B-day Bow

 

[Bow]

Thanks Guys!!

 

[ChaoticAtmosphere]

Welcome to the 45 club Bow! Happy Birthday. (my 45th was in April)

 

[Bow]

Thats it, #1,000

 

[Chicken Patty]

I am 45 years old today, I have the day off from work. Its time for a shot and a beer

Happy bday bro, best wishes to you my friend.

 

[mjkmike]

Have a great Birthday Bow. I'm also 45 but most days still act like I'm 25.

 

[ChaoticAtmosphere]

I find this kind of amusing, apparently my memory didn't like being OC'd to 2000Mhz, the graph below proves it!

 

[BinaryMage]

What the? That is quite impossible!

 

[ChaoticAtmosphere]

What the? That is quite impossible!

That is what I thought until I came up with this theory:

22 wu's returned with instant errors...this may have caused my projected score to go down, then the system sent the units back to me and I sent them back without errors this time. so the graph going back to the 21st of sept. is when the system deducted the erroneous wu's then spiked back towards the 22nd of sept. when they were returned valid.

Kinda makes sense to me but ya it's still pretty weird. Oh and the sharp spike upwards? That's when they sent 24 units over the course of 5 minutes; they kept sending me six more because they completed within minutes and sent back to WCG - with huge errors! :lol:

 

[twilyth]

So basically, your computer went back in time to fix the mistakes?

 

[ChaoticAtmosphere]

So basically, your computer went back in time to fix the mistakes?

That's another possible theory. Maybe I should figure out how my computer does that and get it to take me way back in time to allow me to fix all my mistakes!

 

[Chicken Patty]

I'm not sure what to think of that graph

 

[ChaoticAtmosphere]

Well it was definitely my memory. After I noticed that I had 6 new wu's I'd come back and see they were completed in 2 minutes...I think i noticed this happen 2 times, which is when I shut down and decided to use memtest86 with my ram oc'd at 2000Mhz which Corsair tested and claimed it can do. I used the same timings and voltage as in their specs. 1 pass, 2 errors. That's when I brought the ram oc down to 1667 and it's been smooth sailing since as you can see below.

I also brought down my CPU to 3.0Ghz in case. I wasn't sure if it was ram or cpu errors at that point, so when I get the time I will slowly push back up to 3.5Ghz and leave my memory alone. It seems happy at 1667Mhz and I'm happy too because it's still 600 Mhz faster than my old DDR2 memory and double the capacity.

 

[HammerON]

I find this kind of amusing, apparently my memory didn't like being OC'd to 2000Mhz, the graph below proves it!


http://img.techpowerup.org/110922/Capture304.jpg

That is just funny! I wish my stocks (investments) charts looked like that

 

[Velvet Wafer]

Well it was definitely my memory. After I noticed that I had 6 new wu's I'd come back and see they were completed in 2 minutes...I think i noticed this happen 2 times, which is when I shut down and decided to use memtest86 with my ram oc'd at 2000Mhz which Corsair tested and claimed it can do. I used the same timings and voltage as in their specs. 1 pass, 2 errors. That's when I brought the ram oc down to 1667 and it's been smooth sailing since as you can see below.

http://img.techpowerup.org/110923/Capture1573.jpg

I also brought down my CPU to 3.0Ghz in case. I wasn't sure if it was ram or cpu errors at that point, so when I get the time I will slowly push back up to 3.5Ghz and leave my memory alone. It seems happy at 1667Mhz and I'm happy too because it's still 600 Mhz faster than my old DDR2 memory and double the capacity.

remember, low latency is better than high clocks on AMD, once you start to have clocks beyond 1600

 

[ChaoticAtmosphere]

remember, low latency is better than high clocks on AMD, once you start to have clocks beyond 1600

I did not know that. So you think Corsair's tested timings of 9-10-9-27-51 for 2000Mhz is for Intel system? and if so, do you think I can still manage to get 2000Mhz? If yes what timings would you recommend?

I kept currently have the timings at 9-10-9-27-41 (I can't do a trc of 51, the bios only goes to 41 on this board)

 

[twilyth]

This is an ongoing debate. Personally, I agree with going for lower latency, but the best thing to do is test it out for yourself. You've probably got at least 2 or 3 JDEC settings on your ram. Try the next to lowest setting in terms of latency and clockspeed (since they go hand in hand obviously) and see how it feels. It really depends on what kind of work loads you tend to use the PC for. I have a spread sheet listing total access times for the first 10, 100 and 1000 bytes of memory access and you can see that it varies a lot depending on how big your requests for sequential memory access are. I can post it if you want.

edit: since I dug it up, I may as well post it.

 

[Velvet Wafer]

I did not know that. So you think Corsair's tested timings of 9-10-9-27-51 for 2000Mhz is for Intel system? and if so, do you think I can still manage to get 2000Mhz? If yes what timings would you recommend?

I kept currently have the timings at 9-10-9-27-41 (I can't do a trc of 51, the bios only goes to 41 on this board)

yeah im sure they did only test them on Intel... i think you may be able to do it, but that it would prove more beneficial, especially in your case! 9 and 10 are pretty,pretty high timings, andi guess you could screw the NB a little higher, and the timings much lower, if you just aimed for something around 1866, 7-7-7 CR1!

BTW
This opened my Eyes when i saw it: Tony OCZ´s AM3 Memory Guide

 

[hat]

That's another possible theory. Maybe I should figure out how my computer does that and get it to take me way back in time to allow me to fix all my mistakes!

I wish my computer would let me travel time when I screwed stuff up

 

[Pembo210]

i had a graph that looked like that after my time/date clock was off for a couple days from some tinkering

I guess i have to get my rig back in gear if i want my WGC stars back.. FL summer is hot & almost over, will have it crunchin again soon.

 

[twilyth]

Here are the people who seem to have had the most improvement in their ppd over the past 4 weeks.

Note: This based on taking Free-DC weekly total divided by 7 and then subtracting from that the 28 day total divided by 28. So (wk.total/7) - (28day total/28)

Obviously, this double counts the most current week and therefore isn't as good as say, comparing the daily avg over the past wk to the daily avg over the previous 3 wks, but I think it still works. The main flaw is that it doesn't give you as dramatic of a comparison as you might otherwise get.

F150_Raptor|12,058
GREASEMONKEY|6,994
bogmali|4,440
msgclb|2,431
jellyrole|1,892
twilyth|1,752
JrRacinFan|1,306
Qu4k3r|1,139
ChaoticAtmosphere|899
Niko084|800

 

[BinaryMage]

I wish my computer would let me travel time when I screwed stuff up

It's called Control-Z.

 

[JrRacinFan]

OK guys, remember when I said in another thread where I can't pull more work and a few WU's errored. Well I just primed(4hrs) and linpacked(1hr) my pc and it's not my rig. Now it can't pull anymore wu's from WCG. My 720 Black is pulling and sending work ok. Anyone got insight on this?

EDIT:

Want to add I didn't change any settings, gaming rig in system specs.

 

[ChaoticAtmosphere]

OK guys, remember when I said in another thread where I can't pull more work and a few WU's errored. Well I just primed(4hrs) and linpacked(1hr) my pc and it's not my rig. Now it can't pull anymore wu's from WCG. My 720 Black is pulling and sending work ok. Anyone got insight on this?

EDIT:

Want to add I didn't change any settings, gaming rig in system specs.

when you primed did you do small FFt's with no memory? I suggest doing a memtest86 from a boot flash drive to see if you ram is producing the errors, mine was.