• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

CPU OC with SMP/Bigadv TPF & PPD Database

J

jebo_4jc

[H]ard|DCer of the Month - April 2011
2FA
Joined
Apr 8, 2005
Messages
14,572
How's that for a thread title? :p

I think it might be useful to have a place to double check which CPUs at which clockspeeds produce what PPD.

I am particularly curious what other people are getting with their L5640s in their SR2. My particular setup gets *slower* if I bump up the bclk from 200 to 203. I've heard other people have them higher, so I would like to know who has what.

You should all have full edit rights, so have at it:

https://spreadsheets.google.com/ccc...jJabzNYOXRnUVRzQ3pZVlE&hl=en&authkey=CMWAuLgM

It may not be incredibly useful to input your entire HFM benchmark database in here. It would probably be best to stick with a representative sample (i.e. what your CPU gets on a 6701, 6012, 2684, 2686, and 6900 to cover the bases).
 
Last edited:
lol - I was about to start a thread on this exact topic - I had created a standard benchmark unit, and Musky has been working on making it easy to run with an awesome utility he wrote, and with more project captures.

@musky - are we close to to release?

The issue is given the huge variance in speeds even of the same project, we need to run the same captured unit to be meaningful....
 
sounds great.

The information in the spreadsheet might still be useful, in case somebody doesn't get a chance to run the benchmark WU.

We'll just have to adapt the "WU" category column to reflect the benchmark WUs.
 
damn gpu clients are killing my smp client PPD by about 600 PPD. its lower then a Q6600 @ 3.2ghz. but the gpu clients add 10.6k PPD :D
 
I'm going to see if I can't make a google docs form for this. TPF is H:M:S right?
 
khwain said:
I'm going to see if I can't make a google docs form for this. TPF is H:M:S right?
Click to expand...


its either H:M:S or M:S:MS.. dunno i just copied what jebo put.

holy shit this is funny watching multiple people editing the file at the same time.
 
I entered M:S and it converted it to xx:xx:xx whatever that means
 
yeah i just changed mine to H:M:S. its because you didnt put a number at the beginning to represent hours so it auto corrected it to H:M:S but in the opposite direction.
 
lol 339 people viewing it right now. i want to see what it looks like when 10+ people are editing it at the same time :D
 
So far I'm the only one with a i5 in the list. We need moar numbers!!
 
this spreadsheet has existed for about 36 minutes.

the numbers will come :)
 
MIBW I changed your 2686 entry to reflect the fact that it's a benchmark unit. Maybe we could get a bunch of consistent entries that way.

Once you/musky get the bench finalized, let's try to get some consistent numbers in there.

Also, try to report RAM speed in actual MHz, not ratios. I assume that's why MIBW shows "1333"
 
jebo_4jc said:
Also, try to report RAM speed in actual MHz, not ratios. I assume that's why MIBW shows "1333"
Click to expand...

Mhmm.. I suppose we take the DRAM frequency from CPU-Z and then multiply it by 2? Also does 1T help much? Haven't used 1T since the old DDR1 days.
 
rUmX said:
Mhmm.. I suppose we take the DRAM frequency from CPU-Z and then multiply it by 2? Also does 1T help much? Haven't used 1T since the old DDR1 days.
Click to expand...


it helps but only if its stable on your system. puts more load on the ram. so what might be stable at ddr3 1600 2T might be stable at ddr3 1333 1T. just depends on your ram and if 1T is the rated timing for that ram.
 
This is highly surreal watching it in realtime! Great stuff Jebo :D

A thought - how do we make additional sheets with this data sorted by project and then TPF?
 
may i be worthy said:
This is highly surreal watching it in realtime! Great stuff Jebo :D

A thought - how do we make additional sheets with this data sorted by project and then TPF?
Click to expand...
that might be better left to running periodic reports by dumping the spreadsheet to an .xls file and doing some fancy sorting/filtering

Last time I tried to make a large scale collaborative google docs spreadsheet we broke it by getting too complicated :(

You should just be able to "save as" and download an .xls file.
 
yeah i've been downloading it as an html file every so often in case something happens. i dont have excel or any word processing installed right now.


may i be worthy said:
This is highly surreal watching it in realtime! Great stuff Jebo :D

A thought - how do we make additional sheets with this data sorted by project and then TPF?
Click to expand...


you know theres a chat on there as well, in the top right corner.. just click on the anonymous users and the chat drops down. :D
 
Not to be stealing an idea, but I felt like doing something..."productive." Here's the result.
Here is a link to a google docs form that has all the basic fields of the spreadsheet. I left the tpf/ppd not required since you don't need both of them.
Then, this is where you can view the data. I have it set to read only for now, but I can change that if y'all end up using it. Sheet 1 is where all the data goes in, but it's not sorted. The default sheet is sorted first by CPU model, then by # of processors, then by clock, then by project. Right now it'll keep doing that until it hits 160 rows, then I just have to expand the data set the sort is looking at...

What do you think?
 
khwain said:
Not to be stealing an idea, but I felt like doing something..."productive." Here's the result.
Here is a link to a google docs form that has all the basic fields of the spreadsheet. I left the tpf/ppd not required since you don't need both of them.
Then, this is where you can view the data. I have it set to read only for now, but I can change that if y'all end up using it. Sheet 1 is where all the data goes in, but it's not sorted. The default sheet is sorted first by CPU model, then by # of processors, then by clock, then by project. Right now it'll keep doing that until it hits 160 rows, then I just have to expand the data set the sort is looking at...

What do you think?
Click to expand...


bad ass.
 
khwain said:
Not to be stealing an idea, but I felt like doing something..."productive." Here's the result.
Here is a link to a google docs form that has all the basic fields of the spreadsheet. I left the tpf/ppd not required since you don't need both of them.
Then, this is where you can view the data. I have it set to read only for now, but I can change that if y'all end up using it. Sheet 1 is where all the data goes in, but it's not sorted. The default sheet is sorted first by CPU model, then by # of processors, then by clock, then by project. Right now it'll keep doing that until it hits 160 rows, then I just have to expand the data set the sort is looking at...

What do you think?
Click to expand...
that's a good start. The only thing is, when people are inputting their CPU models or RAM data in the notes they are doing it completely blind. Whereas if they're typing it directly into the spreadsheet they can see the pattern everybody else is using.

One thing you could do is add some validation to the CPU column, but that would take a big list of CPUs.
 
I just need to figure out how to get the script to download the client and get it to work with the saved units. This is the way to do this - everyone running the same units. I have a 6061, a 6701, a 2684, a slower 2685, and a 2686 saved off. More to come...
 
Given that everyone running this will have fah installed anyway, could you include a setup that simply asks for where fah is installed and copy from there?
 
or how about package all the WU's together with the F@H client. then give the user the option to select which WU they want to run.
 
I think that is the problem musky is trying to solve - not legal to package and redistribute the client.
 
sirmonkey1985 said:
or how about package all the WU's together with the F@H client.
Click to expand...
It's against the EULA to redistribute the client. PG takes this very seriously and has busted others in the past. Otherwise, Vijay has no objections to musky putting together this benchmark. I've already asked Vijay about it.
 
What about memory speed and timings? I tightened mine and ppd increased 5%

It'll help explain discrepancies at least between processors with the same clock speed. I'd imagine 1333mhz at cas9 and 2500mhz at cas9 having a big difference.

NEVERMIND!!!! I opened the file lol
 
added my info. Going for tighter timings on top of 2000mhz ram helps A LOT.
+ 5kppd going to 2000mhz from 1600mhz
... + 10kppd getting cas 8 instead of 10 I had before
 
Well, two things - Musky's benchmark is getting closer to release, so I have added my latest times.

Secondly, SR2#3 got a bit faster with some memory fiddling, :D which I will write up in the SR2 Optimisation thread... but now with NUMA disabled there is no longer the huge advantage to 2:10 vs 2:8 with tighter timings. So I have gone from DDR 2050 down to 1640 and gotten a fraction faster, but more importantly, more stable. Looks like it is the 205 baseclock doing the heavy lifting, and 2:8 is a lot easier to keep stable than 2:10. Watch this space.
 
You must log in or register to reply here.
Back
Top