LinkBack URL
About LinkBacksIs my logic about right? I think it would be something like that, going from 3 poor cores to 6 very fast cores at the same tdp must mean double ppd and roughly double ppw for the CPU alone.
Is my logic about right? I think it would be something like that, going from 3 poor cores to 6 very fast cores at the same tdp must mean double ppd and roughly double ppw for the CPU alone.
this is what makes the servers so good for ppw. The bigger servers are using about 370 Watts but average around 30000ppd when using the bigadv switch. Gives an average of over 81ppw. The smaller servers only using around 210watts but are averaging 12-14kppd at the moment so not as efficient.... (yet)Originally Posted by nibbler
I managed to break the 5 Million mark and hit 3 million so far this month. I don't think I'll get to the 4 Million mark as that would need over 7 days from now...
Tattysnuc (26-11-2010)
That's a fairly significant achievement. I seem to remember you saying that you had to take some of the servers down this month for some work you were doing.
Looks like this is Hexus's top scoring month from what I can see! Well done to all involved!
I'd like to thank my friends for supporting me...
Tattysnuc (27-11-2010)
Cheers.
Two of the bigger servers dropped off about two weeks ago, and a couple more this week. You can see the drop from the 1 million points a week mark.
Still, I've pretty much doubled my score this month which I'll take.
And if this isn't Hexus's top scoring month, I'll be very surprised as we are on the way to 6.5 million points in a month.
I wonder if that'll get us on hexus's front page. would be good to drum up some more support.
Yes, being on Hexus' front page would be good
On the other hand, we could just try to coax Lowe and Directhex back
I'll let you bake the cookies and cup cakes while I buy the cloudy lemonade...Yes, being on Hexus' front page would be good
On the other hand, we could just try to coax Lowe and Directhex back
lololol
nibbler (26-11-2010)
Can anyone tell me anything about the -advmethods flag?
Only got given 75k for the work that I was expecting 125k for. I guess one of the work units was "overdue" or something, which would stack up becuase the first day i installed my 980x it wasn't stable which crashed my pc 2 or 3 days in a row while I was at work, leaving the machine sitting idle.
Harrrrummmmppppphhhhh. At least I'll still get to my 1million ppm unless we have a power cut or something.
Only 5k to go!
Harrumph too! Just checked the work server and after completing the 500k WU I picked up when I started the client yesterday I got assigned a 2M. That'll chew up most of the rest of the weekend *sigh*.
Good to hear you going for an efficiency gain nibbler! A 95W PII X6 sounds like a great investment - and if you have a decent air cooler you should be able to throw some overclocking at it to push those WU through that little bit fasterI think you'll be better off with a GTX460 for folding, as it effectively has twice the hardware but doesn't have twice the power consumption, but of course with a PII X6 you could probably do -bigadv units if you wanted...
nibbler (27-11-2010)
Nah don't especially want to, not really an option out of the box and it would require the pc to be left on for 2 days at a time and I value my sleep! The GTS450 seems to me to be a better investment as it's pretty close PPDwise but much less power wise, but we'll see.
I think my motherboard is rubbish tbh, although it should manage a nice OC, so an extra 600Mhz maybe. Got a good old xiggy so it should handle it seeing as my current core temp is 30C under full load and that's for a 95W CPU!
I'm surprised there's that little PPD difference between a GTS450 and GTX460 - where did you dig up the numbers from? Although I have to say I've always wondered if the folding clients aren't actually *that* parallel, since the lower number of shaders in nvidia's cards seem to outgun the larger number of shaders in ATI cards so consistently... perhaps nvidia have now got up to a number of shaders where a lot of them are sitting idle for the folding client...?
Google tbh.
Some people say that it's closer to 12kvs15k all depends on the units. I'm currently getting 6.5kppd with my 9800GTX merely because it's a 353 point unit rather than 494. Roughly 20% extra just because of the WU then.
Apparently using the -advmethods flag means that a GTS450 will hit 14kppd. I'll be doing that then. I reckon the tradeoff of 1kless ppd and 60W less power is worth it.
So apparently there are good units for a G80 core with the GPU3 client and using -advmethods, giving it a go now.
AN article on Bit-Tech that I read about folding performance said that you should only run the GPU3 client on the new fermi architechture, otherwise it runs significantly slower. G80 should stick to using the GPU2 client.
Also, can you confirm what -ADVMETHODS details you've found, because I'm just running my twin 480's at stock at the moment because any sort of overclock returns NAN's for the current batch of OpenMM work units, and this has dropped my production rate per card to 9k. My GTX 285 on air does that, so I miust be able to get more out of them.
Might rebuilt my ITX machine and use it as a heater for the house. The wife didn't should when she came in my room last night and felt how warm it was. I've got a couple of X58 mobos, and a GTX 8800 and a GTX 285 sitting around idle. Wonder if they'd make a difference to the house temp if I get em up an running. These old edwardian(? - 1906) houses take a lot of heating!
There are currently 7 users browsing this thread. (0 members and 7 guests)
Posting Permissions
Reply With Quote
