SPCR Folds Team Blog

aristide1 · Post by **aristide1** » Wed Aug 12, 2009 7:02 pm

Never heard of FAHWeb.com.

Passing thought and applicable to most video cards. but strongly with the GTX260. If you're going to run Ubuntu instead of getting the EVGA 1255 consider the 1257 ($5 more) or even the 1258($25) more. The reason? their default speeds are much higher. Yes the gpu and memory speeds are cranked up and not so useful, but the shaders are also cranked up.

Option if possible - get the 1255, flash it's BIOS to make it think its a 1257 or 1258. Not sure how it's done, and its dangerous as well.

KansaKilla · Post by **KansaKilla** » Wed Aug 12, 2009 7:36 pm

NeilBlanchard wrote:Hi,

My surgeries went very well: I had a small squamous cell carcinoma on my gum, so the first surgery removed 3 teeth and the bone around the roots. The margins were clean, and it is expected to be curative.

Because I had that, they did a PET scan which turned up the tumor on my right kidney. So the second surgery was to remove the tumor (leaving ~2/3 of the kidney intact), and again the margins were clean. Better yet, the tumor was oncocytoma -- it is benign!

So, the number of follow up scan have been greatly reduced. I had the 41 staples out last Friday, and I am up and about and feeling much, much better than I thought I was going to. The worst is definitely over.

Holy crap, Neil.

Oral cavity SqCC's are nothing to even joke about. Sounds like an alveolar ridge lesion. Also sounds like you got the right treatment. Think the Lord you don't have to have adjuvant therapy because radiation to the oral cavity is one of the hardest things a person can go through. Make sure you get and keep a copy of the pathology report, and that the report actually says "negative margins" or the like. I've heard patients tell me that their surgeon told them that he/she got it all when the path report shows close/positive margins. Hearing this makes me think I have to stop my dipping habit. Soon.

Hooray for the oncocytoma! You should be good to go.

You've been through the wringer, bro. I guess now you can concentrate on more folding.

cordis · Post by **cordis** » Wed Aug 12, 2009 9:49 pm

The fahweb.com site is a site that allows you to upload the web page view of fahmon on to an ftp site, and from there you can view it from the web. Here's my page:

http://www.fahweb.com/stats/cordis/

I'm not sure if you have to register to see the pages or not, but I think it should be up for public view. So basically, it's a slightly degraded copy of the normal fahmon view, but it's on the web, so you can check and see how your folding is going when you're away. I'm going on a little vacation in a week and a half, so I was looking for a way to keep an eye on things while I'm gone, and this seems to fit the bill. It doesn't seem very popular, though, so I was curious to hear if it was a bad idea for some reason. I guess it just hasn't been heard about. If you think it's a good site, I can toss it into the useful links, I guess.

And hey Neil, great news, hang tough!

aristide1 · Post by **aristide1** » Fri Aug 14, 2009 6:59 pm

My CPU can't find any work to do. Anybody else in that boat?

cordis · Post by **cordis** » Sat Aug 15, 2009 11:39 am

It looks like the classic non-smp non-gpu clients I have are floundering waiting for new WUs. One of mine did manage to get a WU, but it waited a pretty long time. Checking the fahmon server status page:

http://fah-web.stanford.edu/serverstat.html

It looks like my latest job that's stopped is waiting for a server that's in 'Reject' mode. That doesn't sound good. Oh wait, no, it just picked up a WU, but now it can't connect to the server to deliver any results, it might wind up stuck at 0%. Think I'll kill it, wipe the data files and restart it, that might help. I do hate it when the servers go funky.

Post by **NeilBlanchard** » Sat Aug 15, 2009 5:14 pm

Hi,

Luckily, it was only the PSU that had failed on my X2 4200+ machine. It is now running F@H and SETI @Home again. Whew. I've got 4GB of RAM and an original Ninja modified to fit the mobo, and 3 hard drives installed in an Evercase 4252 (remember them?), so it is good to have it back up and running Ubuntu, at least. WinXP was getting very flaky.

I replaced the PSU with another of the same unit (Fortron Green 300watt) because that is what I have on hand. Hopefully it was something unusual that caused the failure. The UPS that it is plugged into did switch to battery power and the alarm was going off, and I switched off the UPS. If it happens again, I'll shut down the computer first...

aristide1 · Post by **aristide1** » Sun Aug 16, 2009 7:48 am

For those people who built multi gpu machines and have the cpu doing nothing things are changing really fast.

The 45 watt AMD dual core cpus - the 4050e, 4850e, 5050e etc, they are all gone from NewEgg. Places that still have some are selling them for a premium. The 4850e was $55 on the Egg, now you need $100 to buy one.

The AMD X3 710 core is also gone on NewEgg, and it's only been around since the beginning of the year. They sold the last they had for $99. That was a deal.

KansaKilla · Post by **KansaKilla** » Mon Aug 17, 2009 4:28 am

had the same problems as aris and cordis, but both seem to have picked up work units overnight. i'm crossing my fingers.

cordis · Post by **cordis** » Mon Aug 17, 2009 10:05 am

It looks like the servers are running resonably ok now, although I did take a bigger ppd hit yesterday than I thought I would. Oh well, things seem ok now.

aristide1 · Post by **aristide1** » Wed Aug 19, 2009 2:43 pm

So I'm thinking you have a dual core cpu and you're running SMP, which is 4 threads and the 2 cpus are loaded down just fine, the threads outnumber the cores.

Under Windows (maybe not Win 7) quad cores are unevenly loaded. The 4th core does little work running SMP. Solution? Run 2 instances of SMP (affinity or not), especially on a quad core that has virtualization, but still effective if it doesn't. Result? More PPD and again, the work threads outnumber the cores.

But ....

Suppose we had a compromise. What happens if we run SMP under Windows with 3 cores? Yeah, it may take will take longer than 4 cores, but really? A lot longer or just a little? The 4th thread would manage to get passed around to whatever core had a free moment, and since it's Windows there always seems to be at least one.

Thoughts? I'd love to compare an AMD 710 to an 810, but that's not going to happen.

aristide1 · Post by **aristide1** » Wed Aug 19, 2009 2:58 pm

Here's something you don't see everyday.

http://www.newegg.com/Product/ShowImage ... -%20Retail

The BFG Tech GeForce GTX 295 H2OC is a single slot video card, which means there are some Core i7 boards that can hold SEVEN of them.

Can GPU folding handle 14 GPUs on 1 PC?

cordis · Post by **cordis** » Wed Aug 19, 2009 4:52 pm

Hell, imagine the radiator that would take. I'm thinking a 295 dissipates around 200W, seven of those would give you 1400W? I suppose a repurposed car radiator would do it, but still. And you'd definitely need a big psu, or a dual psu case, a couple zalman 1000W ones would be quiet. Hmm....

Lockheed · Post by **Lockheed** » Fri Aug 21, 2009 6:21 am

Post removed

aristide1 · Post by **aristide1** » Fri Aug 21, 2009 2:27 pm

I used to think XbitLabs was a good source for power consumption numbers, but their base platforms vary so unless it's just a video card swap the comparison is apples to oranges.

The cpu power consumption numbers are really questionable because of the same type of problem, which video card is plugged in at the time. They test a low power CPU and you see crazy numbers, because they have an ATI 4870 plugged in. Duh! Some forums have horrendous idle numbers, and then you see they're on an i7 platform. Duh again!

Hey Waldo, you just printed this AMD processor uses 170 watts under load. AMD rated it at 95 and only a few boards can support 140. What's wrong with this picture?

Some advertise "under full load" as when a game is playing. The problem with those numbers are often it's not just the GPU being maxed out, but the CPU and the GPU at the same time. It's a max, but when reviewing just a video card the numbers are misleading. Do this on an i7 platform and your eyes will bug out.

None seem to have the insight to say, "OK, here's our baseline, a xxx system with integrated video. It uses xxx watts. We add a video card, it uses xx watts more at idle and xxx watts more under load." I mean really, how hard is that? The only person I found with such insight is the guy who runs this place - Mike!

My 9600GSO - 20 watts more at idle than a low level ATI card at idle. I start GPU folding it increase 45 watts. I still look for power consumption numbers, but I don't take them too seriously.

cordis · Post by **cordis** » Fri Aug 21, 2009 3:42 pm

Hey, is anyone else seeing these 787 point gpu jobs? I'm getting a ton of them, and they're pretty cool, they seem to utililze the gpus pretty well, my ppd values are up pretty consistently:

gtx 260 -> 7083.00
gtx 275 -> 7555.20
gtx 285 (overclocked) -> 9854.61

I had a whole bunch of those 511 jobs last week, they dragged me down a bunch. But now, good times!

aristide1 · Post by **aristide1** » Fri Aug 21, 2009 4:39 pm

In the good old days my 9600GSO could reach 3800 PPD.

I rarely look at what project they are working on. I can't change it, why frustrate myself?

Lockheed · Post by **Lockheed** » Fri Aug 21, 2009 4:48 pm

Post removed

aristide1 · Post by **aristide1** » Sat Aug 22, 2009 8:14 am

Lockheed wrote:I didn't get the numbers from X-bitLabs so not sure where you got that idea from?

Not an idea, just a statement. It used to be a reference for me.

The numbers can still be useful, I tend to look at how much more (or less) one card uses than another, as opposed to the total power consumned. So no need to shrug, it's all good.

Post by **NeilBlanchard** » Wed Aug 26, 2009 6:37 pm

Hi,

Boy, "aspd" has picked up the pace!

http://folding.extremeoverclocking.com/ ... =&u=463718

aristide1 · Post by **aristide1** » Thu Aug 27, 2009 7:45 pm

I blinked and I am in 6th place.

You know it wasn't all that long ago nobody on our team was bright red.

We've made hugh gains, and we've lost a lot of member to boot.

cordis · Post by **cordis** » Mon Aug 31, 2009 8:20 pm

Damn, we're down to 49th now, passed by that Brasil team. Bummer.

kittle · Post by **kittle** » Wed Sep 02, 2009 3:35 pm

well my ppd dropped off a lot when my GTX 280 died.
since I cant find the receipt, i cant return it for a RMA

so that means i'll save up for the card I wanted at the start - GTX285 w/2GB. But thats going to be a while.

Add to that we have been having hot weather so i leave my main folding PC turned off during the day.

cordis · Post by **cordis** » Wed Sep 02, 2009 9:14 pm

Hey, another San Jose folder, that's cool. Well, summer is winding down, hopefully it'll get cool enough to do some more folding. And there's always the next sale at Fry's, you never know what they'll come up with.

KansaKilla · Post by **KansaKilla** » Sat Sep 05, 2009 7:52 pm

Man, I wish I had a Fry's nearby. Microcenter is okay, but Fry's just rocks. (Not the grocery store Fry's. They're just okay.)

aristide1 · Post by **aristide1** » Mon Sep 07, 2009 9:37 am

A question I posted in the Hardware/CPU thread over on the Stanford FAH forums:

My question would be in regards to quad cores and SMP under Windows. We've seen that the 4th core is less busy, and some people have dealt with that in various ways, like running 2 iterations of SMP on a quad core (I'm not advocating that nor do I want to start a discussion about it.) My question? Given Windows behavior and disregarding the L3 cache differences for the moment, would an AMD X3 710 (3 cores @ 2.6GHZ) fold significantly slower (25% or more) than an X4 810 (4 cores @ same speed)? The idea being to keep all cores as busy as possible, presumably when the threads outnumber the cores?

KansaKilla · Post by **KansaKilla** » Mon Sep 07, 2009 3:39 pm

it will be really interesting to see what the response is, and the ramifications for multiple smp clients on multiple cores.

aristide1 · Post by **aristide1** » Mon Sep 07, 2009 3:53 pm

Follow up: The multi core inefficiency is in the folding core, not in Windows. That's all I heard so far. Given that I wish they would create WUs that have 5 or 6 or even 8 threads. I tried to Google the issue but did not find any clear answers. They were contacting one team in particular that might have an answer.

Looks like the 710 X3 is already toast. The 'Egg hasn't had any for a while. AMD may be pushing the 705e.

KansaKilla · Post by **KansaKilla** » Mon Sep 07, 2009 6:15 pm

aris, you thinking about it for an upgrade to an existing build or as a new build component? given the superiority of hte core processors, this might impact others to choose intel over amd in this regard.

aristide1 · Post by **aristide1** » Mon Sep 07, 2009 6:43 pm

KansaKilla wrote:aris, you thinking about it for an upgrade to an existing build or as a new build component? given the superiority of hte core processors, this might impact others to choose intel over amd in this regard.

Thinking about upgrading the processor on the K9A2. Let's see what happens now that I have all 4 cards installed and running with the shaders cranked up (thanks to EVGA Precision Tools).

aristide1 · Post by **aristide1** » Tue Sep 08, 2009 3:28 pm

Well I decided to glance at some of the "new" P55 motherboards. No more than 2 PCI-E slots and like P45, you got 16 lanes to work with (some also use the 4 extra maintenance lanes in a separate slot, some do not.) So once again you get 2 slots with 8 lanes each at best. Add to that some i5's that cost as much as an i7 920 and the question is "Why bother?"

i5's remind me of AMD processors, followed by even more processors than have been dumb downed for marketing reasons. It's like anti-progress. Or perhaps "con"gress?

SPCR Folds Team Blog

fahweb.com info

yeah

things seem ok

radiator?

btw

sigh

hey