boinc update

---

So, a new project has come into the BOINC world:

Collatz

The Collatz project is designed to prove/disprove the Collatz conjecture through brute force techniques. The project will end when this is done.

Good thing about this project is that there are optimised apps for ATi GPUs and nVidia GPUs. The 64-bit CPU application is pretty good as well, the 32-bit one is a bit crap. But, it doesn't have the single precision requirement like MilkyWay, so I can use my two 4670 cards for it as well.

So I've put them into Dutchie, and Dutchie into the old P4 case I had. My ATi cards in HAL are now working on Collatz too, and when GPUGrid hits 1M, it'll transfer over too. My CPUs will still work on the goals mentioned in my last post. So far, time per WU looks like:

4850:
802.781 sec for 159.75 creds
780.875 sec for 149.11 creds

GTX280:
1359.91 sec for 155.14 creds
1442.56 sec for 157.38 creds
1434.11 sec for 158.53 creds

Q6600 @ 2.85Ghz, 32-bit:
32533.23 sec for 167.23 creds

So yeah, CPU is a bit crap at 2k creds per day per quad core on 32-bit. But 1M Collatz creds should take about 60 days on one 4850. Or 30 on two, 15 on four.

boinc update

---

Ahh, more BOINC progress.

Just hit 13M credits total, nearly 10M in Milkyway, nearly 1M in GPUGrid and nearly 250K in SETI:

I'm hitting all my main targets here. I was wanting to pad out this graphic:

With results at each of the 10K, 25K, 50K, 100K, 250K, 500K, 750K, 1M, 2.5M, 5M and 10M milestones. Each milestone requires a certain amount of projects to reach that level, and SETI will be getting my my 250K one soon and MW will hit the 10M one soon. So that goal is done.

Well, what's left then? Gain more credits?

In respect to the fact that I'll be using a 16 thread machine at work in the near future, I've come up with the following plan:

a) Reach my main goals with respect to the second graphic. This means:
- Milkyway @ 10M
- GPUGrid @ 1M
- SETI @ 250K

b) Work on little projects where I am below 80% in terms of credit. For now I'll limit this to:
- Rosetta
- FreeHal
- Einstein
- Spinhenge
- ABC
- Ramsey
- SIMAP
- Malaria
- RCN
- SETI-Beta
And for each of them I'll get them up to 50K, and hopefully above that 80% line.

c) Now for some bigger goals - get as high as I can on each project to get 1st or 2nd in terms of 'who joined the same day I did'. This means the following projects and targets:
- CPDN @ 250K
- QMC @ 600K / 1.2M
- PrimeGrid @ 250K
- WCG @ 1.25M
- Cosmology @ 1M
- Rosetta @ 500K/1M/5M
For Rosetta and QMC, the numbers gets me into 3rd/2nd/1st but because there is such a gap it might be worth doing it in stages.

All this time, my GFX cards will be doing MilkyWay (get up to 20M or something) and GPUGrid/SETI/Aqua.

I do have a motherboard and chip here (i.e. Dutchie) not in a case or with graphics cards, so I'm deciding whether to build it up into a machine, or try and sell it on as my father is needing a new machine soon.

benchmark update

---

I took a bit of a detour this past couple of weeks. I'm doing a big round up of all the latest Catalyst driver releases for ATI cards on all the 3D benchmarks. There are some significantly surprising results! I'm less than halfway through as the testing is quite rigourous, and the results will be put up in the members section of BenchTecUK forums. Not a member? Join the team at HWbot, put up 3 scoring results (which isn't that hard) and start posting on the forum.

I've actually been quite generous to myself recently, and picked up some more equipment:

ATi 2400 Pro
nVidia 7800GS
5 Socket A processors, from Duron 700 to Athlon 1400
Socket A motherboard and cooler w/VGA port (1.5V and 3.3V)
Various Socket 478 processors from ebay

I took apart my old P4 mobo from its case so I could mess around with it. I took the cooler off, and the P4 proc came off with it! Looks like they'd used 3g of thermal material to seat the proc, and it'd set. I had to douse the area with TIM to help dissolve the thermal material, then use two screwdrivers to seperate the proc from the cooler. But I reseated it the way I wanted, hopefully I can push that beyond 3.6Ghz now.

I bought the Socket A procs and gfx cards from the BenchTec forums. I didn't have a socket A motherboard, but the procs were cheap, so I picked up a mobo/proc/cooler combo off ebay for under 20 notes. Got the mobo combo in the mail yesterday, only to find they'd used EVEN MORE thermal material on the proc... it hadn't even set properly. So I cleaned that up, and I'll test it soon after my Catalyst round up.

As far is BOINC is concerned, I'm atm only running 2 ATI cards on Milkyway - it'll be four when I've finished the Catalyst roundup. Aqua isn't giving out as much work, or credit, as before - but I got enough WUs to push me over 1m this week. GPUGrid climbed over 750k and is onto 1m fairly soon. SETI is being very awkward, and is having trouble getting to 250k. I may swap my nVidia card on GPUGrid over to Seti when GPUGrid reaches 1m - maybe get SETI to 1m. We'll see what happens.

At the minute, Dutchie is laying without power supply or graphics cards. I'll sort that soon.

On a bright note, I'm being commissioned to design a couple of websites. No doubt you can guess where that cash is going.

Also as a side project, as I've been learning CUDA and dabbling in OpenMP, I've been thinking of writing a benchmark program. The guys at BenchTec are being supportive - it probably wont be used at hwbot, but might use it in some form of competition at BenchTec. I'll post it here too :)

Take care, I'm back to benching.

belated update

---

So - not much happening. Until later today.

I'll finally get around to benching that 4850X2. Since then, I picked up the Gainward GS GLH 4870X2 (retails at £420) for £200 on the BenchTec website as well, so I'll be benching that too (and in CF). I expect some global points here, so wish me luck.

In terms of moving computers around, I finally organised the machine I'm selling to a mate - yes, I'm selling HarukaKanata in the CM 335 Elite. As such, my PC situation is being completely moved around (also so I can get my i7 out for benching).

JC now has a 550W PSU and the GTX280 in it. It's now my purely CUDA machine (BOINC and work).

HAL is now up and running - as a 2x 4850 in CF crunching machine AND my network storage:

E6400 @ 2.4Ghz
4GB DDR2
MSI Platinum PowerUp
700W PSU
2x4850 in CF with Akasa Nero Vortexx VGA Coolers
Antec 300

A few points about HAL - with two GFX cards in there always at full pelt, with the extra coolers, the temps are sort of in reasonable levels. The card on top never goes above 60C, however the card at the bottom hits 90C. While better than the 110 I was getting before, I'd like that 90 to be 80. Though I have noticed that if I overclock the cards, they actually use less shader cores on BOINC, and are cooler, but still crunch faster. Weird that. So the top card runs at 800/750 (mem speed doesn't matter for BOINC, so it's underclocked), the bottom at 730/750. Given that 3D speed is meant to be 625/1000, I reckon that's quite good.

In terms of Dutchie, well at the minute it has not got a PSU, and is going in my Verre v770. It'll get a couple of graphics cards (if at some point my 4670s become useful...) and a PSU as and when.

HELLFIRE is going into the Thermaltake Armour. It's big, roomy, and I've been spending money on things like the 4870X2 rather than on a case. I've changed the fan on the CPU cooler to an AC-Ryan 80CFM one rather than the Delta - I think it was going too fast for optimal air contact.

So today, my benching will hopefully consist of:

4850X2
4870X2
4850X2 in CF (i.e. a 4850X2 and something else)
4870X2 in CF (again, a 4870X2 and something else)
4850 in CF if I get around to it...

Maybe sort out my overclock on HELLFIRE, too.

In terms of BOINC update, well the ATI cards are doing wonders at MilkyWay. I'm now 6.3 million credits there, 9mill total on BOINC, top 1000 in world, top 50 in UK. I dabbled a bit with a project called Aqua@Home, which were giving very good credits per CPU time. But now their credit system is being moved around, so I'm doing a bit more on SETI to get it up to 250K, and Aqua to 1M. Would like GPUGrid on 1M too, with MilkyWay on 10M.

boinc update

---

Another week, another not much happening on the benchmarking front.

BOINC on the other hand, is going in leaps and bounds. Now at 3.88 million on Milkyway, I passed the 5 mill total mark, and The Clangers (the team I crunch for) passed 20 million.

As I was having a problem with my 4850s overheating in Dutchie, I've decided to purchase some VGA coolers:

At 11.47 a pop at SCAN, they seem OK to blow the air out the back, rather than just circulate it around.

It's also pay day next week. So along with the VGA coolers, I want a PSU for JC to put the GTX280 in, a PSU to put into my X2-4400 that I'm selling, another case for the X2-4400, and a 4850X2 to put into HELLFIRE to replace the GTX280. Then we'll see what's what on the BOINC front.

In other news, I was playing around with CUDA this evening, and got some things compiled. Though my basic idea of vector = vector * (vector+vector) at 512 threads didn't really see any improvement on speed. I have some things I need to CUDA up, so I'll have a bash at that soon.

boinc update

---

Well, over a week later and lets see what is what.

Over the past few days, averaged around 140k points per day (PPD). MilkyWay are now issuing some work units that are slightly less points per second than before, but at least it's all work!

Gone from 1.59 million to 2.83 million since the last post. At the minute I have one 4850 card in HK, and the other in Dutchie. Having them both in Dutchie was putting making temps of 110ºC!!!

Looking at the cards in terms of cores and core speed, I posted the following over on the MW forums:

Cards that can process MW at the minute:

3850 - 668Mhz, 320cores, GDDR3 @ 828Mhz
3870 - 775Mhz, 320cores, GDDR3/4 @ 900/1125Mhz
3850X2 - 668Mhz, 640cores, GDDR3 @ 828Mhz
3870X2 - 775Mhz, 640cores, GDDR3/4 @ 900/1125Mhz

4770 - 750Mhz, 640cores, GDDR5 @ 800 Mhz
4830 - 575Mhz, 640cores, GDDR3/4 @ 900 Mhz
4850 - 625Mhz, 800cores, GDDR3/4 @ 993 Mhz
4870 - 750Mhz, 800cores, GDDR5 @ 900Mhz
4890 - 850-1000Mhz, 800cores, GDDR5 @ 975Mhz
4850X2 - 625Mhz, 1600cores, GDDR3 @ 993Mhz
4870X2 - 750Mhz, 1600cores, GDDR5 @ 900Mhz

So in terms of RAC, using the 3850 as a normalised value of 1:

3850 - 1
3870 - 1.16
3850X2 - 2
3870X2 - 2.32

4770 - 2.25
4830 - 1.72
4850 - 2.34
4870 - 2.81
4890 - 3.18 to 3.74
4850X2 - 4.68
4870X2 - 5.62

Formula = (cores / 320) * (1+((Mhz - 668)/668)

So a 4870X2 will get you roughly 5.62 times more RAC than a 3850.

Now, my intention at the beginning of the week was to get some 3850X2 cards, for 100GBP each, so crunch and also bench for some points (because not many have been benched...!).

I was going with the following:
HELLFIRE (i7 920) - 2x 3850X2
HAL (E6400) - 2x 4850
Dutchie (X2-5050e) - 2x 4670
HK (X2-4400) - GTX 280
JC (3000+) - 3850X2

But then the following occured to me given the numbers above. 3x 3850X2 cards at 100GBP each give the equivalent of 6x PPD. However, for the same price, 2x 4850X2 cards give 9.36x PPD. So I made a spreadsheet comparing price of the cards, and MilkyWay output:

Read it here.

So the way to go is 4850s or 4850X2s. Which in a sense is good - 4850s will get me some nice Global points at HWbot - though 3850s would have gotten me hardware points.

Also this week, a friend needs a new computer, so I'm going to sell HK, my X2-4400, to them. This will mean Dutchie will more than likely become my PC for network storage, and HAL will become a BOINC machine. It will also fund the above purchases :) :)

Thus the situation will end up:

HELLFIRE (i7 920) - 2x 4850X2
Dutchie (X2-5050e) - 2x 4670
HALBERD (E6400) - 2x 4850
JC (3000+) - GTX 280

It just means I also need to purchase a PSU for JC, and a case for HAL.

boinc work

---

So, the BOINC project MilkyWay has work churning at an alarming rate :) Through a third party, the MW project can use ATI graphics cards with double precision accuracy to work on their science.

For the past couple of months, the users have been limited to 13 million credits a day - this meant for my two 4850s, they got around 15k a day because they were out of work for so long (and also there's an issue with cores to GPU ratio, as you could only get 6*cores worth of work units backlogged to crunch).

Now before when they had work all the time, each was getting about 60k a day (they were both in dual core machines). As of two hours ago, and because of a slight shift in their science over the weekend, the project is sending out work quicker than people can crunch it. So my ATi cards haven't stopped.

The change in science over the weekend caused some ATi cards to stop, meaning more work for the rest of us. As a result, I was getting 50k a day from my cards. Now hopefully I'll get 120k from the two cards, maybe even more:



Thanks to the ATi card development, MilkyWay is my top project, with 1.53 million credits as of 6pm today (it's 7 hours later, and now up to 1.59mil, which extrapolates to 205k/day).



At the minute I have several main aims in BOINC.

1) To increase the following graphic, by getting as many projects as required to certain credit levels - 50k (9); 100k (8); 250k (5); 500k (2); 750k (2); etc



2) To get into the top 80% of as many projects as possible
3) To beat my highest world position of 3185, which means getting 3.5million credits

As a result, my main projects are:

MilkyWay (for ATi cards)
GPUGrid (for nVidia cards)
SETI@Home (can use nVidia)
World Community Grid
PrimeGrid
ClimatePrediction
Rosetta

boinc update

I just hit 2 million total BOINC cobblestones today :)



That's just under 1 million this month alone :)

boinc update

I made 42,347 today, now at 1.6m:



Still chugging away at MilkyWay. Also trying to get up to 80% on all my projects - currently WCG is getting some credits. Also QMC is - trying to boost that up to 100K.

For the benchmarking, I moved the 4850 ATi card that was chugging on MW into JC, my brothers AMD 3000+ PC. Maybe there the overall power usage will be lower, but as it's only single core, it'll only pull 6 work units at a time, meaning it could be out of work more. Fingers crossed the longer work unit part of MilkyWay will be up soon!

boinc

You may have noticed that HALBERD is Crossfire compatible, but I've stuck only one 4850 in it, and the other 4850 into HarukaKanata. This makes sense in two circumstances - both are being used for games, or I'm using BOINC.

What is BOINC? It's a program which allows you to help scientists around the world with computational problems.

The project I'm helping with at the minute is MilkyWay@Home, as it uses ATI clients as well as CPU clients. However, work is distributed to a max of 6 'work units' per CPU core in each machine. Each work unit takes 30 seconds on a 4850 (compared to 15 minutes on the Q6600), and it's sporadic to get work, so by seperating out the GFX cards I can do more work.

They are working on a sub-project so the ATI cards can do longer work units (up to 8 hours each), and when that happens I'll consolidate both 4850s into HarukaKanata.

So, what is MilkyWay@Home? Well the project is developing an accurate 3D model of the Milky Way. They are focusing on strips of the sky, currently working on the Sagittarius cluster. Given the whole MilkyWay is a large place, there'll be lots of work to do!

From the website:

The goal of Milkyway@Home is to use the BOINC platform to harness volunteered computing resources in creating a highly accurate three dimensional model of the Milky Way galaxy using data gathered by the Sloan Digital Sky Survey. This project enables research in both astroinformatics and computer science.

In computer science, the project is investigating different optimization methods which are resilient to the fault-prone, heterogeneous and asynchronous nature of Internet computing; such as evolutionary and genetic algorithms, as well as asynchronous newton methods. While in astroinformatics, Milkyway@Home is generating highly accurate three dimensional models of the Sagittarius stream, which provides knowledge about how the Milky Way galaxy was formed and how tidal tails are created when galaxies merge.

How does a BOINC project get help from people like me? Because depending on how much work we do, we get credits for it. Credits are awarded on a system such that (from what I found), a Pentium III 800Mhz system will get 100 credits a day. I used to run a BOINC group called BOINC@Hull (from when I was in Hull university) for two years, and in that time and given that I had 11 computers, I made 800,000 credits (so 8000 days worth on a P3 8Ghz).

However, since then we now have dual and quad core computers (or 6x with the Dunnington core) that run at 2-3Ghz and compute 50-100 GFLOPs. Or take an ATI 4850, with 800 stream processors, that can run 800-1000 GFlops. So with that output, in the past month, I've already made 700k quite easily. In fact, a 4850 at full tilt all day can earn around 80k credits.

However, given the work situation at MilkyWay, currently that's around 20k. That is set to change with the new MilkyWay side project thing.

I currently 'crunch' for a team called The Clangers - a nice bunch of people. If you want to crunch, come join us and meet us on the forums! As a team, we have 14 million credits:


Our current push is on MilkyWay, however there have been other pushes before on Rosetta (protein folding) and SIMAP (protein matrix calculation).

Personally my score on MilkyWay is skyrocketing thanks to these 4850s:

This graph shows the last 60 days. Got my first card about 20 days in, and the second card around 20 days after. Time dictates I should get another?? =P I could stick it in JC. (The recent tail off is because of me benchmarking :) )