Maniacken [US-Distributed] 2004-06-06 16:39:25 | I finally got it. 1.004391 (2064.6 Mpts) [v4.41b] {67D7FA734D1938F889F8D650} happy hunting a better result to everyone. |
Pollock[Romulus2] 2004-06-06 18:36:47 | Nice Hit! 1.025061 (472.4 Mpts) [v4.41a] {87CB495C828C47EEC597CD70} This is waiting in the recheck line, but needs a little more work on another parameter. It will get better when the 10 test runs lined up in the queue are finished. One parameter was going in the wrong direction. Phaserot was getting stuck. One parameter was improving, but several others were going in the wrong direction. I'm surprised that it grew to 1.26, the backward trend started at 1.10 and just continued. The bunch of 1.30 runs I sent should keep growing and point it in the right direction. Also working on a second (more efficient) breed at 1.20 and improving rapidly. It should be ready soon. |
Maniacken [US-Distributed] 2004-06-06 19:13:27 | nice going pollock. the first hit of that result was a 1.02. it then went 1.015, .98 and .97. the results are just jumping arround. i look forward to your test results, but ill still let my computers guess and check. just wondering if you could explain how to set up a queue for multiple tests. |
AySz88 2004-06-07 05:36:29 | Don't forget that what you think is the 'wrong direction' might not actually *be* the wrong direction..... People might be introducing too much user interference. That said... Can someone explain what each parameter controls and the function of the parts? |
Pollock[Romulus2] 2004-06-07 09:09:50 | quote: Simple solution: Do not use the sample file or remove all of the user-created results. If they do not have a #gen=x; value, they are user-created. My personal opinion is that the sample file does more damage by focusing the clients on only the higher-scoring results. Many slow developing (and maybe better) breeds are lost because of this. My results.dat file contains nothing (zero) results from the sample file. The reason that my tweaked results are similar is that the breed at the top of the list is one that I created. I have three different breeds in my list, but two are lagging behind. They may eventually be better, but they never make it to the sample file because it favors the top scores. Not my fault. quote: My (ignorant) interpretation of the parameters: 'Solenoids Only' (first section) used in both the Phaserot and Chicanelinac simulations. tantalumrodr=000; Tantulum Rod 'R' - The rod angle of the tantulum rod where the initial 'burst' takes place. tantalumrodz=000; Tantulum Rod 'Z' - No real clue! Apparently it is the force (energy) of the initial burst. s2f=999; Solenoid 'F'ield - The electrical field setting for the solenoid. s2l=999; Solenoid 'L'ength - The length of the solenoid tube. s2r=999; Solenoid 'R'adius - The diameter (height) of the solenoid tube. d1l=000; Drift 'L'ength - The length of the space between the solenoid tubes. Phaserot only: phaserotcells=000; - No real clue! Possibly the size of the phase rotation cells. ps2f=998; 'P'hase Rotation 'S'olenoid 'F'ield - Same as above. ps2l=000; 'P'hase Rotation 'S'olenoid 'L'ength - Same as above. pd2=000; 'P'hase Rotation 'D'rift Length - Same as above. prf2p=225; 'P'hase 'R'otation 'F'ield 'P'hase - The electrical field setting for the phase rotation cells. prf2v=475; 'P'hase 'R'otation 'F'ield 'V'oltage - The electrical voltage setting for the phase rotation cells. Chicanelinac only: The bending chicane (middle section) in the simulation is not adjustable. linaccells=000; - No real clue! Possibly the size of the accelerator cells. ls2l=000; 'L'inear Accelerator 'S'olenoid 'L'ength - Same as above. ls2f=916; 'L'inear Accelerator 'S'olenoid 'F'ield - Same as above. ld2=000; 'L'inear Accelerator 'D'rift length - Same as above. rf1v=000; 'RF' Frequency 'V'oltage - The electrical voltage setting for the accelerator cells. rf1p=366; 'RF' Frequency 'P'hase - The electrical phase setting for the accelerator cells. |
Pollock[Romulus2] 2004-06-07 09:23:08 | quote: Simple. Put all of your TEST runs in the queue.txt file with a line-break after each one. It should look the same as a results.dat file if done properly. Not sure if there is a limit on how many will run, but it will work with ten results for sure. |
Stephen Brooks 2004-06-07 12:46:27 | phaserotcells and linaccells are master genes: if you increase them, you increase the number of repeats of the basic pattern in that part of the accelerator, and hence you will actually produce more other parameters to choose from. Now, the reason why you reducing them has so far made things better is that if you increase them without filling in the additional parameters, you will get random values appearing for those parameters, meaning it might mess up the result. Theoretically, I'd think high values of phaserotcells and linaccells would be capable of the highest muon transfers, but perhaps I'll have to modify the algorithm in some way to prevent the program taking the 'easiest' route in these cases... tantalumrodz is just the Z-coordinate of the tantalum rod: it can also move back and forth within the solenoid. |
AySz88 2004-06-07 15:39:09 | Maybe instead of making a non-specified parameter random, you could make the program match the parameter of the previous cell? |
Maniacken [US-Distributed] 2004-06-07 16:23:33 | pollock do you mean more efficient as in less mpts to get the result? If so i have noticed a drop from about 500 mpts per result to arround 415 to 420 mpts for my highest muon rates. |
Maniacken [US-Distributed] 2004-06-07 16:25:10 | hey pollock i got one brewing also 1.033560 (415.1 Mpts) [v4.41b] {F8338CCCEC91BA7522CAE187} 1.033560,1.019831,1.035282 (1246.7 Mpts) [v4.41b] {17BB6D3C13CBC0E076D1F1EE} |
kitsura 2004-06-07 17:40:18 | I haven't been using any of the sample files and I realised that all my chicanelinac results are stuck at -1.7x I wonder how it suddenly just jumped to -0.0x from -1.7x is it a calculated result or a simulated one? |
Maniacken [US-Distributed] 2004-06-07 18:12:59 | maybe you got lucky with a random trial type. |
kitsura 2004-06-08 05:23:42 | Impossible, random trial types are more often than not very low yields, most of the better results are independantly optimised or manually optimised. |
Herb[Romulus2] 2004-06-08 06:09:13 | Just look at your -0.0 result what the gen=x says if it's 0 then you've proven possibility of progress with randoms, if not it was a pretty good selection of any of the other methods, especially if gen=7 (extreme) was chosen. |
kitsura 2004-06-08 10:23:37 | Heh, maybe I didn't make myself clear in the first post all my results are stuck at -1.7x so I don't have any -0.0 result. But I was wondering how the overall results as seen in the graph managed to jump such a huge gap. |
Pollock[Romulus2] 2004-06-08 12:31:09 | quote: Yup, it looks like the latest bunch at the top of the chicane is a shorter, more efficient breed. It is obviously different. If you open the results.dat file in WordPad and turn off text-wrapping, then scroll to the right, ot is obviously much shorter than the previous bunch. It looks like both breeds are competing well. This discussion gave me an idea to aim in a different direction. The first experiment (early-development)with phaserot was a success. 1.359739 (2265.3 Mpts) [v4.41a] {94DC129F4A8D87C21572F6CB} Hopefully, the next generation will get even closer to the baseline. It looks promising. Then it should work just as well on the chicane. Update: The second generation was a roaring success. The first run: 1.431589 (459.5 Mpts) [v4.41a] {A3D1CA5D82E44284969D1029} quote: The one that made the big jump for me was a hand-made result. At the time, the stats were down for a few days and TA[Z] reached the same point with a client-generated result. The answer to your question is that both methods worked. If you want to improve your scores without 'tweaking' then try setting your config.txt file to run more 'extreme' simulations. The 'extreme' runs do basically the same thing as manual tweaking. 'Random' runs usually do suck, it is not impossible for them to generate a good result, but it is not very likely. |
Maniacken [US-Distributed] 2004-06-08 14:40:19 | any idea when you are going back to chicanelinac pollock? |
Pollock[Romulus2] 2004-06-08 16:23:02 | quote: Not Sure! I really did not expect such a major jump with the mods that are running now. The current phaserot run will be at least a 1.55xxxx when the final recheck finishes. It still needs some adjusting. If i send it in, it will change the direction of the sample file because it is a major modification. It will not be sent until I have at least two breeds to compete with each other. That may take some time with only one computer to run them on. The chicane will likely take even longer, but I may set up a list before I leave for work tonight. It will be interesting to see if it works as well as phaserot. If you want to try some simple tweaking, my plan was to work on a different section which would be easy to change. The current values for: tantalumrodr=000;tantalumrodz=422; Those just do not seem right to me. I would suggest changing the 'z' value to 000 or 999, then adjust the 'r' value to 100,200,300,etc. until something works. The flow of particles into the bending chicane still seems to just clog it up with a heavy loss. Those changes may even out the flow of particles into the chicane. The part that I am working on is totally unrelated to those settings. |
Maniacken [US-Distributed] 2004-06-08 17:13:41 | thanks for the ideas pollock. i was wondering if you had the time could you explain what parts you have worked on, and how those changes went. also possible the methods you are using to tweak your results. also how do you determine that you have at least two breeds. you had said "breed at the top of the list is one that I created. I have three different breeds in my list, but two are lagging behind" can you explain how you use this to better the results. if you dont want to go into detail on the forums and give away your ideas drop me a line at maniacken@hotmail.com or is there a way to do personal messages on these forums. maybe we can bounce ideas off of each other. |
Maniacken [US-Distributed] 2004-06-09 05:46:02 | here is the results i got last night. first i changed tnatulumrodz base run tantalumrodr=000;tantalumrodz=442;1.026??? (415.2 Mpts) [v4.41b] {277DCCC6C7D510337A58E3FE} tantalumrodr=000;tantalumrodz=000;0.181833 (233.6 Mpts) [v4.41b] {277DCCC6C7D510337A58E3FE} tantalumrodr=000;tantalumrodz=100;0.298363 (279.6 Mpts) [v4.41b] {277DCCC6C7D507BAE191B3FC} tantalumrodr=000;tantalumrodz=200;0.499475 (324.7 Mpts) [v4.41b] {277DCCC6C7D4F5228402FCA5} tantalumrodr=000;tantalumrodz=400;0.974538 (411.7 Mpts) [v4.41b] {277DCCC6C7D4DE5079FA3E90} tantalumrodr=000;tantalumrodz=500;0.978766 (416.5 Mpts) [v4.41b] {277DCCC6C7D4CDD8D0243754} tantalumrodr=000;tantalumrodz=600;0.728968 (386.9 Mpts) [v4.41b] {277DCCC6C7D4C35F1ECABC41} from this i concluded between 400 and 500 is the best i then changed the angle tantalumrodr=200;tantalumrodz=000;0.161057 (234.4 Mpts) [v4.41b] {277DCCC6C7D62E6D3E1AA65C} tantalumrodr=300;tantalumrodz=000;0.176091 (238.6 Mpts) [v4.41b] {277DCCC6C7D0BBBB66F48F1B} tantalumrodr=400;tantalumrodz=000;0.178179 (234.5 Mpts) [v4.41b] {277DCCC6C7D344C8828DE436} tantalumrodr=500;tantalumrodz=000;0.177135 (228.1 Mpts) [v4.41b] {277DCCC6C7EDD015DB6B4BD4} tantalumrodr=600;tantalumrodz=000;0.188062 (224.9 Mpts) [v4.41b] {277DCCC6C7EC5D22A03ADF60} tantalumrodr=600;tantalumrodz=000;0.176265 (224.4 Mpts) [v4.41b] {277DCCC6C7EC5D227D83C1D2} tantalumrodr=700;tantalumrodz=000;0.184756 (217.9 Mpts) [v4.41b] {277DCCC6C7EEEE70B342A55A} tantalumrodr=800;tantalumrodz=000;0.167565 (205.7 Mpts) [v4.41b] {277DCCC6C7E97B9DE916F6B3} early results showed the angle didnt really change much except higher angles seemed more efficient. i then changed tantulumrodz to see if more particles would swing the outcome. tantalumrodr=100;tantalumrodz=442;0.964080 (423.4 Mpts) [v4.41b] {277DCCC6C7D7610C965C6D10} tantalumrodr=200;tantalumrodz=442;0.954319 (434.2 Mpts) [v4.41b] {277DCCC6C7D1F25A7BB60A9F} tantalumrodr=300;tantalumrodz=442;0.899612 (438.2 Mpts) [v4.41b] {277DCCC6C7D07F676D830647} tantalumrodr=400;tantalumrodz=442;0.853954 (430.8 Mpts) [v4.41b] {277DCCC6C7D308B4D427D636} tantalumrodr=500;tantalumrodz=442;0.767301 (415.9 Mpts) [v4.41b] {277DCCC6C7ED95C00112B365} tantalumrodr=600;tantalumrodz=442;0.723173 (400.3 Mpts) [v4.41b] {277DCCC6C7EC26EF2285950B} tantalumrodr=700;tantalumrodz=442;0.599579 (383.7 Mpts) [v4.41b] {277DCCC6C7EEB23C4008E081} tantalumrodr=800;tantalumrodz=442;0.519834 (357.0 Mpts) [v4.41b] {277DCCC6C7E93F4994455C5E} this appeared to show less angle was better. |
Pollock[Romulus2] 2004-06-09 07:10:18 | quote: Nearly every thing has been adjusted at least a little bit. Both were done the same way. I started with an old 'Solenoids Only' front section from v4.32 which was proven to be a good one. The rear sections ,linear accelerator and phase rotation channel, were just random ones that were generated by the client. The rear sections were modified by trying 000,500,999 as values for each group of parameters. Then they were adjusted in the most likely direction.The phase and voltage settings for both models are dependent on each other and are the hardest to find. That is why they were both set to 500 in the early results and refined later. Then the tantalum rod values were modified to slow the flow of particles into the bending chicane so they would not all hit the chicane at once and 'leak' as badly. I still think the 'best' result will have a different value for tantalumrodr= than the current 000 but that may change later. quote: Example: one breed of chicane results has a longer (29 solenoids) front section. The current top result has 25 solenoids. I added solenoids to the longer one as an attempt to slow down the flow before the bending chicane. It eventually evolved in a different direction, then unfortunately began to follow the shorter breed. The two are nearly identical now. Phaserot is a different story. quote: If there are two obviously different results with different sets of parameters, it can encourage the client to try a wider variety of parameters. Sometimes they will merge together (chicane) or move in different directions (phaserot). The sample file leans toward only the highest results, then adds a bunch of low (mostly junk) results. The results which are slightly lower than the highest score are usually ignored. The two lower scoring breeds have never appeared in the sample file, but I have at least 100 different versions of them in my results.dat file. quote: It would be wonderful to see others competing. With only one computer, it is a sure bet that when the final results are in, I will not even be in the top ten. There is too much computing power in this project for me to compete in the fine-tuning when the results are nearly maxxed out. I just look at the models as a puzzle. The only thing that I am doing now is putting the pieces in order. Making them fit perfectly takes more computing power than I could hope to generate. |
AySz88 2004-06-09 12:51:46 | Am I the *only* person that actually reads the technical stuff? http://stephenbrooks.org/ral/report/2003-1/ <-- most recent that has explanations The range of the tantalum rod's angle is between 0.00 and 0.50 radians (if it's the same as in previous optimizations) - the previously optimized value was 0.09359 radians (tantalumrodr=187). However, there's a graph also that says 0.45 radians (tandumrodr=900) would produce the least loss, so I guess some experimentation is needed. Also, the optimized tantalum rod's z location was 0.2252 metres (range was 0.00 to 0.45, so that would be tantalumrodz=500 ). Notice also that solenoid lengths and radii were *not* at their maximums for the chicane. Length = 0.463 meters; snl=657 Radius = 0.402−0.003n meters, or snr=(1000-7.5n) when n > 7 One thing to note is that this project seems to only count on-energy particles (I think). The chicane is designed to intentionally toss out any particles that are not in the correct energy range. So, this time, the optimizer will want to try to keep as many *on-energy* particles as possible, instead of just any random particles - thus, too much manual adjustment might not produce features that no one's thought of (the purpose of this project). Also - a shorter muon front-end (the part after the chicane) might not result in enough acceleratiion, although more particles will reach the end. I can't give any advice for the new parts of the optimizer, though (I can't find any detailed technical explainations). Good luck on those. |
Maniacken [US-Distributed] 2004-06-09 21:27:01 | aysz88 i think i will read the technical paper tomorrow as i graduated from college earlier tonight. pollock i am only running tests on one of my computers. one of my computers is running trial and error tests, and the other two computers are on distributed folding. i wish i could see what my computer is doing but since 4.41b i keep getting direct draw error on all my computers. DirectDraw error 80070057 @ makebank - creating shad |
AySz88 2004-06-09 22:09:23 | quote: ... congratulations! |
kitsura 2004-06-10 07:58:23 | I believe that given enough time and enough processing power (NEC Earth Simulator anyone?) the client is capable of evolving on its own to produce a very efficient design. That's why I think that the computed baseline design will definitely be surpassed sooner or later. |
Pollock[Romulus2] 2004-06-10 13:33:11 | quote: That is great to hear. I'll second the congrats. |
Pollock[Romulus2] 2004-06-10 14:20:43 | quote: Odd timing! Before I had read your post, an idea popped into my head about the tantalumrodr=000; value. It will grow higher, but it requires a major overhaul of the solenoid channel. The current run is at tantalumrodr=075; and 1.547448 (475.7 Mpts). The next batch of experiments should raise it higher. It should eventually surpass the original breed at 1.589764 (2385.3 Mpts). The 1.547448 is a radical redesign of the entire result from the first to last parameter and should grow much higher after some fine-tuning. These mods work easily with phaserot, but applying them to the chicane will be a massive undertaking. In a few small trials, they seemed to work, but the major changes will need to wait until I have some extra time. I agree with the assessment of the chicane and *on-energy* particle. I also believe that if a huge bunch of *on-energy* particles enters the chicane at the same time, many 'good' *on-energy* particles are forced out due to the huge volume. Sort of like a traffic jam when one lane is blocked. The future 'tweaking' involves speeding up the particles but allowing them to spread out, rather than my original philosophy of slowing them down. I also have some future ideas about the tantalumrodz=403; value. If the value for sil= were changed, that would require an adjustment of tantalumrodz=403; to compensate. It may not work, but is worth trying. Current values: s1l=822 tantalumrodz=403; Trials: s1l=900,700,etc., possibly even 000 tantalumrodz=??? If anybody wants to try it, feel free to do so. No guarantees it will work. It will be quite a while until I can try it. Those values should be dependent on each other, but the sif= value will likely not change. |
AySz88 2004-06-10 20:06:00 | Is there any utility that creates test lines (to ensure that every parameter was set)? |
[DPC]Stephan202 2004-06-10 22:26:11 | Never seen that one. It wouldn't work either/be useful because not every test has to contain every parameter. At least, that's what I vaguely remember. Have no time now to sort it out. |
AySz88 2004-06-11 05:38:08 | Actually, you do need to set every parameter (or all the omitted parameters become random - too much noise for an efficient test) BTW, Pollock - it's s1l=, but you typed sil a couple of times (I hope you didn't use "sil" when you were testing...) |
Herb[Romulus2] 2004-06-11 08:19:15 | Quote "Is there any utility that creates test lines" Not a utility, but with some smart usage of Excel and a hex-editor I set up numberous testlines in a few minutes keeps da box busy when I'm sleeping The hex-editor is needed because we exceed the 256 columns with parameters, so results have to be flipped vertically and reverse. The largest queue.txt had 36 values to test so far in one go. I've added up the queues during runtime already without problem. You can also alter results.dat at any time. Just ensure the client will not write any time soon to these files. With seeded values, there's always a good timespan left. Altering the results.dat makes sense, when the client will go into normal mode any time later. |
AySz88 2004-06-12 13:14:34 | Well, does anyone know the current ranges of each parameter? (Apparently, they've changed.) It would be nice to have something to convert "real" values (i.e. distances in meters, etc.) into three-digit numbers. [edit] Stephen's reply: [17:25:57] scrutney_mallard: that's defined in the lattice file [17:26:01] AySz88: Oh. [17:26:17] scrutney_mallard: but it's kind of complicated since the ranges vary a bit [17:32:38] scrutney_mallard: most of the solenoids are -4 to 4 Tesla in field, 0.2 to 0.6 in length 0.1 to 0.4 in radius and the drifts are 0.5 to 1 [17:32:52] scrutney_mallard: Oh, actually in the new one the lengths go 0.2 to 0.9 so you'll need to rescale [17:32:55] scrutney_mallard: but the rest are linear scaled [18:03:57] AySz88: What does the "phases" mean? (the blue/orange)? [18:04:30] scrutney_mallard: whether it is accelerating phase or decelerating phase at the time [18:04:35] scrutney_mallard: orange I think accelerates [18:04:47] AySz88: So orange = good? [18:04:55] AySz88: or a combination? [18:06:31] scrutney_mallard: it wants the energies within a certain range so in the case of chicanelinac you need mostly orange but with some blue [18:06:37] scrutney_mallard: phaserot needs an equal amount [18:07:15] scrutney_mallard: but it's not quite that simple as whether it hits on the rising or falling edge determines whether the beam is focussed or stretched longitudinally ... He went offline at this point, though, so I'm not sure exactly how streteched it is wanted. I do remember something like the beam needs to fit within a __ MHz curve. |
Pollock[Romulus2] 2004-06-12 21:59:44 | Sorry AySz88, I only wish that I had some clue to what all of that means. Maybe it will make sense after it sinks in for a while. Until that time, I'll leave the accelerator settings to you. |
Stephen Brooks 2004-06-13 06:19:20 | Isn't there a -list switch of Muon1 that will print out the component parameters from a given genome (I think the last one in results.txt or something?). Use the -? switch to get help on the other commandline switches... |
AySz88 2004-06-13 12:20:19 | Are you talking about 'muon1 -l' and 'muon1 -L'? The program crashes if I attempt those. |
Pollock[Romulus2] 2004-06-13 13:13:28 | No, they only work in command-line. muon1 -c -l Besure to leave a space between each switch. This is the list: Muon1 Commandline Switches: -b Background mode -c Console output mode -g Graphics mode, show every timestep -l Save component list from last recorded result, -L also shows coords -q Complete what is in queue and quit -r Recheck last result in results.txt once, -R forever, -r123 times -s Send results and quit -scr Screensaver mode -/123 Divide number of particles by 123 -series_loop=5 Do 5 steps in series for each particle (may be faster) -cfg:net -cfg:nonet Apply pre-defined configuration sets -init Halt after initialisation phases (at commandline, for debugging) -once Do one run and quit -outerloop Outer-loop debugging -parse Lattice file parse debug output -rfsync Print RF cavity synchronisation times to rfsync.log -version Print version identifier -view Lattice viewing mode (paused, no simulation) |
AySz88 2004-06-13 22:26:24 | That crashes for me, too... C:\Documents and Settings\*snip*\>muon1 -c -L Loading genomes... 1, done. Making new genome, TrialType=SetFromFile... Done Interpreting lattice file 'ChicaneLinacB'... Done Beamline consists of 404 units Adding components to simulation space Building proximity grid 6x2x50 (600 cells)... Done Tracking central particle to synchronise RF phases... 23.0% *crash* muon1.exe has encountered a problem.... Current contents of results.dat: tantalumrodr=000;tantalumrodz=500;d10l=000;d11l=000;d12l=000;d13l=000;d14l=000;d15l=000;d16l=000 ...and queue.txt: d10l=000;d11l=000;d12l=000;d13l=000;d14l=000;d15l=000;d16l=000;d17l=000;d18l=000;d19l=000;d1l=00 |
Stephen Brooks 2004-06-14 01:23:01 | It needs to have a result in results.txt... |
AySz88 2004-06-14 09:28:22 | It has several results there.... should there only be one? (I don't have the program here to test right now...) |
Pollock[Romulus2] 2004-06-14 17:38:13 | Problem solved. Use a lower case -l when using the switch. I tried with an upper case -L and the result was a duplicate of your crash. muon1 -c -l Then look in your muon directory for a file named listcomponents.txt for the output. |
AySz88 2004-06-14 19:49:02 | Yep - found the same thing but forgot to edit... Here's everything relevant to the 'original' best optimization (without the linac at the end), such that people can tack it onto their own tests. No guarantees, though; I probably screwed up *somewhere*. (Don't forget to remove the line breaks) I'll start running a couple of tests on my own tomorrow. d1l=000;d2l=000;d3l=000;d4l=000;d5l=000;d6l=000;d7l=000;d8l=000;d9l=000;d10l=000;d11l=000;d12l=0 00;d13l=000;d14l=000;d15l=000;d16l=000;d17l=000;d18l=000;d19l=000;d20l=000;d21l=000;d22l=000;d23 l=000;d24l=000;d25l=000;d26l=000;d27l=000;d28l=000;d29l=000;d30l=000;d31l=000; s1f=999;s1l=779;s2f=999;s2l=697;s2r=523;s3f=808;s3l=462;s3r=976;s4f=942;s4l=375;s4r=966;s5f=999; s5l=375;s5r=956;s6f=999;s6l=511;s6r=946;s7f=999;s7l=375;s7r=936;s8f=999;s8l=375;s8r=926;s9f=999; s9l=375;s9r=916;s10f=999;s10l=375;s10r=906;s11f=999;s11l=375;s11r=896;s12f=999;s12l=375;s12r=886 ;s13f=999;s13l=375;s13r=876;s14f=999;s14l=375;s14r=866;s15f=999;s15l=375;s15r=856;s16f=999;s16l= 375;s16r=846;s17f=999;s17l=375;s17r=836;s18f=999;s18l=375;s18r=826;s19f=999;s19l=375;s19r=816;s2 0f=999;s20l=375;s20r=806;s21f=999;s21l=375;s21r=796;s22f=999;s22l=375;s22r=786;s23f=999;s23l=375 ;s23r=776;s24f=999;s24l=375;s24r=766;s25f=999;s25l=375;s25r=756;s26f=999;s26l=375;s26r=746;s27f= 999;s27l=375;s27r=736;s28f=999;s28l=375;s28r=726;s29f=999;s29l=375;s29r=716;s30f=999;s30l=375;s3 0r=706;s31f=999;s31l=209;s31r=696;s32f=833;s32l=378;s32r=686;tantalumrodr=187;tantalumrodz=438; |
AySz88 2004-08-24 09:43:22 | Congrats all. Chicane is now better than baseline, with a result of 1.360418 by [DPC]Meloentje, who currently has 2328 results. |
Nexus[Free-DC.org] 2004-08-24 16:18:31 | Congrats all. Now lets see how high we can push this sucker. |
Stephen Brooks 2004-08-26 06:42:51 | Heh. It's doing this with 2/3rds the amount of linac equipment as the original design too. That should save some £££ |