Jump to content


Sophist Member
  • Posts

  • Joined

  • Last visited

  • Speed Test

    My Results
  • Days Won


Everything posted by Blako

  1. Mud: I found a quadcore client to out perform 4 single core clients. Hyperthreaded cpus are not recommended unless there are more then 2 physical cores, else double the work gets done in half the time and science slows waiting for your results. I would recommend against running 5 threads on a 4 thread machine as collisions of "is cpu busy? Yes" would happen all the time, unless your already taking this into account. In fact on the hyperthreaded 4 physical 8 logical core i7s, seven cores at full load out perform eight cores at full load. TriRan: Only the smp client with its large memory usage and high system requirements has an early return bonus. The idea of "early bonus" came from bigadv units that required 20-100 megabyte uploads verses 1mb results file from your radeon. Points FAQ said: Vijay Pande said, donors really pick up in the winter months while folding machines are turned off in the hot summers.
  2. When version 7 launches on start-up it cant find previous work. If I use a shortcut to fahcontrol, its as if the program runs on a different user profile and it can find prior work given that the shortcut (fahcontrol) had a work unit before. For some reason it says time per frame: 2.9 seconds, point per day: half a million., instead of the actual TPF of 1 min 59 sec. Lorne! Nine jobs worth 12k points on Dec 17th? Brownie for you and that Fermi card me thinks your cooking. Cool I got a work unit designed by a graduate student: 7809 Roughly speaking every time you double your cores, you triple your (smp) points. So you guys can use arrow keys to move around the text box? For me it just moves the page. http://www.youtube.com/watch?v=XtGf0HaW7x4&feature=related Stanford uses cpus to great effect in problems where a series of instructions must follow each other in order. The gpus are used in problems that benefit from parallel independent instructions.
  3. Im having trouble getting V7 to continue after turning the computer back on. So far I think I lost 2 units, no joke cant find them, and turned in 1 as anonymous user. Any ways TriRan: Theory 1: Gpu waiting on Cpu The ATI GPU2 cores require cpu power to feed the video card, can you go to task manager and set "FAHcore_11.exe" > set affinity > "below normal priority" where core_a4 should already be at "idle" priority. The core_11.exe uses a lot of cpu time (reports range 5%-20%) and your already in that range mmm. Can you confirm with gpu-z that your running 3d clocks at 50% gpu load? Theory 2: Old core Your Radeon is working an old unit (core 11 for 3xxx series) instead of the new openCL (core 16) available by using flag of "client-type" "advanced" How to add flags in version 7: http://foldingforum....hp?f=67&t=18187 "April 2011 Warning: the beta tag is there for a reason, the core is not finished and may crash more often than usual or otherwise degrade system performance." Theory 3: Its working properly Check the point value of the current gpu unit (is it big?). Is the 5870 in the area of 8000ppd? Lorne: You need a very stable overclock. For games one minor visual glitch per hour is ok. For folding you need no errors for a month. Make sure several projects are successful before overclocking some more. Sources: Core 16 5870 Problems folding Stanford Points FAQ
  4. I don't get it and I'm not familiar with forum code. Where does that go? What is it wrapped in? Does it work out of signature? The forum doesn't allow straight html so whats required? My Sig is already full by the way
  5. Congrats to Lorne45 for completing his 10th smp workunit and now receiving early return bonuses. Lols, nice mudmanc4, mmm wait did you test your code? Is it just me or does the use of arrow keys deselect the text box for you all too? http://folding.extre...age.php?t=44352 So I found and installed the FaH version 7 beta. Installation took under 2 minutes. So it opens up and I click configure then I notice my computer fans are speeding up. The thing automatically found my quad core and video card, downloaded the required cores, downloaded a workunit for each, and started running! Holy Pizza Batman!
  6. Well "a lot less" no, your 5870? is producing 8000 ppd compared to your Phenom which can return 1000-5000 ppd depending on early return bonus. I would say that your Phenom could produce 15-35% of your theoretical max output. So while smp is less then gpu its definitely more then none. It appears that the points earned during December 11th server outage might be saved after all... I think it was Dec 12th that I wanted to show you a 1k point project with additional 4k early return bonus but that is when I saw Mud's post of "server down". Lets see how it pans out.
  7. O, I remember now, I'm using flags of: " -forceasm -verbosity 9 -local". Where "-verbosity 9" makes the console more "talkative". All's Good, your client does say Project 5733 compressed to 250kb. The link says 511 points. For only 1.5 hours thats sweet. I'm debating trying the FaH version 7. Some say its great, easy to install, and easy to understand. Others say the beta bugs are still being squashed. Cheers
  8. TriRan, what are those jobs worth being pumped out by your Radeon? Did you happen to see the resulting upload size? I'm just curious. Good Job for the team [11:28:07] + Attempting to send results [December 14 11:28:07 UTC] [11:28:07] - Reading file work/wuresults_03.dat from core [11:28:07] (Read 2508622 bytes from disk) [11:28:07] Gpu type=3 species=21. [11:28:07] Connecting to [11:28:28] Posted data. [11:28:28] Initial: 0000; - Uploaded at ~116 kB/s [11:28:28] - Averaged speed for that direction ~102 kB/s [11:28:28] + Results successfully sent [11:28:28] Thank you for your contribution to [email protected]
  9. Top left corner of http://folding.stanford.edu/ is the news. Yes the stat page/records are down but at least the servers to get and return work units are online.
  10. Wait a second! Are you running SMP or GPU? Depending on the game, the SMP client at 98% cpu load on idle priority hurts my frame rate on average by 0% - 10%. I could not tell the difference from SMP on or off in assassins creed (44 fps vs 43 fps). In contrast, running the GPU client cuts my game frame rate by 0% - 60%. Some games 60 fps when gpu folder on and 60 fps with gpu folder off. Well then theres minecraft with 100 fps when its off verses one frame per two seconds if running gpu folder. You could play with task manager core affinity to separate game/folder on the cpu. I dare you to gather some results with fraps or whatnot and produce something like this: Blake's Folding while Gaming results: SMP client cuts game frame rate by 0%-10% on avg (idle priority seems to wait its turn leaving my games uneffected) GPU client cuts game frame rate by 0%-60% on avg (studders and pauses noticed in some games)
  11. Superb man, the (common for me) 580-ish point projects were taking my Phenom x4 3.6 ghz about 12 hours My smp client yesterday (dec 9th) was assigned project "6098" worth 1593 points (before early return bonus) and its estimated to take 35-38 hours. I have only gotten one of these big projects before and it took 4 days of the 7 day deadline to complete. TriRan, you have a passKey to qualify for bonus points after your 10th completed smp project right? When I return jobs early i get 2-4 times the point value per job. If your cpu is faster and on more hours per day then my computer I expect you to surpass me. Hooah! Well, mudmanc4 the classic single core clients are designed for people that have old computers or don't have them on 24 hours a day. Thus the projects have 14-40 day deadlines and are worth little. The SMP, GPU, and the upcoming Big Advanced projects have short deadlines to promote quick turn around from results of "step 1" to assignment of "step 2". Please read: Folding at Home version 7 for windows (if your using windows)
  12. That video is talking about the wavelengths used by wireless devices crowding the available frequencies. You got: AM FM wireless networks blue-tooth CB-Radios Aviation communication Naval communication Wireless home phones Professional Wireless microphones Global positioning satellites AND remote controls All these devices are sharing a very limited range of frequencies available to be transmitted effectively in earth's atmosphere. http://electronics.howstuffworks.com/radio-spectrum1.htm NinjaGeek asked The costs to build a fast wireless infrastructure on a city wide scale are ridiculously expensive so there exists a point where providers would have to limit coverage, or speed, or raise prices. (can you afford a cell tower every 1-2 miles) http://mobileopportunity.blogspot.com/2011/06/truth-about-wireless-bandwidth-crisis.html Compare transmitting at the speed of sound, speed of electricity, and speed of light. (just an example not accurate velocities)
  13. Cheers Tommie, look around learn the market. Then if a sale comes along you will know if its an opportunity to buy. My Dad was looking at computer monitors for a few months then a Fry's Electronics anniversary sale advertised a lg e2260 at 70-60% of retail price. No more 50 pound, two foot deep monitor sitting on this desk, hoorah! Its your call deciding if this falls into the "must have" or "would be nice" categories. The new screens have to use less power then the projection heavy weight design. Who would scrap (harvest) your old tv?
  14. Reminder: glass skyscrapers have many attributes of a greenhouse. This would not be for residential homes with 5% of their exterior covered with windows. Homes (mostly) have more roof space, on which to put solar panels, then window space. Skyscrapers have more windows then roof. True, there would be more return placing them in Las Vegas with 210 clear days a year verses Billings, Montana with 89 clear days a year Dubai's wind turbine powered skyscraper
  15. EaglePC, Here is your last post which got replys. and on the left hand side of your profile page is your recent content: https://testmy.net/ipb/index.php?app=core&module=search&do=user_activity&mid=41354
  16. The old quad core client would launch 4 folders.exe while the new version has a single folder.exe with as many threads as you specify. Interesting Mud, my phenom at 3ghz is more then twice as fast as my Pentium 4 at 3ghz in single threaded apps. Can you confirm if you have 4 physical processors? (and not a pair of hyper-threaded single cores) Here is the benchmark client list: CPU: Pentium 4, 2.8ghz = 110 PointsPerDay SMP: 2 x Woodcrest 5140 dual cores = 1760 ppd SMP2: i5-750 = 6189 ppd SMP2 Big Adv = 9284 ppd? before early return bonus *requires 8 logical cores and 0.5GB of ram per core (recommend 8+ physical cores, .75GB ram per core) PS3: 900 ppd GPU2: ATI Radeon 3850 = 1500 ppd GPU3: GeForce GTX460 = 9787 ppd If heat, power usage, or processor load is a factor on a multi-core machine you can always use the classic uni-core client.
  17. Now that you know that testmy.net can detect problems where some other tests can't, read this: Also note the timewarner network status page in my sig. On the timewarner test site I get 31mb/s of my advertised 10mb/s While at testmy.net I get 8 of 10 mb/s. 80% efficiency, I can live with that.
  18. RTB, sounds like a jitter test. Although for a ping of 200ms it "only" runs 5 ping a second. On a voip test im getting 20% packet loss when transferring at high speed. hummm http://www.whichvoip.com/voip/speed_test/ppspeed.html
  19. A quad core with early return bonus far out does four classic clients. If you have a quad core at 1/4 usage then its only working in one thread as you stated. The smp client defaults to 4 cores but if you forgot the -smp flag then it becomes a single cpu client. If you downloaded the single core client then the -smp flag has no effect. Do you remember visiting the High Performance Client page? Look for the bold font " Windows: V6 Beta SMP2/CPU clients " For those special people who have a 16 thread server: (lol I can dream, 128 threads ) The smp client with the -smp flag will use 4/16 of power and also report 25% cpu usage. In that case change "-smp" to "-smp 15" A 10% overclock on the gpu gave a 10% increase in production. If left on for 24 hours which it can't, gtx460 at 10k point per day stock, bumped to 11k. The X4 965 if left on 24/7 returns 1k point per day without bonus and 5k ppd with early return bonus. I use something called Process Lasso, with probalance (autoTuning) turned off, where cpu folding is left unchanged at idle priority and gpu folding is set to below normal priority on startup. You can even set your work on core 1 and your play on core 2. PS. Where is the "view replys to your posts" button? I didn't know you posted a reply Make sure you have a passkey to qualify for bonus points.
  20. I was hoping for more info... sigh I got problems within my network. (very unsteady transfer speed) Link: VanBuren's Cablenut settings
  21. Wireless over distance also is quite susceptible to interference. Out in the world you got wireless phones, garage door openers, radiation both natural and man-made (microwaves), radio, and blue-tooth. All these frequencies including those emanated from a florescent light are flying through the air you breath. If your WiFi (a wave) gets affected by interference (another wave) the data can be corrupted and need to be resent. The walls of your home, the 2x4's, the insulation and wiring, can to some degree prevent your WiFi (wave) from reaching your antennae. Windows and mirrors even reflect the signal. If your data is lost and needs to be sent a second time it will take twice as long and testing sites will report half the speed. TriRan implied: If someone is using half the capacity you get half left.
  22. This will point you in the right direction for custom Cablenut file. http://www.j79zlr.com/cablenutXP2k.php
  23. If your internet provider is in this list then yes records have been kept. If your not in this list then you must be looking at the average speed of "unknown" providers. https://testmy.net/hoststats/
  24. I am also timing out on the third to the last hop of a testmy.net trace route! I know some servers deny a trace route for security reasons and yes its possible that this hop of the trace is being blocked, but I cannot reach 60% of my advertised 10mpbs connection on Google chrome and very rarly reach 80% of advertised on firefox.
  25. I am on a 10 mbit connection. If I start a download from a server thats a little busy and is only sending me data at 6mbit/sec then the other computers in the house have 4mb/s available. The faster your connection the harder it is for you to find a server that can fully utilize your capacity. If I get a server sending me data at 99% of my capacity the other users have 1% left. The TCP optimiser (which changes RWIN) has been explained in the following way: Consider you have to move a pile of dirt from one side of the yard to the other. If you fill your wheelbarrow to the brim you will make less trips but if you loose control and spill its going to be a big mess. On the other hand you can make many small loads where any accident is going to be relatively small. RWIN is the amount of data in bytes that can be downloaded withOUT confirmation of arrival at your end. By increasing RWIN you receive more data before saying "I got that delivery, send another one". If your on a 30mb/s line then open TCP optimizer move the slider to 30mb, select the network adapter your currently using (wireless/NIC), then at the bottom select optimal and apply. Mind you that if your provider is have problems then all the tuning in the world won't fix their messed up network. What company are getting your internet from? speedtest.net will tell you your best case while testmy.net will tell you when there is a problem.
  • Create New...