00:23.48 | *** join/#openjtag ka6sox (n=ka6sox@nslu2-linux/ka6sox) |
01:29.47 | *** join/#openjtag ByronT (n=byron-po@nslu2-linux/ByronT) |
01:34.44 | [g2] | lennert I think you'll want a Black Dog |
01:37.06 | ka6sox-office | I saw them at LWE |
01:54.33 | [g2] | ka6sox-office what did you thing ? |
01:54.35 | [g2] | think |
01:58.26 | *** join/#openjtag bullet (n=bullet@166.219.203.62.cust.bluewin.ch) |
02:13.43 | ka6sox-office | [g2], I liked it..a friend of mine has one. |
02:25.36 | [g2] | it's got an FPGA with the 405 hardcore no ? |
02:26.09 | ka6sox-office | [g2], yes it does. |
02:27.15 | [g2] | ka6sox-office do you know what they are doing for the USB controller, is it a soft contorller on the FPGA ? |
02:27.37 | [g2] | contorller :) heh |
02:30.59 | ka6sox-office | I don't know yet..but I think that the schematics are available. |
02:54.43 | [g2] | mine should arrive next week sometime |
03:14.51 | dyoung | arf |
03:44.48 | *** join/#openjtag ChanServ (ChanServ@services.) |
03:44.48 | *** mode/#openjtag [+o ChanServ] by irc.freenode.net |
05:16.29 | *** join/#openjtag ka6sox (n=ka6sox@nslu2-linux/ka6sox) |
09:27.42 | *** join/#openjtag odoc (i=odoc@is.borntobooze.de) |
10:23.59 | *** join/#openjtag vmaster_ (i=vmaster@p549B69C5.dip.t-dialin.net) |
15:19.47 | key2 | anyone knows how If there is a way to get 128GB of RAM for a NFS on a computer ? is there any appliance doing that ? |
15:27.40 | [g2] | key2 I think there are some big servers that support lots of memory |
15:28.24 | [g2] | there are the 36-bit extensions |
15:28.37 | lennert | intel x86 is 64G max |
15:28.45 | [g2] | which only 16*4 or 64G |
15:28.49 | vmaster | opteron server could possibly use that much |
15:29.12 | lennert | you'll need 64 DIMM slots |
15:29.24 | lennert | you'll probably get 7 other CPU sockets with it |
15:29.45 | lennert | 'for free' |
15:30.16 | [g2] | right, I don't understand why you'd put that on 1 box unless it has 10G |
15:30.41 | [g2] | 1G top out under 128MB/sec |
15:31.28 | [g2] | and my nfore4 Ultra would probably serve that up with a good PCI-E card |
15:31.43 | [g2] | it does 600Mbs+ with a $19 card |
15:31.56 | lennert | with 10G it'll still take 100 seconds at 10G line rate to fill your DRAM |
15:32.14 | [g2] | nod |
15:32.52 | [g2] | so unless it's one big DB file that all transations are using, it could probably be split across boxes |
15:33.18 | [g2] | but I think there are some Opteron boxes out there |
15:33.44 | lennert | 128G is useful to keep your entire database/fileset in ram, for low latency access |
15:33.45 | [g2] | or lennert can whip you up an FPGA that has 4 PPCs on it and supports 128G memory :) |
15:34.13 | lennert | [g2]: the problem here is pin count :) |
15:34.35 | lennert | and drive strength |
15:34.39 | [g2] | they make some pretty big FPGA |
15:34.52 | lennert | yup |
15:34.54 | [g2] | I'm sure you'd be cascading some of the memory |
15:35.05 | lennert | nallatech's boards are only $30k :) |
15:35.51 | [g2] | well what's a 1G of 400Mhz DDR go for these days ? One would probably want ECC |
15:36.12 | [g2] | like for nallatech ? |
15:37.16 | [g2] | I'm at nallatech.com |
15:37.48 | lennert | i think the BenBLUE-V4 was $30k or so |
15:38.04 | lennert | i have their price list somewhere.. nothing is under $10k, pretty much |
15:38.40 | [g2] | I'm surprised they don't have PCI-E |
15:38.52 | key2 | vmaster: and with a FPGA, it wouldnt be possible to make a board with 128 RAM slot and make an RAMDISK NFS of 128GB ? |
15:39.17 | lennert | [g2]: 1gb pc3200 ddr400 400mhz - $61 |
15:39.51 | key2 | so it's fine |
15:39.54 | lennert | [g2]: hmm, i thought they had pci-e.. |
15:41.38 | [g2] | ok so $61*128= 8K |
15:42.46 | vmaster | key2: why would you want to spend that much money just to have a nfs share on a ramdisk? |
15:43.04 | [g2] | lennert I think you'd have an extra 14 cores :) |
15:43.28 | [g2] | http://extended64.com/Article30.x64 |
15:44.12 | [g2] | http://www.theinquirer.net/?article=23518 |
15:44.17 | [g2] | that's the one I remember |
15:44.48 | lennert | [g2]: looks like a microsoft astroturf site |
15:44.57 | lennert | [g2]: "see we have a community too" |
15:45.10 | vmaster | http://h18004.www1.hp.com/products/servers/proliantdl585/ |
15:45.22 | [g2] | lennert I just meant the iwill board |
15:45.35 | lennert | oh |
15:46.39 | lennert | 128GB at 266mhz, or 48GB at 333mhz, or 32GB at 400mhz |
15:48.34 | [g2] | well it's not like the 266mhz would matter on NFS unless you've got 10G pipes |
15:48.46 | [g2] | in which case you've got other issues |
15:48.59 | lennert | [g2]: key2 didn't say yet what he wants to do with it :) |
17:50.12 | key2 | sorry |
17:50.13 | key2 | back |
17:51.37 | key2 | It was basically for a Video On Demand service, a hard disk would be too low for sending 300 streams to 300 different people |
17:52.30 | lennert | how about 16 machines with 8G each? |
17:53.17 | key2 | do you know how much that costs ? |
17:53.35 | vmaster | i'd bet it's a lot less than one machine with 128G |
17:53.56 | vmaster | how much bandwidth does one stream take? |
17:53.58 | key2 | vmaster: I need 200 of those machine |
17:54.03 | key2 | 3Mb |
17:54.16 | vmaster | as in megabyte? or millionbit? |
17:54.24 | ka6sox | 3Megabits/sec |
17:54.25 | key2 | bit |
17:57.51 | vmaster | shouldn't a disk array be fast enough to handle this? i mean, even the most expensive disks are going to be a lot less than the memory you're thinking of |
18:00.17 | key2 | like how many disks |
18:00.46 | ka6sox | vmaster, not really |
18:00.46 | ka6sox | if everyone was watching the SAME content yes.... |
18:00.55 | ka6sox | and if they started at the *same* time. |
18:01.08 | key2 | no |
18:01.13 | key2 | they don't start at the same time |
18:01.21 | ka6sox | having to have the physical heads move all over the disk is the problem. |
18:01.26 | vmaster | hmm, yeah |
18:01.30 | key2 | basically |
18:02.17 | key2 | I wanna have at least 400 movies |
18:02.42 | key2 | and all should be watched at the same time |
18:02.44 | key2 | how can I do |
18:04.41 | ka6sox | get a Z900 mainframe and go from there. |
18:04.56 | ka6sox | they are optimized for bandwidth. |
18:06.01 | key2 | Z900 |
18:06.02 | key2 | ? |
18:06.21 | ka6sox | ibm System 390 Z mainframe computer |
18:07.32 | key2 | it would cost more than a single board I would made with 128Gb of ram |
18:07.38 | key2 | and few fpga |
18:07.41 | key2 | wouldn't it ? |
18:07.57 | key2 | -made + make |
18:08.20 | ka6sox | one BIG ram disk eh? |
18:08.34 | key2 | well today 1Gb of ram doesn't cost shit |
18:09.04 | key2 | and since I need a lot of those device (+200) and need them to be quiet small (2U) |
18:09.13 | key2 | i would think about makin my own |
18:09.56 | key2 | unless there is a better solution cheaper |
18:10.28 | ka6sox | you are streaming to settop boxes? |
18:10.46 | key2 | yeah |
18:10.53 | key2 | and I can't put them all in a datacenter |
18:10.59 | key2 | it would make too much of trafic load |
18:11.05 | key2 | i better put one or two on each DSLAM |
18:12.12 | ka6sox | yep |
18:12.22 | ka6sox | at 3Mb/sec you would have to. |
18:12.45 | key2 | I mean our backbone is made with Gb optic fiber |
18:13.13 | key2 | but sending all those video from one point through the whole network is stupid |
18:13.16 | key2 | it would slow it down |
18:13.43 | key2 | it's better to have those very big in each telco center and send once for good the movie when it's needed |
18:15.53 | lennert | key2: what company you working for? |
18:16.05 | key2 | i'm in france |
18:16.09 | key2 | so probably don't know it :) |
18:16.13 | key2 | it's called free |
18:16.23 | key2 | we do 24Mb dsl for 29.9Eur/month |
18:16.56 | lennert | ah, i think i've heard of them |
18:17.32 | *** join/#openjtag bullet (n=bullet@85.57.77.83.cust.bluewin.ch) |
18:21.10 | key2 | now I have to figure out how much a virtex would cost :) |
18:21.37 | key2 | and If I have to put them all on the same bus or make different buses to speed up the thing |
18:23.31 | lennert | a virtex is useless for you |
18:23.46 | lennert | fpgas won't help a bit for what you're trying to do |
18:25.42 | lennert | you need i/o bandwidth and lots of spindles |
18:26.03 | ka6sox | or lots of RAM disks to eliminate latency |
18:26.37 | lennert | lots of RAM is impractical if you ever want to reboot |
18:26.50 | lennert | if you have 128G of movies in-memory, and you want to reboot.. it'll take you an hour to get everything back in RAM |
18:27.08 | ka6sox | thats why you only have 2 movies in any one box. |
18:27.09 | lennert | (assuming you can read them back in at 35 meg/second) |
18:27.13 | ka6sox | 10GB of ram/box |
18:27.18 | lennert | yeah |
18:27.29 | lennert | either that or a few big boxes with lots of disk |
18:27.55 | ka6sox | ram disks will be more expensive but faster and very low latency |
18:28.04 | lennert | if you're seek-bound (and you will be, for a certain number of clients), it's better to have 1000 9G disks than 30 300G disks |
18:32.24 | key2 | lennert |
18:32.59 | key2 | let say I have only 8 movies of 800MB each |
18:33.09 | key2 | and 5000 people are watching each of them |
18:33.14 | key2 | how do I have to deal |
18:33.23 | key2 | do you think the Head of the hard disk can go that fast ? |
18:33.56 | key2 | and on only 2U |
18:34.03 | key2 | you won't be able to fins cheaper |
18:34.14 | key2 | it doesn't matter if it takes 1hour to load it back |
18:34.23 | key2 | it's not supposed to reboot every hour |
18:34.40 | ka6sox-away | sure..just use WinCE |
18:34.53 | key2 | wince ? |
18:35.02 | lennert | 128G of ram is $8000 while 128G disk is less than $60 |
18:35.36 | key2 | lennert: but you can't have 100 people watching the same thing on the $60 hard drive |
18:37.03 | key2 | so you can't deal with hard drive unless you have 1000 of them |
18:37.09 | key2 | which takes lot of room |
18:37.31 | key2 | you can't go in each DSLAM center and put one full cabinet of hard drive |
18:37.55 | key2 | plus the power |
18:38.04 | key2 | plus the temperature |
18:38.05 | key2 | ... |
18:38.09 | key2 | it's a pain in the ass |
18:38.12 | key2 | relaly |
18:38.50 | key2 | why do you think most companies that do Pay Per Vew do multicast streamin the same movie every 10min |
18:38.59 | vmaster | wouldn't you have to used buffered ram? |
18:39.12 | key2 | what u mean |
18:39.20 | vmaster | if you put 128G in one machine |
18:39.27 | key2 | yeah |
18:39.30 | key2 | what would be the pb ? |
18:39.42 | vmaster | buffered ram is even more expensive |
18:39.43 | key2 | it would be a big FPGA dealin with htose |
18:40.09 | key2 | buffered ? |
18:40.12 | key2 | for what reason |
18:40.12 | key2 | ? |
18:40.42 | key2 | a DDR should make it |
18:41.59 | vmaster | as far as i know there's a problem with line drive strength when having too many ram modules on a single bus |
18:42.08 | vmaster | that's why you use buffered memory |
18:42.13 | vmaster | with a buffer for every line |
18:42.36 | vmaster | reducing the capacity that needs to be driven |
18:43.40 | key2 | what do you call line drive strenght |
18:44.13 | key2 | that the power is not enough ? |
18:44.37 | vmaster | mhh, not power, but current |
18:44.57 | vmaster | [16:34] < lennert> [g2]: the problem here is pin count :) |
18:44.57 | vmaster | [16:34] < lennert> and drive strength |
18:45.02 | vmaster | lennert mentioned that already |
18:45.19 | key2 | well pincount shouldnt be a probleme |
18:45.30 | key2 | if they are all on the same bus and multiplexed |
18:45.49 | key2 | but drive strenght |
18:45.55 | key2 | dunno what would be the probleme |
18:46.19 | bullet | key2: check out servers. they use registered (buffered) dimms. |
18:47.13 | key2 | bullet: but why wouldnt it work |
18:47.22 | key2 | if you have a chip, you can chose it with the CE |
18:47.23 | vmaster | if you use buffered ram it could work |
18:47.39 | key2 | and what's the difference |
18:47.54 | key2 | if you put 4 of them or 100 of them |
18:48.06 | vmaster | the buffer - the memory controller doesn't have to drive as much current as it would have to |
18:48.12 | vmaster | without the buffers |
18:48.35 | bullet | http://en.wikipedia.org/wiki/Registered_memory |
18:49.49 | key2 | ok |
18:50.31 | key2 | so it means that when you have that amount of ram on the same time, all of them would use a certain amount of the current even if the CE is not enabled ? |
18:51.38 | key2 | so the controller would have to deliver more current to "talk" to all of them |
18:51.39 | key2 | ? |
18:52.00 | vmaster | i think so |
18:52.30 | key2 | well it doesn't cost that much to put a little buffer in front of them |
18:52.36 | key2 | just to bring the current hight |
18:52.38 | key2 | higher |
18:53.29 | lennert | the spartan3 can drive like 24 mA per pin before it goes pop |
18:53.49 | key2 | lennert: ok but let's take it like that |
18:54.08 | key2 | if you put a little buffer in front of the line for every 4 of them |
18:54.11 | key2 | then it's fine |
18:57.05 | key2 | but then I might get some timing issue |
19:02.09 | key2 | INPUT LEAKAGE CURRENT |
19:02.09 | key2 | Any input 0V = VIN = VDD, VREF pin 0V = VIN = 1.35V |
19:02.09 | key2 | (All other pins not under test = 0V) |
19:02.09 | key2 | Command/ |
19:02.09 | key2 | Address, RAS#, |
19:02.10 | key2 | CAS#, WE# |
19:02.12 | key2 | II |
19:02.14 | key2 | -36 36 |
19:02.18 | key2 | so 36uA |
19:02.49 | key2 | 24mA is a lot compared to the 36uA each input line uses |
20:28.21 | *** part/#openjtag odoc (i=odoc@is.borntobooze.de) |
20:28.34 | *** join/#openjtag odoc (i=odoc@is.borntobooze.de) |
21:05.07 | *** part/#openjtag odoc (i=odoc@is.borntobooze.de) |
21:42.58 | *** join/#openjtag ByronT-Away (n=byron-po@nslu2-linux/ByronT) |