Nitro and others, What do you think of these: http://radio.weblogs.com/0105910/2004/08/30.html http://www.orionmulti.com/ Are you aware of any alternatives? I'd like to build a 10+ node cluster with any modern x86 processor, only 256 to 512 mb per node, no physical HD per node, boot off the network. Power dissapation might not be a problem with winter coming up. Maybe gang together some miniITX MBs with a central power supply and cooling like this: http://www.mini-itx.com/projects/cluster/ Any suggestions?
Prophet, I took a quick look at the Orion DT12 spec sheet. The only tangeable software feature seems to be:"It comes with cluster software development tools installed, including libraries and a parallel compiler that allows you to spread one multiple-file compile to all the nodes in the system." Truly great! Many linux users do this routinely on their home linux computer-farms (clusters if you like). In fact, I do this everytime I compile using the standard distcc and distccmon modules (samba.org). Very impressive! http://www.gentoo.org/doc/en/distcc.xml nononsense
I thought the DT-12 specs were quite good for $10K. It may actually use less power than my dual Opteron 242.
Perhaps, but for the $10K cost of a single DT12 I can put together quite a cluster-farm for doing the things much better the DT12 spec sheet claims to be doing. There are possibly other benefits, but they are not visible on the Orion sheet.
Hi nononsense, Putting asside power consumption, what components do you recommend for maximum performance/cost? Is there a web site that provides good recommendations? I don't mind if the CPUs are 1, 2 or 3 GHz as long as the overall performance/cost is maximized. Thanks.
Almost $1700 for the mobo! NOT Too bad.. and memory going for about $500 per stick, x8 sticks....Plus Opterons.. R U paying cash?
Hi Prophet, My answer may not be what you are looking for as each one's approach to computer trading is likely to be highly different. As I explained, since the P4 era I gave up on my earlier frustrating programming efforts for SMP machines. I currently run two P4's: 2.4 and 2.8GHz. I also keep an older 2*PIII mainly to test out linux communication things. As you see, nothing very planned, simply what the past left me. The speed of the P4's is very adequate for what I am doing, except fot lengthy linux compiles (KDE, Gnome or Xorg), but luckily you don't do this every day! This is when the distcc clustering comes in very handy. As motherboards I only used Intel over the years because they let me down only once - and this may have been my fault. This does not mean that I had no gripes, but these were things I managed to solve usually through bios updates. I have now 850 and 875 chipsets. I almost forgot, I have laptop running continuously doing all market data acquisition and recording. This data is broadcast on my LAN to the other users. At night I will transfer this data to the main db's for historical keeping. This setup allows me to do anything I want on my fast machines - I have to stop trading in this case. So for me no split-nanosecond headaches although I most certainly would fit into what popular wisdom calls a scalper. I don't mean that my MB choice is the "best" one. I followed many stories and talked to a lot of people. I stayed with what worked for me since the first 66MHz board. Disk choice, the same story. Till about 3 years ago, I only used scsi. Everything is now ATA100. All drives 200GB, 7200rpm. No SATA yet. It doesn't really bring you anything, except for the troubles of the past. I will use it when NCQ will be supported from all sides. (My current "#hdparm -tT" performance hovers between 55 and 60 Mb/s). I have a keen eye on clustering but don't require it right now. I can less or more guess what nitro is trying to do from his posts long time ago. My approach is different. Clustering seems to be the way how state of the art engines are now making their marks. One day this will filter down, it is probably going already. Of course, the whole linux/windows thing is quite a pain in the neck. You simply can't always escape the windows grip 100% - time will solve this. For me I'll stick with linux to avoid being trapped as in GG's saying:"You are walking around blind without a cane, Pall". So Prophet, my answer is certainly not in reply to your wish having best cost/performance. For me, I think I did my best. Be good, nononsense
You supply the knowledge: a programmer is not a trader. Highly skilled programmers and systems designers that are familiar with financial systems at the highest levels are available by the dozen. Even the most esoteric fianancial engineering applications are now commoditized. If you would like to hire dozens then just get out of your chair and make a few phone calls. These people are cheap and plentiful. .. And I hope you dont actually think that programming and trading are one in the same job: they are not. If you think that they are then you are designing a recipe for disaster within your business.