![]() If you are going to run CPU and GPU under load at the same time choose >= 750W. Choose a Corsair, Seasonic, BeQuiet, CoolerMaster and such brands with 80PLUS GOLD efficiency rating at >= 500W if you are going to run only CPU tasks. ![]() Pretty much any HDD/SSD >= 120GB for OS install, BOINC and its data directory. ![]() Good value for money right now would be something like:ĪMD Ry+ and OC (or 2700X if you do not want to apply OC yourself)Īsus ROG Strix B450-F Gaming (Decent VRM for this price class)ġ6GB RAM CL15-16 3000Mhz are pretty cheap (2x8GB for Dual Channel preferred) So either the motherboards of choice must have something like BIOS-flashback to be able to flash a new BIOS without CPU installed or it must come with a newer BIOS at delivery, or you must have a 2000-series CPU that is compatible with the board at an older BIOS version to then be able to flash all the boards you buy to a newer version and then swap the CPU for the 3000 series one. however, it is possible to install a 3000-series CPU in an x470 motherboard (cheaper), but you need to have a correct newer BIOS version for it to boot with those newer CPUs. 3700X is more expensive, but has higher performance per core and the new motherboards are more expensive due to PCIe 4.0 implementation in the x570 boards. 2700 is of previous generation but very nicely priced right now and has cheaper motherboards. I would probably go for AMD Ryor 3700X (both 8C/16T). If space and such is not of a concern, then you might want to go for sweet-spot price/performance per machine and many machines instead. Also, get a good motherboard with decent VRMs so that they wont wear out fast due to high temperature. CEP2 project that ran on WCG earlier had huge RAM requirements for instance. The more threads your CPU will run, the more RAM you need. Beefier cooling will also be required to tame such beasts att 100% load and perhaps some OC. I would go for AMD ThreadRipper systems running for instance AMD TR 2950X (16C/32T), 2970WX(24C/48T) and 2990WX(32C/64T). If you go for few machines but more powerful ones, then you probably will pay a price premium for the CPU to get a lot of cores (you will spend beyond the sweet-spot for price/performance ratio) but instead fewer machines will take up less space and you will need fewer motherboards, PSUs, cases etc. for our understanding we had to extend the backend in a massive way to bring our jobs working => to configure the jobs in BOINC we had to write a lot of new automating codeī.How large of a farm did you have in mind? Client->Server bidirectional communication needed – no chance to bring it online through a NAT (VM Server)Ī. For single functions it looks great, but for the complete run like we have in Jenkins => Deploy and collect results looks complexī. We already investigated ideas how to solve this problem:Ī. And main problem: Jenkins will start the execution of parallel pipeline part-jobs only after finishing adding all to the queue. When the queue contains multiple hundred jobs adding new jobs took much longer – will become even worse by increasing queue. ![]() The limitation in Jenkins is currently with massive parallel jobs ( ). We use the Jenkins Label Expression to lets Jenkins choose which job can be run on which node. We used the parallel steps in pipeline and lets Jenkins queue and execute the application. It feels like we have the same problem like any video rendering farm – each picture of a video should be calculated independently and parallel – but it is not rendering.Ĭurrently we tried to execute it with Jenkins and Pipeline jobs. One run of this application can take up to multiple hours. We need to run a windows application (single instance only working) 1000 times with different input parameters. We are currently struggling with the following task.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |