I have used something like Nvidia Tesla P4 T4 Compact Fan Attachment | eBay, though if you have a 3d printer you can search and print that part yourself.
Purchased a few things (and even returned one) to Tugm. Even though everything is from over seas, both purchase/return and communication has always been top notch with them.
I would assume 100-130w, I have a 3400-24 port (no PoE) and it sits about 100w with 15 connections.
POE versions are normally a bit noisy though, so might wanna watch out for that or replace the fans. If you just need 12 1g ports you may be better off getting something different unless you want...
Future proof? Conduit my friend.
run conduit where your drops are and then you can easily change in the future. For me, I would go copper due to most consumer devices using it out of the box. Personal desk/office? Run both.
Looks like another listing, same pics, but slightly more
https://www.ebay.com/itm/256118305160?mkcid=16&mkevt=1&mkrid=711-127632-2357-0&ssspo=Io5_9CyfSTa&sssrc=4429486&ssuid=Io5_9CyfSTa&var=&widget_ver=artemis&media=COPY
** GONE **
Have some gear to offload to anyone local in Las Vegas, I won't ship because it's just too pricy to find proper boxes and ship 2U servers.
Dell r7910 (basically a dell r730) with 2x E5 2686 v3 and 32GB of ram w 2x10g 2x1g card, no drives. Not parting out.
What is your CPU usage on both sides during a long-ran test (where you get 13-15Gbps) - sounds more like the boxes are not able to send that amount of traffic vs cable/nic issue. Netdata is also a great app to install on linux to view potential bottlenecks in real time.
You have any NMS or snmp monitoring on the switch to see port/system statistics over time? Sounds very much like STP or an internal process crashing/restarting but I know some vendors log more/less then others to catch such a thing.
I would not worry too much about CPU speed, being a network lab the more RAM the better imo for such a workloads and there is little traffic to process.
Good number of cores and tons of ram is what I would target, once the devices boot they should use little processing.
Personally I like dell...
And that is exactly the type of Memory I used ;) 2400 Hynix. The 3200 stuff was almost double the price, but in the future will target that on this picky CPU/Mobo if I need to go past 256GB
On a side note, the 2400/2133 32GB sticks I had lying around overclock on this platform to 2933 w/o a...
Ah thank you, will stay away from LRDIMM's for now then. Its a lab box so really doubt I will need the capacity - they were just a good price ($80/each) :p
Does anyone have success or tried using LRDIMM's on this series of boards? I grabbed 4x 64GB quad ranked 2400 sticks and can only ever get one or two of them to show up on boot (boot takes forever with them in, 1-2min to start) while swapping back to RDIMM's 6x 32GB works like a champ...
Shopping around for some RAM on a new build I am throwing together. This is using an EPYC 7502(32c/64t) CPU. I understand RAM speeds affect the Infinity Fabric, and 3200 is the target for best performance, but just how much performance increase would we be talking about when comparing:
64GB...
I want to say IO shields generally don't work on 1u and are commonly built into the chassis. Depending on your budget you can find them for 700-800 on ebay (full unit). Sometimes cheaper for older generations of xeon-d lines.
ex: SuperMicro X11SDV-4C-TP8F 1U Half Depth x1 D-2123IT 2.5" 512GB...
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.