I'm always plugging these, but a Dell T5810 can be kitted out with 64GB of ECC ram and a 14 core Xeon E5-2697 V3 for about $350-ish for the entire rig.
For another $80 or so, you can step it up to a Dell T7810 with twice as much ECC ram and dual sockets, for 28 cores total.
I've tried running ABEC on AWS, and running it locally is much nicer. It really seems to like a LOT of cores and a LOT of ram. I have some 8 core and 6 core Intel I7s and I9s here, and my nearly ten year old T5810s outperform them on ABEC.
Surprisingly quiet too! I have one in my living room tucked behind a couch and you can't even hear it at all.
There are newer options from AMD and Intel, but the resale market of AMD is fairly high, and Xeons have barely been improving during the last ten years. (This is why Intel stock price has gone up 20% since 2014 while Apple is up 700%.) So even a 10 year old Xeon is surprisingly capable, and they're so cheap you can buy one for little more than the cost of lunch. I think I paid $15.
I can't resist and must share something (not mine). It's a preliminary measurement, obviously not anechoic and on-axis only, but anyway:
You wouldn't have guessed: it's B&C DE250 with ATH280EX-MK2.
We can make bets whether it's possible or not.
You wouldn't have guessed: it's B&C DE250 with ATH280EX-MK2.
We can make bets whether it's possible or not.
Last edited:
Hi All - are any of these 3D printed horns usable as is? I know we are mostly testing the performance of the geometry, but is the idea to ultimately build the horn from more solid material (I.e., wood, concrete, epoxy filled, etc.). I’m just curious if they can be used in their 3D printed state at loud enough volumes (- maybe with high enough infill density?
I got one on your recommendation. Made a huge difference. That's why I was able to simulate that OSSE inside an enclosure. Before that I could only simulate free standing in circsym or lower frequencies at low resolution in an enclosure. I haven't used it very much yet but the ability to quickly iterate is instructive with regard to how enclosures work with or against OSSEs.
Yes. Works very well as far as I'm concerned. You could use 3D printed forever or as a temp while you build one out of a material you want for the sake of cosmetics.
BTW, with a bit different mouth shape, I was able to reduce the waveguide diameter from 520 to 460 mm while keeping the performance virtually the same down to ~600 Hz:
(light gray = A520G2, color=A460G2)
It may not seem as much but those 6 cm are a big difference visually.
(light gray = A520G2, color=A460G2)
It may not seem as much but those 6 cm are a big difference visually.