I see a lot of people writing about how expensive the hardware to run these local models is - but see no mentions of the Intel Arc Pro B50/B60/B70 which seem like decent value if you're not interested in Apple kit (as much as anything can be decent value in the current status quo).
I just got a B70 with 32GB RAM for the equivalent of $1200 (incl. sales tax and import duties to my non-US location, so presumably it could be cheaper elsewhere). The memory bandwidth is 608 GB/s. For M5 Max (32-core GPU) it's 460 GB/s and for M5 Max (40-core GPU) it's 614 GB/s. A 3090 is still faster at ~900 GB/s but you're getting 32GB VRAM for a lot less than equivalent Nvidia cards. It's about 1/3 the bandwidth of a 5090 for 1/3 the cost, but with the same 32GB VRAM. If you're interested in being able to run bigger quants with some context and stay on a lower budget then it's an appealing trade off.
I'm still exploring using these local models so don't want to spend the equivalent of $5 000 - $10 000 just to test it out. I don't mind slightly slower perf to do some experimentation more affordably.
I actually got an B50 16GB (with meager 70w TDP!) first to test an Intel card with my stack - it worked easily with Ubuntu & Vulkan. I'd read a lot about hassles and people writing them off as unusable but it seems like these are often with SYCL which doesn't even seem to outperform vulkan and so why bother? (The B50 was just $370 inclusive tax and duties). Literally `apt install` the vulkan libraries and it worked with default xe driver in 26.04 and the vulkan build of llama.cpp. The SR-IOV PF/VF also just works with qemu/kvm, no tricks required. Since I got it fwupdmgr has updated the firmware twice so Intel is presumably actually trying to support these products.
I saw another article - can't find it now - with statements to the effect that transmission of Andes strain hantavirus is typically only between very close contacts, like amongst partners. But already the original couple who traveled in the endemic area of the Andes virus prior to boarding the ship have apparently spread it to another couple on the ship, the ship's doctor, 2 crew members and additionally the person who has tested as positive in Switzerland.
I'm the developer of the ExchangeRate-API.com service.
Obviously it's upsetting to have our API used by a scammer, but our service couldn't have been involved in this hack beyond fetching a JSON-formatted response of up-to-date exchange rates because that's the only functionality our service/domain provides.
My guess is that the scammer implemented a call to our API to fetch up-to-date exchange rates in order to make their fake wallet seem more plausible & real. Interestingly my API doesn't even support any exchange rates involving cryptocurrencies and so the scammer would have had to additionally integrate with a different API to get something like the up-to-date exchange rate between BTC and USD.
The API is a very simple service - it's just a few endpoints that supply JSON formatted exchange rates over HTTPS. Anyone with an email address can sign up to use the service for free and there are even some totally "open access" endpoints that don't require any authentication. One of these has been used in the GNU `units` converter software for a while.
With regard to proving it's a legitimate service, this is the point where I wish I had made more progress with the landing page update that emphasizes social proof I've been working on recently! The API is used by ICs/teams at hundreds of recognizable companies. There are tens of thousands of free users including some that have used the API consistently for free for over a decade. I guess you could check many instances of the service being archived on the wayback machine? https://web.archive.org/web/20240000000000*/https://www.exch... I'll definitely admit the domain does look a bit odd but back in 2010 when registering it the "Exact Match Domain" bonus was a big factor for SEO. The site has been a top 3 Google result for "exchange rate api" pretty consistently - presumably also how the scammer ended up using the service.
I've used Cloudflare since approx. 2019 and their "cloudflared" tunnel infrastructure since approx. 2021 to secure servers against DDoS.
I'll contact popey to see if we can get more details on the exact path/request they saw being made to our domain and if that leads to any further information or logging from our side.
I just got a B70 with 32GB RAM for the equivalent of $1200 (incl. sales tax and import duties to my non-US location, so presumably it could be cheaper elsewhere). The memory bandwidth is 608 GB/s. For M5 Max (32-core GPU) it's 460 GB/s and for M5 Max (40-core GPU) it's 614 GB/s. A 3090 is still faster at ~900 GB/s but you're getting 32GB VRAM for a lot less than equivalent Nvidia cards. It's about 1/3 the bandwidth of a 5090 for 1/3 the cost, but with the same 32GB VRAM. If you're interested in being able to run bigger quants with some context and stay on a lower budget then it's an appealing trade off.
I'm still exploring using these local models so don't want to spend the equivalent of $5 000 - $10 000 just to test it out. I don't mind slightly slower perf to do some experimentation more affordably.
I actually got an B50 16GB (with meager 70w TDP!) first to test an Intel card with my stack - it worked easily with Ubuntu & Vulkan. I'd read a lot about hassles and people writing them off as unusable but it seems like these are often with SYCL which doesn't even seem to outperform vulkan and so why bother? (The B50 was just $370 inclusive tax and duties). Literally `apt install` the vulkan libraries and it worked with default xe driver in 26.04 and the vulkan build of llama.cpp. The SR-IOV PF/VF also just works with qemu/kvm, no tricks required. Since I got it fwupdmgr has updated the firmware twice so Intel is presumably actually trying to support these products.