nanobot

You know this would actually be really handy feature, especially for the configuration note you mentioned. You can add "identifiers" to your tests right now but I think it would be nicer to have a 200-500 char memo field we could add notes to and see on the tables with our own results.

My Win10 system boots in like 15s (with the SAS bootstrapping), my iMac takes like 35s with the "Fusion Drive" (it's a late-2015). To be fair the Win10 machine is 3 SSD's for the OS / data, and an enterprise SAS card for the other data. I haven't done the Big Sur update but Catalina is slow as f- on this thing. I dunno if it's because I open a bunch of stuff that basically stays open forever, or maybe Docker is being an arse, but seems like I'm always at like 20% CPU usage with iTerm2, Mail, Messages, Brave, Atom and VS Code open. I do think Docker is involved, though. (No running containers but docker is a POS anyway.)

@spudler_t What gear are you using? I'm thinking about getting a Fanatec wheel + pedal kit for Assetto Corsa Competizione, any thoughts?

So the 1.3MB/s you are seeing is 10.4Mbps, which is definitely way-low. Based on the Speedtest.net results, it's obvious that you are seeing a good link to your ISP. From there though, the link degrades. There are two common issues that would make sense: 1. ISP is oversubscribed, meaning they sold more bandwidth than they have. Say they oversold bandwidth by 5-times, meaning they have a 50Mbps link, but sold 250Mbps worth of bandwidth. If everyone is trying to download data at a time, you're likely going to see bandwidth / 5 throughput, or in your case, 10Mbps. 2. ISP is having peering issues. Because you can connect to the ISP cleanly, this also makes sense. Basically, peering is how one ISP connects to another. No single ISP transfers all the traffic for the internet, so they connect to each other at various locations. If one or more of these locations is having trouble, you could see lower throughput. In either case, you'll want to run a good amount of data via the TMN testing, and then reach out to your ISP about it. There may be something they can see or do, there may not be.

You are seeing the real speed. Your 3Mbps is what speed you were able to achieve when downloading data from Testmy.net. The 1.2MB/sec you are seeing on the download in your screenshot is what you are able to achieve when downloading from that particular site, and is actually 9.6Mbps. You are seeing different units for the metric: MB/s = Mbps / 8 Mbps = MB/s * 8 If the Speedtest.net results were accurate, the 50Mbps would be about 6.25MB/sec for your downloads. You are seeing roughly 1/5 that. This may be an issue with your ISP peering, or it may be an issue of routes between you and where you are downloading your data from. I recommend you try switching TMN test servers and see how that affects your metrics.

It's not "faked" so much as the server that Speedtest chooses is deliberately the closest and fastest server. In some cases, it might still be within your ISP's network, which means what you are testing is you -> ISP, not ISP -> world.

Team has dropped pretty substantially. I haven't been folding as much lately as I've been gaming with my PC again, so I turn it off for 4-8 hours a day. We're #961 right now, I suspect that will drop further before it goes up.

I don't knock the Sim racing. I run GT Sport races a lot, and even that is challenging, I can't imagine having a good setup. (I am trying to compete using a controller...lol) I'll have to check their YouTube channel out as well. Been getting my fix by watching reruns of old F1 races, and old IMSA races, but it's not the same as watching 'live'.

Given that the FIA and IMSA have postponed and rescheduled the seasons, I may have to start watching this to get my racing fix in. Granted, my team stands no chance in F1 this year, but I still want to watch them lose lol

That's not surprising. I'm seeing the same at my data-center. They're running several 100Gbit connections, but they also host a bunch of infrastructure for Netflix and Hulu, and they told me that "due to the significant amount of traffic to several of our customers, you may see bandwidth impacts." I got another ticket this week that said they're recarving up their bandwidth to limit those two in particular. Going to be interesting to see how that goes.

Just got to thinking about this after using the "latency" test. When I test latency I can hit multiple servers, and it cycles through them. I would like to be able to do that for uploads / downloads too. That is, select a few servers, have it run the test on each and give me all the results. It would be useful to be able to compare to see if traffic to certain geographic regions are affected. I.e. if I've been seeing some sites slower than others, well testing Dallas + LA + Miami + Frankfurt would give me a view of whether I'm seeing bad traffic to certain regions or not.

So speedof.me definitely uses the first technique: pick the fastest server it can. I can't tell, but based on the results I saw it looks like it uses the second technique as well: throw "outliers" away. I'm not sure on Google, but Google has such a wide network they probably geographically "multithread" by default. I'd have to confirm that one.

The two Supermicro's are getting WU's again! My GPU picks WU's up sporadically, so I'm only counting on it for ~half the PPD. CPU's here should be good for about 125k PPD. The GPU should be 700k to 1.5m depending on how WU's play out.

What other sites are you using? If it's anything that uses the 'speedtest.net' / 'Ookla' system, the reasons those are likely unaffected is because they use two techniques to make things look more favorable for your ISP: 1. They use geographically close servers, i.e. as physically close to your network as possible; 2. They typically throw away the two 5% extremes, i.e. the 5% worse and 5% best parts of the test; The difference is TMN does neither. Since the pandemic started, internet traffic is way up. People are streaming, gaming, and in-general downloading more. The core of the internet is seeing the effects of this, even if your local nodes are not.

So I've finally hit the point where my GPU is having trouble picking up WU's. Earlier this week two of my servers started having trouble, so I think we're exhausting the FAH WU's again.