You know, using all those points that got dumped into Intuition when I was rolled, I looked up the reviews for Corel Aftershot Pro over at Amazon yesterday, just… to see. What people were saying, you know? Using my intuition, you understand. Kind of tired of Lightroom and looking at that competition!
What did i see? Exactly what i expected. “OMG i have to download my software from the Internet! Why can’t i have a hard copy? Why isn’t there a printed manual?? I cannot use this software because i refuse to put my work computer on the internet. 0 Stars would never buy again.”
Just saying :). Most of this stuff really has gotten better!
Last time I updated my BIOS was about five years ago, and there was a power failure halfway through. This is in Toronto Canada where the power goes down for, on average I’d say, 20 seconds a year. At the time I was like “are you freaking KIDDING me!?”
Amazingly, the system recovered and continued the install. So they have at least some protection or capability built in these days.
I think the only real tricky thing to updating BIOS, especially if you do it through Windows and not through a USB key, is to disable (disable, not decrypt) Bitlocker if you’re using Bitlocker before you run the installer.
Both the forum and frontpage servers are now updated with meltdown fixes. The forum has a ridiculously overpowered CPU so it don’t even stress, yo, and the frontpage is just wordpress now so it’s sitting pretty.
May I ask what brand m/b you have? I’m curious to know which companies respond with BIOS updates and which ones don’t, to keep in mind for my next gaming rig build.
I don’t know. It’s an Alienware Aurora R5 desktop I bought in Nov. 2016, if that’s any help. I’ve been very happy with it though.
I just went to Dell’s website, where it automatically detected my model, and then displayed all of the current drivers for it along with the dates they were released. I clicked a button, and it downloaded the 4 that were out of date. It then presented the install button, which I clicked, and it installed 3 of them automatically. The BIOS however, I had to click a separate button for it to install. But once I clicked that, it installed itself and rebooted automatically.
Wow, if motherboards need Bios updates this is going to be a disaster. It’s unusual to see random motherboard vendors making updates 3 to 8 years later. I highly doubt a large percentage will get around to it.
The nopcid instruction was added with AVX2 support when Haswell was new, which would seem to imply that Intel CPUs pre-Haswell might face larger penalties than chips post-Haswell.
And
Computers with an Intel Haswell processor or newer have a PCID (Process-Context Identifiers) feature that will help the patch perform well. Computers with older Intel CPUs may see a greater decrease in speed
You’re still not saying where you’re getting that from, but it seems to be somebody making an educated guess about both the mititgation mechanism and the performance impact (“would seem to imply”). In the other hand the “Haswell or older” vs “Skylake, Kabylake or newer” split comes directly from Microsoft’s announcement, based on their benchmarking of the actual Windows patch.
It seems kind of obvious to me which of these two sources one should believe.
Why the discrepancy? Maybe there’s something wrong with pcid on
Haswell, and they couldn’t use it. Or maybe the big hit isn’t from the page table isolation but e.g. the branch prediction changes.
I may have missed this in the earlier discussion (which I did try to read), but are there existing or new chips that do not have this problem? My computer is getting older (going on four years), and I could see this being a push for me to just get a new one. I could also see it being a reason to wait if there are no chips out now that do not have this issue, but will be in the next six months to a year.
The main problem–Spectre–is endemic to basically any chip you could buy today. Since it’s so deeply embedded in the architecture, I’m not even sure we can assume it’ll be a quick fix in the next set of die shrinks and chip revisions, but on the other hand, the processor companies have known about this longer than the public has.
Chips that do not support speculative execution are not affected by either of the issues, but they also have lower performance. Not sure how many of those are available for general consumer consumption, though; they tend to be specialized and used for automation and the such.
That’s also why the Raspberry Py was not affected - the CPU they use doesn’t have speculative execution.
Three separate sources on the PCID thing. One says haswell and later, one broadwell, and one skylake. Microsoft said skylake, but that may only pertain to their specific mitigation strategy.
