> To address the specific needs of Ethereum mining, we’re announcing the NVIDIA CMP [Cryptocurrency Mining Processor] product line for professional mining. CMP products, which don’t do graphics, are [… ]optimized for the best mining performance and efficiency. They don’t meet the specifications required of a GeForce GPU and thus don’t impact the availability of GeForce GPUs to gamers.
hahaha
So it seems like it's just the driver, not the GPU itself.
@forest The GMPs are literally video cards with no output, and a special driver without directx... :D
So apparently people buy old ones, change a registry entry, and use them for virtualisation or just push the output to integrated graphics...
@forest "the f" is bitcoin miners and scalpers buying up so many video cards that it might mess up nvidia's long term gaming market
So nvidia now will start churning out old cores, remove components, and sell them to bitcoin miners at a premium that's still cheaper than an old card...
@JourneymanGeek I know. They're also pretty bad at hash cracking. The issue, as far as I'm aware, is that Nvidia uses a smaller number of very complex cores, compared to AMD which uses a huge number of relatively simple cores. I have a small hash cracking rig that exclusively uses AMD. I'd probably have to pay twice as much if I used Nvidia, not to mention all the extra costs in power.
If I understand correctly, /dev/shm is often preferred over /tmp because /dev/shm stuff is stored in the RAM instead of the disk, leaving minimal trace of stuff. Is that right?
@nobody I won't call them heroes... Ryuk crippled a company I formerly worked for, and they called me to help them rebuild their infra from scratch. Working 16h per day for weeks isn't funny at all, and they did nothing wrong to anyone to deserve being attacked.
they had backups in place, but restoring a couple TB of data over a not so fast connection took days. Every computer had to be reimaged, every password replaced, all API keys revoked and recreated, it was a lot of work.
I have no sympathy for ransomware gangs, even if they attacked my government that I pretty much dislike.
@ThoriumBR I'm fine with ransomware groups targeting LE. For pretty much anything else, I have no sympathies for them either. They are just greedy criminals, nothing else.
And the reason I'm fine with them hitting LE is that it is the same thing as members of one gang shooting members of another gang. I'll applaud both for taking each other out, good riddance.
I found several issues: - sandboxing: of course there is sandboxing... have they never heard of Mandatory Access Control? chroot? LXC? Even Docker? Sandboxing on Linux exists since always. It's not really user friendly, but with some effort you can sandbox almost anything. - "selinux have `execmem` but it's rarely used". Windows have multiple users but everyone runs as admin... - "powerful root": it's because linux is *my system* and I can do anything I want. On Windows I can do what MS wants me to do.
on windows side, the OLE security model is a joke... there are so many ways to execute code using OLE embedding. And what about Office macros? and Powershell? And the security of passwords on ActiveDirectory? Unless you disable some older password schemes (LM hashes?)
the issue is that the balance scale is different for windows and linux... most of the issues on linux side have something similar on windows, and linux have some protections that windows does not.
yes, there are issues on the linux side, but there are so many issues on the windows side that the author does not mention
True, so takeaway is, security is shit everywhere.
But on linux, you can make things better if you want to and know how to
And BTW, Powershell used to be a very good tool for attackers, not so anymore. But office macros are just stupid, why the hell would you not sandbox them?
I like to say that security is a tradeoff, and nothing can be absolute secure... it's the difference between hardening yourself for not being mugged on the street by a random crackhead and protecting yourself against a profession hitman hired by a couple million bucks to kill you.
if you have secure habits, drive-by downloads and random physhing attacks can be detected and you can protect yourself.
but if the TLA (Three Letter Agency) is after you, they WILL own you, it's only a matter of time.
@ThoriumBR I'd like to give them a challenge anyways. (Not that I currently can, but I hope to learn enough at some point in my life that the TLAs don't get me on their first try)
But about Qubes, Spengler is basically saying it is totally useless, gives no benefit over normal security measures. Yet plenty of people continue to recommend it as a secure alternative. The two things don't seem compatible, I want to know which one is right.
Qubes can protect you from a lot of threats, but cannot protect you against a targeted attack
like windows, beOS, QNX, MSDOS...
by what I read, X on Qubes is a "channel" between the VM and the hipervisor, and drivers on the VM can exploit the hipervisor and run code on kernel... So take Qubes as another layer of defense, not a plug and play unassailable security castle
using Qubes, malware would have to know you use it, and attack the browser, run code on the VM, escape the VM, execute code on the host, and read data on host, returning that data to the guest. Possible, but not trivial as the article may suggest.