• Computers
  • Intel will no longer have the edge? (p.3)
2018/01/05 17:21:23
Jim Roseberry
The patch is out.
Backup your machine before installing.
You can compare an audio stress-test pre/post patch.
We're seeing extremely minimal to no performance difference.
2018/01/05 17:33:39
Sycraft
Jim Roseberry
The patch is out.
Backup your machine before installing.
You can compare an audio stress-test pre/post patch.
We're seeing extremely minimal to no performance difference.



I don't have any good benchmark for audio performance, but I tested gaming, DPC latency, and disk IO and found no difference. Well to be completely accurate I found a slight increase in gaming performance post-patch, but it was well within the margin of error.
 
Now this is the OS level patch, I do not have the updated microcode yet (doesn't seem like any vendors have released it) but so far, I'm seeing no change for user type workloads. VM servers may be another thing, I'll be testing those this weekend, but not really relevant to people on this forum.
 
It is really nothing to worry about performance wise so far as I can see for the general public.
 
tunedeaf
"He sold after the announcement, pretty quickly after the announcement. Not sure they can touch him."
 
If he put his sell plan in place after the issue was made public he is okay. I honestly don't know when the public was made aware that this is a significant issue (one that could very well affect stock prices).



It looks like that's what he did. I mean this was discovered last June, I don't know when Intel was brought in on it but it was early in the process. They've known for a long time. Presumably he planned to sell, and waited until the earliest he could legally do it.
2018/01/06 01:04:12
tlw
According to this report, Intel knew of the problems in June, the share sale was readied in October and shares sold November. We all only got to hear about Meltdown over the last few days, and I doubt Intel’s many shareholders got any notification before the rest of the world did.

https://www.cnbc.com/2018...ve-security-flaws.html

The effect on Intel share prices hasn’t been as drastic as if could have been. A 2% drop followed by a bit of a recovery. In terms of the total amount of money involved a 2% drop wipes a few million of Intel’s value but that’s a drop in a buicket as far as Intel are concerned.
2018/01/06 05:25:03
Sycraft
Long term it isn't likely to be a big drop for four reasons:
 
1)  This isn't some kind of negligence on their part. This is a completely unforeseen consequence of speculative execution. That is a technology, widely used, that has been around for decades. It isn't like this was something they were ignoring, this is something security researchers only recently figured out. So while they will get sued, since everyone sues over everything these days, it isn't that likely to succeed.
 
2) This is not an Intel only problem. Despite early reports, this is something that affects basically all modern chips. The Specter attack (reading in-process memory) is something that is probable to work on every processor. That it hasn't been demonstrated yet on every CPU is a matter of what the researchers focused on, not that the others are invulnerable. It'll happen in time. MalwareJake is pretty confident this is only the beginning.
 
3) The performance impact is not near the big deal it was made out to be. On normal systems, it is tiny if even present. Even on big I/O systems, it doesn't seem to be as bad as first made out. So while there is lots of nerd rage, it is hard to legitimately get that mad about it. Couple that with the fact that Intel never made any specific performance claims on their CPUs guaranteeing a level of performance.
 
4) It isn't that big a security hole. It is not the cataclysmic event that some people were (and still are) making it out to be. It is a covert data ex-filtration channel. That sucks, but we have been dealing with that in security forever, and they don't go away. You just have to work to make them difficult to use and to monitor for it happening.
 
So while it will cost them some money, it isn't likely to be an amount such as to hurt stock price.
2018/01/06 19:28:19
tlw
Sycraft
3) The performance impact is not near the big deal it was made out to be. On normal systems, it is tiny if even present. Even on big I/O systems, it doesn't seem to be as bad as first made out. So while there is lots of nerd rage, it is hard to legitimately get that mad about it. Couple that with the fact that Intel never made any specific performance claims on their CPUs guaranteeing a level of performance.


I think the media coverage has tended to over-exagerate any performance issues because journalists and editors generally don’t understand the difference between “protecting against the exploit means a few specific computing operations will now require between 1.05 and 1.3 times longer to complete, which on a cpu that’s clocked in the gighertz range amounts to nothing much” and “everything the computer does will now take 30% or more longer to do”.

Journalists are notoriously inclined to “dumb down” reports concerning science or technology, and in an age of clickbait journalism sub-editors have an eye for creating alarming headlines.

As for the legal implications, Intel and other cpu manufacturers sell and warranty cpus on the basis that they will perform as their publicly released datasheet indicates. Nowhere do they say “we warranty our products will be invulnerable to all security flaws, both known ones and any that might be discovered in the future”.

For the same reason anti-malware software isn’t sold on the basis that it will provide 100% protection for the same reason. Known malware can be checked for, but all that can be done to protect against new malware is to make reasonable checks for malware-like behaviour, block it and flag it up as a potential issue for a human to take a look at and investigate further. Which works most of the time but sometimes doesn’t.

It’s still the case that the only truly secure computer is one with no network or wireless connections that’s placed inside a heavy duty Faraday cage inside a sealed concrete bunker with no entrances. And then only if the computer is never switched on.
2018/01/06 22:33:19
slartabartfast
tlw
I think the media coverage has tended to over-exagerate any performance issues because journalists and editors generally don’t understand the difference between “protecting against the exploit means a few specific computing operations will now require between 1.05 and 1.3 times longer to complete, which on a cpu that’s clocked in the gighertz range amounts to nothing much” and “everything the computer does will now take 30% or more longer to do”.



You realize that you are posting this quite sane observation on a forum that routinely carries reports of how users are overclocking with water-cooled systems, shutting down dozens of irrelevant processes, and following pages of laborious tweaks to get their audio systems an insignificant improvement?
2018/01/10 17:31:27
DrLumen
The conspiracy theorist in me is screaming that intel and the NSA knowingly integrated it into the processors.
 
I find it odd that intel, amd and arm are all affected in some form. Did they cross license that tech or something? And, if so, why didn't they catch it before now?
 
Plus, if I'm not mistaken, the machine still has to be infected and the code executed. I can understand and glad they are patching the problem but wouldn't an anti-virus prevent the exploits as well?
2018/01/11 13:42:30
Jim Roseberry
DrLumen
Plus, if I'm not mistaken, the machine still has to be infected and the code executed. I can understand and glad they are patching the problem but wouldn't an anti-virus prevent the exploits as well?



As with Malware, you're not going to be exposed/exploited running DAW software.  
It wouldn't surprise me if this "design flaw" was at some level intentional.
2018/01/11 14:30:24
gbowling
Sycraft
VM servers may be another thing, I'll be testing those this weekend, but not really relevant to people on this forum.
 

 
Actually in a round about way, it is totally relevant to me. I wanted to keep my DAW pristine, no installation of other software/drivers/etc. that might impact my DAW operation.  But I also didn't want two machines.
 
So I have a virtualbox machine on top of my DAW. My hardware machine has all my audio software and drivers and virtualbox installed on it, nothing else. My virtual machine has all my other software installed on it. 
 
I love the way it operates, when I want to do a session I just shut down the virtual machine and my box is pretty much a dedicated DAW. The only issue I have is that audio in the virtual machine isn't very good (actually almost unusable), so listening to spotify, itunes, or youtube on my virtual doesn't really work well. But I'm not about to install them on my base machine, sans youtube which runs in a browser. 
 
gabo
2018/01/11 16:39:01
TheSteven
gbowling
Sycraft
VM servers may be another thing, I'll be testing those this weekend, but not really relevant to people on this forum.
 

 
Actually in a round about way, it is totally relevant to me. I wanted to keep my DAW pristine, no installation of other software/drivers/etc. that might impact my DAW operation.  But I also didn't want two machines.
 
So I have a virtualbox machine on top of my DAW. My hardware machine has all my audio software and drivers and virtualbox installed on it, nothing else. My virtual machine has all my other software installed on it. 
 
I love the way it operates, when I want to do a session I just shut down the virtual machine and my box is pretty much a dedicated DAW. The only issue I have is that audio in the virtual machine isn't very good (actually almost unusable), so listening to spotify, itunes, or youtube on my virtual doesn't really work well. But I'm not about to install them on my base machine, sans youtube which runs in a browser. 
 
gabo


I've been considering doing something similar but slightly more drastic - also putting my DAWs in a virtual machine.
So only having the base operating system on the machine and everything on separate virtual machines - one for my DAWs, one for my programming apps, one for 'normal' PC usage (email, etc.).
 
In the past 2 years for reasons beyond my control (motherboard failure, OS corruption, etc) I've had to 3 times setup & reconfigure my system from scratch.  My audio setup takes the most time. Each time it's a long drawn out process of getting everything installed, activated and configured.
I haven't run a proof of concept yet (putting Sonar on a virtual machine and checking it out) but hoping that on a decent i7 processor, with ample RAM and SSDs that it proves viable.
© 2024 APG vNext Commercial Version 5.1

Use My Existing Forum Account

Use My Social Media Account