Breaking news: Security flaw supposedly found in Intel's hyperthreading implementation
- Thread starter arcas
- Start date
They will probably get a fix for it in its software implemantation...
Remember HT is merely making an inefficient design work a bit more efficient. HT doesn't add anything other then try to compensate for a penalty that is imposed by the excessively long pipeline and branch mispredictions. A processor like AMD with its improved branch prediction and shorter pipeline as well as the Dothan will benefit very little from HT as it is implemented today......
The more I start to understand HT the less and less I have been impressed with it...maybe the realization of true dual cores (not virtuals) as me liking the reality versus a quick trick to fool the eye.....It had its moments for me but I have seemed to be missing it less and less of recent.
How many ppl would this have really affected??? I dont think I would have been vulnerable....
Remember HT is merely making an inefficient design work a bit more efficient. HT doesn't add anything other then try to compensate for a penalty that is imposed by the excessively long pipeline and branch mispredictions. A processor like AMD with its improved branch prediction and shorter pipeline as well as the Dothan will benefit very little from HT as it is implemented today......
The more I start to understand HT the less and less I have been impressed with it...maybe the realization of true dual cores (not virtuals) as me liking the reality versus a quick trick to fool the eye.....It had its moments for me but I have seemed to be missing it less and less of recent.
How many ppl would this have really affected??? I dont think I would have been vulnerable....
You might want to reconsider your opinion. If an unprivileged process can read from a privileged process' memory, then it must be able to escalate its own privileges. That would be a very bad thing. Yes, it appears that there are ways in software to prevent this from being dangerous, but it's still looking like a terrible gaffe on Intel's part. Nobody else's SMT implementation seems to suffer from this.
Then again, maybe we should reserve judgement until we actually hear the details.
I dont have one anymore!!!
But explain more if you could a type of attack say against someone like me running Mozilla Firefox with AV/ spyware/ and firewall set using the p4....I P2P a bit getting CSI episodes (not into DVDs or mp3s)...
But explain more if you could a type of attack say against someone like me running Mozilla Firefox with AV/ spyware/ and firewall set using the p4....I P2P a bit getting CSI episodes (not into DVDs or mp3s)...
Lets all remember, HT is more of an elegant fix to a problem, rather then a added feature, although I do admire it ..
Zebo
Elite Member
- Jul 29, 2001
- 39,398
- 19
- 81
And? Price went up.
Lemme puts this in plain terms right now since this whole thread and inq is into conspiracy tone.
A "sell off" is market speak which consititutes more shares sold than bought, usually at a signifigant and percipitus rate. This results in a lower share price at the end of the day.
That's not what happened here at all. Lots sold, however lots more bought ending up higher than the day started in the first place.
Institutional investors arnt stupid. They know Intel is a good buy. With a war chest of billions of dividens ready to distribute once tax code turns favorable. With a house-hold name. With cash flow unmatched. With good ratios. With good prospects. And a relativly low share price.
uOpt
Golden Member
- Oct 19, 2004
- 1,628
- 0
- 0
Go FreeBSD 
Well, people worked abound a number of bad processor bugs in OS kernels, lets hope there's a fix here.
As for an attack, this is a break of the separation of local user IDs. I can imagine a Windoze user running IE with administrator rights is not impressed.
But in the multiuser and hosting world that is a big deal. You get a hook into Apache - bad but on halfway. If you can use the hook into the userid running apache to read arbitray memory out of other user's processes that allows for a number of "interesting" uses, as Colin explains the stealing of private keys. Or listen to sshd accepting people's passwords (go ssh-agent!) as administrators log in.
Well, people worked abound a number of bad processor bugs in OS kernels, lets hope there's a fix here.
As for an attack, this is a break of the separation of local user IDs. I can imagine a Windoze user running IE with administrator rights is not impressed.
But in the multiuser and hosting world that is a big deal. You get a hook into Apache - bad but on halfway. If you can use the hook into the userid running apache to read arbitray memory out of other user's processes that allows for a number of "interesting" uses, as Colin explains the stealing of private keys. Or listen to sshd accepting people's passwords (go ssh-agent!) as administrators log in.
- Aug 22, 2001
- 32,090
- 32,629
- 146
Zebo
Elite Member
- Jul 29, 2001
- 39,398
- 19
- 81
Yea I should STFU.. Hav'nt made a dime in 6 years:| shows what I knows.
I'll say one thing.. Intel processors have obsure bugs (so does AMD's) that could result in a hard lock....always have..has'nt hurt them. HT itself has been sucessful 2 yrs running and some very minor obscure HT bug all of a sudden is going to cause the stock to collapse. GMAB! These people don't even know what a CPU looks like let alone what HT is. All they care about is P/E's, profits, valuation, etc etc which Intel has in spades.
I'll say one thing.. Intel processors have obsure bugs (so does AMD's) that could result in a hard lock....always have..has'nt hurt them. HT itself has been sucessful 2 yrs running and some very minor obscure HT bug all of a sudden is going to cause the stock to collapse. GMAB! These people don't even know what a CPU looks like let alone what HT is. All they care about is P/E's, profits, valuation, etc etc which Intel has in spades.
Wow:Q, EF...EF...EF Hutton with the accent on the second T. Damn, I think the last time I saw that I was sportin' Jams, listening to my brand new Sony Walkman cassette, and chasin' the blossoming young felines.
Perhaps the OP wasn't using "market speak". Sell off by simple definition means, "To get rid of by selling, often at reduced prices". The Inq reported just that & the OP probably inferred that there could possibly be a connection between the HT security flaw report & the irregular amount of stock suddenly sold within that short (brief) period of time.
If that's what the OP meant (which I believe is so), perhaps he could've worded it a bit better for ppl like you but the general meaning of it (IMO) is there. In any case, thx for breaking down the marketing term for "sell off" for us simple folk.
- Aug 22, 2001
- 32,090
- 32,629
- 146
I am old schoolOriginally posted by: TStep
Wow:Q, EF...EF...EF Hutton with the accent on the second T. Damn, I think the last time I saw that I was sportin' Jams, listening to my brand new Sony Walkman cassette, and chasin' the blossoming young felines.
Glad some one else can remember the ads too, 1/2 this crowd ain't even old enough to shave yet I remember the commercials as well!!! The guy standing in the loby talking to his buddy them all of a sudden all the ppl around him hush and lean over as if they are trying to get some juicy gossip....Yeah I may have seen those while playing with my Legos and watching Brady bunch on network TV!!!
GuitarDaddy
Lifer
- Nov 9, 2004
- 11,465
- 1
- 0
You guys make me feel old
When I was playing with LEGO's and Lincoln Logs (remember those), TV was black and white:frown: And when EF Hutton talked nobody listen
When I was playing with LEGO's and Lincoln Logs (remember those), TV was black and white:frown: And when EF Hutton talked nobody listen
Elcs
Diamond Member
- Apr 27, 2002
- 6,278
- 6
- 81
Even if HT is 'an elegant fix to a problem', as an AMD-exclusive owner so far (due to price/performance a the time of purchase, ive been VERY impressed with HT.
Anything that improves performance in real-world, useful software without a massive impact on price or introduces many problems is good in my opinion.
HT has been out for how long? and they've only just found a problem with it? I cant remember any major security flaw in HT prior to this.
I say Intel have done great with it and 1 security flaw found after so long is something to be proud of.
Whilst HT may be significantly less useful in the near future on the bleeding edge of technology but it has served its purpose well.
Anything that improves performance in real-world, useful software without a massive impact on price or introduces many problems is good in my opinion.
HT has been out for how long? and they've only just found a problem with it? I cant remember any major security flaw in HT prior to this.
I say Intel have done great with it and 1 security flaw found after so long is something to be proud of.
Whilst HT may be significantly less useful in the near future on the bleeding edge of technology but it has served its purpose well.
i think ht is just another one of their marketing schemes. like, "wow, our processors go twice as fast as amd, so they are twice as good" and "we have HT and they dont" i dont know what has happened to intel in the past few years. its kinda sad.
LOL, the blows just keep coming for Intel.
Q: How many Intel HW engineers does it take to fix a problem?
A: None. We'll fix it in software.
Q: How many Intel HW engineers does it take to fix a problem?
A: None. We'll fix it in software.
lol I value it too, why do you think I use my 2.8 C for ripping/playing games/P2P/norton at the same time.
Ask my athlon to do that and it'll go on strike!
Thats wrong, HT is not just a marketing scheme and it does have a real impact on performance. Basically what it does (correct me if I'm wrong) is tell windows there are two processors and then take the second processor's instructions and run them in the gaps in the first processor's instructions. It helps with multitasking and with anything that is multithreaded.
Right. Some of IBM's POWER chips do the same thing as HT, as do some of Sun's SPARCs, and neither do this just for show. Theirs don't seem to suffer these kinds of bugs either.
Here's the abstract from the paper:
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 23K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter