Intel Hyper-Threading Accused of Being a Security Threat

[rantoc]

Bet the new 9xxx will be just as screwed HW wise, when will we see REAL HW fixes rather than mitigations from an exploit that's been known for well over a year. Intel are releasing a new CPU family soon, only with mitigations - Not fixes vs this mess.

 

[newsonline5000000]

What we need is a totally new CPU design from the ground up with security in mind while designing the new chip.

Leave the X86 already and use security as an excuse . this is Very old technology already.

 

[Footloose]

Oh please. The RISC vs CISC argument is older than the hills and total crap. There may have been an advantage to RISC 15 or 20 years ago but that is no longer the case. The CISC based CPUs from Intel and AMD are architected with the same elements that originally made RISC faster than CISC in the distant past.

 

[artk2219]

Intel should take this as an excuse to move to ARM/RISC CPU and ditch This old technology they have been using since forever ...

Apple Moved from IBM CPU's to Intel in 2 years . I think Intel and the PC should do the same

Did you forget the /s at the end or something? Apple's move isnt even remotely comparable, especially since Apple didn't own any patents or IP on either of the solutions it picked, also they were a TINY part of the PC market at the time (they're still not a huge part). Not to mention the headaches this would cause on pretty much every technology sector, troll be trolling? I suppose theres future potential in RISC V, but thats not going to be the case for many many years, they at least need to finish milking their multiple different flavors of Skylake first.

 

[Trevor98]

I am curious if this also effects AMD processors. Some of the vulnerabilities that this is connected to are not on AMD, and I know their implementation of SMT is different, but it doesn't sound like they tested AMD at all, no mention of AMD at all in the article. So far, it's looking like AMD has a huge opportunity in the data center with so many vulnerabilities being found on Intel, in addition to the other issues with 10nm etc. Add hyperthreading to that and you suddenly see why AmD is going in strong with 7nm epyc.

A very thorny problem remains with AMD, and it persists with Ryzen: AMD's lackluster single core CPU IPC ("Instructions Per Cycle"). They just cannot match Intel's IPC to save their life. It's been this way since about 2007. I think AMD needs to hire better mathematicians and computer scientists.

 

[dudmont]

A very thorny problem remains with AMD, and it persists with Ryzen: AMD's lackluster single core CPU IPC ("Instructions Per Cycle"). They just cannot match Intel's IPC to save their life. It's been this way since about 2007. I think AMD needs to hire better mathematicians and computer scientists.

Yes and no. The first part of your statement is 100% spot on. As to the second part, have you ever seen the amount of money Intel puts into R&D? It absolutely dwarfs AMDs total annual revenue. But, like most things in the computer world, to get that last little bit of advantage/performance costs many times in money.
Here's a question for anyone interested. Does anyone think some of the reason Intel is delaying 10nm might have something to do with maybe reworking some of the new chips for Specter/Meltdown mitigation? Not saying it's so, but merely a possibility.

 

[newsonline5000000]

Intel should take this as an excuse to move to ARM/RISC CPU and ditch This old technology they have been using since forever ...

Apple Moved from IBM CPU's to Intel in 2 years . I think Intel and the PC should do the same

Did you forget the /s at the end or something? Apple's move isnt even remotely comparable, especially since Apple didn't own any patents or IP on either of the solutions it picked, also they were a TINY part of the PC market at the time (they're still not a huge part). Not to mention the headaches this would cause on pretty much every technology sector, troll be trolling? I suppose theres future potential in RISC V, but thats not going to be the case for many many years, they at least need to finish milking their multiple different flavors of Skylake first.

If intel wants to make a new CPU design from the ground up , it is better they do it outside the X86 ... This is What I am saying. This is their perfect Chance to do it .

 

[TerryLaze]

If intel wants to make a new CPU design from the ground up , it is better they do it outside the X86 ... This is What I am saying. This is their perfect Chance to do it .

Yeah making a CPU that is unable to run anything that's on the market today...now that would be a real smart move.

 

[Peter Martin]

LOL, yes, we need compatibility. you can't ditch everything out there, business could not run, etc.. LOL

 

[newsonline5000000]

If intel wants to make a new CPU design from the ground up , it is better they do it outside the X86 ... This is What I am saying. This is their perfect Chance to do it .

Yeah making a CPU that is unable to run anything that's on the market today...now that would be a real smart move.

Okay be stuck with X86 until Aliens come and laugh at us in 300000AD

 

[JamesSneed]

If intel wants to make a new CPU design from the ground up , it is better they do it outside the X86 ... This is What I am saying. This is their perfect Chance to do it .

Yeah making a CPU that is unable to run anything that's on the market today...now that would be a real smart move.

Okay be stuck with X86 until Aliens come and laugh at us in 300000AD

I don't think he was suggesting that. It's just small minded to say x86 needs to go because there is RISC. Most of the benefits are already baked into x86 these days and there really wasn't a huge difference between the two anyhow. With that said x86 will die off eventually not because some slightly faster chip design but because something order of magnitudes better comes along. Like say some new instruction set to run on top of quantum computers.

 

[newsonline5000000]

If intel wants to make a new CPU design from the ground up , it is better they do it outside the X86 ... This is What I am saying. This is their perfect Chance to do it .

Yeah making a CPU that is unable to run anything that's on the market today...now that would be a real smart move.

Okay be stuck with X86 until Aliens come and laugh at us in 300000AD

I don't think he was suggesting that. It's just small minded to say x86 needs to go because there is RISC. Most of the benefits are already baked into x86 these days and there really wasn't a huge difference between the two anyhow. With that said x86 will die off eventually not because some slightly faster chip design but because something order of magnitudes better comes along. Like say some new instruction set to run on top of quantum computers.

It has been long enough .. I remember 10 years ago , The same stupid excuse was used : No Software will run ...

This will never stop btw ..

Intel and MS should start Moving on , it will take at least 10 years for companies to update their software and I dont want to Die Using X86 ...

MS and Intel Should have PARALLEL project behind the scene ...

and no , it is not slightly faster , it can be 30 folds faster ...

we need New faster design , Maybe also Nivida and AMD can Join the team and bring the best CPU Design.

 

[JamesSneed]

@newsonline5000000 , please read this: https://www.eetimes.com/author.asp?section_id=36&doc_id=1327016

Then realize you can stop worrying about x86.

Honestly there is virtually no real RISC in use anymore anyhow. It stands for reduced instruction set however even RISC based cpu's have a ton more instructions baked in these days.

Unless you are in your 20's maybee 30's can expect to die using x86.

 

[newsonline5000000]

@newsonline5000000 , please read this: https://www.eetimes.com/author.asp?section_id=36&doc_id=1327016

Then realize you can stop worrying about x86.

Honestly there is virtually no real RISC in use anymore anyhow. It stands for reduced instruction set however even RISC based cpu's have a ton more instructions baked in these days.

Unless you are in your 20's maybee 30's can expect to die using x86.

It was the Cheap X86 platform that made the RISC disappear , not which is better . and the software houses followed the market .

I am not insisting on RISC by the way , Just stating an example because there are no Designs in the market outside the two today.

Actually I wish for GPU makers to unite with intel to make the next CPU ...