-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Socket AM2 Has A Secret Weapon.
- Thread starter chovi
- Start date
Associate
- Joined
- 14 Aug 2004
- Posts
- 1,979
- Location
- In a bar, gettin' ******
Careful now!Dolph said:
OCUK has been featured in a favourable light more than once recently on that site.
Or were they just spouting tripe?
If I recall correctly, those articles were in regards to pricing. Pricing is a numerical fact and can be backed up simply by visiting the site.Digital Punk said:
Most of the other stuff The Inquirer says, on the other hand, is purely speculation and usually goes along the lines of "we heard on the wire" or "some guy's cousin who cleans the house of one of the caretakers at (insert company name here) said" etc.
Associate
- Joined
- 12 Jan 2003
- Posts
- 2,037
well aside from wether its true or not, i have found the Inquirer to be fairly accurate in all so i don't really get where all these remarks from people come from about them making stuff up.... maybe they do, maybe they don't but as said, i've found them pretty accurate on the whole.
lets wait and see.
lets wait and see.
The Inquirer article seemed okay to me, it was the last paragraph that threw it all out and made me wonder if its author/source really had any idea what they were talking about...
mad:!) comes about it will behave almost completely transparently from the OS and certainly won't need some "driver" to give the CPU instructions on what can be parallelised and what can't be. But then, maybe there are some serious flaws in its implementation which has forced AMD to add an "opt-out" function so it can be disabled for certain types of software which would otherwise have a big performance hit.
Erm... riiiight. Seriously, when this "reverse hyperthreading"
mad:!) comes about it will behave almost completely transparently from the OS and certainly won't need some "driver" to give the CPU instructions on what can be parallelised and what can't be. But then, maybe there are some serious flaws in its implementation which has forced AMD to add an "opt-out" function so it can be disabled for certain types of software which would otherwise have a big performance hit.
Last edited:
NathanE said:The really funny thing about this is that it's very likely a bluff, playing into the current demand for "pro-AMD" tech' news. The irony is that Core 2 architecture _does_ have some form of hyperthreading built into it (though disabled in the first wave of chips) and Intel seems very reluctant to talk about it
The HT present in NetBurst chips was just one form. There are many different forms of HT. I'd hazard a guess that Core 2 may have some variation of SOEMT - a type of HT that works well on small-average length pipeline CPU designs.
Of course. Even the HT in NetBurst had a large overhead. Anything that adds delay to instructions reaching the pipeline will. However the aim of all types of HT is to provide a performance boost large enough to negate that overhead and then some.
Also note that this type of "inverse hyperthreading" (god I wish people would stop calling it that...) has been done before on mainframe chips (e.g. PPC AS400) for donkeys years.
But Mainframe chips (and anything beyond the basic x86 in general really) tend to run specific optimised code, you pair your processing/backend choice with the specific function you want it to provide. As such it's much easier to implement specific things (such as core multiplexing, which is the correct term for this 'reverse hyperthreading thing) in with much more efficiency. Most mainframe chips aren't used for general purpose computing.
Implementing somehting like this (or even working out whether it's worth implementing) is much easier when you know exactly what it will do at all times.
Associate
- Joined
- 15 Sep 2005
- Posts
- 869
remember that on several occasions ATI/Nvidia have launched a new graphics chip that all the pre-release benchmarks show to be easily the best card only for Nvidia/ATI to announce a new set of drivers a day later that suddenly unveils an extra 20% of performance in their ard that had previously been hidden so that all they grab all the launch day hype. Maybe AMD have a similar trick in mind?
Unfortunately, you can't just trickle out a driver update for CPUs the way that you can for GPUs, or it would be a possibility. I'm pretty sure that when it comes to CPUs, what you see is what you get, and what we're seeing is £180 Core Duo E6400s spanking £700+ FX-62s.deshepherd said:
If AMD think they can steal Intels thunder with 'Reverse Hyperthreading' they are going to get a big shock, if its true that Conroe XE's already have core multiplexing built is, as some are suggesting.
If thats true, intel can just push core multiplexing to the standard edition processors just like they enabled HT on all P4's to improve general performance.
Anyway the performance of multiplexing in a general 'desktop' environment is not yet proven, probably depends on compiler optimizations, and developers often dislike running optimizations which help one platform, but have a penalty on another.
If thats true, intel can just push core multiplexing to the standard edition processors just like they enabled HT on all P4's to improve general performance.
Anyway the performance of multiplexing in a general 'desktop' environment is not yet proven, probably depends on compiler optimizations, and developers often dislike running optimizations which help one platform, but have a penalty on another.
Methinks July 27th is the date that Intel discloses its new Hyperthreading technology hidden inside the Core architecture. At least, there are a bunch of NDA documents on Intel's FTP site that will be published on that date regarding Core 2 
This new HT tech' is only available on the higher end Core 2 processors.
A research paper coauthored by Intel many many years ago.
http://www.princeton.edu/~rblee/ELE572Papers/DynamicMultithreadingProc_akkary.pdf
It gives some fairly large insights as to what is behind the Core architecture.
This new HT tech' is only available on the higher end Core 2 processors.
A research paper coauthored by Intel many many years ago.
http://www.princeton.edu/~rblee/ELE572Papers/DynamicMultithreadingProc_akkary.pdf
It gives some fairly large insights as to what is behind the Core architecture.
Last edited:
Soldato
- Joined
- 18 Oct 2002
- Posts
- 12,257
- Location
- England
Intel also in on reverse Hyper-Threading....
http://www.tweaktown.com/news/5891/is_intel_also_in_on_reverse_hyper_threading/index.html
Seems you can do Reverse HT also on the new Conroes thru bios updates/new mobos
http://www.tweaktown.com/news/5891/is_intel_also_in_on_reverse_hyper_threading/index.html
Seems you can do Reverse HT also on the new Conroes thru bios updates/new mobos
HighlandeR said:http://www.tweaktown.com/news/5891/is_intel_also_in_on_reverse_hyper_threading/index.html
Seems you can do Reverse HT also on the new Conroes thru bios updates/new mobos
Yeah, they've half-inched that diagram from the thread linked earlier in this one at XS. I do wish people would stop calling it reverse hyperthreading though, it's no such thing and it's such a bad description for the idea.
It's also worth noting that the diagrams were user created and do NOT come from intel.
The full discussion of the core multiplexing technique at XS does actually give a lot more information.
However, Intel are certainly looking into this as a defnite idea, as presumably are Intel.
http://www.intel.com/technology/magazine/research/speculative-threading-1205.htm
Read through that and it is really the speculative threading we are looking at. The idea that instead of the code having to be written multii-threaded, putting logic on the chip to allow it to decide on threading and so on.
Associate
- Joined
- 29 Nov 2005
- Posts
- 433
- Location
- Milton Keynes
Dolph said:
Could you just clarify which one was supposed to be AMD? The first Intel or the second one?
Soul Rider said:
The second one, mainly because while I'm sure both parties are working on it, Intel release a lot more in the way of tech papers, studies and discussions on what they are up to than AMD do.
It's also worth noting that the Intel approach discussed in that article is not purely a hardware based solution. It's a combination of hardware and compiling software designed work together to effectively generate multiple threads out of single threaded programs by using guesswork on dependancys and the like.
