· Content
· News
· Articles
· Mailinglists
· Knowledgebase
· Trouble Tickets
· Files
· Glossary
· Links
· Compatibility Lists
· Forums
Welcome to our website
To take full advantage of all features you need to login or register. Registration is completely free and takes only a few seconds.
"Hyperthreading Good!" -- Intel
Posted by: Jim on: 08/28/2001 11:17 PM [ Print | 17 comment(s) ]
Chris was kind enough to throw this tidbit my way. C|Net is reporting that Intel is preparing to unleash it's 'hyperthreading' technology on the world. Two processors in one?
Technically, hyperthreading takes advantage of additional registers--circuits that help manage data inside a chip--that come on existing Pentium 4's but aren't used. Through these registers, the processor can handle more tasks at once by taking better advantage of its own resources. The chip can direct instructions from one application on its floating-point unit, which is where the heavy math is done, and run parts of another application through its integer unit.Yay?
A chip with hyperthreading won't equal the computing power of two Pentium 4's, but the performance boost is substantial, Poulin said. A workstation with hyperthreaded Xeon chips running Alias-Wavefront, a graphics application, has achieved a 30 percent improvement in tests, he said. Servers with hyperthreaded chips can manage 30 percent more users.Click here. Read stuff.
« Accelenation digs into Tyan's Tiger MP! · "Hyperthreading Good!" -- Intel
· Intel News Galore »
Comment
purice Registered User Posts: 68 Joined: 2000-11-12 |
![]() so is this a new name for SMT or what? tyan thunder 2 supermicro p6dls msi 694d asus a7m266d |
Comment
BBA Resident Modding Freak Posts: 1390 Joined: 2000-02-12 |
![]() It sure sounds like a reason for major software makers to make all software SMP capable though. I like the idea...Maybe AMD will compete with something better ![]() BBA |
Comment
The Emperor The dark side of smp Posts: 116 Joined: 2000-09-17 |
![]() Better than smp? Blasphemy!!! To live is to die. |
Comment
Vuke69 Bitpimp Posts: 341 Joined: 2001-03-16 |
![]() So what happens when you have dual, or quad, hyperthreaded processors? Sounds like fun. Did anyone else notice in the article that they said Intel showed off a 3.5 GHz P4? I think we'll see 4GHz sooner than anyone thinks. Moore's law says 9 months, I bet It'll be more like 6, depending on marketing strategies of course. Right arround the same time Intel starts showing off its 7 Ghz Chip. I personally am not that fond of AMD, but I am realy glad they are finally giving Intel a run for their money. Competition is good for all of us. It means better products, sooner, and for less cash. I'm a capitalist pig, and love it. |
Comment
Jizzler Multi-Core Zealot Posts: 2040 Joined: 2000-10-28 |
![]()
He he ![]() |
Comment
Jizzler Multi-Core Zealot Posts: 2040 Joined: 2000-10-28 |
![]()
Yeah, maybe Intel will stop being such -censor- and start enabling SMP in their lower end CPU's again ![]() |
Comment
Jig Registered User Posts: 114 Joined: 2000-09-11 |
![]() i noticed the 3.5GHz. sheesh. |
Comment
Cory <font color="#E22000">Do you like my hat? It's made of MONEY!</font> Posts: 413 Joined: 2000-05-19 |
![]() woo! 3.5 GHz. That should run Outlook and IE really fast, no? Jim_ and I were sitting around over here tonight, and I'm like, "Hey you wanker, boot that box under the desk up." So 10 minutes later, we're sitting there with Mozilla, mIRC, WinAMP, etc. running (under Windows 95 no less) on this Pentium 250 MMX (83 MHz FSB, ph33r) with 64 MB of EDO RAM and a gay 6 GB, 5400 RPM hard drive, and Jim_'s like "Hey, good enough for me." I agree. Unless I'm gaming, I don't need anything more than that. 3.5 GHz? riiiight. [url="http://www.2cpu.com"]www.2cpu.com[/url] [url="http://www.cory-grimster.org/boxen.html"]Boxen[/url] |
Comment
DJS Your Reliable Partner Posts: 364 Joined: 2000-06-19 |
![]() Kinda amusing headline at Electic the other day...I think they were serious. "Industry Fears old PC's are still good" |
Comment
Vuke69 Bitpimp Posts: 341 Joined: 2001-03-16 |
![]()
I cant wait for that day to come again. I have a BP6 with dual cel 466s, if it were not for that board, and celerons being SMP enabled, I probably still be one of those wusses with only one CPU. |
Comment
spoozilla SMP Misfit Posts: 47 Joined: 2000-12-05 |
![]() Has there been any mentioning if the upcoming McKinley processor is going to utilise SuperHyperMegaUltraThreading - X ![]() That could be rather 'juicy(tm)' 16 way 64bit loving, capable of 32 simultanious instructions. ![]() Spooky www.int13h.com Evil Genius... The most merciful thing in all the world is the inability of the human mind to correlate all of it's contents. |
Comment
BlueCoder Registered User Posts: 109 Joined: 2000-04-22 |
![]() I finally get SMT. It allows another thread to run while another thread is waiting on memory. Cache intensive applications like dnet MD5 wouldn't benefit at all though. But what about SMP hyperthreading? How does that work? How does the operating system manage more than two processors with hyperthreading? In a single processor system you can just let it look like two processors. But with SMP systems not all the virtual processors are equal and hence not symetric. There would need to be some extra logic to handle hyperthreading to provide the means for the OS to balence threads in cooperation with applications beyond current API's. |
Comment
netnic30 Hardware Hacker Posts: 9 Joined: 2001-08-19 |
![]() As long as the code is correctly written to support 2 way smp the same code should work correctly with N way smp. Although from the description of this this sounds like a lightweight k thread implementation. |
Comment
BlueCoder Registered User Posts: 109 Joined: 2000-04-22 |
![]() Exactly. Light weight threads. So the scheduling requires more logic. Prioritizing neglected threads for instance. Seems like a pretty complicated balencing act to me compared to real SMP. Managing processors of different speeds is one thing but this is far worse as I don't see how the threads are gauranteed processing with hyperthreading. Seems to easy for a one thread to hog the processor on moment and then the other thread hods it. Not symetric at all. |
Comment
netnic30 Hardware Hacker Posts: 9 Joined: 2001-08-19 |
![]() It turns out that lite threads are just context limited threads limited register usage memory and so forth. Keep in mind that this would not be a general threads package but one more used by the o/s and well designed API's available to the user. Very doable "OpenVMS AXP" has them today, so it is possible |
Comment
BlueCoder Registered User Posts: 109 Joined: 2000-04-22 |
![]() As I expressed. Requires a new API to use it.It's difficult to express the concept. Maybe a better term would be simease threads (at least while it's running); one and seperate simutaniously. One thread could thouretically hog the processor one moment and then could share equally the next. So there is no gaurantee of symetry. The answer is the OS would prioritize one simese thread so that it gets priority and load the next schuled thread and let it let it get a head start on the processor. Just loading the context while the processor is running should improve performence with systems running many threads. But apart from the new scheduler there would be Apps specifically coded to thread in pairs so the processor wouldn't have to worry about symetry or balencing the processing. |
Comment
PsycoBob[Q2] 5w LED caselight? Mmm... Posts: 726 Joined: 2001-04-04 |
![]() Hyperthreading might make P4's even with Athalons (you can tell I don't like AMD- I can't even spell their processor names...), clock fer clock. If I understand it right, you'd get a huge performance boost out of having a larger cache, so hyperthreaded Xeons would kick ass. Is this like nVidia, releasing those kick-ass drivers right when ATI announced the Radeon2 test cards were being sent out? Make poor-performing P4's, but simply software-disable the hi-performance features that's make them smoke Athalons..... then announce how to turn them on when your stock price needs a boost. ![]() I like the idea of everyone with P4's suddenly having a nice 'Gift' on the windows update page- a free 20% performance boost. Damn. Oh well, on to plan B I guess... <sets fire to head> You're not paranoid if they're REALLY after you. Errors? The cause of the problem is Interference between the keyboard and the chair. |