• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Intel Penryn will use HyperThreading

zekrahminator

McLovin
Joined
Jan 29, 2006
Messages
9,066 (1.28/day)
Location
My house.
Processor AMD Athlon 64 X2 4800+ Brisbane @ 2.8GHz (224x12.5, 1.425V)
Motherboard Gigabyte sumthin-or-another, it's got an nForce 430
Cooling Dual 120mm case fans front/rear, Arctic Cooling Freezer 64 Pro, Zalman VF-900 on GPU
Memory 2GB G.Skill DDR2 800
Video Card(s) Sapphire X850XT @ 580/600
Storage WD 160 GB SATA hard drive.
Display(s) Hanns G 19" widescreen, 5ms response time, 1440x900
Case Thermaltake Soprano (black with side window).
Audio Device(s) Soundblaster Live! 24 bit (paired with X-530 speakers).
Power Supply ThermalTake 430W TR2
Software XP Home SP2, can't wait for Vista SP1.
Penryn, the 45nm successor to Conroe on the Intel roadmap, will have several new features the previous lacked. One of these is Intel HyperThreading, which would give a computer four logical cores on a dual-core processor. Intel's Penryn may also have up to 6MB of L2 Cache, and other than the die shrink, will have a lot of minor performance tweaks.


View at TechPowerUp Main Site
 
Last edited by a moderator:
i believe hyperthreading is just a hackjob to help windows along...when i had my northwood ht i just kept it disabled...
 
Im guessing this will be a more refined version of HT. I actually noticed the difference between a P4 with HT and one w/o
 
It doses not actually create 2 extra logical cores by using hyperthreading just parts of threads can run simultaneously so its faster than 2 cores but slower than 4 cores.
 
And the O/S and software must both support hyperthreading to actually see a performance gain. Hardly any mainstram software is effectively written for hyperthreading at this stage.
 
Anything written to take advantage of multiple cores will take advantage of HT. So there is actually a lot of stuff out there that is written for it at this stage.

I personally don't see the need for this. Back in the P4 days HT was very nice to have, it gave many of the advantages of having multiple processors without the huge cost of a multiple processor setup. However, now that dual-cores are very affordable, I don't see a need for HT on multi-core processors.
 
Actually there are hardly much software that support dual core, multi-core is simply in your dreams. LOL
 
Actually there are hardly much software that support dual core, multi-core is simply in your dreams. LOL

My system currently runs 49 processes, that's what Windows shows anyhow. Add to that that there is plenty multithreaded software available, there has been for many years.

I think it's a good think for Hyperthreading to return, on the contrary of what some think it does make a nice difference. I've been using Xeons with Hyperthreading for quite some time and compared to just 2 cores 2 ores with Hyperthreading is quite noticeable under heavy load.
 
I knew HT would be back for the dual core. Then HT quad cores. Next thing will be Core4duo2 HT Extreem Edition... :wtf:
 
My system currently runs 49 processes, that's what Windows shows anyhow. Add to that that there is plenty multithreaded software available, there has been for many years.

I think it's a good think for Hyperthreading to return, on the contrary of what some think it does make a nice difference. I've been using Xeons with Hyperthreading for quite some time and compared to just 2 cores 2 ores with Hyperthreading is quite noticeable under heavy load.

Yes but all those processes are not using system resources, certainly not at the same time otherwise there would be an argument for having a chip with 49 cores!I have a single core and in 3D Mark 2006 for example the difference in score between leaving all processes open and closing almost all of them is negligable but to be fair there is a slight improvement, whether that improvement warrants the outlay of upgrading the CPU however I very much doubt.
 
I guess Intel never really wanted to get rid of Hyper threading. This could be cool. A cheap alternative to quad core :)
 
at school they use P4 with HT and w/o and the ones w/o run a lot slower starting up and compiling programs i think its good tech but thats just me
 
at school they use P4 with HT and w/o and the ones w/o run a lot slower starting up and compiling programs i think its good tech but thats just me

HT was primarily created to not only begin the transition to dual cores, but take advantage of the deep pipeline in a P4, because flushing it was so costly, being able to swap to another thread while being flushed could be.... effective.

I'm curious to see how the core 2 duo would take a handle on HT...

I am, however, doubting it would be nearly as effective as it is on a P4... but I'm sure it will be tweaked and tuned for the core 2 rather, anyways.

Will be interesting, having heard the AMD side of rumors to things like this including "reverse HT" and possible HT flags enabled.
 
Will be interesting, having heard the AMD side of rumors to things like this including "reverse HT" and possible HT flags enabled.

But that's Hyper-Transport, not Hyper-Threading...
 
HyperTransport is a type of bus system made by AMD to allow faster connections between the RAM and core while HT is Intels core technology that allows multiple threads to run semi-simultaneously.
 
Yes but all those processes are not using system resources, certainly not at the same time otherwise there would be an argument for having a chip with 49 cores!I have a single core and in 3D Mark 2006 for example the difference in score between leaving all processes open and closing almost all of them is negligable but to be fair there is a slight improvement, whether that improvement warrants the outlay of upgrading the CPU however I very much doubt.

There is a very big difference when you are doing something else in the background and try to run 3DMark06. If I disable one of my cores on my Core 2 Duo and try and encode a DVD and run 3Dmark06 my 3Dmark06 score is drastically reduced, and when playing games I can actually feel the affects(and yes I have done this test to prove to a friend that dual cores are useful). However, with both cores enabled the DVD encoding is hardly noticeable in both the 3Dmark06 score or actual gameplay.
 
There is a very big difference when you are doing something else in the background and try to run 3DMark06. If I disable one of my cores on my Core 2 Duo and try and encode a DVD and run 3Dmark06 my 3Dmark06 score is drastically reduced, and when playing games I can actually feel the affects(and yes I have done this test to prove to a friend that dual cores are useful). However, with both cores enabled the DVD encoding is hardly noticeable in both the 3Dmark06 score or actual gameplay.

In order to use the other core to its fullest potential programs need to be built with multi threading in mind. This allows the CPU to split up the threads and run them simultaneously. So you will not notice a difference unless the program is multithreaded.
 
In order to use the other core to its fullest potential programs need to be built with multi threading in mind. This allows the CPU to split up the threads and run them simultaneously. So you will not notice a difference unless the program is multithreaded.

You don't seem to understand what I am saying. Windows and most OSes are built to use multiple processors/cores/threads. So while you might be running two programs that aren't build with multi-threading into them, the OS will split the work load up between the two cores.

Say for instance you have 2 CPU intensive programs(3DMark06 and DVDShrink). Both use just one thread. You start them both running at the same time. You now have 2 CPU intensive threads running simultaneously. One thread will then use one core, and the other thread will then use the other core. Neither program is designed to use multiple threads, but both will recieve the benefits of having a mutli-threaded capable system and the difference will be noticeable compared to doing the same thing on a system that isn't multi-threaded capable.
 
There is a very big difference when you are doing something else in the background and try to run 3DMark06. If I disable one of my cores on my Core 2 Duo and try and encode a DVD and run 3Dmark06 my 3Dmark06 score is drastically reduced, and when playing games I can actually feel the affects(and yes I have done this test to prove to a friend that dual cores are useful). However, with both cores enabled the DVD encoding is hardly noticeable in both the 3Dmark06 score or actual gameplay.

Thats surprising when you say with both cores enabled and running a DVD also the difference in score in 3D Mark 2006 is hardly noticable since 2006 is VERY dual core biased, I would have thought that if only one core was working on it then the score would not be much better than a single core score.
 
You don't seem to understand what I am saying. Windows and most OSes are built to use multiple processors/cores/threads. So while you might be running two programs that aren't build with multi-threading into them, the OS will split the work load up between the two cores.

Say for instance you have 2 CPU intensive programs(3DMark06 and DVDShrink). Both use just one thread. You start them both running at the same time. You now have 2 CPU intensive threads running simultaneously. One thread will then use one core, and the other thread will then use the other core. Neither program is designed to use multiple threads, but both will recieve the benefits of having a mutli-threaded capable system and the difference will be noticeable compared to doing the same thing on a system that isn't multi-threaded capable.

OK, I get your meaning so tell me, in some benches/tests why dual cores dont outperform single cores? I was always led to beleive that it dont matter how many cores you have, if there is only one thread available in the app at a time then its only one thread that is going to be processed at a time. Take 3D Mark 2005/NBench/ScienceMark/Aquamark and the list goes on, its raw speed that wins the day there, in some cases system performance also, but using your theory would mean that a dual core would normally win the day because if they are processing 2 threads for every one a single core can manage then thats twice the speed surely?
 
You still seem to be missing what I am saying. If you just run a single benchmark by itself it isn't going to show a real performance difference. However, it is when you are running more than one program that it gets noticeable.
 
You still seem to be missing what I am saying. If you just run a single benchmark by itself it isn't going to show a real performance difference. However, it is when you are running more than one program that it gets noticeable.

The main problem these days is going to be who gets to play the balancing act.

Are AMD/Intel going to try to delegate it to hardware? i.e. HT style? perhaps a custom reverse HT to allow multiple CPU's to work on one thread?

Is microsoft going to try to be the slave driver and delegate work and play teedertodder?

Thats whats going to make a difference in the future.

OS X handles multiple cores beautifully, XP has made great strides.

Now wheres all the multi threaded windows games? I've got my mac versions.. :banghead:


But that's Hyper-Transport, not Hyper-Threading...

I was talking about hyperthreading. Nowhere in my post did I reference HTT.
 
XP already handles deligation, and Vista is even better at it. If your Mac games are multi-threaded and exist on Windows also, then they Windows version is almost certainly multi-threaded.
 
Back
Top