Hyperthreading moves into hyperspeed throughout 2003, with the Canterwood and Springdale chipsets forming the backbone of Intel’s desktop roadmap. According to Intel roadmaps seen by ExtremeTech, Intel’s desktop processors will extend up to and possibly beyond 3.6-GHz by the beginning of 2004, with Celerons cresting 2.5-GHz by the same timeframe. Await soon an article regarding hyperthreading, here on OSNews.
Hyperthreading Moves Into Mainstream In 2003
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
And here I am wondering if I should upgrade my 500Mhz Pentium 3 to 1Ghz..
And here I am wondering if I should upgrade my 500Mhz Pentium 3 to 1Ghz..
I have a PIII – 450 and love it. It performs great for all tasks that I need to do. I have also considered greatly upgrading to a PIII – 1.2 GHZ. It will just be difficult since I bought one of the first PIII that ran socket 1.
Pros:
– No need for a new motherboard
– Low power consumption
– The PIII actually outperfoms the PIV MHz to MHz
Wow…
Looks my P4 2.8GHz with 845PE will be old soon
And with every tiny, incremental enhancement of CPUs these days, fewer and fewer people upgrade. I’m saving up for that 8 CPU SMP system with 4 GB RAM. Maybe then I’ll be able to compile my kernel even FASTER!
Having threading hardware on the chip essentially cements native threading as the way to get better performance in your application.
HT will be a powerful driver for single-threaded GUI’s to upgrade their code to be thread-friendly.
Overall, HT is great technology with the great upside that it will give developers an incentive to write better code.
I guess if I want that ‘Hello World’ program I show at the beginning of every semester when I teach a Java 101 course … I better make it threaded for better performance
(though, how would that work? ‘hello’ in one thread and ‘world’ in another … lol)
-Micheal
looks @ Intel Roadmap
looks @ AMD Roadmap
looks @ PowerPC970
– looks @ own Athlon that’s is slowly dying (Athlon 1.2 @ 1.4, now down to 1.3 after stability problems).
– looks @ bank account
Moore’s law hurts the wallet.
Ah well… Hopefully all the software including the OS itself can actually utilize all the extra powah!
Well If you were doing a GUI Hello world in Beos it would be 3 threads i belive, Maybe this is true for any platform if you make the effort.
before any program start.
IIRC:
-one for the garabage collector
-one for compiling to platform
-one to maintain all activities
Can you put multiple Pentium III-M’s onto a single ATX motherboard? I’ll take one of those, or a quad VIA C3 (if it were possible).
I read somewhere that Intel won’t be as clock speed oriented this year as it was last year. I think they will be spending some time getting more out of the chip in terms of tweaking hyper threading, FSB speeds, CPU cache etc.
Prescott should be released with the following changes
90 micron process (Northwood is 130).
13 new instructions (2 are HT specific)
Improved Hyper Threading (see above)
800MHz FSB (from 533)
16KB L1 cache (was 8KB)
1MB L2 cache (was 512)
I don’t know what clock speed it will be released at. My guess is at least 3.4GHz.
You serious … is this for home/desktop use or for server (like a sub xeon processor)?
I would like to see a review between an Intel HT processor and an SMP Athlon system. I imagine that you could setup these systems at a similiar price-point, and I’d love to see the comparison.
The fact that I can spend thousands on a dual-xeon setup doesn’t excite me.
Intel has had hyperthreading in their Xeon processors for a long time. It’s not a guaranteed performance boost, especially on older apps. However, new apps that have been tuned for hyperthreading pick up a substantive boost.
What will be more interesting than 1 HT Intel chip vs. 2 full AMD CPU’s will be how the Barton-based 512k cache Athlon MP’s stack up against the current Xeons.
After Pentium moves to 1MB on-chip cache and an enhanced version of hyperthreading, you’ll see the same technology on Xeon. This is to compete with AMD’s Hammer which also offers 1Mb cache.
As for your original comparison, you’ll probably find the usual results for comparing a single high-speed processor vs. a dual system. The single processor system will win many benchmarks, but the dual system will feel smoother in actual usage. You may also find that the AMD-based dual is less reliable than an Intel chipset system.
–ms
Which mainstream are we talking about ? According to me, that word means that hyperthreading will be available to those who are not willing to pay more than $150 for a processor.
Call me cheap but I can’t picture myself coughing up $600 for something that will help me encode video in 10 minutes instead of 30.
If VIA were to make a 16, 32 or 64 way system for dirt cheap with those C3 processors (modified to do SMP of course) they’d have a killer product. At 6-7W each, that’d be still less power than the dual MPX systems we have here at work.
If you could get a whole system for 2-3k it would definitely make a name for VIA in the system market.
For now I’ll got buy a dually MPX based AMD system and still be cheaper (and more useful) than the single CPU Intel system.
Athlons are the best processors as far as I’m concerned. Our dual cpu Athlon 1800 systems blow our dual 2.4 Ghz P4 Xeon systems out of the water. And that’s with taking full advantage of the hyperthreading on the xeons, as much as Linux can anyway. Mind you all our jobs consist of mainly floating point computations. And maybe things will change as soon as Linux gets better support for hyperthreading in both compilers and the kernel. But for now, Amd is way to go.
“I would like to see a review between an Intel HT processor and an SMP Athlon system. I imagine that you could setup
these systems at a similiar price-point, and I’d love to see the comparison.”
Yes I would like to see the same tests, but have you noticed that all those uber reviews on TomsHW and all the others all run the exact same stupid single threaded single tasks that I don’t or never will run. The only relevant test might be mp3 coding, but I wouldn’t be doing that full time test mode but as a background op.
It would be in Intels HT & AMDs SMP interests if they really want to push these, to get Tom Dick & Harry to come up with more meaningfull tests to show the benefit of multithreading. Like say running VMWare and having big jobs run on 2 OSs simultaneously, that would be a million times better than running 5000fps of some idiotic game.
The fact that I can spend thousands on a xGHz P4 setup doesn’t excite me either, 2x speed ups isn’t what it once was.
But the company I work for should be buying these fastest machines for engineering, but as often as not the secretary is more likely to have a faster PC than the engineers.
“If VIA were to make a 16, 32 or 64 way system for dirt cheap with those C3 processors (modified to do SMP of course)
they’d have a killer product. At 6-7W each, that’d be still less power than the dual MPX systems we have here at work.
If you could get a whole system for 2-3k it would definitely make a name for VIA in the system market.”
I had been thinking the same thought, but then again how much time does any one person really spend in an app that can be threaded over more than a very few cpus. The C3 would probably not make a good SMP design, its a basic no frills FPU poor design, their engineering effort is tiny in comparison to Intel/AMD.
Those miniITX boards could probably make a nice little integer cluster box for running “some” distributed apps. Their FP sucks so that limits it right away. Their small cool size would make packaging pretty easy, each with own 12V supply powered off a medium ATX PSU unit.
If I had 4 in a box, I might keep 3 or 4 OSes live at same time on a KVM switch, but then again the MPX with VMWare would probably blow that away.
With my SMP AMD setup I have really enjoyed the advantage in CPU intensive processes like DVD-Divx, Ogg encoding, running CubaseSX and Video editing. Now with Intels push of HT which is just SMP on chip, I will be getting Il2 FB taking advantage of SMP so I can now enjoy flying my WW2 simulation and have it run smoothly with all features like graphics and AI bumped up to the max.
SMP is great as brough to us by the likes of Be Inc. Now if only MS and other OS providers gave us SMP utilisation along those lines the consumer would be laughing. Nah, I don’t think it is in their interest.
“I had been thinking the same thought, but then again how much time does any one person really spend in an app that can be threaded over more than a very few cpu….”
A design like this would be more interesting for a time share system (usermode linux anyone?) or for some heavy duty IO intensive processing that could be trivially parallelized, or as an awesome test machine for OS research (Hurd comes to mind).
Scalable SMP interconnects (anything NOT intel) are way faster and more efficient than network interconnects. In this case a bunch of trash processors (ie, C3) in parellel would perform well against a single HT or dual athlon processor.