3.06Ghz Pentium 4 to support HYPERTHREADING!!!

[SSXeon5]

Originally posted by: dexvx
Just like you made the thread that said that was a picture of a Prescott when it was in fact a Tidewater (or a new form factor P4, i forget)?

When there is a question mark after something are those concrete facts? Maybe you should go reread that thread ... here you go:

Intel Prescott (.09um) Pictures! (UPDATE)

SSXeon

 

[yodayoda]

Originally posted by: SSXeon5

Originally posted by: dexvx
Why is SSXeon is interested in Prescott? Is that your name or something? You've posted a lot of "news" about Prescott and about 90% of them happened to be wrong.

Way to be an ass

And I do have facts that they were running the prescott @ 4ghz if you want me to bring the thread back up...

SSXeon

what's the point? your name is Xeon, you do nothing but rant about P4 systems, hyperthreading, and stratospherically priced future Intel chips, yet you are sporting a great relic of 20th century computer architecture, a 100% intel compliant (wtf?) Pentium !!!. put your money where your mouth is and buy some of the horsesh*t you are trying to convince others to purchase.

 

[SSXeon5]

Originally posted by: yodayoda

Originally posted by: SSXeon5

Originally posted by: dexvx
Why is SSXeon is interested in Prescott? Is that your name or something? You've posted a lot of "news" about Prescott and about 90% of them happened to be wrong.

Way to be an ass

And I do have facts that they were running the prescott @ 4ghz if you want me to bring the thread back up...

SSXeon

what's the point? your name is Xeon, you do nothing but rant about P4 systems, hyperthreading, and stratospherically priced future Intel chips, yet you are sporting a great relic of 20th century computer architecture, a 100% intel compliant (wtf?) Pentium !!!. put your money where your mouth is and buy some of the horsesh*t you are trying to convince others to purchase.

CLICK HERE and order a box of midol .....

SSXeon

 

[Duvie]

After reading the link from sonoran it does appear that the programs need to be compiled to take advantage og the hyperthreading technology...I wonder how long that will take given most things are yet to have sse2 encoding....

Also in the article when the guy brought the 3ghz to its knees by watching a high detail movie in windows media player and encoding an audio file....That seemed strange to me!!! The movie was jerjy to point of not watchable....Is this merely a trick of setting the audio file to high priority??? I mean I watch movies (divx with take a lot of cpu cycles due to the compression factor) and encode and while connected to the internet and the movie is perfect to watch...What is different??

 

[dexvx]

Thats odd... your Intel system doesnt include an Intel NIC, wtf is up with that? AMD vs Intel you can argue over, but Intel NICs are #1

And FYI, I read that thread before you started to edit things.

 

[SSXeon5]

Originally posted by: dexvx
Thats odd... your Intel system doesnt include an Intel NIC, wtf is up with that? AMD vs Intel you can argue over, but Intel NICs are #1

And FYI, I read that thread before you started to edit things.

Um wtf are you talking about, take my link to midols site and quiet down. And I edited what I have under edit, not what i said before

SSXeon

 

[sonoran]

Originally posted by: Duvie
After reading the link from sonoran it does appear that the programs need to be compiled to take advantage og the hyperthreading technology...I wonder how long that will take given most things are yet to have sse2 encoding....

Not quite sure where you got that impression. My understanding is that optimizing the code can maximize the potential gain from hyperthreading, but that hyperthreading makes a significant difference any time you have multiple programs (or threads) executing at once.

I watch movies (divx with take a lot of cpu cycles due to the compression factor) and encode and while connected to the internet

Not sure why their machine seemed slower than yours. Maybe they were using SDRAM? The way you're describing using your PC is exactly the type of usage for which hyperthreading was designed.

 

[Duvie]

multitasking maybe....but in the article the gentlemen is clearly pushing to the software guys to have them compile their programs to take advantage of this....

If hyperthreading works out of the gate in multitasking then it would be a great benefit to me...waiting until individual apps compiles later would be fine....


I am not sure when I would get a 3.06ghz cpu though....I rarely ever buy the latest and greatest...Buy the time I want to spend the cash on it, p4's will likely be in the 3.5-3.7 range!


too bad there wasn't a way to enable it in the older chips!!!

 

[Kell]

Originally posted by: sonoran

Originally posted by: Duvie
After reading the link from sonoran it does appear that the programs need to be compiled to take advantage og the hyperthreading technology...I wonder how long that will take given most things are yet to have sse2 encoding....

Not quite sure where you got that impression. My understanding is that optimizing the code can maximize the potential gain from hyperthreading, but that hyperthreading makes a significant difference any time you have multiple programs (or threads) executing at once.

I watch movies (divx with take a lot of cpu cycles due to the compression factor) and encode and while connected to the internet

Not sure why their machine seemed slower than yours. Maybe they were using SDRAM? The way you're describing using your PC is exactly the type of usage for which hyperthreading was designed.

If an app isn't hyperthreading-optimized, enabling hyperthreading might actually hurt performance. AnandTech's last two server benchmarks demonstrated that, though I can't be arsed to dig them up atm.

 

[mechBgon]

I was wondering if I'd be the first to point out that in Anand's real-world tests, HT doesn't necessarily increase the performance, but Kell beat me to it. Here's a quote from the first article, where HT gave a 1.5% boost in single-CPU performance, but a 5% drop in dual-CPU performance:

While we started off the analysis on the previous page by talking about the single CPUs and moving on to the dual configurations, we can't do the same here. If you look at the results of the older 1.7GHz Xeons you'll quickly realize that a single Athlon MP 2000+ can outperform the pair in this test. Once again, the newer Xeons manage to perform just about on par with the Athlon MP 2000+. The 2.2GHz part is 1.3% faster and enabling Hyper Threading gives it a 1.5% improvement, both not enough to truly separate the two.

Looking at the faster dual CPU solutions, the dual Athlon MP 2000+ is running on par with the dual 2.2GHz Xeon platform. What's interesting is that there is actually a 5% reduction in performance when Hyper Threading is enabled with two physical CPUs installed in the system. Although this isn't as huge of a hit as is possible when enabling Hyper Threading in desktop situations, the fact that it did occur makes it clear that although Intel recommends turning it on for servers you really have to look at the type of transactions your server is handling before making the final call on the feature. Luckily a 5% penalty is small enough that if accidentally enabled, it won't cause too much harm but it's something to be aware of nevertheless.

Looking at the results for a single-CPU system with HT off versus on, best-case results in Anand's tests were about 12% improvement, which is a good thing but still leaves them trailing the AthlonMPs in this type of application. Worst-case results for single-CPU systems were gains of about 1.5%. It would be interesting to see how HT can do in a suite of typical home-computer enthusiast apps.