The Pentium D? (Pentium Different)

[@md0Cer]

Info from the INQ about the new Pentium Different!

http://www.theinquirer.net/?article=22306

Now this information could be true, but I just see this as another pile of bull feces plopped down by the INQ to have something to fill those news pages.

Pentium D XE. The "D", we're told, stands for different,

I was under the impression the "D" stood for "desktop" or even perhaps "dual" as in "dual cores." It fits quite nicely with the Celeron. You have a Celeron M and D, and you have a Pentium M and D. The Pentium 4 then works kinda like a middle product like the Athlon XP did for AMD between the K8's (Athlon 64 and FX) and the Duron. I find it quite hard to beleive Intel would call this the "Pentium Different," or even if so, I find it hard to beleive they would tell the INQ or anyone else as it sounds incredibly stupid. How does this sound "Dude! I'm gettin a Dell with a Pentium Different!"

Intel Pentium D XE criticised for poor hyperthreading

Now a chap has claimed online that overall performance for the Pentium D XE is poorer with HT turned on for this chip than when two separate CPUs are used.

He wonders if Microsoft has plans to patch the XP threads scheduler for dual core HT CPUs

I can't see any proof of this. Why would the hyperthreading be lacking anyway? It would give you a gain, but you do have two cores already.

The INQ shows absolutely no evidence of how it is "lacking." In the link to the so called "online chap," "online chap" just says he/she is "playing" with one and "noticed" it was lacking. No evidence shows that the performance is lacking. From what I get out of the "online chap's" newsgroup post or whatever that may be, it seems as though he says the OS has trouble associating and sending stuff on its way to 4 CPU's. If I am not mistaken, no one has really complained about it with quad CPU systems or SMP Xeon's with HT. Now it is possible the HT in these CPU's is flawed, but unlikely considering Intel knows what they are doing and the HT shouldn't really have changed.

Anyway, thats my take on the article about the "Pentium Different" What's yours?

 

[VolcanoWarez]

that names does suck.. I don't see how hyper-threading could slow you down, I do think Intel should have implemented HT on all duals, and maybe increased the FSB and cache of the EE to give it the edge.

 

[@md0Cer]

that names does suck.. I don't see how hyper-threading could slow you down, I do think Intel should have implemented HT on all duals, and maybe increased the FSB and cache of the EE to give it the edge.

Agreed. I was just commenting on how the Inquirer is very likely full of **** on this article as it is unlikely that:

1.) The Hyper-Threading is defective on Extreme Edition Pentium D's or from that newsgroup post, the OS can't associate efficiently with the 4 logical cores (at least enough to slow it down)

2.) Intel would intend the "D" to stand for "different." It is however, quite likely that like the Celeron, it stands for "desktop" or "dual."

Anyway, my take on the article is this is complete BS.

As for the XE, I think both of your ideas are good, but may not be likely at the moment. I beleive the FSB is about as fast as you can go already without using insane ram (for a stock CPU), and possibly already pushing how much the chipset can handle (already has to take 2 CPU's worth, but im not an expert on this). As for cache, if they did 2mb per core, that may pose quite a heat issue as you have that many more transisters adding on to an already hot CPU.

 

[illogical06]

I was under the impression the "D" stood for "desktop" or even perhaps "dual" as in "dual cores." It fits quite nicely with the Celeron. You have a Celeron M and D, and you have a Pentium M and D. The Pentium 4 then works kinda like a middle product like the Athlon XP did for AMD between the K8's (Athlon 64 and FX) and the Duron. I find it quite hard to beleive Intel would call this the "Pentium Different," or even if so, I find it hard to beleive they would tell the INQ or anyone else as it sounds incredibly stupid. How does this sound "Dude! I'm gettin a Dell with a Pentium Different!"

If it really is Pentium "Different" then whoever thought of that should be shot on sight...

I can't see any proof of this. Why would the hyperthreading be lacking anyway? It would give you a gain, but you do have two cores already.

http://www.legitreviews.com/article.php?aid=188

This is some pretty good evidence from a well respected tester. But of course, one source alone is not to be trusted.

1.) The Hyper-Threading is defective on Extreme Edition Pentium D's or from that newsgroup post, the OS can't associate efficiently with the 4 logical cores (at least enough to slow it down)

Valid point, and probably the main reason...but the hit it takes is pretty ridiculous with HT enabled.

 

[Stratcat]

Agreed. I was just commenting on how the Inquirer is very likely full of **** on this article as it is unlikely that:

1.) The Hyper-Threading is defective on Extreme Edition Pentium D's or from that newsgroup post, the OS can't associate efficiently with the 4 logical cores (at least enough to slow it down)

I've got to respectfully disagree, concerning the possibility of an OS compatibility issue.

The OS having thread scheduling probs might not be as far fetched as it seems. Windows 2000 Pro is certified for (2 cpu) SMP support, but not SMT support (single cpu, or otherwise). This means its certified fine for dual procs, but not certified for an HT'd enabled proc. And this has been borne out in the real world. W2KPro will work w/HT, but in some instances it works fine, and in others, HT will actually cause a significant performance loss. Yet W2KPro will perform fine in a non-HT, but dual-proc environment.

I've heard this repeatedly from I.T. guys, including one whose opinion I highly respect. His advice: Always disable HT on W2KPro due to its propensity to cause a performance loss w/some applications. This subject of W2KPro having SMP, but no SMT certification, has come up in another forum I frequent, which has a lot of I.T. guys in it, also with the same aforementioned advice garnered from experiences working in the real world.

So yes, even though XP handles single cpu SMT well, and also dual cpu SMP, I believe it is possible that XP, as it currently stands, could have specific performance issues regarding multi-proc SMT, or more accurately, multi-core SMT. But, in all fairness, and to play Devil's advocate to my post, I'd expect to see the same issues when running XP & dual Xeon Noconas that are HT enabled, too.

I'm not saying this is absolutely happening, but do believe it is not outside the realm of possibility, perhaps occuring only under specific apps & conditions.

Strat

 

[illogical06]

So yes, even though XP handles single cpu SMT well, and also dual cpu SMP, I believe it is possible that XP, as it currently stands, could have specific performance issues regarding multi-proc SMT, or more accurately, multi-core SMT. But, in all fairness, and to play Devil's advocate to my post, I'd expect to see the same issues when running XP & dual Xeon Noconas that are HT enabled, too.
Strat

Thats what i was saying in lehman's terms....or at least attempting to. I personally don't believe that its defective, i meant that the OS isnt constructed to handle it but i wasnt clear.

 

[Sjaak]

2.) Intel would intend the "D" to stand for "different." It is however, quite likely that like the Celeron, it stands for "desktop" or "dual."

Are you kidding? The D in pentium D stands for dual core