Intel 3.2 G Dual Core Performance Preview

[Duvie]

The did that at a site called trustedreview.com....only like 4 test but you can see the xeon with 533fsb bus at 3.2ghz beat the EE and regular P4-D....This thing not only loses to 800fsb prescotts, it will likley also lose to single core p4 northwoods and prestonia xeons in single non multithreaded apps...

They seriously need more bandwidth...edited..better yet no shared memory bus

notice the 4200 in sandra at HardOCP and they said Intel was saying 4700mb/s should be the norm...This was below that of current prescotts at same speed......

 

[mamisano]

I am wondering, when HT came out back before XP took off, many folks had problems with slower Win2K performance. Will there need to be a patch to XP, maybe the scheduler, to better support Dual Core HT, or does this not matter as we already have Dual Xeon HT rigs out there?

My guess is that with 4 virtual cores, the bus simply can not supply data fast enough.

Also, with the crossbar utilized by Intel to connect the 2 cores to the bus, there HAS to be some increased latencies. I wonder if anyone will run some latency tests on these new chips.

It is going to be an interesting Q2 with Dual Cores, and ATI R520 announcement

 

[Markfw]

Originally posted by: Duvie
The did that at a site called trustedreview.com....only like 4 test but you can see the xeon with 533fsb bus at 3.2ghz beat the EE and regular P4-D....This thing not only loses to 800fsb prescotts, it will likley also lose to single core p4 northwoods and prestonia xeons in single non multithreaded apps...

They seriously need more bandwidth...edited..better yet no shared memory bus

notice the 4200 in sandra at HardOCP and they said Intel was saying 4700mb/s should be the norm...This was below that of current prescotts at same speed......

Exactly. I have been saying for weeks, that Intels Dual-core problem will be bandwidth. The AMD64/Opteron's have that all over the P4, and the core-core communications over HT will kill them in a true SMP application.

 

[imported_Ged]

Originally posted by: mamisano
I am wondering, when HT came out back before XP took off, many folks had problems with slower Win2K performance. Will there need to be a patch to XP, maybe the scheduler, to better support Dual Core HT, or does this not matter as we already have Dual Xeon HT rigs out there?

My guess is that with 4 virtual cores, the bus simply can not supply data fast enough.

Also, with the crossbar utilized by Intel to connect the 2 cores to the bus, there HAS to be some increased latencies. I wonder if anyone will run some latency tests on these new chips.

It is going to be an interesting Q2 with Dual Cores, and ATI R520 announcement

Picked this up somewhere. Linkage

Talks about a possible problem with the Windows Scheduler for HT enabled Dual-Core CPUs.

 

[Duvie]

Good find...interesting...I am goig to follow that....

Are we sure if anyone has been testing these on winxp pro or the home edition??? The xeons have been around with 4 possible cpus for awhile and this seems liek it would be a similar issue unless the OS is in fact having issues with what is a logical and physical cpu...

 

[Hyperlite]

Originally posted by: Jeff7181

Originally posted by: geforcetony
Well, seeing as how Prescott's didn't perform well at low speeds (below about 3.6GHz), with their insane 31-stage pipe, this isn't a surprise at all. The Prescott architecture needs high speeds to REALLY shine, and seeing as how Intel hit a brick wall as far as clock speeds go, its really pointless.

To me, the real "showing" of dual-core will be with a dual-A64, and dual-Pentium M (Dothan) setup. With dual-Prescotts, its just blah.

Good point... I wonder if they had to increase cache latency even more for dual cores...

Thats an excellent point, i hadn't even thought about that.

I was wondering about XP's ability to handle Dual core HT processors in another thread, i suggested that might be the reason why Intel left HT out of the initial release of the dual cores.

That was great review by AT, but i have one gripe. They should have included the FX55 in the last test Anand did with all the apps and switching back and forth.

Have there been any reports yet of what the clock speeds will be on the AMD DC releases?

 

[imported_michaelpatrick33]

If AMD releases a 2.4 dualcore aimed at the server and workstation market (where there is a bigger need and money to buy) how in the world will a 3.2 Prescott dual core even come close in performance even with hyperthreading. A single FX-53 compared to a Prescott 3.2 isn't even a contest. With AMD designed from the groundup (basically) for dualcore on the 64 line (using less power and producing less heat to boot) no wonder Intel is releasing (paperlaunches) a dualcore to John Q public. I imagine the server and workstation administrators are looking at the cost, heat, and power requirements (and new motherboards to boot) and laugh weakly with a nervous glance at the budgeting department.

 

[carlosd]

Now HT seems to be really oudated compared to dualcores performance, we just have to see how the Pentium D outperforms the EE in a lot of benchmarks, getting too little benefit from HT. I mean, dual cores CPU have really huge performance over the HT crap. Also I wasn't impressed with intel dual cores performance it seems that AMD will have much better dual core CPUs.

 

[jbh129]

Another winner from Intel. They are really great

 

[clarkey01]

Originally posted by: jbh129
Another winner from Intel. They are really great

FELIX !.


Please try and conribute more to the thread then a simple " Yep they won"

 

[Capt Caveman]

Originally posted by: michaelpatrick33
If AMD releases a 2.4 dualcore aimed at the server and workstation market (where there is a bigger need and money to buy) how in the world will a 3.2 Prescott dual core even come close in performance even with hyperthreading. A single FX-53 compared to a Prescott 3.2 isn't even a contest. With AMD designed from the groundup (basically) for dualcore on the 64 line (using less power and producing less heat to boot) no wonder Intel is releasing (paperlaunches) a dualcore to John Q public. I imagine the server and workstation administrators are looking at the cost, heat, and power requirements (and new motherboards to boot) and laugh weakly with a nervous glance at the budgeting department.

"Intel is releasing (paperlaunches) a dualcore to John Q public" - How is this a paperlaunch? People have the chip in their hands and providing reviews.

And I think comparing a 2.4 AMD dualcore is not apples to apples when it comes to this Intel 3.2 dualcore, since they are for two different markets. It's been mentioned the Intel 3.2 dualcore will be priced less than $100 more than a regular Intel 640. And obviously the 2.4 AMD dualcore will cost an arm and a leg based on the market. Plus, Intel will be releasing EE and server dualcores w/ HT. My guess is that this would be a nice feature having 4 logical cpus handling all the processes a server has to deal with it.

Hopefully, some more extensive reviews will be out. They all seem to be missing one thing or another.

Also, hope AMD will have some cpu's for review soon.

Interesting stuff...

 

[five40]

Did anyone read Anands review of this???? Holy crap is all I have to say. I don't care about any of the stupid benchmark crap. Reading his "every day work" test results made me want one SOOOO bad. I do the stuff he does in the article every single day. I would love to see performance increase like that. In one test he couldn't even put the FX on the chart cause it was so slow compared to the Dual Core.

 

[PetNorth]

Originally posted by: Capt Caveman
How is this a paperlaunch?

paperlaunch = not available for a while time to buy it at all.

Originally posted by: Capt Caveman
And I think comparing a 2.4 AMD dualcore is not apples to apples when it comes to this Intel 3.2 dualcore, since they are for two different markets.

nop... I think future A64 DC 2,4 2x1MB L2, same price range that future Pentium XE DC.

And I think something like an A64 DC 2.2, 2.0 and 1.8 2x512Kb L2, same price range that Pentium D 3.2, 3.0 and 2.8 respectively.

 

[jbh129]

All I can say is Bravo Intel. Even if AMD is better this should hopefully force them to market in a shorter time.

 

[Jeff7181]

Originally posted by: five40
Did anyone read Anands review of this???? Holy crap is all I have to say. I don't care about any of the stupid benchmark crap. Reading his "every day work" test results made me want one SOOOO bad. I do the stuff he does in the article every single day. I would love to see performance increase like that. In one test he couldn't even put the FX on the chart cause it was so slow compared to the Dual Core.

Which test was that?

 

[Capt Caveman]

Originally posted by: PetNorth

Originally posted by: Capt Caveman
How is this a paperlaunch?

paperlaunch = not available for a while time to buy it at all.

Originally posted by: Capt Caveman
And I think comparing a 2.4 AMD dualcore is not apples to apples when it comes to this Intel 3.2 dualcore, since they are for two different markets.

nop... I think future A64 DC 2,4 2x1MB L2, same price range that future Pentium XE DC.

And I think something like an A64 DC 2.2, 2.0 and 1.8 2x512Kb L2, same price range that Pentium D 3.2, 3.0 and 2.8 respectively.

So, you don't think Intel will be producing and selling dual cores? If no, then it can be considered a paperlaunch. But it's known that they will be coming. Second, the public has been given cpu's to review. Thus, I think you have your definitions incorrect.

If you read my whole paragraph and just not snip the first sentence, server/workstation market is much different than the desktop market. Intel EE dualcores and server dual cores will have HT. These are apples to apples to AMD's server dual cores. "I think" - no concrete facts, paperlaunch by you.

 

[five40]

Originally posted by: Jeff7181

Originally posted by: five40
Did anyone read Anands review of this???? Holy crap is all I have to say. I don't care about any of the stupid benchmark crap. Reading his "every day work" test results made me want one SOOOO bad. I do the stuff he does in the article every single day. I would love to see performance increase like that. In one test he couldn't even put the FX on the chart cause it was so slow compared to the Dual Core.

Which test was that?

Here ya go

Quote: "The most surprising is how poorly AMD did in this test. We actually had to exclude them from the graph as it distorted the bar lengths too much. AMD weighed in at over 27 minutes; from actually using the system"

 

[MDE]

Originally posted by: Capt Caveman

Originally posted by: PetNorth

Originally posted by: Capt Caveman
How is this a paperlaunch?

paperlaunch = not available for a while time to buy it at all.

Originally posted by: Capt Caveman
And I think comparing a 2.4 AMD dualcore is not apples to apples when it comes to this Intel 3.2 dualcore, since they are for two different markets.

nop... I think future A64 DC 2,4 2x1MB L2, same price range that future Pentium XE DC.

And I think something like an A64 DC 2.2, 2.0 and 1.8 2x512Kb L2, same price range that Pentium D 3.2, 3.0 and 2.8 respectively.

So, you don't think Intel will be producing and selling dual cores? If no, then it can be considered a paperlaunch. But it's known that they will be coming. Second, the public has been given cpu's to review. Thus, I think you have your definitions incorrect.

If you read my whole paragraph and just not snip the first sentence, server/workstation market is much different than the desktop market. Intel EE dualcores and server dual cores will have HT. These are apples to apples to AMD's server dual cores. "I think" - no concrete facts, paperlaunch by you.

It isn't even a launch, much less a paper launch. What part of Intel Dual Core Performance Preview Part I: First Encounter doesn't compute in your heads?

 

[Hyperlite]

and you expected.....?

thats nothing for the intel guys to brag about, everyone knew that would be the outcome. (well, that AMD would lose anyway, i dunno about that huge margin =/)

 

[Duvie]

You ppl...paperlauches are when the reviews proceed anyway for us the consumer to buy these chips...Now I heard no official release yesterday of this from INtel. They may be in route to OEMs fiorst like Dell...If theses chips are available for sell in 1-2 weeks then this was not a paper launch. If not it was....

"How is this a paperlaunch? People have the chip in their hands and providing reviews."

This means nothing.....ever heard of engineering samples?? Ever heard of the Barton 3200+ amd 300+ launch??? reviewers had reviewed them several weeks before they were even avaialble to preorder in low stock. Ther eused to be a time INtel would have NDAs where all the reviews cam eout on the day they offical launched it and ppl could have one in their hands by the end of the week....


I admit Intel looks good, but it would be foolish to jump when we haven't seen AMD's version. How many are going to buy the EE model?? Most will not...If you read ATs review you heard it was a still nice feature to have for responsiveness. dual core AMD and dual core INtel will be identical in "average work use". I for one can wait since I have the platform ready for AMD.....dual core 1.8 to 2.0ghz will be very nice for me so since a week or so back some one sneaked a peek at a 2.4ghz dual core amd chip I can tell you in the limited test it had a slightly better scaling then Intel's P4-D w/o HT. 2.4ghz will likely be high priced but if they come in with 2.2's, 2.0's and 1.8s expect the lower ones to be 240 range...

 

[Loxias]

I'm not sure you could call this a 'loss' for AMD as their DC tech hasn't been (p)reviewed yet.

 

[Duvie]

"If you read my whole paragraph and just not snip the first sentence, server/workstation market is much different than the desktop market. Intel EE dualcores and server dual cores will have HT. These are apples to apples to AMD's server dual cores. "I think" - no concrete facts, paperlaunch by you."

Oh how little you know.....INtel is not rushing to the server market first. I dont think they are going to try to push this starving 2 core prescott there. They will need to market some good chipsets and dump this shared memory bus. In a review at one of the other sites (handpicked by Intel, not the "public" LOL!!!) it showed a 533fsb dual xeon at 3.2ghz outperforming the dual core and EE version....

 

[Capt Caveman]

Originally posted by: MDE

It isn't even a launch, much less a paper launch. What part of Intel Dual Core Performance Preview Part I: First Encounter doesn't compute in your heads?

I don't think Launch was suggested or used in any context by either of us. Please read more carefully.

 

[Capt Caveman]

Originally posted by: Duvie
"If you read my whole paragraph and just not snip the first sentence, server/workstation market is much different than the desktop market. Intel EE dualcores and server dual cores will have HT. These are apples to apples to AMD's server dual cores. "I think" - no concrete facts, paperlaunch by you."

Oh how little you know.....INtel is not rushing to the server market first. I dont think they are going to try to push this starving 2 core prescott there. They will need to market some good chipsets and dump this shared memory bus. In a review at one of the other sites (handpicked by Intel, not the "public" LOL!!!) it showed a 533fsb dual xeon at 3.2ghz outperforming the dual core and EE version....

???? That's why it's not apples to apples. Thanks for your input and validating that these cpus are for two different markets.

 

[jbh129]

By coming out first, Intel will dominate the dual core desktop market