Canterwood benchmark and P4 800 FSB benchmark real

juin · Apr 14, 2003

http://www.hardware.fr/
The same website that have offe the 1 real NW benchmark.

[-peep-] french

eden · Apr 14, 2003

Is it me, or is PAT i875 a major disappointment with so much overhyping?

--
This post is brought to you by Eden, on a Via Eden, in the garden of Eden.

Twitch · Apr 14, 2003

I want to see a test that includes more benchmarks, including things like SATA performance. Also, English would be nice.

<-----Insert witty sig line here.

juin · Apr 14, 2003

http://www.xbitlabs.com/

a others like

yes not much benchmark i guess they have go for a smaller benchmark to be the 1 to have it online.

Pat do it job offer a 3% extra it appear in UT Win rar and all kind of game that like bandwith.A pure GPU or CPU bound cannot be chnage by memory but a memory bound will see a big increase from Pat a others will see no gain.Some thime it the extra bandwith that will help also dont forget RDRAM is faster by itself so PAT may just conter the fact that DDR is slower

[-peep-] french

juin · Apr 14, 2003

I have see up to now no website have bench I850E vs canterwood.Does intel have ask to not bench I850E vs canterwood

[-peep-] french

Twitch · Apr 14, 2003

What happened to 3.2 ghz Northwood? I only see the 3 Gig...

The X-Bit Labs benchmarks show the i875 in a much better light than that French site you linked to first. Especially considering the 3.0 gig beats the 3.06 in almost every test, and in some cases very easily, despite being clocked lower. This demonstrates the real-world value of the extra bandwidth--in some cases only a 3-5% improvement, and in other benchmarks, a 15-20% improvement.

I think it's about what I expected. I still want to see more benchies though...

Especially, I want to see a SATA bench...I want to know if SATA on the ICH5 Southbridge provides a material improvement in performance of SATA devices.

<-----Insert witty sig line here.

HardWareBoss · Apr 14, 2003

The bench marks for Canterwood should be out this week. Since it releases 14 April. Why no benchmarks on Intels web site? I checked today. They have nothing on Mobos for Canterwood or Springdale. One would think since they make the new chips they would have benchmarks on thire site. Why do we have to go to france of all places?

HardWareBoss · Apr 14, 2003

When is 3.2 ghz Northwood going to be released? May?

Twitch · Apr 14, 2003

Canterwood and Springdale benchmarks here:

<A HREF="http://www.xbitlabs.com" target="_new">http://www.xbitlabs.com</A>

<-----Insert witty sig line here.

Twitch · Apr 14, 2003

Hey! The THG review of Canterwood is up already!

Looking good!

<-----Insert witty sig line here.

slvr_phoenix · Apr 14, 2003

Is it me, or is PAT i875 a major disappointment with so much overhyping?

Overhyping? I hadn't even heard it talked about in conversation until now. I had barely even read anything about it before hand. What little I did read was pretty simple and straight forward. There hadn't really been any hype as far as I could tell.

Seriously though, it's just meant to be a counter to AMD's 'on-die' memory controller. AMD claims that putting the controller on the die is the only way to squeeze the last bit of performance out of DDR. Intel meanwhile showed that it's still perfectly possible to get more performance without going to an on-die controller, and thus offering more flexability in the complete system than AMD's Hammers.

Slimming down a bit more of the latency of DDR is always nice, but nothing spectacular. (Especially in light of how many people still use the cheaper CAS2.5 or even CAS3 RAM in the first place.)

<pre>I'm proud to be an American,
who served my country in the US Air Force,
to protect the rights of my fellow Americans,
to hold protests against others like me.</pre>

slvr_phoenix · Apr 14, 2003

Having been busy hiking this weekend, I only just got around to checking out the benchmarks of the new P4 and chipset. I have to admit, I'm impressed.

The new P4 3.0GHz really blows away the P4 3.06GHz in some of those benchmarks. Even with it's 66MHz less clock, the only benchmarks it seems to really lose are the the theoretical ones. In all of the real world apps it ties at worst and wins significantly at best. Not bad for being slower.

Intel is definately working hard at increasing the IPC of the P4 systems. To do that and still be increasing the clock speed as well... Well the P4 just sure isn't a Nilly-Willy anymore.

I had been waiting to see how the new P4s perform on a Canterwood before I decided to go with a P4 or an Athlon. I'm afraid to say that my next system might actually be an Intel after all. If Intel can release a 2.8GHz (or even better, a 2.6GHz) P4 with HT and an 800MHz FSB, I'm 100% sold. Even if not, I might have to struggle to see if I can afford a 3GHz P4 then. After another speed release, it should be more affordable.

AMD's only hope to be my next platform is if they can somehow wow me with a move to a 400MHz FSB (not likely to be all that wow though) or wow me with the Athlon64. Considering that I'm looking to buy my system in July though ... it doesn't look good for AMD being my next purchase at all. (Which is funny since just a week ago it was going to be an AMD system.)

<pre>I'm proud to be an American,
who served my country in the US Air Force,
to protect the rights of my fellow Americans,
to hold protests against others like me.</pre>

bikeman · Apr 14, 2003

Yesyesyes ... This does look good!! The only thing that bothers me is that the pictures of the last benchmarks aren't visible on tom's. Anantech does give all the benchmarks and with their system setup there appear to be some problems with the drivers. On some tests the i875p prooved to be significantly slower than the i845pe ... Also, they tested the same CPU on a i850e and an i875p and sometimes, still, the RAMBUS-platform was just slightly faster. Though that could be caused by those driver-problems, off course ... But I'm up to reading Anands review of the new P4, where they will be using the 800 MHz FSB ... I'll be back!!

Greetz,
Bikeman

Then again, that's just my opinion

bikeman · Apr 14, 2003

Ok, now all the benchmarks are visible at Tom's. And though I haven't got the time to study it in detail (I've got to study, once again ...) it seems as if Tom had had the same problems as Anand. So I guess the definitive judgement will only be made after those new drivers from Intel will emerge. Though I doubt they will cause the i875p to outperform the i850e in all benchmarks, which is something I was actually hoping for ...

So now, back to the books ...

Greetz,
Bikeman

Then again, that's just my opinion

vacs · Apr 14, 2003

If Intel can release a 2.8GHz (or even better, a 2.6GHz) P4 with HT and an 800MHz FSB, I'm 100% sold.

P4 2.4Ghz and above with HT will be available during Q2 2003 (don't know about FSB though, probably only 533MHz).

BTW, yes the new P4 with the Canterwood really looks nice. Seems like Intel is taking their challenge against the Athlon64 very seriously. As far as I remembered the Athlon64 sticked with a 9800 Pro Inquirer.net tested scored around 15000+ points in 3dmark2001. The new P4 with the same graphiccard scored over 17000+. Not a small difference...

I can't wait for my P4 Prescott with a nv35

slvr_phoenix · Apr 14, 2003

Anantech does give all the benchmarks and with their system setup there appear to be some problems with the drivers.

Minor speed issues. It isn't really a 'problem' so much as a lack of speed optimization at the moment. But then Intel has also said that they're going to release an update later that will grant RAID1 support. No doubt those drivers will also perform better.

Also, they tested the same CPU on a i850e and an i875p and sometimes, still, the RAMBUS-platform was just slightly faster.

That's to be expected. RDRAM is superior to DDR SDRAM in some ways, even still. That doesn't make DC-DDR worth any less though, as it is still superior in other ways.

Still, it's a shame that Intel is completely ditching RDRAM. A dual-channel 32-bit PC800 RDRAM system would have been interesting and low-cost. Heck, they could have easily designed a dual-channel 32-bit PC1066 RDRAM system had they wanted to. PC1600 isn't a necessity for the continuation of RDRAM.

After reading AnandTech's articles though, I have to admit that I'm rather unimpressed. I mean besides getting a number of things wrong (I.E. "The first 800MHz FSB CPU from Intel is the Pentium 4 3.0C which, as you can probably guess, runs at an even 3GHz - just 6MHz shy of the previous champ." That should be 66.66MHz shy, not 6MHz shy.) they also hand-picked some rather silly hardware to use. (I.E. "2 x 256MB PC800 Samsung RIMMs" Why not use PC1066 like any normal person would?)

Heck, even in AnandTech's review of the 875P motherboard they used a "Intel Pentium 4 3.06GHz" instead of the new P4C 3.0GHz. According to THG the PAT enhancement "only works with the 875/ Canterwood and 200 MHz FSB." So unless THG is wrong, AnandTech's review of the 875P motherboard never included the PAT enhancement because they ran it with a 133MHz CPU.

I'm not saying that THG's reviews are perfect, but I've found that AnandTech's reviews are often marred slightly with hardware selections that seem to be made on purpose to put specific hardware choices into a less-than-ideal light so that their favorites end up looking better.

One thing that still bugs me about all of the Canterwood reviews that I've read though is that no one is even trying to compare a Canterwood mobo to a Granite Bay mobo or a SiS655 mobo. One would think that comparing DC-DDR platforms would be as much of a priority as comparing DC-DDR to RDRAM. (And certainly more important than comparing DC-DDR to SC-DDR.)

<pre>I'm proud to be an American,
who served my country in the US Air Force,
to protect the rights of my fellow Americans,
to hold protests against others like me.</pre>

Twitch · Apr 14, 2003

Yeah...I wasn't crazy about Anand's decision to split up the i875 and P4 3.0/800 reviews the way he did. Everyone else has combined them into a single review, why not Anand? Because of the PAT technology (by the way, other review sites are also saying PAT only works on 800 FSB) slapping a 3.06/533 onto an i875 doesn't tell the whole story. I prefer Anand's reviews of graphics cards, but I think I prefer THG for core logic and processors.

I want to see benchmarks of i875 vs. SiS655 using the fastest possible hardware on both setups. I also want my SATA benchmarks. THG did a couple teaser benches on SATA, but I want to see a comparison of 2 identical SATA hard drives in both single and RAID 0 configuration, first on the best SATA controller, and then on the i875 integrated controller. According to THG, there does seem to be some performance enhancements, but how much?

Oh well, I guess all the hardware sites were racing to get their reviews up, so we can't expect every single feature to have been extensively reviewed.

<-----Insert witty sig line here.

Edited by Twitch on 04/14/03 12:03 PM.

slvr_phoenix · Apr 14, 2003

Yeah...I wasn't crazy about Anand's decision to split up the i875 and P4 3.0/800 reviews the way he did. Everyone else has combined them into a single review, why not Anand?

I didn't mind that so much. Splitting it into a motherboard review and a processor review is find with me, since they are two different products. However, the reviews did end up leaving much to be desired in the end. That I minded. That and the nearly endless list of processors that were tossed into the systems, as if we need that many for a comparison. It was senseless.

(by the way, other review sites are also saying PAT only works on 800 FSB)

Thanks.

It's good to know. THG isn't always perfect, but it seemed an odd thing to throw out there, so I figured it was probably right. It's kind of strange though. You would think that if it could be used for an 800MHz FSB, it'd work for a 533MHz FSB just as well... I guess Intel just wanted to give people yet one more reason not to stick with a 533MHz FSB CPU. So it was probably done for marketing reasons, not for technical reasons? I guess? Still, it is strange.

It's kind of sad about AnandTech. I mean I hear a lot of people prefer their reviews because they're supposedly better. I've personally yet to be conviced of that. AnandTech seems to frequently obfuscate reality. (Though admittedly not nearly as much as uber-biased places.)

I want to see benchmarks of i875 vs. SiS655 using the fastest possible hardware on both setups.

I still say throw Granite Bay into that as well (and also throw in slower memory speeds) so that GB owners know if an upgrade is at all worth it and we can compare what is optimal for our price range.

I also want my SATA benchmarks. THG did a couple teaser benches on SATA, but I want to see a comparison of 2 identical SATA hard drives in both single and RAID 0 configuration, first on the best SATA controller, and then on the i875 integrated controller.

I agree. It'd be nice to see more done. I'd especially like to see more done on RAID1 and RAID01/10/whatever. (And if anyone has RAID3 or RAID5 for SATA yet.) I'm personally far more concerned about data redundancy than about just a pure speed boost.

And one review that I think is incredibly lacking is a real-world gaming simulation. Set up a networked gaming server and client with sound and RAID0. Let's see the CPU performance hits for running everything together (especially for on-board devices that dump their processing onto the CPU) and also see if any of those theoretical PCI bottlenecks occur in real use of systems where everything is run through the PCI bus. (And consequently if doing things like running the ethernet on it's own bus makes a difference.)

While we're at it, I'd like to see if running this new ethernet bus significantly improves the performance of clustered systems.

<pre>I'm proud to be an American,
who served my country in the US Air Force,
to protect the rights of my fellow Americans,
to hold protests against others like me.</pre>

Twitch · Apr 14, 2003

In regards to Anandtech's reviews:

The main thing I like about Anand's reviews of graphics cards is that he includes Minimum FPS scores in almost every review. I wish more reviewers would do that. I tend to think Anand is a bit of an Intel honk, but that may just be because he was a keynote speaker at IDF.

Regarding reviews of SATA, my understanding is that you'd have to use a seperate SATA controller on the i875 if you wanted to compare anything other than RAID 0/1. Of course, it would still be interesting to see if Canterwood offers a benefit for SATA apart from the ICH5R integrated SATA.

I totally agree with your statement that we need real-world gaming benchmarks. For example, we know from THG's recent review that nForce2 Sountstorm APU uses the least CPU time of any of the sound solutions. In that light, it would be interesting to know how AMD/nForce2 stacks up against an equivalent P4 when stereo sounds is enabled, and when 3D positional sound is enabled, etc. That would be a very interesting review, since there are many, may possible configurations. It would be especially informative if minimum FPS scores were also included.

Here's an example of one of the things about which I'm curious:

Theoretically, i875's integrated SATA bypasses the PCI bus freeing up other resources. So, if you have an i875 board and use a single SATA hard drive on the integrated SATA controller, will this improve gaming performance with sound on? How about if you use 2 drives in a RAID 0?

We should start our own hardware site!

I will be an "idea man" and maybe a grammar editor, since I am totally unqualified to do hardware reviews. Hehe...

<-----Insert witty sig line here.

Kemche · Apr 15, 2003

I tend to think Anand is a bit of an Intel honk, but that may just be because he was a keynote speaker at IDF

I don't think Anand at Anandtech.com has ever been a keynote speaker at IDF. I think the Anand who was the keynote speaker at IDF was some person who heads the Mobile Group at Intel. He actually works at intel. So please don't confuse your Anand(s).

KG

"Artificial intelligence is no match for natural stupidity." - Sarah Chambers

juin · Apr 15, 2003

It Anand cherkera (something like that) mobile platfrom director intel CORP nothing to do with Anandtech, also X-bitlab editor name is anand.

[-peep-] french

Twitch · Apr 15, 2003

My bad.

<-----Insert witty sig line here.

eden · Apr 15, 2003

Reading THG's article also overall left me disappointed.

To Slvr, because you're not always around the forums, you didn't see the amount of hype. Hell the articles regarding Canterwood always left us with "Oh man, this is gonna be so better than Springdale". I didn't know it was a mere 3% (one cycle saved, wow).

While tests like the Q3 left me wowed, the rest revealed that Intel also can play the Barton game, where you release lower clock speed CPUs and yes, even with the better IPC, they can end up losing to their former models.
I don't know though if the new platform cost is justifiable. Think about it, 180$ for 2*256 quality RAM for CAS2, then add to the incredible cost of Canterwood chipsets, the mainboard costs should be that of the lowest RDRAM mainboard costs. While the new CPU costs VERY LITTLE, I admit, you are reminded of the rest, which may end up costing what a normal Northwood high-end combo costs. Who knows, though.

However, it still leaves me dry, when I see that the best performance enhancements, of the 15% order, are the result of 50% higher FSB and memory bandwidth, PLUS a Performance enhancing chipset.
When looking at that, perhaps AMD's 400MHZ FSB CPU does stand a chance after all, although very little, just enough to reclaim one or two benchmarks.

Twitch is right about sound+game, all game benchmarks are lies in fact. They reveal what a deaf person would get in performance. When turning on sound, which we all do, especially EAX or higher, unless we use Audigy 2s or nForce 2s, we're bound to lose 10% in performance. And if the integrated audio or the Audigy or SB Live used for Pentium 4 systems isn't that good, in the end, it could be that AMDs perform better in these tests. I hope not. With that said, I hope they can try testing deeper in Sound and Gaming, as I know it shouldn't be done in the CPU tests, if we're looking at pure CPU enhancements.

I guess indeed, RDRAM 6.4 would've been killer. I hope to god that the SiS QDR RDRAM will allow 6.4 with 800MHZ FSB, it should be VERY interesting.

--
This post is brought to you by Eden, on a Via Eden, in the garden of Eden.

eden · Apr 15, 2003

Whew, lotsa Indians in the editing tech sector!

--
This post is brought to you by Eden, on a Via Eden, in the garden of Eden.

Twitch · Apr 15, 2003

Yeah! I never new there were so many Anands around. :/

I guess Anand must be the Indian equivelant of "Bob."

<-----Insert witty sig line here.

Canterwood benchmark and P4 800 FSB benchmark real

Distinguished

Champion

Distinguished

Distinguished

Distinguished

Distinguished

Splendid

Splendid

Distinguished

Distinguished

Splendid

Splendid

Distinguished

Distinguished

Distinguished

Splendid

Distinguished

Splendid

Distinguished

Distinguished

Distinguished

Distinguished

Champion

Champion

Distinguished

Share this page