Granite Bay

[Toymaker]

for those who don't normally keep up with this stuff

http://hardware.earthweb.com/main/article/0,,12108_1484881,00.html

http://www.gen-x-pc.com/new_rambus_and_ddr_info.htm

http://www.gamepc.com/labs/view_content.asp?id=thunder7500&page=1&bhcp=1

 

[IntelConvert]

And for those who think that Granite Bay will be the only game in town next month, think again...

Here comes SiS 655 Dual Channel DDR333 - with all of GB's other goodies (comparable features to SpringDale, with a 6-month head start)!

Behold the Asus P4SDX -- Features and Pictures

And the 'frosting on the cake' is that SiS 655 boards will surely cost less than GB boards!!!

 

[TheCoz]

Thanks for the linkage IC.

Looking at the P4SDX I notice that there's only 2 output inductors and that means it uses a 2-phase VR design. Their Granite Bay board (P4G8X) also uses 2-phase VR. Everyone else seems to be using 3 or 4-phase VR these days so why are Asus taking a step backwards? At least they're using 3 FETs-per-phase I suppose. It seems like a long time ago everyone was heralding the use of 3-phase VR on the Abit KT7 and here are Asus running 2-phase VR on a board designed for use with 3GHz+ CPUs!! Epox and Abit are both using 4-phase VR on their Granite Bay boards so why are Asus going with two? I just wish I knew some basic electrical engineering so I could understand this better. I hope their (Asus) DCDDR P4 boards aren't going to turn into another couple of P4T533 disasters!:Q

I'm looking forward to Evan's 3-way P4 DCDDR shootout - Intel E7205 vs VIA P4X600 vs SiS 655.

 

[IntelConvert]

Originally posted by: TheCoz
Thanks for the linkage IC.

Looking at the P4SDX I notice that there's only 2 output inductors and that means it uses a 2-phase VR design. Their Granite Bay board (P4G8X) also uses 2-phase VR. Everyone else seems to be using 3 or 4-phase VR these days so why are Asus taking a step backwards? At least they're using 3 FETs-per-phase I suppose. It seems like a long time ago everyone was heralding the use of 3-phase VR on the Abit KT7 and here are Asus running 2-phase VR on a board designed for use with 3GHz+ CPUs!! Epox and Abit are both using 4-phase VR on their Granite Bay boards so why are Asus going with two? I just wish I knew some basic electrical engineering so I could understand this better. I hope their (Asus) DCDDR P4 boards aren't going to turn into another couple of P4T533 disasters!:Q

I'm looking forward to Evan's 3-way P4 DCDDR shootout - Intel E7205 vs VIA P4X600 vs SiS 655.

Coz, you're very welcome... just my way of saying that those of us interested in DCDDR shouldn't ignore SiS 655 (better features and less cost than GB)!

Re Asus' 2-phase voltage regulation, you really have to examine the components more closely before making negative conclusions about their design. Theoretically speaking, with more phases the power load is spread across more MOSFETs and of course, less load on a component means less heat. But 2-phase may actually be better then a 3 or 4-phase if the 3 or 4-phase implementation uses cheap components and the 2-phase implementation uses quality (high-rated) components!

It would be nice if Evan would examine and compare each board's voltage regulators (and their components) in future reviews/shootouts... Evan, can do?

 

[Cheval]

Originally posted by: IntelConvert

It would be nice if Evan would examine and compare each board's voltage regulators (and their components) in future reviews/shootouts... Evan, can do?

I'd like to second that!

 

[Zap]

Originally posted by: IntelConvert

Originally posted by: senior guy

Originally posted by: Evan Lieb

Originally posted by: Mr. Crisis
Ack, but Evan we can't wait so long!
Could you at least give us a hint about the performance relative to the 850E with Rambus 1066? Just a tiny hint...!

It's faster.

Evan, I don't see why that should be. ...how do you account for that?

seniorguy~ While DCDDR and RDRAM provide comparable bandwidth, DDR can do it at higher FSB. Whereas PC1066 tops out at around 150MHz FSB and RIMM4200 at around 160MHz FSB, DCDDR can do 200MHz FSB. For low FSB overclocks, I would expect RDRAM and DCDDR to perform close to one another, but for high FSB overclocks, DCDDR has a clear performance advantage!

Something I recall hearing about... besides bandwidth issues (which DDR is catching up to RDRAM) DDR has lower latencies than RDRAM. Did I recall right, or do I need to lay off the wacky weed?

 

[TheCoz]

Originally posted by: Zap
Something I recall hearing about... besides bandwidth issues (which DDR is catching up to RDRAM) DDR has lower latencies than RDRAM. Did I recall right, or do I need to lay off the wacky weed?

Yes, DDR latencies are slightly lower than RDRAM on current P4 systems but it's pretty close.............

(Figures from Ace's Hardware)

64/128 byte stride (clockcycles)
P4 2.8 DDR333 - 78/252
P4 2.8 PC1066 - 73/318

Of course, these figures don't mean that DDR has lower latencies than RDRAM per se - but simply that the DDR rig in the test has lower latencies than the RDRAM one did. Also, the figures for DDR are with an asynchronous FSB and memory clock (133:166), they will be synchronous with GB and therefore probably better than those figures. RDRAM latencies are lower the faster it goes so PC1200 RDRAM would probably have lower latencies than DDR. Confused? My work is done.

 

[First]

Yeah, I've been comtemplating whether or not to include some info on voltage regulators as well as some IRQ stuff. We'll see.

 

[Toymaker]

Originally posted by: IntelConvert
And for those who think that Granite Bay will be the only game in town next month, think again...

Here comes SiS 655 Dual Channel DDR333 - with all of GB's other goodies (comparable features to SpringDale, with a 6-month head start)!

Behold the Asus P4SDX -- Features and Pictures

And the 'frosting on the cake' is that SiS 655 boards will surely cost less than GB boards!!!

Ah, yes
more good reading
http://www.neoseeker.com/news/articles/headlines/Hardware/1871

 

[Lizardman]

Too many good choices for a GB mobo. I also cant wait to see what SIS and VIA have to offer. Maybe they can get it right for once and beat out intel.

 

[semitech]

Two questions... any predictions?
1. When will I be able to buy a Granite Bay motherboard (Intel and/or other mfg)?
2. How much will it cost (Intel and/or other mfg)?
I've waited this long to build a new system, and now it seems like waiting a few more weeks is a smart move, but I really want to know when I can pull the trigger and what size hole it will put in my wallet.
Anyone?

 

[TechMasterChen]

I think in this thread someone said the NDA expires in or around the Nov. 14th for some of the Granite Bay boards. I would think the boards would be somewhat widely available by then if Intel has been flooding their chipsets down the distribution channel. We get the Granite Bay announcements on Nov 14th then Nvidia on November 18th! My wallet is already hurting...

 

[senior guy]

Originally posted by: semitech
Two questions... any predictions?
1. When will I be able to buy a Granite Bay motherboard (Intel and/or other mfg)?
2. How much will it cost (Intel and/or other mfg)?
I've waited this long to build a new system, and now it seems like waiting a few more weeks is a smart move, but I really want to know when I can pull the trigger and what size hole it will put in my wallet.
Anyone?

1. First availability: Mid-Nov; supplies increasing late Nov - early Dec. Same time-lines for SiS655 (Asus and ECS boards are ready).
2. My guess: Less featured GB boards $130 - $160; fully featured GB boards $175 - $220 (SiS655 boards should run about 15% less).

 

[oldfart]

I'd like to see max stable overclock performance of a given processor comparison between 850E RDRAM and DC DDR. For instance, use the 2.26 P4 that you used in the 845PE tests.

Granite Bay test
2.26 @ 166 FSB, 2.8 GHz, DC DDR333

850E test
2.26 @ 166 FSB, 2.8 GHz, PC1328 (I doubt this is possible) more likely PC996

 

[degeester]

I'd like to see a comparison on the onboard sound chips likely to be used in Granite Bay MB's. C-Media vs. Realtek vs. Analog Devices vs. anything on the horizon. It will help us decide on spending for Creative Audigy, etc ... also any conflicts in using PCI sound cards.

 

[Willy9Boy]

Reviews are usually handled very professionally. There are some things I would like to see added to the Granite Bay reviews.
1) List the MB Bios revision and if it has been flashed using the MB website files.
2) Compare the GB to the SIS 655 and any others
3) I have seen some discussion about memory. Will the MB support 1, 2, or 3 sticks? Will it support different speeds ? PC2100 in one slot and PC2700 in second slot? What are performance advantages/disadvantages to using CAS 2.0 versus CAS 2.5 etc ? Can faster memory be overclocked more than slower memory?

 

[mcveigh]

Welcome Willy!!

 

[Maverick]

Evan, can you make sure that the GB isn't affected by the AGP 8X problems that the Via and SIS boards seem to be having with the Radeon 9700Pro? Thanks.

 

[Kajpin]

Well, I'm glad I ran across this topic

I plan to build my first computer based on this MB after putting up with system builders for the last time. With that said, I'm hoping to gleen a lot of information from these coming reviews.

First, I'd like to see a comparison of the boards (including boards based on the SIS 655 if possible) stock features and performance at stock speeds (no overclocking on anything). I'd like to see you use the AGP 8x 9700 too, since it's stability on current 8x boards is reportedly quite poor.

then, on to the real fun...

Now I'd like to see how well these boards overclock. Basically what I hope to be able to do is bump the FSB to 166 in sync with the memory. Giving me a system with dual channel ddr at 333(pc2700) speeds running in sync with the cpu.

Now I'm not real interested in how well these boards do this with stock cooling. I'm pretty sure anybody that is going to overclock will use some form of advanced cooling. What I'd like to see is the test setup use a fairly standard well ventilated overclockers case with a fairly standard overclockers cpu heatsink/fan. Nothing really extreme, but a basic setup for any intelligent overclocking.

Then see how far you can go with fsb while remaining stable. Some people will want to see how far they can go. For me, I want to see which board gives me the most headroom so I can maintain a comfortable cushion.

Thanks

 

[jaeger66]

Regarding the 2-phase regulation-Intel's 845E board uses 2 phase and the voltage is absolutey stable. Vcore is dead on at 1.50v and doesn't drop so much as .01v under full load. Its about quality, not parts count.

 

[senior guy]

As IntelConvert explained (very clearly I might add)...

Re Asus' 2-phase voltage regulation, you really have to examine the components more closely before making negative conclusions about their design. Theoretically speaking, with more phases the power load is spread across more MOSFETs and of course, less load on a component means less heat. But 2-phase may actually be better then a 3 or 4-phase if the 3 or 4-phase implementation uses cheap components and the 2-phase implementation uses quality (high-rated) components!

 

[THUGSROOK]

Originally posted by: senior guy
As IntelConvert explained (very clearly I might add)...

Re Asus' 2-phase voltage regulation, you really have to examine the components more closely before making negative conclusions about their design. Theoretically speaking, with more phases the power load is spread across more MOSFETs and of course, less load on a component means less heat. But 2-phase may actually be better then a 3 or 4-phase if the 3 or 4-phase implementation uses cheap components and the 2-phase implementation uses quality (high-rated) components!

yea but ask ANY experienced Asus board owner about how well it regulates voltage....
not very well

 

[senior guy]

Originally posted by: THUGSROOK... yea but ask ANY experienced Asus board owner about how well it regulates voltage....
not very well

I don't believe IntelConvert was saying that Asus had quality voltage regulators, just that you can't 'count on it' by the number of phases.

 

[jaeger66]

Originally posted by: THUGSROOK

yea but ask ANY experienced Asus board owner about how well it regulates voltage....
not very well

Hmm, my 2-phase P4B533-V regulates perfectly fine. Not quite as good as the Intel, but it's certainly no problem. It's another one of those things, like IRQ sharing, that people obsess about for no good reason.

 

[IntelConvert]

Originally posted by: senior guy

Originally posted by: THUGSROOK... yea but ask ANY experienced Asus board owner about how well it regulates voltage....
not very well

I don't believe IntelConvert was saying that Asus had quality voltage regulators, just that you can't 'count on it' by the number of phases.

Exactly my point... I'm sure that some MBs have marginal voltage regulators, but it's flat-out wrong to believe that they do just because it's a 2-phase design!