Intel D975XBX (Bad Axe)

[Sandog]

Originally Posted by bingo13

I will have hard numbers available early next week along with a couple of other surprises. The Intel 975x is a very good chipset in the hands of the right motherboard supplier.

We're all looking forward to some news on this front and I can't wait to see what the new board looks like. I think it shows how enthusiastic the 975X chipset is by the way companies like Gigabyte and even Abit are betting on it. These are two of your top OC mobo companies who appear to have thrown the kitchen sink into building their 975X chipset board.

[Intel_User]

I bet "with platform refresh" means its like 865/875 and Prescott situation, where you couldn't run Prescotts with earlier 865/875 mobos but with the updated versions. So, basically it can be said that 975X supports Conroe, but not until the updated versions are out. I believe 975X is just a more feature rich version of the 955X chipset with fixes to the errata and such. The really new chipset I think is the Broadwater chipsets which are being named 965, possibly coming with better memory controllers. Only superiority of 975X over 965 is that 975X would support dual graphics and more enthusiast options, probably 965 is better with memory performance though.

As you see here: http://www.anandtech.com/mb/showdoc.aspx?i=2605&p=2

The 975X chipset only supports DDR2-667 while Broadwater is gonna support DDR2-800.


Presler EE is 5-10% faster than Smithfield EE.


So I recommend people if they can wait they should wait for Conroe support or just get Broadwater.

[Intel_User]

The article about Presler EE and 975X performance is at Inquirer. The site linked also tested temperatures and amperage measurements. Presler at same clock speed is around 10-15% less in power consumption and 6-7 degrees C less than Smithfield. Presler EE has around equal power consumption as Smithfield EE and around 5C lower temps at load.


Don't see this as Intel's 65nm problems, its just that Netburst based CPUs(Pentium 4, D) is not Intel's concern anymore, 65nm should truly shine with Conroe. There seems like almost no concentration to lower power on Presler's. If people did not know, optimized 90nm(Smithfield) will be not much worse than unoptimized 65nm(Presler) in clock speed scaling and power consumption, as a CPU that works well in one process tech isn't guaranteed to be good on the smaller process tech.

[Sandog]

I agree Pressler is not the main focus at Intel but it will be a while before Conroe is out and I will get a Presler until that time comes. The Presler may be slightly faster and cooler but I think it will clock higher. I might even try VT

[TQ]

Originally Posted by Sandog

I agree Pressler is not the main focus at Intel but it will be a while before Conroe is out and I will get a Presler until that time comes. The Presler may be slightly faster and cooler but I think it will clock higher. I might even try VT

I may be behind the times a bit, but will the 975 support Conroe? If, as you say, Conroe won't be out until '06 year's end will the 975 offer an upgrade path from Presler to Conroe?

[Sandog]

The 975 chipset looks like it will offer a future upgrade path to Conroe but it may not be able to on the intitial boards. We will just have to wait and see. Intel is changing its schedule and seems to be putting products out when it can which is earlier than expected sometimes. It is possible to see Conroe before the end of '06 or the middle. I have a feeling though that by the time Conroe comes out there will be the Broadwater chipset available and this chipset will work better with Conroe I believe.

[JoeBar]

Originally Posted by bingo13

I will have hard numbers available early next week along with a couple of other surprises. The Intel 975x is a very good chipset in the hands of the right motherboard supplier.

Anxious for them!

[Intel_User]

It is absolutely ridiculous for Intel to make a CPU that costs a thousand dollars american, while having to upgrade, CPU, and mobo EVERY generation. Why can't a super expensive CPU have a platform upgrade path??

This means initial 975X users are screwed when coming to upgrade to Conroe. And plus there is absolutely no reason why Conroe can't be supported on even current 955X chipsets.

[Intel_User]

I don't see why you guys have to get dual channel DDR2-667 for your 800FSB CPUs when single channel DDR2-667 is very close to FSB bandwidth. Also, DDR2-533 is sufficient for 1066FSB when its dual channel.

Since there is increase in latency associated with going to dual channel I wonder if going to single channel would give better performance especially if BW requirements are met??

[TQ]

Originally Posted by Intel_User

Since there is increase in latency associated with going to dual channel I wonder if going to single channel would give better performance especially if BW requirements are met??

I may be wrong, but I don't believe it's single or dual channel that would cause a latency issue with DDR-2, but only that DDR-2 takes longer to read.

[Intel_User]

Well, of course DDR2 has higher latency than DDR, and that latency is in terms of how fast the RAM can be accessed. Dual channel has higher latency as evidenced by not 2x the bandwidth increase from single channel. Single DDR400 can do 3080MB/s while Dual Channel DDR400 can do 5000MB/s. Rather than Dual Channel DDR2-533, wouldn't single channel DDR3-1066 be better??

[TQ]

Maybe, but not being an engineer it's hard for me to say, but with both Intel and AMD's chipets designed with dual channel in mind there must be a reason for it.
(I've just got to find time to get to all those white papers that have been piling up...)

[Sandog]

See if you can follow this link.

[TQ]

Why, yes...yes I can.

[wicked71]

Do you still think board(d975xbx) is goin to be out in 2/06? Since Intel released the chipset earlier than expected to first-tier mobo companies, could it be out before the end of the year?