New chipsets?

[tormentor22]

when would the new chipsets be available? im about to buy a new PC but i can wait if it is comming soon..
i was lately surfing the intel website when i noticed that the G35 chipset is ranked as a "Performance chipset" exactly as the 955X and 975X, while the P35 chipset is a "Mainstream" so does that mean that the G35 is better than P35 in performance or is that only coz the G35 got a DX10 integrated video card?

[Leeghoofd]

Prolly the latter, the X38 will replace the P35 soon ( mid september) as the performance chipset... if you buy a seperate vidcard or already own one plz go with P35 and not G35 ( as it will cost extra and performance is pretty crappy in comparison with a decent gfx card , logic ofcourse )

[Start]

I thought the X38 would replace the 975X and the P35 is replacing the P965.

[Lazy Dog]

I thought the X38 would replace the 975X and the P35 is replacing the P965.
Me too...:confused:

[zechraffie]

I thought the X38 would replace the 975X and the P35 is replacing the P965.

i think you're right.

[Leeghoofd]

With replacing the P35 I meant it will be the fastest chipset of the two, only thinking in terms of performance not witch chipset is going to make another abundant.... it would be nice if it was SLI and Crossfire compliant... Think Intel couldn't keep up with producing them....

[Richard Dower]

It's a pity X38 won't ship with ICH10...that would make it even more worthwhile to make the switch.

[hitokiri69]

I'm curious about this myself, so I finally decided to post something... :D

I'm currently on E6400 @3.2 and am planning to update into quad sooner rather than later. I need a new HTPC server rig to handle HD material, and 6400 is more than adept to be recycled for that purpose.

The problem I see right now is, Intel released P35 without PCIE 2.0 support.
This is really bugging me, since otherwise I would've already shelled out for UT P35-T2R and G0 quad. That would've been a future proof update path, since GPU is the only part I usually replace every six months or so. The rest, every three years or so. Even if G92 doesn't yet fully saturate the first gen bus, I'm sure G100 will already make 2.0 a necessity.

Now it appears the x38 is the first PCIE2, followed by nVidia's C72 and C73.
I however could not care less about SLI, since I'm only gaming @1920x and even my current 8800 GTS 640MB handles most stuff admirably.

Also, by now we know that x38 will be a very complex and hot chip, and I'll bet we won't be seeing too many 500FSB oc's, where as the P35 goes 500 flying, with stock cooling. You'll probably need dry ice to break 500 with x38. I hope I'm very, very wrong about this... :shrug:

And to OC the ~2.3GHz Yorkfield (only chip in a feasible price range on launch) to anywhere close to such stellar clocks we are currently witnessing e.g with G0 Q6600, you are going to need some hefty fsb to push the chip. I'm just not seeing this happen with x38...

Hopefully we'll see a respin of P35 soon, or some similar "mid range" chipset with DDR2 support. I don't really see DDR3 as a viable option, until you can just shell out around $500 for 4GB PC3-16000 :p:

[Leeghoofd]

Did you ever see a big performance gain going from AGP 2/4X to AGP 8X ? or even AGP Pro ? (what a marketing hype that was) It's all theory (giving nice numers and stuff...) , I don't even know if the AGP bus is saturated with current vidcards ? I rather think it's more of a power related problem, as PCI 2.0 will be able to deliver more watts to the vidcard... but they need to do something about the wattaes drawn and heat output

[tormentor22]

I thought the X38 would replace the 975X and the P35 is replacing the P965.

Me too...:confused:
Me three!!!:D

Well u ppl didnt get w i mean...
what i wanted to say which is better in performance without talking about discrete cards, i know a discrete is much better than integrated, but is the overall chipset performance in the G35 is better than it is in the P35?? well if u take a look at the previous generations P965 is better than G965 in performance but it seems its not the case with the new 3 series??

[Leeghoofd]

Just marketing mate on Intel website, if this would be the case then we all would use G35 chipsets wouldn't we, to overclock and have fun...

P35 will be superior to G35 be sure of it... and I believe X38 will be better than P35 clock for clock, maybe not such a great overclocker (hence why I stated it will replace the P35 in terms of performance, but you guys got that part wrong, prolly due to my bad english)

[hitokiri69]

I don't even know if the AGP bus is saturated with current vidcards...

That's actually a good point. Been wondering about this my self lately.
It would be very interesting to be able to set up some sorta reference test platform, with exactly the same type of contemporary pc hardware, with the exception of other MB sporting with specially crafted single 8800GTX in AGP 4x/8x and another MB with typical PCIE setup...

Otherwise the gains going from AGP 4x to 8x eg. is hard to put into perspective, since for me that meant moving up from GeForce2 Ti to Radeon 9800 PRO with a new MB and the works...

[T2k]

Did you ever see a big performance gain going from AGP 2/4X to AGP 8X ? or even AGP Pro ? (what a marketing hype that was) It's all theory (giving nice numers and stuff...) , I don't even know if the AGP bus is saturated with current vidcards ? I rather think it's more of a power related problem, as PCI 2.0 will be able to deliver more watts to the vidcard... but they need to do something about the wattaes drawn and heat output

It's EASILY saturated in games.

[Leeghoofd]

Got any proof there T2K ? , I recalled with my 9800pro we had to put AGP from 8X to 4X to stabilize the system with zero performance hit... the AGP Pro was the next big thing as AGP bus was gonna be saturated, all theoreticall and marketing bull... best thing to do is to put a 1950 pro AGP against an PCI express version withe same CPU (ofcourse ram and co will be part of the benchmark) but I'm pretty sure there's not much gain/loss there....

[tormentor22]

Well yes everything theoretical, but when u talk about 100% increase in bandwidth in theory that would be giving at least 40% more in true, this is exactly the same as ddr2 speeds, i tested 533 667 and 800mhz rams but didnt feel any change, these dont get u sensible change when ordinary use, but may make ure games go a bit smoother, or let some programs which require long time to finnish (24h+) to minimize one hour which is really isnt bad.

[icon57]

It's a pity X38 won't ship with ICH10...that would make it even more worthwhile to make the switch.

agreed! but i am making the switch anyways, seeing as how i missed the p35 mobos...

these days, i really dont believe any hardware, is future proof...we are lucky if we get 6 months out of the new releases we buy todaY.

[tormentor22]

Also, by now we know that x38 will be a very complex and hot chip, and I'll bet we won't be seeing too many 500FSB oc's, where as the P35 goes 500 flying, with stock cooling. You'll probably need dry ice to break 500 with x38. I hope I'm very, very wrong about this... :shrug:

i doubt that, since as we saw in the computex pre X38 boards MSI placed an ordinary heatsink on the chipset not a rollercoaster heatpipe cooler as its the case with the P35, i believe they must have found if cooler to do such a thing

[Blauhung]

Also, by now we know that x38 will be a very complex and hot chip, and I'll bet we won't be seeing too many 500FSB oc's, where as the P35 goes 500 flying, with stock cooling. You'll probably need dry ice to break 500 with x38. I hope I'm very, very wrong about this... :shrug:

And to OC the ~2.3GHz Yorkfield (only chip in a feasible price range on launch) to anywhere close to such stellar clocks we are currently witnessing e.g with G0 Q6600, you are going to need some hefty fsb to push the chip. I'm just not seeing this happen with x38...i doubt that, since as we saw in the computex pre X38 boards MSI placed an ordinary heatsink on the chipset not a rollercoaster heatpipe cooler as its the case with the P35, i believe they must have found if cooler to do such a thing

The X38's major boost is probably going to be in the category of lower latencies and better MCH timings. As with most every high end Intel chipset (955/975) you see much better performance on the better chips, so I might be dissapointing a few people by saying this, but this chipset might end up in a similar position of lower max FSB in exchange for lots of stability and tiny latencies/timings on high multiplier chips. You will probably need a high multiplier cpu in order to make this chipset shine.

of course if the bulk of these timings are bios controllable, then the high end of the FSB range might be possible. But keep in mind, this will be the extreme chipset, and the silicon itself is optimized for the extreme cpu's, not budget ones.

[tormentor22]

so what u think Blauhung? im going to buy 6550 or 6750 and im wondering shall i wait for the X38 to be reliezed or buy a P35 now?
BTW isnt there any news about the september price cuts? as i remember when the July 22 price cuts happened we knew what would be the prices, so what about this time?

[T2k]

Got any proof there T2K ? , I recalled with my 9800pro we had to put AGP from 8X to 4X to stabilize the system with zero performance hit... the AGP Pro was the next big thing as AGP bus was gonna be saturated, all theoreticall and marketing bull... best thing to do is to put a 1950 pro AGP against an PCI express version withe same CPU (ofcourse ram and co will be part of the benchmark) but I'm pretty sure there's not much gain/loss there....

Umm what proof you need? Max out everything in Doom 3 or any newer game in 1920x1200 w/ 16xQAF + 4xAA and start playing. :cool:

[T2k]

i doubt that, since as we saw in the computex pre X38 boards MSI placed an ordinary heatsink on the chipset not a rollercoaster heatpipe cooler as its the case with the P35, i believe they must have found if cooler to do such a thing

Also it could be a better overall ASIC design so temperature envelope can stay the same.

[Leeghoofd]

Umm what proof you need? Max out everything in Doom 3 or any newer game in 1920x1200 w/ 16xQAF + 4xAA and start playing. :cool:

You think the PCI express version is gonna perform better than the AGP ? and most AGP vidcards don't even perform fluid at those resolutions. I really doubt there's a performance difference between eg the 1950 pro on AGP or PCI express... all marketing mumbo jumbo... it's the vidcard itself the bottleneck not the bus...

[Loony]

It's a pity X38 won't ship with ICH10...that would make it even more worthwhile to make the switch.


When is ICH10 supposed to debut?

[tormentor22]

Also it could be a better overall ASIC design so temperature envelope can stay the same.

Yea thats also a possibility, actually even intel is using the same heatsink they used for the bad axe mobo, didnt seem bigger as i saw in a picture for the mobo, so even usual heatsink may still be able to do it efficiently if they were good

[Leeghoofd]

rumours of an X48 being the overclockers dream as the enthousiast chipset, X38 for mainstream users

[Loony]

rumours of an X48 being the overclockers dream as the enthousiast chipset, X38 for mainstream users


When?

[tormentor22]

rumours of an X48 being the overclockers dream as the enthousiast chipset, X38 for mainstream users
When?

According to this site: http://www.fudzilla.com/index.php?option=com_content&task=view&id=2749&Itemid=37 it will be the same time X38 is introduced.
i believe the X38 will be meant to be a performance chipset while the X48 will be the extreme one

[tormentor22]

i doubt that, since as we saw in the computex pre X38 boards MSI placed an ordinary heatsink on the chipset not a rollercoaster heatpipe cooler as its the case with the P35, i believe they must have found if cooler to do such a thing
Also it could be a better overall ASIC design so temperature envelope can stay the same.

Well it seems they have changed there mind! X38 will actualy use heatpipes much similar to the one used in the P35 version take a look here (http://www.tweaktown.com/news/8079/index.html)