But if you take the 2x4gb you will not use triple channel architecture for the RAM...Also regarding memoery overclock, it is correct that you will not push the ram faster if you overclock using ONLY multiplier...But you ll get better overclocking if you raise the bclk clock instead of multiplier..by increasing the bclk you push also ram and uncore componenets (like L3, IMC etc)....I could nt get my i7 to 4.0 by 191x21 multi, so i settled with 173x23 and i am very happy .....for the time being of course :)))

1155 only supports dual channel RAM. Also, there is no bclk overclocking on 1155 - overclocking is limited to k series CPUs in which you use the multiplier to overclcock. So you're left with either the i5-2500k or the i7-2600k. Luckily, k series CPUs should only be about $20 more than their locked counterparts.

So I guess we'll just have to wait and see what sort of results people get with the 2600K - lets hope it does in fact overclock to close to 5GHz on air. For those that have read the difference between the i7 950 chip, what other difference is there between the 2600K that may slow it down? Its a smaller architecture, 8MB of cache I think, what other differences are there? Given that FSX is so CPU speed dependent, I would hope that we see some good results at these high speeds.James

Sandy Bridge overclocks are not going to be limited by heat like current i7-9xx chips are, but rather by voltage. Because of this, I don't think it's going to matter if you're watercooling or air cooling. Either way, you're going to hit a voltage wall regardless of how cool you can keep the chip. I'm expecting 4.6 to 4.8 to be a realistic 24/7 overclock. Water won't be necessary!Sandy Bridge chips are supposed to be 10 to 20% faster clock for clock. Consider that a realistic 24/7 overclock for an i7-950 is probably about 4.2GHz and Sandy Bridge holds a .5GHz advantage in addition to the 10 to 20% clock for clock advantage. The clock for clock advantage comes from the new optimized architecture. The new cache system is supposed to be a lot more efficient.Another potential advantage of the new Sandy Bridge mobos is that the PCIe lanes operate at twice the bandwidth - granted their are far fewer lanes than on a 1366 mobo. That will probably prove real advantageous for those of us running single GPUs - but not till GPUs utilize that much bandwidth.

Another potential advantage of the new Sandy Bridge mobos is that the PCIe lanes operate at twice the bandwidth - granted their are far fewer lanes than on a 1366 mobo. That will probably prove real advantageous for those of us running single GPUs - but not till GPUs utilize that much bandwidth.

1155 comes with PCIe 2.0 lanes just as 1366 & 1156, where does the extra BW come from please?

1155 comes with PCIe 2.0 lanes just as 1366 & 1156, where does the extra BW come from please?

I know - it's really confusing. Apparently, while current intel chipsets use "PCIe 2.0", they still only operate at 250MB/s. I would say we've been getting screwed all along, but so far no cards to my knowledge can saturate the lanes anyway. PCIe 2.0 on the new Sandy Bridge mobos will be true PCIe 2.0 and support the full 500MB/s. PCIe 3.0, set to release late 2011, will support up to 1000MB/s per lane.SourceFurther Reading

http://en.wikipedia.org/wiki/LGA_2011

I know - it's really confusing. Apparently, while current intel chipsets use "PCIe 2.0", they still only operate at 250MB/s. I would say we've been getting screwed all along, but so far no cards to my knowledge can saturate the lanes anyway. PCIe 2.0 on the new Sandy Bridge mobos will be true PCIe 2.0 and support the full 500MB/s. PCIe 3.0, set to release late 2011, will support up to 1000MB/s per lane.SourceFurther Reading

Ok, those 8 PCIe lanes @ 2.5GT/s I think you're talking about are the P55 lanes dedicated to lower performance devices (4x, 2x, 1x). Those are the 8 PCIe lanes that are connected to the south bridge / PCH and have nothing to do with the graphics 16x PCIe slots. Think of it this way, there's no way 8 lanes could handle a single GPU at 16x or 2 at 8x. The second row in that table "CPU PCI config: 1 x 16 or 2 x 8 PCIe 2.0" is the one used for graphics. Exactly the same for both P55 and P67. And you are completely right, PCIe 2.0 is more than enough for a single card and even 2 cards at 8x8x for 99% of the games out there and for years to come, interestingly enough, not FSX, that will bottleneck the PCIe bus at 8x. PCIe 3.0 is just another useless thing that will lure people into buying the newer 2011 socket CPU's just like quad channel and SATA 3.0 (ok, SATA 3.0 is not exclusive to 2011 just another example of things we don't need at all and still will make us think we'll get better performance when it's simply not true)

So you don't think the 645 new pins that handle those new PCI express 3.0 lanes won't have the same accelleration as we saw when Intel moved the DDR3 lanes from the northbridge to the CPU for LGA1366?

So you don't think the 645 new pins that handle those new PCI express 3.0 lanes won't have the same accelleration as we saw when Intel moved the DDR3 lanes from the northbridge to the CPU for LGA1366?

In single GPU configs at x16 PCIe 2.0 no current card is even close to clog the PCIe bus.Dual at 16x16x PCIe 2.0 in 1366 provides also way more BW than needed so no need for PCIe 3.0It's like SATA 3.0. You'd need dozens of drives to saturate SATA 2.0Integrating the PCIe controller is nothing new really. P55 has that and in X58 the QPI link provides the necessary BW between the CPU and the north bridge anyway so that's no problem (just some extra lantency)

FSX in no way tapped out the memory bandwidth of FSB/Northbridge.... moving that functionality to the CPU in 1366 provided even more unused bandwidth. But still at the end of the day no one with an i7 is willing to go back to a Core Duo...so even if it improves things by 20% for the 100% change in bandwidth, it's still quite a bit of progress.

http://www.com3.es/2...ation-released/

Rumour has it that products based on this new specification could be out as early as June of next year. Intel might adopt PCIE 3.0 in its server products by the end of next year, as well. Then again, the benefits of increased performance might not be that significant to end users actually. As of current state, there’s no need for a PCIE 3.0 specification, as current PCIE 2.0 will not bottleneck current graphic cards at all. The thing here is, all your textures are loaded in the VRAM before you even start rendering, and that’s where your game actually starts getting processes. Once loaded, there won’t be any huge data transfer in and out of your graphic card. Current generation graphic cards will never maximize the PCIE 2.0, and unless next generation graphic cards will have higher power supplied to the card (which is quite impossible as they’re restricted), it is quite impossible to bottleneck current PCIE 2.0. The only possible usage of PCIE 3.0 is SSD on PCIE, and that has yet to lift off in a big way right now.

This sort of articles are the ones on which I base my opinion about PCIe 3.0. What I don't know is if PCIe 2.0 8x8x in 1155 will still be enough in the future for SLI/XFire

Wow it took me like 4 links to get back to the original source, Might explain the information loss. I think you need to consider bandwidth versus actual throughput.. a 20mbit modem is still twice as fast as 10mbit right even if I only send 512 bytes?