Monday28 November 2022

PCIe GEN 4.0 to be released in 2014-2015, aimed at improving tablets

Reading time is around minutes.

744px-PCI_Express_logo_svgWith PCIe 3.0 just coming on to the scene the PCI special interest group (PCI-SIG) just announced that they are planning on releasing PCIe 4.0 sometime around 2014-2015. Now that is assuming we all survive the Mayan apocalypse, but if we all do then it looks like the interconnects between the devices on our systems will get quite a bit faster. According to the press release, the new specification can move 16 “Gigatransfers” (is that even a word?) per second.

Now, the question that most people are asking is; why is this coming out now? As we mentioned PCIe 3.0 has just been released and there are very few components that can support it. We know that MSI has shown some of their boards can and Gigabyte can support is as long as you are in single card mode (it supports PCIe 3.0, just not in native mode).  Meanwhile Intel is said to be adding in PCIe 3.0 support into their Ivy Bridge CPUs that should hit the market around April-May 2012. So why give PCIe 3.0 less than two years for its life cycle?

One of the reasons is that the PCI-SIG is looking at the future of the computer industry. As it stands right now many of our fastest components are limited by the PCIe interconnects that are present. One of these is the current (and anticipated) SSD technology. Most chipsets do not offer more than 1-2 PCIe lanes for data transfer each lane of the PCIe 2.0 specification is 500MB/s. if you have one lane of PCIe per device (or two for your typical SATA 3.0 controller) you are still missing 100MB/s of potential data transfer. Now there is the argument that you will never see the full 600MB/s that SATA 3.0 is supposed to offer, but the simple fact is that with the current way boards are built there is a true physical limitation to this. PCIe 4.0 will help to remove this barrier, not only for the existing SATA specification, but also for the next generation.

Another issue comes in mobile and embedded devices where you do not have the options to double up on lanes or throw in multiple bridge chips to add more PCIe lanes. Here you still want to have high bandwidth, but are space and power limited. If you can drop in the same number of lanes but double the speed available you can significantly increase the performance of tablets. This move means that the tablets available in 2015 could potentially be as much as 2-4x faster than what we will see even in 2013. This opens the way for 1080p and higher resolutions on tablets. Can you imagine gaming at 1080p on a 10.1-inch tablet and sill having headroom for an SRS surround sound CODEC?

As for PCIe 3.0 it is a step in the right direction, but even PCI-SIG seems to be admitting it is not a long term solution. While it does deliver twice the bandwidth of PCIe 2.x (offering 1GB/s Vs 500MB/s)and removes the 20% data overhead per lane. There are other factors that have to be taken into account to get it running using the current technology which the PCI-SIG said was possible. In fact Gigabyte and MSI ended up in a fairly public argument over what constitutes PCIe 3.0 support.

With PCIe 4.0 we could potentially see many of the system bottlenecks opening up and allowing systems to take advantage of all of the high speed components that are connected. Although most consumers only think of PCIe in terms of GPUs, PCIe is needed to sort out the increasing demand we are putting on our systems as USB 3.0, SATA 3.0 and even newer technologies become native on the boards. Pretty soon we will run out of bandwidth and despite improvements in CPUs and Memory, our expensive computers, laptops and tablets will begin to slow down.

Read the full press release here

Discuss this in our Forum

Last modified on Thursday, 01 December 2011 07:16

Leave a comment

Make sure you enter all the required information, indicated by an asterisk (*). HTML code is not allowed.