|Although the Lynx platform, namely Llano desktop, is slated for next month, there were of course many socket FM1 motherboards on show at Computex. They were all based on the Hdson-M3 or the A75 chipset which supports USB 3.0 and SATA 6 Gbps natively. While several have two PCI Express 16x ports, the second is generally cabled at 4x, connected to the southbridge and only functional if the 1x ports are not used. Only Gigabyte and Jetway currently have a model with two ports cabled at 8x and connected staight to the Fusion APU.
Given the native support for USB 3.0 and SATA 6 Gbps, the manufacturers don’t need additional controllers. When they support eSATA however, one of the SATA ports is monopolised for this purpose which limits you to 5 ports.
Most manufacturers told us that they had been getting ready more variants than were on show but were still very vague about which market will be targetted for this solution. Mid range? Low end? Mini-ITX?
The Asrock A75 Extreme6 is an example of quite a complete version of the A75 chipset: two PCIE ports in 16x format but one of which is cabled at 4x, five SATA 6 Gbps ports, VGA, DVI and HDMI connectivity, THX certification and Firewire.
Asus is proposing several models. The first, the F1A75 is in ATX format, has a UEFI bios and one PCIE port connected to the APU at 16x, one 4x port, two PCIE 1x ports… and no video connectivity. Either Asus is planning to move it onto a bracket or not use the APU’s connectivity, which is strange to say the least.
Next, the F1A75-M is a micro-ATX version of the F1A75 which has VGA, DVI and HDMI video connectivity.
Finally, the F1A75-I is a mini-ITX model in a totally different design of course. It has just one PCIE 16x port and four SATA 6 Gbps ports. It has DVI and HDMI connectivity and also Displayport and, in the Deluxe version, Bluetooth. Note that the fixture system of the CPU cooler, but not the socket, has been rotated by 90° compared to all the other motherboards.
The Biostar TA75A+ has a PCIE 16x port, a 4x and two 1x port, six SATA ports and VGA, DVI and HDMI connectivity. The TA75M+ is similar but in micro-ATX format.
ECS were showing the A75F-A in ATX format with a standard configuration: one PCI 16x port, a second 16x cabled at 4x, two PCIE 1x ports, five SATA 6 Gbps ports and VGA, DVI and HDMI video connectivity.
The A75F-M is in micro-ATX format and only retains the PCI 16x port connected to the APU. It does however have an extra SATA port, making 6 in total, as it doesn’t have eSATA connectivity. Finally, the A75F-M2 is a more compact version of the A75F-M with just two memory slots and no DVI connectivity.
Foxconn was showing two relatively basic models, one in ATX fomat and the other in micro-ATX. Strangely there’s no HDMI connectivity on the ATX version.
In addition to the GA-A75-UD4H, which supports two PCI Express 8x connectors connected to the APU, Gigabyte were showing the GA-A75M-UD2H in micro-ATX format and whose second PCIE 16x port is cabled at 4x via the chipset. As with numerous other models, it has five SATA 6 Gbps ports, the sixth being used as an eSATA.
The Jetway JHA16 covers more bases with two PCIE graphics ports that can work at 2x 8x, six SATA 6 Gbps ports, VGA, DVI and HDMI connectivity and a mini-PCIE slot.
MSI were showing two A75 models. The first, in ATX format, is the A75A-G35. It has a PCIE 16x connector, three PCIE 1x and one PCIE 16x ports cabled at 4x but which prevent the simultaneous use of the 1x ports. While the card does have six SATA 6 Gbps, it only has two memory slots, which is strange for an ATX model.
The A75MA-G55, although in micro-ATX format, is higher end than the first model. It has a better quality power stage and four memory slots. Although CrossFire X support has been announced, it looks as if it’s provided via a second graphics port connected at 4x to the chipset.
Finally the Chinese manufacturer, Weibu, was showing an A75 model with WB-A75+PWR which offers full video connectivity but audio connectivity limited to the AC97.
|Expected for the end of the year, socket 2011 is here en masse at Computex, each manufacturer showing at least one motherboard model equipped with an X79 chipset. Overall all the manufacturers seem to be turning towards a similar design that consists in placing the four memory channels in pairs on either side of the CPU socket. Given that this chipset still doesn’t offer native USB 3.0 support, dedicated controllers are of course present… but not in all models! Some are still in prototype form of course and don’t yet include some of the refinements that will be present in the final versions, such as radiators or the colour of the PCB. Note that the PCI Express ports are in 3.0 format here and therefore one 8x port on X79 is similar to a 16x port on the previous platforms. Note also that, with just one exception, there aren’t any models offering numerous PCI Express 16x ports via a switch such as the NVIDIA one, as these are currently limited to PCI Express 2.0.
The Asrock X79 Extreme 4 has 3 PCI Express ports in 16x format. While two are cabled at 16x, the third is limited to 4x and connected to the chipset. An additional controller is included for the support of two USB 3.0 ports.
Asus is showing no fewer than 6 versions of an X79 prototype motherboard. The standdard CIX79 and the, LE, EVO, PRO, PLUS and DELUXE. They all have the same base but come with varying details: digital audio out, FireWire, Power buttons, debug led, USB 3.0…
The ECS X79R-A seems almost finalised. It’s a high end model with four PCI 16x ports (cabled at 8x when they are all used), 2 USB 3.0 controllers and Bluetooth.
Gigabyte are showing the GA-X79A-UD3. The card has five PCI Express 16x ports, two of which are actually cabled at 16x. One of these drops to 8x when the third port, also at 8x, is used. The last two are at just 4x. An NEC controller takes care of the USB 3.0 support.
MSI is showing three different cards based on the X79 chipset. The first, the X79A-GD65 has three PCI Express connectors in 16x format, one of which seems to be cabled at 4x. Two NEC controllers manage the USB 3.0 support. The X79A-GD70 is similar but employs better quality components, mainly in terms of the CPU power stage which is a good deal sturdier. Finally, an X79A-GD80 prototype has the particularity of having no fewer than four PCI-Express 1x ports in addition to the 16x format ports.
The Sapphire PB-CI7X79N is probably the highest end model on show. In contrast to the other manufacturers, Sapphire has decided to use an NVIDIA NF200 PCI Express switch although it’s limited to PCI Express 2.0. As a result, of the six 16x format ports, only two are PCIE 3.0 compatible, one at 16x and the other at 4x. The four other parts share 32 PCI Express 2.0 lanes, either at 2x 16x or 4x 8x. USB 3.0 is of course supported via four ports at the back of the card and via an additional connector for a bracket or ports at the front.
Finally, a model a little bit different, the P270WM from Clevo, which aims to offer a (trans)portable socket 2011 very rapidly.
|The various motherboard manufacturers have taken the opportunity provided by the Intel Z68 Express release to launch various motherboard models. Here’s what’s coming:
- 3 ASRock
- 5 ASUS
- 11 Gigabyte
- 2 MSI
Gigabyte has obviously been the most active and is bringing out lots of (too many?) models. They have in fact rolled out the whole standard P67 range as Z68 Express versions, without however including IGP connectivity. As they don’t have Virtu either, the only advantage of these models is SSD Caching and you have to wonder what these motherboards really bring to the table given that the Gigabyte has announced that it won’t be stopping P67 Express production either.
They have in any case taken the opportunity to boost the overall specs of their range a bit, with the UD3, UD4 and UD5 now getting a Marvell SATA 6G controller that only the UD7 had on the P67 range. FireWire has also been rolled out on the UD3 and UD4 in the Z68 range. On the P67 boards it was only included on the UD5 and UD7. Four other motherboards have video outs for the IGP, two ATXs and two micro ATXs.
ASUS has also chosen to bring out cards without video outs for the IGP – two out of five in the Z68 range, but they do get Virtu for anyone wanting to use QuickSync. Only the P8Z68-V Pro is available immediately and we’ll have to wait until the end of the month for the others, the specs for which aren’t therefore definitive. Note that two Maximus range motherboards are on the list, the first, a Micro ATX, will be affordable (€169) but the Extreme-Z with its 4 PCI-Express x16 slots will cost €349.
ASRock is bringing out three cards, one of which is a Micro ATX. They all have the connectivity needed for direct IGP support. Lastly MSi seems the least convinced by the Z68, with just two models, which isn’t necessarily a bad thing as long as they cover most of the bases, as they do with the high end Z68A-GD80 and the more limited and less expensive Z68MA-ED55. On this last model however it’s a shame that the second PCI-E x16 port is cabled at x4. If you’re going to include two ports you might as well go for x8/x8!
|While the LGA 1155 processors were universally well received at launch in January 2011, they marked a new stage in Intel’s segmentation strategy. On LGA 1156 you could overclock all CPUs via the bus whatever the chipset. On LGA 1155 however, only the P67 chipset and a processor with Turbo (Core i5/i7 but not i3), what’s more in the K series, would enable processor-based overclocking (using the multiplier).
Impossible then to overclock on an H67 motherboard as you could with H55/H57. Indeed you absolutely had to choose between overclocking the CPU and using the IGP and this at a time when the IGP was becoming a more powerful piece of hardware with QuickSync, acceleration of H.264 transcoding.
Intel has now launched a chipset as part of the Intel Series 6 range, the Z68 Express which combines the qualities of the P67 and H67 Express, namely:
- P67: Overclocking CPU via multiplier
- P67: Option to use 2 PCI-E x16 (cabled at 2x8)
- H67: Access to the IGP (+ overclocking IGP)
As a way of really underling the enthusiast nature of the Z68 Express, Intel has allocated it the latest innovation introduced by the Intel Rapid Storage Technology 10.5 drivers, the Intel Smart Response Technology, which allows you to combine a hard drive and an SSD. Also rolled out on the mobile HM67 and QM67 chipsets, this technology isn’t available on the rest of the range, which is a purely software limitation.
The principle is similar to the one used for hybrid hard drives. Smart Response analyses hard drive accesses so as to place the most frequently used data on the SSD. This data is therefore placed on a separate physical peripheral but the two (HDD and SSD) make up just one peripheral, the size of the hard drive, as far as the system is concerned.
The gains announced in practice are impressive and the level of performance is close to what you get on an SSD, but as is often the case with hybrid drives, the gains aren’t necessarily representative. Of course, when you launch the same test several times, the data is placed on the SSD and accessing it is then a good deal faster.
The only thing is that it’s objectively difficult to say what sort of performance a user will get in practice apart from that it will be somewhere in the interval separating HDD and SSD performance, closer to the first if you’re processing a lot of data and closer to the second if you’re not processing all that much. In that case it will make more sense to go straight for an SSD for the system and most used applications, especially as the secondary hard drive could then be kept in standby mode. The advantage of Intel SRT therefore seems questionable… unless you already have a system SSD and a second SSD lying around unused which can then be recycled for this purpose, probably a pretty unusual situation.
Note that Intel has launched a new SSD for this technology, the 311 Series SSD. With a capacity of just 20 GB, it uses 34nm SLC flash which gives it excellent endurance and above all decent write speeds (115 MB/s) where a 320 Series 40 GB SSD will only give 45 MB/s. The only problem is it costs $110, which is what a 60-64 GB MLC SSD costs.
Underlining the high-end multifunction aspect of the Z68 Express, Intel is also highlighting its use of Lucid Logix Virtu technology. This additional software layer allows you to connect the screen to the IGP (i-Mode) or additional GPU (d-Mode) physically and move from one to the other on the fly, depending on the application, which allows you to benefit from the IGP’s features ( such as QuickSync) without being limited by its 3D capacities.
Note that Virtu has a 5 to 10% impact on PU performance in 3D if you’re using the IGP, an inevitable cost given that the rendering carried out in the IGP frame buffer has to be recopied before it’s displayed. Moreover, the application must have a profile in Virtu, otherwise it will run on the IGP! As this configuration doesn’t really give any gain in terms of energy consumption, given that during 2D processing on the IGP the graphics card won’t be completely turned off, it makes more sense to virtualise the IGP with the GPU for QuickSync encoding as the impact on performance is then negligeable and the encoding applications managing the MediaSDK less numerous than the games requiring a good GPU! d-Mode is also available in the absence of a graphics connector for the IGP on the motherboard, which will be the case for some models.
Whether they like it or not, motherboard manufactures, already heavily affected by the bug on B2 chipsets, will launch multiple Z68 Express motherboards. These cards will of course be more expensive than the existing P67 and H67 motherboards as the mark-up Intel is asking for on the Z68 will have to be added to the initial model price as well as costs linked to additional video connectivity as well as any charge for the Lucid Logix Virtu license and of course some margin to make up for the inherent launch costs. All said and done the Z68s are likely to cost €10 to €30 more than the H67s/P67s!
As you’ll have understood by now, our early feelings on the Z68 Express are pretty lukewarm. The old adage better late than never may come to mind but instead of simply correcting its abusive segmentation between plaforms, what Intel has done is add complexity to its range by positioning what is at the end of a day nothing more than an H67+ as the crème de la crème in Sandy Bridge chipsets. It would have been more elegant to release a new H67 chipset that allowed CPU overclocking and offer Smart Response Technology across all Intel Series 6 chipsets with a driver update.
Thankfully Intel seems to have decided to change its policy on the forthcoming Series 7 chipsets, codename Panther Point, slated for 2012. According to the latest rumours
the three versions, H77, Z75 and Z77, should allow overclocking and benefit from IGP support, with the differentiations coming uniquely in terms of the Intel Smart Response Technology and the routing of the CPU’s PCI-E 3.0 lanes between 1, 2 or 3 graphics ports. It remains to be seen if QuickSync 2 included in the forthcoming Ivy Bridges will be accessible without going via the IGP or a third party software layer such as Virtu, as this really would be a plus. In the meantime, we’ll have to make do with the current state of affairs, which is far from satisfactory, and pay the not negligeable additional cost if you want to have access to the IGP for one reason or another at the same time as overclocking. Something worth thinking about and which means the “standard” P67 and H67 soluitions still have an important part to play!
Copyright © 1997- Hardware.fr SARL. All rights reserved.
Read our privacy guidelines.