Corsair P3: SATA 6 Gbps from Corsair CES 2011: filling up on Fusion notebooks Marvell 88SE9130 & HyperDuo 1st Vertex 3 Pro tests CES 2011: Nvidia launches the GeForce 500Ms CES 2011: Microsoft Windows and ARM CES 2011: Nvidia is preparing an ARM CPU

Janvier 2011 L M M J V S D 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 You can read previous news using our archives. [ Print all news of this day ]

<< Previous day's news January 4, 2011

Next day of news >> January 7, 2011

Corsair P3: SATA 6 Gbps from Corsair   Posted on 06/01/2011 at 20:44 by Marc

	Corsair has announced a new range of SSD, the Performance 3. Like the Crucial C300 and its successor, the Crucial M4 (equivalent of the Micron C400), it is based on the SATA 6 Gbps Marvell 9174 controller. It should be available in 256, 128 and 64 GB versions with the respective sequential read/write performances of 480/320, 410/210 and 365/110 MB/s. Here then, Corsair is taking things on from the Crucial M4, which runs at 415/260 MB/s max. It remains to be seen of course what random access performances will be. The Corsair P3-256, P3-128 and P3-64 should be available come end of January.

CES 2011: filling up on Fusion notebooks   Posted on 06/01/2011 at 17:30 by Damien

	AMD seems to have won over a good few manufacturers with its entry level APU and the Brazos platform, which offers the opportunity of developing netbooks beyond the standards of those based on Atom. Before the opening of this year’s CES, we saw many different designs on the AMD stand, some of which are ready already, proving the attraction of the platform. This puts AMD in an interesting position going forward with a product that differentiates itself from the Intel offer in this segment, something we haven’t seen for a long time and perhaps never in the domain of mobile computing! Brazos laptops of every size. Brazos from Acer, HP and in the Lenovo ThinkPad range.

Marvell 88SE9130 & HyperDuo   Posted on 06/01/2011 at 13:46 by Marc

	Marvell has just announced a new Serial ATA 6 Gbps controller, the Marvell 88SE9130. This additional chip, like its predecessors, is connected to a PCI-E 2.0 lane and offers two additional SATA 6 Gbps ports. It also supports a new technology called HyperDuo. The concept of HyperDuo is simple: it combines an SSD with an HDD to get the best of both worlds. Both peripherals are visible as distinct volumes to the end user, with HyperDuo managing distribution of files to the SSD or the HDD according to how often they’re used. You can also chose where files go manually. Advanced users may well prefer to retain control of the distribution of their data and use the two distinct volumes as they please, bypassing HyperDuo. HyperDuo works in two modes, with Safe Mode mirroring the same data on the SSD and the HDD. The capacity of the logical volume grouping the HDD and SSD then corresponds to the capacity of the HDD. Capacity Mode combines the capacities of the two drives into a single logical volume but without duplication of data. HyperDuo works on the same principle as hybrid drives and extends it to two distinct peripherals. ASRock, ASUS, Gigabyte and MSI are likely to be using the 88SE9130 in forthcoming motherboards. It remains to be seen if the HyperDuo drivers will also support TRIM, something we’re still waiting to hear from Marvell on. Note moreover that the future Intel RST 10.5 drivers are likely to integrate a similar technology, named RST SSD Caching.

1st Vertex 3 Pro tests   Posted on 06/01/2011 at 11:06 by Marc - source: AnandTech

	OCZ has taken advantage of CES to show off its first SSDs based on the new generation SandForce controller. These are the Vertex 3 Pro and Vertex 3 EX. Both are based on the SF-2582 and will become available in the second quarter of the year. The pre-version of the Vertex 3 Pro on show uses Toshiba 32nm NAND but the definitive versions will be based on 25nm IMFT NAND, in MLC or eMLC versions. The Vertex 3 EX will use SLC. While SandForce says that its new family of SF-2000 SATA 6 Gbps controllers will enable sequential speeds of 500 MB and 60K IOPS in 4 KB random accesses, OCZ goes further and is talking reads of 550 MB/s, writes of 525 MB/s, 80 IOPS SLC and 75K IOPS MLC for the Vertex 3s. AnandTech have carried out a few tests on this pre-version of the Vertex 3 Pro. The first figures obtained are pretty impressive, with read speeds of 493 MB/s, almost double what you get on an SF-1200/1500 120 GB. The same goes for writes which vary between 518 MB and 262 MB/s depending on how compressed the data is. Random writes are up to 40% faster, while random reads are tripled, though this figure does seem suspect.

CES 2011: Nvidia launches the GeForce 500Ms   Posted on 06/01/2011 at 08:53 by Damien

	NVIDIA is following AMD’s example and has announced its new range of GeForces designed for notebooks. Coming as they do, just a few months after the GeForce 400Ms, don’t hold your breath for any major innovations: what we have here is a renaming of the old range, with clocks adapted for some models. AMD and NVIDIA both market products under new names when each new range of laptops comes out. Although they both point the finger at those dastardly OEMs who supposedly force them to act in this way, we shouldn’t let the wool be pulled over our eyes, this renaming is first and foremost a deliberate sales strategy which allows them to maintain higher pricing without losing too much ground when new products aren’t ready in time for a new cycle. The first stage for NVIDIA is to convert their entry-level cards to the GeForce 500M family: A particularity of all these models is that either GDDR3/DDR3 or GDDR5 may be used. GDDR5 gives twice the bandwidth, which will have a significant impact on performance. You’ll therefore have to be attentive to the choices of differenct laptop manufacturers, likely as they are to go for the cheaper DDR3. The GeForce GTX 445M has been renamed as the GT 555M but is unlikely this time to be released with a complete 192-bit memory bus and come with faster GDDR5 memory. The other models are all based on the GF108, NVIDIA’s entry-level DirectX 11 GPU. The clocks on the GeForce GT 550M and 540M have been revised upwards in comparison to the GeForce GT 435M and the GeForce GT 525M is situated mid-way between the GeForce GT 435M and 425M. Note however that though the GeForce GT 520M is clocked higher, it has half as many processing units as the GeForce GT 420M. In addition to these GeForce 500Ms, NVIDIA has introduced a GeForce GTX 485M based on a full GF104 GPU, similar to the one we should have with higher clocks on the forthcoming GeForce GTX 560: This model will be a far superior replacement for the GeForce GTX 480M based on such a cut down high-end GF100 (so as not to go beyond the already very high thermal envelope of 100 watts) that its performance will be lower than this mid-range GPU. Note that all these GPUs still retain Optimus support, which allows the system to juggle transparently between an IGP and a dedicated GPU so as to give optimum performance / energy consumption. Moreover, NVIDA has announced that 80% of laptops which use these cards will also have Optimus technology as most manufacturers have this time accepted to pay the premium billed by NVIDIA for its activation.

CES 2011: Microsoft Windows and ARM   Posted on 06/01/2011 at 08:03 by Damien

	This is no doubt the most important announcement of this year’s CES: the next version of Windows will exist in an ARM version as well as x86/x64 versions. A revolution in the world of PCs is on its way. With the multiplication of SoCs (System on Chip), the explosion of their capacities and the new usages which result (such as tablets), Microsoft could no longer ignore them. Microsoft CEO, Steve Ballmer unveiled a strategy which will consist in providing a version of the next Windows optimised for both ARM and x86 SoCs, which will mean that AMD and Intel on one side and Qualcomm, Texas Instruments, NVIDIA and plenty of others on the other, will be fighting for the same terrain: PCs. Of course, all the software and drivers will have to be adapted to support the ARM version, which will require a considerable amount of work for all the players on the market, justifying Microsoft’s early notice on this forthcoming support. Microsoft itself will be porting Office and all the other smaller Windows applications. Indeed Microsoft already has a version of Windows running on ARM. This one is still basic and is based on Windows 7, but works, as we were able to observe in the demo on Qualcomm’s Snapdragon, Texas Instrument’s Omap and NVIDIA’s Tegra 2.

CES 2011: Nvidia is preparing an ARM CPU   Posted on 06/01/2011 at 00:00 by Damien

	There have been regular rumours of the development of an NVIDIA CPU for some time now. Jen-Hsun Huang, NVIDIA CEO, has used CES to make these rumours official and mark what could be a very important development for the company. This CPU, known under the name ‘Project Denver’, has been under development for several years now and is based on ARM rather than x86 architecture as some people had been thinking. NVIDIA says this choice is down to the explosion of the ARM ecosystem, which is giving increasing importance to the architecture, especially as Microsoft has just revealed that Windows 8 will be ARM compatible! There is however a problem with ARM architecture that prevents it from being used more generally: it mainly targets mobile peripherals for which battery life is of course more important than performance. This is where NVIDIA comes in. With project Denver, NVIDIA has announced the development of a cutomised ARM core which targets high performance, without however mentioning any of its architectural specifics. If these ambitions become a reality, they could open a whole new world up to ARM architecture: NVIDIA is talking smartphones and PCs, servers and consoles! Might x86 be under threat? It’s difficult to say yet, especially as no precise dates are in place and the fabrication process will be a crucial factor in this new war declared by NVIDIA.