Asus P7P55D Premium

ASUS delivers feasible SATA 6Gb/s solution on this new high-end P55 motherboard.

This new motherboard from ASUS has one thing going for it and that is the "proper" - or better - implementation of a SATA 6Gb/s chip for the bandwidth hungry SSDs of this world.
Below, what ASUS calls "Other Storage Solutions", friendly words for bad storage solutions.


ASRock is one manufacturer providing a SATA 6Gb/s add-on card with their top P55 board and is also using the Marvell 9123 chipset.


The PCI-e 2.0 x1 interface provides 500MB/s, still 100MB/s shy.
This can be good or bad, depending on how you look at it. While you can't deliver more than 500MB/s to the CPU/RAM, you can move data between two HDDs/SSDs at the full 600MB/s. The bottleneck is just between the controller and the P55 PCH, other than that everything will be fine. If you're hoping to get that much data to the CPU, you have to settle at 500MB/s and even using RAID on SATA 6Gb SSDs will yield you no more than that.
Same goes for ASRock, the controller can move data at 600MB/s between drives connected to it, it just can go further than 250MB/s to the CPU.


344MB/s in burst speed and not much more than that. It's still 100MB more than current SATA 3GB/s ports but not the improvement that everyone was expecting. A few more months, one upcoming AMD SB800 southbridge and faster SSDs and that will be radically different.

The pricing of the P7P55D Premium is expected to be close to $250.

Source: XFastest

AMD's new chips shader count revealed?

Charlie Demerjian is reporting the the upcoming AMD Radeon HD 5800 cards will feature 1600 shaders, double that of the current shader count of the 4850/4870 series cards.
With such an increase in raw processing power, the card should indeed feature the rumored 384bit bus - like the G80 - although AMD's R600 and R700 chips weren't really that bandwidth starved when compared with Nvidia's latest cards. As for processing power, it will surpass 2.5 Tera FLOPS per second easily, although sustained numbers will be harder to achieve.

AMD's replacement for the brilliant AMD Radeon HD 4770, which will be either called RV830 or RV840, will feature a bump to 800 shader processors and probably an even bigger increase in clock speed. I find it strange that AMD will be releasing a mainstream part this early, so we might see it fit the 5600 series instead of the 5700 series, leaving that for an update later in the life of this generation of DX11 cards.

Both parts look very appealing and will see performance figures towards the end of next week, although I'm very cautious about performance if AMD adopts a 256 bit GDDR5 memory bus instead of a 384bit one.

Source: SemiAccurate

Shuttle X50 All in One

Shuttle unveils an Atom 330 desktop computer with Linux inside.

The Linux PC – available in black or white – is fitted with Intel’s Atom 330 dual-core processor (2x 1.6 GHz), 1 GiB DDR2 memory, 160 GB hard disk, 15.6" touchscreen (39.6 cm, resolution 1366x768), 1.3 megapixel webcam, card reader, microphone and stereo speakers.

Shuttle's X500V All-in-One PC has an array of additional expansion options with rapid Gigabit network, WLAN (Draft-N), 6-channel audio, VGA output and five USB ports. It can be fitted to monitor arms or wall brackets with the VESA mount. The Kensington Lock provides anti-theft protection.

The new All in One is priced at 450eur, without VAT, in Europe. It's an interesting price but it should pack at least 2GiB for this price. Other than that, it seems like a very interesting non-gaming machine.
The Atom 330 is about equivalent to a Core based Celeron at 2GHz, or a similar Sempron, but does outperform them in some applications. Its advantage is lower power consumption from both those parts while still being able to handle 1080p video without a problem.

The bundled Linux version is OpenSUSE 11.1, which I have already tested on an Acer One netbook and which has revealed itself as a very decent OS. It's supported for 2 years, after which you should update it to the next (free) version of the OS from Novell.

Linux compatible hardware is something still hard to come by these days and a price premium may justify itself if you don't get any problems. I've had more than my share with motherboards from ASUS.


Source: Shuttle computer

AMD Phenom II X6 coming in 2010

Advanced Micro Devices is preparing a desktop processor with six processing engines, sources familiar with the company’s plans revealed. The new central processing units (CPUs) will not be available this year, but are likely to boost performance of AMD’s desktop platforms sometime in 2010.

"Thuban", as it is codenamed, will be released in the AM3 socket and will feature a total of 3 MiB of L2 cache and the same 6 MiB of L3 cache as "Istanbul" based hexa-core Opterons. "Istanbul" cores top out at 2.8GHz, not very far from the 3.4GHz of the AMD Phenom II X4 965.
2010 is a bit late, considering AMD has been shipping hexa cores for a while and that they can fit AM3 sockets without a whole new spin. Don't expect it to break ground on the gaming front but this new Phenom II X6 processor may become the chip of choice for high-end workstations and cheap scientific computing clusters.

Update: new details about the new processors here.

Source: Xbitlabs

Powers Of Ten, Powers Of Two, GB and GiB

One of the more common misconceptions on the IT and consumer electronics market is that there's only one measure of how a computer can store information. In fact, there are two: the SI prefix and the SI binary prefix. This new designation was standardized ten years ago, in 1999.

A "Ki", or kibi, is the prefix for 1024, or a KiB, if we're talking about bytes. The KiB symbol corresponds to a kibibyte whereas a GiB corresponds to a gibibyte. The new prefixes are contractions of two words: kilo binary and giga binary.
This distinction between byte and binary bite is the reason why sometimes my posts have GiB instead of the more common GB, especially when referring to the systems or graphics card memory, where they are indeed MiB or GiB and not MB or GB. While I try to keep things right, I do sometimes interchange between the two.

Typically, hard disks come with capacities of multiples of 10 while anything with RAM or Flash related comes with capacities that are always multiples of 2 - like most everything in a computer, including 3 bits per cell NAND, which will yield a power of 2 number of values, 8 different values per cell.
Back in late 90s, I thought hard drive manufacturers were ripping me off by stating less actual capacity then what they were selling, these new suffixes didn't even exist but nowadays that situation hasn't changed, the world around it has. HDDs still come with base 10 SI prefix but manufacturers are being just plain honest, it's the rest of the world has still not woken up to the new reality of binary prefixes.

Let's have a look at a modern SSD, a new and mostly unused OCZ Solid with 30GB:

~ $ sudo fdisk /dev/sdcCommand (m for help): p

Disk /dev/sdc: 32.0 GB, 32044482560 bytes

As we can see, the device actually has 32GB of FLASH memory inside. The label should have read "OCZ Solid 30GiB", right? Almost:

32044482560/1024/1024/1024 = 29.84375 GiB (the real)
or
30*1024*1024*1024 = 32212254720 = 30GiB (the ideal)

One can't complain can it? While OCZ is selling the drive as 30GB unit, it actually has 32.044GB, or 29.84GiB.
It was nice of OCZ to do this and they also sell both Vertex and Agility models the same way.

When it comes to RAM, you're actually always getting something with MiB or GiB of memory and not MB or GB. Much of the data that gets stored on RAM is a multiple of 32bit (4 bytes) that gets written in memory, be it either the system's memory or the graphics' card memory. Information that is not, typically text, that is counted in characters each taking up one byte, is usually not.
So, much information is packed in multiples of two: textures in games are 512x512, 1024x1024, 2048x2048, etc. This is useful because pages in memory also are store in multiples of 1024, 4KiB and 2 or 4MiB. If you're not using a size that fit pages in an efficient way, then you're wasting space that could have gone better used but there are also other reasons for this.

Flash is an example of a storage technology that should always come be counted with binary prefixes and not regular base 10 prefixes. Flash can either store 2 bits per cell, 4 bits peer cell or 8 bits per cell, in the above mentioned upcoming MLC chips. How can you then sell a device with precisely a GB when it's flashed based? It's not easy, hence the "not a Gigabyte nor a Gibibyte" actual measure of the OCZ Solid* storage capacity.

Operating systems, most notably Linux, have been counting information on binary bytes for some time and I've stumbled with ever increasing exposure to the new SI also on web applications. It's still too little but there's some adoption.
Starting today, Silicon Madness will also take that route and will only mention capacity by it's right SI. Even while as strange as a Gibibyte might sound, we won't restrain ourselves from using that term.


* The device hasn't got enough usage that I can believe the lack of space to make up 30GiB is due to lost space in damaged flash cells but more this actual inability to produce a capacity in base 10 from a base 2 type of device. Notice the last 4 numbers of the actual size of the device: 2560. 2560 is 1024*2+512.

AMD and the APU, 785G performance shortcomings

Intel is moving the GPU next to the CPU, starting with the 32nm "Westmere" architecture based "Clarkdale" processors, which will span, at least, Core i3 and Core i5 models by the end of the year.

As I mentioned on the article about Nvidia's missing chipsets for the LGA 1156 platform, Intel needs "Clarkdale" desperatly if it wants to have "Nehalem" based CPUs with integrated graphics available in laptops. There simply isn't an easy and cheap way to fit an integrated graphics core on the chipset anymore, now called PCH(Platform controller HUB).

AMD is also moving in the same direction, probably more by need than to just claim be the first to integrate both the CPU and GPU, or even to come up with something truly innovative - currently, AMD is calling it's integrated CPU/GPU the APU(Advanced processing unit).
The need for AMD to integrate is similar to the motives that also drive Intel: the integrated memory controller sits on the CPU now and the ever increasing bandwidth that it provides needs to be fed to a far place, the chipset. AMD relies on HyperTransport to do this, which has been enough up to now but is starting to become a bottleneck to the integrated graphics core on the chipset.

Theoretically, HT 3.0 is able to provide 10.4GB/s when running at 2.6GHz(5.2 Giga Transfers/s) on the 16 bit HT that AMD currently uses. AM2, AM2+ and AM3 platforms have been engineered to deliver up to that HT clock and no more. That is still enough to serve as a processor interconnect but AMD still hasn't scaled that high: it's still at only 2GHz(4GT/s) for the higher end Phenom II X4 processor.

2GHz @ 16 bits is enough to provide 8GB/s, enough bandwidth to keep dual PCI-e 2.0 x16 slots fed but not much else. Remember that the current push for SATA 6Gb/s will push another 600MB/s per SATA channel to the CPU - over the HT interconnect - and then bottlenecks will become more apparent than they are today. While that wasn't an issue while HDDs were around, SSDs have been pushing the envelope and are already saturating SATA 3Gb/s ports.
AMD could use 32 bit wide HT links but those would increase the costs of AMD motherboards, which is definitely an advantage to go AMD rather than Intel right now.
Another issue is PCI-e 3.0, which will push 8GB/s per x16 slot, a bandwidth target that not even a 5.2GT/s link will be able to sustain if you need to transfer information to both cards at the same time. This is not a big issue for games but is much more apparent when using GPUs to offload computation and it frequently becomes a bottleneck when doing synchronization to more than one GPU at the same time. My experience coding CUDA for the last 6 months has brought this to my attention more than what I'd previously thought and there is a dire need for the move to PCI-e 3.0 when doing GPGPU.

When you think of the platform like this, it becomes apparent why AMD might have refrained itself from upgrading the graphics core in the new RS800 series chipsets, comercially known as AMD 785G - while the shader model support is now compliant with DX10.1 requirements, the performance hasn't increased much as the GPU still retains the same 40 shaders that the 780G did. AMD might have chosen to stick to the rather old 40 shader core due to lack of available bandwidth to feed the GPU rather than for die space and cost constraints.
Just look at bandwidth: AMD's integrated controller can deliver around 9GB/s in AM3 form. Right now, you can't even feed that much to the integrated core if you'd like to(except with OC, of course) - the current AM3 HT 3.0 implementation tops out at 8GB/s.
AMD needs to move the GPU closer to the core soon, where high-bandwidth interconnects are cheaper to build. Before AMD can do that, it will be very hard to see another iGPU push the envolope like the 780G did when it was released, there simply isn't enough bandwidth. You can simulate this yourself, just mess around with the HT bus speed and see the effect it has on performance of the integrated core. Hopefully, I'll be able to provide some benchmarks of that since I've already found some unexpected situations(XBMC, for instance), where the integrated GPU can be starved for bandwidth doing simple things like browsing through the interface.

In the end, while AMD can still try and push for a higher bandwidth with HT 3.1, which could deliver 25.6GB/s, running 6.4GT/s @ 32bits, it would still have to suffer from more expensive motherboards and a whole new platform. At this point, why not just move a GPU closer to the core, in an Multi Chip Module(MCM) solution, just like what Intel is doing with "Clarkdale"? The MCM solution isn't planned but the APU is. With either route, performance is expected to go up considerably and Intel is looking like it will be the first to capitalize on the early move, courtesy of the MCM solution.