New IGP for Socket 775

Nvidia is apparently planning to start selling the GeForce 9 based integrated graphics chipset 730i for Intel in August.

There will apparently be two versions:

• MCP7A-U based on GeForce 9400 - 580MHz core / 1500MHz shader
• MCP7A-S based on GeForce 9300 - 450MHz core / 1200MHz shader

Both will support:

• 1333MHz FSB
• dual channel DDR2 / DDR3
• PCIe 2.0
• 16 stream processors
• HDMI
• dual link DVI
• DisplayPort
• D-sub
• Hybrid SLI
• PureVideo
• Cuda

Mobo has extensive cooling and a nifty BIOS

It looks like Asus is going to bring their Republic of Gaming branding up another level with the release of the upcoming X48 chipset motherboard, the Rampage Extreme.

According to reports coming from other side of the world (specifically: Taiwanese technology site Hard Spell), with the Rampage Extreme you can overclock the FSB to 710 MHz (2,840 MHz effective). For comparison's sake, generally most new motherboards are doing well if the can overclock to the 500 MHz mark.

From the picture below, you can see that this overclocker has a E8400 running at 710 * 6, which translates into a system running at 4.26 GHz.

Besides awesome overclocking potential, the Rampage Extreme also features a BIOS that can be adjusted on the fly. That's right -- you don't even have to reboot to overclock your system for a gaming session. Called the "TweakIt", the new motherboard also has a LCD display right on the board with a small little joystick that you can use to adjust your clocks. 

To support these high speeds, this Asus board reportedly has high quality capaciitors, and extensive heat sinks on the North/Southbridge.

For further information or more pics, follow the source link below.

Nvidia caves in, will allow SLI on upcoming x58 motherboards for Intel

Nvidia blinked first.

Intel did not want to give Nvidia a CSI license because Nvidia would not allow SLI on intel motherboards with multiple GPU slots.

Nvidia did not want to give Intel an SLI license because it wants to sell motherboards... it would REALLY prefer if people could only get SLI if they bought Nvidia motherboards.

Unless there is an unannounced backroom deal giving Nvidia CSI, Nvidia caved, and there will be SLI Nehalem motherboards - with an Intel x58 chipset and an Nvidia nForce 200 chip that has "patented SLI technology for graphics bandwidth management and multi-GPU peer-to-peer comunications both required to optimize graphics performance" - which is marketspeak for we would only let Intel do SLI if we got to add a small chip for which we will charge an arm, leg and reproductive organs for.

Technically, without driver games/restrictions (read DRM) there is NO reason why SLI should not work on ANY multi-slot PCIe motherboard that supports at least 8x on the slots. None. Zip. Zilch.

Remember the SLI patches that were floating around on the net for non-Nvidia chipsets a while ago? Legal threats and driver changes killed those.

Actually, the nForce 200 chip WILL have one technical result: it will slow down the GPU's - instead of talking to 16x PCIe 2.0 from the x58 directly, a single 16x PCIe 2.0 slot will be "splt up" by the nForce 200 to two PCIe "16x" slots, just like on the 780 motherboards.

More powerful cpu's...

Digitimes reports that Intel's upcoming that three Bloomingdale variants of Intel's upcoming Nehalem architecture will be launched at the end of this year. The new names have yet to be decided on, the processors are currently only known by code names.

• XE @ 3.2GHz
• P1 @ 2.93GHz
• MS3 @ 2.66GHz

The all three devices are:

• for LGA1366 socket
• 130W TDP
• 8MB L3 cache
• quad core
• SMT support

Intel will also be launching the new X58 chipset at the same time, which will use the ICH10 south bridge; and X58 uses the new QPI (QuickPath) HyperTransport-like interconnect instead of an FSB.

The X58 features known at this time are:

• Four PCIe 2.0 8x slots
• Quad CrossFireX support
• no SLI support yet, unless Intel and Nvidia make up

AMD may be throwing their hat into the emerging UMPC and MID marketplace.

Over the past few days this slide has been floating to various hardware websites with no specific origin to the slide leak, but it seems to indicate that AMD may be gearing up for a battle against Intel in the UMPC and MID marketplace, meaning that future eee PC's and netbooks might see some variety down the road.

Right now we're seeing the Intel Atom processor come in with a 2.5w rating for Silverthorne and a 4w - 8w rating for Diamondville processors, the catch here is that we're only seeing the CPU power consumption itself and not the chipset for the CPU. When taking into account the 945 chipset the power rating changes quite a bit with the chipset consuming 22w of power. This changes the overall consumption for Silverthorne to around 24.5w and the Diamondville to around 26-30w consumed right now.

What the slide is apparently telling us is that this arrangement will feature the cpu + memory controller + the northbridge all in a total power consumption level of around 8W TDP, the only thing that seems to be missing in that case is the GPU but given the work of AMD into low power north bridge features they might just be excluding that, even if not it seems they still have quite a leg up on the Atom + 945 combo in market right now for power consumption if they had to include a seperate gpu to the mix.

This could be some interesting news though for the consumer since if AMD is bringing a fight to Intel in the low power market then it means we'll see a war in prices and an overall benefit to battery life which already at this stage is already going to see a huge boost with such low power usage. One can hope that AMD does step it up though with higher frequencies if they do plan to battle in the UMPC market since the Atom is rumored to be on the path to 2.2GHz in dual core format and presently already has a 1.6GHz chip on the market.

One can really only hope that this ends up appearing on an upcoming roadmap from the company as it seems that it would bring good things to those pondering a step into the UMPC market and frankly I'd love to be able to not worry about my battery life at all while working on things..

Uses PS3 technology

A newly constructed American military supercomputer has surpassed the computing speed record at more than 1.026 quadrillion calculations per second, reaching the long-sought petaflop measurement, and sooner than expected. At this point, it doubly exceeds the speed of the previously fastest machine, I.B.M's BlueGene/L, who also made this, the aptly named 'Roadrunner.'

Part of the design includes 12,960 chips which function as an improved version of the company's Cell microprocessor, the parallel processing chip originally made for the PlayStation 3. The chips are used as 'accelerators' for portions of calculations.

Thomas P. D’Agostino, administrator of the National Nuclear Security Administration, says if all six billion people in the world used hand calculators 24 hours a day, seven days a week, it would take them more than half a lifetime (46 years) to do what the Roadrunner is capable of in one day. I'm betting the Roadrunner itself performed that calculation.

The supercomputer cost $133 million (that's it?) and will be used "principally to solve classified military problems to ensure that the nation’s stockpile of nuclear weapons will continue to work correctly as they age." Here's the part I love, though: it will also be used to study scientific problems like climate change. I'm having trouble comprehending that proposition meeting.

Peter J. Ungaro, chief executive of Cray, a maker of supercomputers says “it’s a sign that we are maintaining our position," though pointing out “the real competitiveness is based on the discoveries that are based on the machines.”

Of course, there's always something left to achieve, and the next goals in mind are the exaflop, (one quintillion calculations per second), then the zettaflop, yottaflop and xeraflop. Many flops.

They apparently want to sell off low profit lines and keep premium (high profit) lines

Digitimes reports that Abit is thinking of getting out of the motherboard business, and MSI is thinking of getting rid of its low profit OEM lines.

Apparently Abit shipped less than half of its goal of motherboards so far this year, and it is waiting to see what happens with its P45 based boards before deciding to get out of the market.

MSI apparently wants to get out of the OEM business due to lower profits than it wants on OEM products, and only keep the premium branded motherboards.

Geez, MBA thinking or what?

<climbing on soap box>

In theory, its fine and dandy - even logical - to get rid of product lines that don't return enough profit. The problem is deciding what enough profit is.

I've personally seen companies go under when the management becomes convinced that they should only concentrate on high profit products and services, due to their higher return, and get out of the lower profit "bread and butter" products and services. Sounds great in theory - lowers costs, increases profits.

Unfortunately for freshly hatched MBA's, and business people who don't think enough, this only works for a while - because they are missing a HUGE detail:

There is a LOT MORE lower profit business than higher profit; they have to use the Henry Ford model - sell LOTS at low margin and actually make more money than selling a few at a high margin.

Let's take a quick look at motherboards.

Of course Abit, MSI, Asus etc will have a higher profit margin on a $250 "top of the line with bells, whistles and cow's" motherboard than on a $100 workhorse or a $50 cheapie special.

Let's say they make $150 profit on the high end (remember, distributors, dealers etc want their cut, and it does cost money to make)

Let's say they make $30 profit on the medium board, and only $2 on the cheapie.

Okay, some guesses

Let's say they sell 100,000 high end boards - that $15M profit.

Let's say they sell 1,000,000 medium boards - that $30M profit.

Let's say they sell 5,000,000 cheapie boards - that $10M profit.

Guess what?

If they only sold the high end board, they would only make $15M profit instead of $55M by having a diversified portfolio!

Sure, my analysis is too simple, and it costs more to make three types of boards, but the greater profit allows more R&D to make the high end boards, and often the medium boards are somewhat cut down versions of the high end boards - with less fancy heatsinks etc., so they do save a lot on development.

And they are forgetting that people will buy a LOT more mid/low range boards than high end... I doubt ANY company could survive on just high end boards.

<end of rant>

Not being satisfied with battling for consumers dollars two motherboard giants take their battle to court

Basically, the story goes something like this:

- Asus makes claims about how good its EPU "energy saving technology" is

- Gigabyte questions those claims during a media event

- Asus says Gigabyte just does not understand Asus' engineering design

- Gigabyte shows some blown up capacitors during an event, Asus claiming that Gigabyte was pointing fingers at it

... and so on ...

so now Asus has gone to the FTC (fair trade comission) in Taiwan saying Gigabyte made false accusations to damage it, and apparently also files a lawsuit.

Meanwhile, gamers yawn, and just want more performance, not caring about energy savings.

testtest

Post news here

fgfggf

bvbv

vbbv

bvbv

More CeBIT news.

Abit has some new products to show:

• N8500GT2 and N8600GT2 low/mid range video cards
• Abit X48 Max Tigerforce motherboard - large coolers, lots of Sata ports, diagnostic LED's, five phase power
• Abit IX38 GT3 - a DDR3 board supposedly only for Europe
• IP45 GT3 - P45 enthusiast board with heatpipes

Asus has introduced some new products:

• Ares - a gaming PC in a large unusual case; it will be sold in various configurations and comes with a new style keyboard with a round swiveling game key and a customized Logitech mouse
• Essentio PC - a small form factor HTPC with an Nvidia 8600GT and a Blu-Ray rom drive; again it will come in various configurations.
• MK241H monitor - 24" monitor with HDCP and optional webcam and speakers.
• G70 gaming laptop - dual 8700GS GPU's, two HD's, 17" 1920x1200 screen... decent specs, but a desktop with high end GPU will eat it for breakfast.

HD decode capable IGP chipset from AMD.

Today at CeBIT AMD has announced the availability of the 780 series of integrated graphics motherboard chipset.

The AMD 780 features:

• DX10 support
• Hybrid Crossfire support
• "full HD experience" support for VC-1, MPEG-2 and H.264
• supports DVI, HDMI, DisplayPort

Click here is the press kit on the chipset

AMD is also demonstrating a desktop and a server 45nm processor made with the "K10.5" enhanced K10 (Phenom) core.

Not cheap - I think ATI may dominate the low end, mass market motherboard market.

Digitimes reports that Intel will be releasing a P43 chipset - basically a P45 that does not support dual PCIe 8x slots, and limits DDR3 speeds to 1066MHz. The lobotomy will net motherboard makers a $6 savings per motherboard; personally I think that smart motherboard vendors will pass on the P43, DDR3-1066 usually performs a bit worse than DDR2-800.

Here are the new prices:

• $34 for G43 & P43
• $40 for G45 & P45 & Q43
• $44 for Q45
• $70 for X48

Interesting. Past Foxconn boards I've tried were not exactly great overclockers.

Hmmm... maybe we will have to start covering Foxconn motherboards again.

The Inquirer is running a story that Foxconn intents to go head-to-head with Asus for the enthusiast market. Apparently Foxconn has established a "Quantum Force" brand for its enthusiast boards, and it apparently intends to ship a killer x48 based board - that is outfitted from the start for your choice of four types of cooling:

• passive heatsink
• heatsink + fan
• water cooling
• liquid nitrogen cooling (!)

While I love to overclock chips, I tend to stop at water cooling, as I don't really like having dry ice or liquid nitrogen around - but its nice to have the option for say overclocking contests.

Quantum Force also reportedly is intending on building boards with many more slots - four PCIe x16, and up to six other slots.

Intel tries to reduce consumer confusion

Intel has apparently decided to call the upcoming Montevina mobile platform the "Centrino 2" platform.

Reportedly consumers have been confused by the changing nature of the Centrino platform - which has gone through "Carme", "Sonoma", "Napa" and "Santa Rosa" incarnations - the confusion arising from not knowing from certain which generation of mobile technology one was getting in their Centrino notebook.

By naming the new platform "Centrino 2" customers will know that they are getting a Montevina platform instead of an earlier Intel platform.

Intel will also reportedly launch six 45nm mobile processors - the new processors will all apparently come with a 1066MHz FSB and core speeds between 2.26 and 3.06GHz. Prices are reported to be between $209 and $851 in 1000 unit OEM lots. Seven more SFF mobile processors may be launched in Q3.

Motherboard makers let the cat out of the bag.

Digitimes reports that the Intel X48 chipset (the followup to the X38 with supposedly the only major difference being official 1600MHz FSB support) will apparently be shipped in the middle of March.

The X48 was originally scheduled to ship in January, however shipment was held back to allow motherboard manufacturers tosell up their X38 inventory.

Apparently the Core 2 Extreme QX9770 quad core processor may become available at the same time as the X48 chipset.

Asustek  has announced has announced its X48 based P5E3 Premium WiFi-AP motherboard, supporting up to DDR3-2000 dual channel memory,  dual 16x PCIe 2.0 GPU lots, and Express Gate BIOS resident web browser and VOIP.

Two CPU slots allows up to eight cores of processing power.

A new motherboard came out today that's a little bit different. Some would even say its special.

Intel's 'Skulltrail' platform has the ability to use two CPUs simulatenously. This high-end hardware is aimed at the hardcore out there -- because now, If you have the cash, you can have 8 cores of processing power at your disposal. Additionally, the motherboard has 4 PCIe 16x slots. Both SLI and CrossFire can run on Skulltrail, which is another first. 

Skulltrail has a FSB of 1600 MHz, and can handle Penryn CPUs. A 1000W+ power supply is recommended for the Skulltrail.  

Expect record benchmarks to begin breaking left and right with this mammoth monster from Intel. 

You can read a review of the Skulltrail over here.

More IGP motherboards for low end machines.

Digitimes reports that Nvidia will launch MCP7A and MCP7C integrated video chipsets in August to better compete against Intel integrated offerings.

This is apparently a response to the lackluster sales of the MCP73 compared to the Intel 3x series of IGP solutions; and Nvidia will apparently release the MCP7A in April with the MCP7C to follow in August.

Sources indicate that the MCP7A-U and the MCP7A-S will both feature 1333MHz FSB, DX-10 support, HDMI, DVI, HDCP, HyrbidSLI and PureVideo HD, and unlike the MCP73 that featured only a signle memory channel, they will feature dual memory channels. Supposedly MCP7A-U motherboards will be around $80 FOB and the MCP7A-S will be $5-$10 less expensive.

The MCP7C will apparently be a cut-down MCP7A, and will only have one memory channel - but motherboards based on it are expected to be priced $60-$65 FOB.

And up to 32GB of RAM

Specifications for the new Mac Pro have been released: and it appears this machine is going to have a little bit of processing power handy.

A pair of Intel 45-nano, 12MB L2 cache'd, quad core 5400 series Xeon processors are going to supply 8 cores of parellel CPU power, running at speeds up to 3.2 GHz. The Mac Pro will also support up to a (whopping) 4 TB of interal storage, and 32GB of 800 MHz DDR2 ECC FB-DIMM RAM.

“The new Mac Pro is the fastest Mac we’ve ever made,” according to Philip Schiller, an Apple Product Marketing guy. “With 3.2 GHz 8-core Xeon processing, a 1600 MHz front side bus and 800 MHz memory, the new Mac Pro uses the fastest Intel Xeon architecture on the market.”

Standard with the Mac Pro is also a HD 2600 XT -- but the system can handle up to 4 video cards.

The price for all this power is around $2700, for 2.8 GHz model. And that's without a monitor. You can read more about this fancy new Mac here.

Gigabyte is running an overclocking contest with a $2500 top prize.

Angela from Gigabyte sent us a heads-up about an overclocking contest they are running:

To actively engage hardcore PC enthusiasts in all its glory and express appreciation to those helping build GIGABYTE into the powerhouse it is today, GIGABYTE will be awarding over $12,000 in cash and prizes to its overclocking contest winners. Participants will be competing with hardcore PC enthusiasts on tweaking GIGABYTE Intel® P35 and Intel® X38 Series Motherboards to unleash exceptional performance of GIGABYTE Motherboard, adding a record-breaking score to Overclocking Hall of Fame. In addition to its over $12, 000 cash and prize, Hall of Fame Champion will be awarded with GIGABYTE Dream Machine GA-X48-DQ6 to experience the next generation Intel® X48 chipset right after its first debut. Overclocking video will be published at YouTube to show off one’s ability to overclock PC beyond its limit. The top 10 video winners will also take home GIGABYTE wind breakers. OMEGA pen and laptop protectors are also given out to add excitement to the competition.

We have reviewed Gigabyte X38 and P35 based motherboards, and they really are great overclockers.

So get out your thermal paste, heatsinks, roll up your sleeves, and enter the contest!

Personally I can't wait to see the results!

680i owners might be really annoyed

A post by a moderator in the EVGA support forums has delivered the unfortunate news that current owners of EVGA (at probably all other) 680i motherboards will not be able to run Intel's upcoming 45 nano quad core Penyrn CPUs.

For those of you who shake your head at this statement in confusion, current owners of NVIDIA chipset 680i motherboards had been promised support for upcoming Penryn CPUs in future BIOS updates. Many (probably most) expected this support to be for both Wolfdale (dual core) and Yorkfield (quad core) Penryn CPUs. But, alas, it is not to be: it seems the only Wolfdale will be compatible with the 680i chipset.

Many will probably suggest that this solely an artificial limitation impressed on this chipset, while NVIDIA will probably suggest that there are some technical hiccups that are precluding Yorkfield support.

Many will also probably also suggest this limitiation was decided upon to help the sales of the soon-upcoming 780i chipset, which will support Yorkfields.

The 780i chipset is, a rehash of  the 680i chipset, with no earth-shattering improvements, but a few minor ones, such as ESA support (for better monitoring and tweaking of system components), PCIE 2.00 support (which doesn't help gamers very much at all right now), and support for 3-way SLI (for the hardcorest).

Update: the moderator has reposted, describing a hardware limitation that prevents Yorkfields running on the 680i:

"We can tell you it is a hardware limitation on the 680i board.  With the release of a new CPU sometimes it is not just a simple die shrink, even if the CPU draws less power there could be other specs and requirements.  The P35 chipset is Intel’s first chipset to support thier Yorkfield CPU’s.  The P35 was released June of 2007, the 680i motherboard launched Novemeber of 2006.  The 700 series motherboard will have the necessary changes to fully support the Yorkfield CPU’s.  I know this is not the best news we can bring you but we are doing everything we can at this time.  I will answer all questions as best as I can."

Hardware limitations or not, many 680i owners will still be saddened to hear that they will not be able to fully engage the power of the 4 core Penryn.

Intel launches Core micro architecture based 7300 series quad core processors and Craneland quad socket chipset.

The new 7300 series Craneland chipset provides four separate 1066MHz FSB's, one for each of four Xeon processors that it supports, allowing for a significant boost in agregate bandwidth. With this "individual FSB per processor" approach Intel is able to extend the viability of its server platform until it brings its hypertransport like CSI to market. The 7300 chipset also includes 28 PCIe lanes, and supports up to 32 FB-DIMM's.

Intel will also apparently launch six quad core Xeon 7300 series processors, and two dual core Xeon 7200 series processors.

• X7350 @ 2.93GHz, 8MB L2 cache, 130W TDP at $2310
• E7340 @ 2.40GHz, 8MB L2 cache, 80W TDP at $1980
• E7330 @ 2.40GHz, 6MB L2 cache, 80W TDP at $1391
• E7320 @ 2.13GHz, 4MB L2 cache, 80W TDP at $1177
• E7310 @ 1.60GHz, 4MB L2 cache, 80W TDP at $856
• L7345 @ 1.86GHz, 8MB L2 cache, 50W TDP at $2301
• E7220 @ 2.93GHz, 8MB L2 cache, 80W TDP at $1177
• E7210 @ 2.40GHz, 8MB L2 cache, 80W TDP at $856

All the processors are based on the 65nm Tigerton core.

Designers of Cray-2 aggravated

Anyone who has ever touched a computer a with 10-foot pole (or shorter) knows that the price of computational power falls fast. And this has once again been proven this weekend by a grad student and professor from Calvin College: together, they built a desktop sized supercomputer from off-the-shelf parts, for just under $2,500.

The design is somewhat reminsecent of a clubhouse sandwich:

The supercomputer's name is Microwulf. It is a variety of 'Beowulf cluster.' Cluster meaning a networked computers running in parallel; and Beowul denoting a cluster that is a "homemade, cost-effective" solution that "liberates researchers from expensive commercial options for super-computing, much like Beowulf of the Old English poem liberated the Danes from the tyrannical rule of Grendel." 

The hardware involved includes four MSI K6N6PGM-F motherboards, four Athlon 64 X2 3800+ AM2 CPUs, 8 sticks of Kingston DDR2-667 RAM, four PCI-E Intel network adaptors, and some Zalman case fans and a single 250 GB Seagate SATA HD. 

As for the software: They chose open-source distro-darling, Ubuntu Linux.   

For a price-tag of $2,470 USD, the Microwulf's performance is rated at 26.25 Gflops ( [10^9] floating-point-operations-per-second). In comparison, only 10 years ago, IBM's chess Grand Master-squashing Deep Blue supercomputer was built, at a cost of somewhere around the neighborhood of $5 million dollars, and its performance was rated at 11.38 Gflops.

More Nvidia chipset goodness for Intel processors...

There is a report that Nvidia will release five chipsets in the MCP 73 family for Intel platforms on September 25.

Four include an integrated GPU:

• GeForce 7150 nForce 630i - 1333MHz FSB, single channel DDR2-800
• GeForce 7050 nForce 630i - 1333MHz FSB, single channel DDR2-800
• GeForce 7050 nForce 620i - also presumably 1066MHz FSB, single channel DDR2-667
• GeForce 7050 nForce 610i - 1066MHz FSB, single channel DDR2-667

All four IGP chipsets support HDMI, HDCP and PureVideo HD

And a non-GPU chipset:

• nForce 630i - 1333MHz FSB, single channel DDR2-800

All of the chipsets support PCI express and Intel CPU's from Celeron D's to Core 2's.

The single channel memory is going to hurt...

New features for AMD 690G based motherboards...

A new BIOS from AMD adds a plethora of new features to motherboards based on its 690G chipset: with IGP

• 720p video playback support
• 1080p video playback
• voltage adjustments
• memory timing settings
• multiplier settings for CPU clock
• multiplier settings for HyperTransport clock
• IGP GPU overclocking

You might want to improve the stock chipset cooler if you overclock the chipset a lot :-)

It sounds like with the new BIOS, the 690G will make a neat little HTPC platform!

Cell Computing Board combines Cell and RSX GPU for non-gaming applications

Sony has been eager to tout the benefits its high-flying Cell processor outside of PlayStation 3 gaming, and to this end the company is planning to demonstrate a prototype "Cell Computing Board" at Siggraph in San Diego next week. The Cell Computing Board will naturally include the Cell processor, as well as the RSX GPU also seen in the PS3 console. Sony is hoping that developing such a computing platform will allow the Cell to see use in a greater variety of applications, including scientific.

The board is designed for computers that handle multimedia applications, such as graphics and video, and scientific computations that require processing of massive amounts of data, Sony said. IBM, Sony, and Toshiba, jointly developed the Cell processor.

By incorporating the RSX chip built by Sony and Nvidia, the new technology can reach computational speeds beyond the 230 gigaflops, or floating point operations per second, of the Cell processor alone, Sony said. The Cell Computing Board, which uses up to 400 watts of power, can be embedded in a 1U-size server and mounted on a 19-inch rack.