Saturday, December 29, 2007

Most AM2 Motherboards Not Phenom Ready

AMD's new quad core cannot compete with Intel's high end processors - despite being clearly faster when comparing individual cores including the entire cache hierarchy. Hence AMD adjusted its product strategy and points at the value of the Phenom processor together with the AMD790 chipset (Spider platform) as well as Phenom being an excellent upgrade option for the majority of socket AM2 systems on the market. Such an upgrade scenario is very likely, so we grabbed nine different socket AM2 motherboards to see if Phenom would really work with them.

In theory, a motherboard should provide a suitable core voltage to a processor automatically, and since the HyperTransport interface hasn't changes as well (with the exception that Phenom supports the faster HT 3.0), chances are that a Phenom processor could boot on a compatible motherboard even without having updated the BIOS. We've seen this in the past when AMD released faster Athlon 64 X2 models or stepping upgrades. Some motherboards would boot and display generic processor information together with a clock speed. Though this is not ideal due to missing support for processor features and probably improper processor settings together with missing micro code updates, it can help to check whether compatibility is there. We don't recommend to run a system with a processor not officially supported by the motherboard.

The more reliable approach is to look for a BIOS update on the website of your motherboard manufacturer. We've compiled a list of links to the download pages of the most popular motherboard manufacturers; you'll find it at the end of this article. Make sure that your desired processor is supported once you find an updated BIOS version for your motherboard. The BIOS update process depends on the motherboard, as there are multiple approaches to flash the firmware. We will discuss the upgrade options first and have a look at the nine socket AM2 motherboard, which we tried to upgrade with a Phenom 9600.

Read the full Story here

Friday, December 28, 2007

NVIDIA Asks Card Makers to Reduce Manufacturing Costs of 8800 GT Cards

NVIDIA recently contacted its graphics card partners asking them to reduce the number PCB board layers used in GeForce 8800 GT-based graphics cards from ten to six in order to reduce manufacturing costs and so lower the card's ASP (average selling price) in the market. The redesign would allow the NVIDIA cards to compete in terms of pricing with AMD's Radeon HD 3800 series products. Although the Radeon HD 3800 series was launched three weeks later than the GeForce 8800 GT, Radeon 3800 demand has started to pick up, bringing the market shares of NVIDIA and AMD from 90% and 10%, originally, to 70% and 30%. If the PCB layers are reduced from ten to six, graphics card makers are expected to save more than US$10 for each card, which would allow the NVIDIA products to go into price competition with those of AMD. Despite the cost benefits, some graphics card makers are unhappy with NVIDIA's suggestion, pointing out that the chip maker is in effect asking them to do the job of improving the price/performance ratio of its products, while preserving its own profit margins. NVIDIA responded in saying that the redesign is only a suggestion which it believes is the best solution to meet the current market conditions. Card makers will not be forced to implement the change, the company stressed.

Saturday, December 22, 2007

Samsung, Hitachi sign license deal on hard drives

Samsung and Hitachi have agreed to drop all pending patent-infringement lawsuits against each other and instead have decided to sign a license agreement covering their hard disk drive patents. Under the deal, both companies will be able to share HDD-related patents registered before 2003 permanently and others until 2011.

The deal also covers patents from IBM, whose disk drive business was bought by Hitachi in 2002. Hopefully, the agreement will bring further innovation to the hard drive market in the form of energy-efficient drives and larger capacities, which these days come from new developments in how data is recorded onto the surface of a disk to improve density rather than from just adding platters.

Intel GPU/CPU combo in 2009?

Really integrated graphics may be coming in 2009

Penryn is almost here; Nehalem is due next year.

But Intel may be planning to follow them up with new chips integrating a GPU with the CPU in 2009 - at least according to the Reg.

Bloomfield will apparently feature a 1366 pin socket (ouch) and feature "QuickPath" - Intel's answer to AMD's "HyperTransport" along with up to three channels of DDR3 memory. The new core apparently will have 8MB of L2 and use QuickPath to connect to a "Tylersburg" northbridge that will provide PCIe 2.0 lanes.

Strangely enough, less than six months later there will apparently be a "Lynnfield" going to a 1160 pin socket with four cores and hyperthreading, 8MB of L2, dual DDR3 channels, using PCIe as a chip to chip bus. Frankly, this sounds weird to me. Why on earth would Intel go to a 1366 pin chip, to go to a 1160 pin chip less than six months later?

The only way that this would make sense is if one of the chips was destined for the low end integrated market.

Let's peek into our crystal ball... to reduce costs, you have to simplify designs. What if Lynnfield integerated the GPU? With PCIe integrated onto the chip, it is not farfetched to consider that Intel may also consider integrating a couple of SATA channels and some USB2.0 channels... then you could have a nice little quad core "SOC" (System on a Chip) allowing very low part count integrated low end designs.

Such an SOC really would only be suitable for low end machines as the GPU would have to share memory bandwidth with the four core / eight thread CPU portion, so even a hypothetical dual channel 2000MHz DDR3 system with close to 32GB/sec bandwidth would be hard put to exceed the performance of an 8600GT - and it would starve the processor for memory bandwidth at that.

To make matters even more confusing, the referenced roadmap also shows a mainstream "Clarksfield" part using a 989 pin rPGA socket. UGH.

PLEASE PEOPLE: We don't need so many sockets!

Wednesday, December 19, 2007

Carbon one step closer to becoming viable alternative for silicon

Researchers at Princeton University have discovered a potentially viable way to use carbon as a replacement substrate material for silicon. The team has figured out how to build transistors on tiny patches of the substrate, called graphene, just a few atomic layers thick. The research indicates that switching speeds could be 10x faster than silicon.

Graphene is a 2D layer of the common material, graphite.

ImageThe big hurdle was in overcoming the limited size of a graphene sheet. Whereas today's silicon wafers come in 8" or 12" (200 mm or 300 mm) circles, the largest graphene sheet that could be created in the lab was only a couple millimeters square, too small for today's several hundred square millimeter CPUs. So, what the researchers did was essentially create a patchwork of tiny graphene sections, only 1/10th mm square. These small patches are overlaid on a traditional silicon substrate wherever logic circuits are required. The end result is like having a football field covered with 1" square pieces of paper. The entire football field is covered with paper, and yet no individual piece is very big.

To create the final substrate, the graphene source material could be thought of like sheets of cookie dough. A type of stamp comes down and presses against the graphene. When the stamp is extracted, the graphene sticks to the stamp. It is then moved onto the final position on the destination substrate, whereby it is pressed onto it. The act of pressing transfers the graphene from the stamp to the substrate, where it adheres and resides through the stamping force.


Earlier efforts involved using a type of "Scotch Tape" to randomly pick up sheets of graphene for deposition. This resulted in a haphazard layout, resembling the images shown in this document. There are no images of the new process, but it could be easily envisioned as a series of squares covering larger surface areas.


While it seems like a lot of work, the payoff potential is in processing speed. One measurement of the potential advantages of carbon over silicon comes from something called an "electronic hole" measurement. In such tests, the carbon circuits performed 10x faster than silicon. According to the researchers, there would be an almost immediate benefit in radio technology, such as cell phones and wireless devices with high power outputs. Devices could be smaller, use less power, and produce greater signals due to the properties of carbon-based transistors.

One of the researchers believes the carbon technology could be adapted in a few years. The current 1/10th mm size is the first step in the proof of concept implementation. Now that the team has proved it can be done, the next step is to scale it up for larger applications. Eventually, entire CPUs could be created that would be 10x on the same amount of power as is seen today.

The work was carried out by professor of electrical engineering, Stephen Chou, and graduate student Xiaogan Liang at Princeton University. No word was given on the source of funding, however Princeton University receives funding from DARPA, the U.S. Department of Energy, and other U.S. government sources.

Tuesday, December 18, 2007

How To Hack Your Graphics Card

In the bad old days, you'd buy a video card that ran at some standard clock frequency. The core clock was the core clock, the memory clock was the memory clock, and that was that.

Then a couple of companies decided to try their hands at selling and supporting video cards clocked higher than the reference clocks. In a commodity market, selling an overclocked card was one way to eke out a little more profit margin and distinguish yourself from the competition. At first, the GPU makers frowned on this, but not for long. In fact, Nvidia and ATI have begun embracing the idea. They allow companies that resell cards, that are often indistinguishable from each other, to compete on something other than software bundles and prices.

When Nvidia launched the 8800 GT, they told the press they left lots of headroom in the card, so card makers could, if they so desired, pump up clock rates and give buyers a little more value. Of course, that added value comes at an extra cost. As we noted in our recent 8800 GT roundup, the factory overclocked cards cost quite a bit more than the standard cards. These higher clock rates can make a big difference in some games, and less so in others.

Of course, you can simply take control of the situation by overclocking the card yourself. This article shows you just how to do that.

Read the full story Here

Sunday, December 16, 2007

AMD Hints at 32nm Test Shuttles, Claims 45nm Samples Ready in January

AMD has had more than its share of problems over 2007 with possibly the worst issue being with its new Phenom and Opteron processors. With the current problems with AMD's K10 processors, now the question becomes, "Will AMD make its deadline for the next processor?"

ChannelWeb Network sat down for a phone interview with AMD Executive Vice President of the Computing Products Group Mario Rivas for some more information on the bug and how AMD plans to recover from the torrent of bad press that has resulted.

Rivas says that the bug started as an observation and it wasn’t until mid-November that it actually turned into a more serious bug. Rivas also said that the company tried to do BIOS workarounds and patches with a 90% success rate.

In a closing comment, Rivas details some heavy information about the company's next-process chips. "We have 45nm on the way. We will have initial samples also in January. I'm fairly confident that those puppies are going to boot, and then we can have a follow-up conference call and I'll tell you, 'The sucker is booting.'"

Typically, when a processor is first taped-out, a operating system boot is one of the markers of a successful design. Intel's 45nm Penryn, for example, loaded the Windows XP operating system on the first spin. 11 months later, the processor began shipping for volume.

According to AMD’s John Pellerin, director of logic technology and development, AMD plans to ramp production of 45nm high-k dielectric chips in the first half of 2008. Pellerin says that AMD is more concerned with finding customer applications for its processors rather that a process race with rival chip makers.

Pellerin said in an interview during the International Electron Device Meeting that AMD’s 32nm high-k parts were solidly in the development phase showing that AMD is already looking down the road and hoping to learn from the problems encountered with its early 45nm process.

Rivas closes, "We also have 32nm advance work in SRAMs, which as you know is the initial step. So we will be a fast follower again, and as long as we have architectural advantage, our 45nm will be as good as the other guy's 32nm."

Intel announced its 32nm test shuttle just three months ago in September 2007. If AMD does have a 32nm test shuttle already, the year differential for 45nm may shrink considerably with AMD's 2010 architecture launches.

Saturday, December 15, 2007

ATI demonstrates hybrid-chip CrossFire graphics tech

We've already seen laptops like Sony's Vaio SZ include integrated graphics alongside much more powerful (and power-hungry) dedicated chips, but AMD's looking to make such setups all the more commonplace with new chips capable of hybrid CrossFire. AMD recently demoed the tech to PC Perspective, showing off a 2.2GHz Phenom machine with both unreleased RS780 integrated graphics and a RV620-based card labeled HD Radeon 3450. Running games like Call of Duty 4 and Unreal Tournament 3, frame rates jumped from 30-35fps to around 55fps when hybrid mode was enabled. That's pretty respectable, although the system is limited to speedups of the slowest chip times two, so bigger gains are probably not in the offing. However, there can still be benefits to using chips of drastically different horsepower: the integrated chips can power down the heavy hitter to save power when not needed, and totally switch over when required. That's a pretty solid compromise, we think -- and with AMD aiming for the initial batch of hybrid CrossFire-capable cards to be priced around $50, it looks like we'll be seeing these setups a lot when AMD starts shipping these early next year.

Guys there is a bad news also, AMD has revealed that they don't have any plans for releasing new family of Graphics Processing Units in Year 2008,bad luck nuh?. You can sneak peek(:P)into the full story here.
Thank goes to Anusha. :D

Do you need 3 way SLI with 3 x 8800GTX

Today NVIDIA lifts it's NDA on 3 way SLI. Although it seemed to have rushed it out ahead of it's competitor AMD+ATI, it doesn't seem to appeal to most gamers. This is largely due to the fact that you need to use at least 3 x 8800GTX to achieve 3 way SLI. You can do that with lower end cards like the more affordable 8600 series.

NVIDIA marketing is silly to launch things at year end when Christmas is round the corner. Press can't get 3 cards and 680i board for review too. So, what is the point ? Even you can get 3 8800GTX cards, you will have to look for a 3 way SLI bridge. Of course you can choose to connect the 6 connectors crossing one another as shown in the review (ref) we posted earlier.

Even with 780i, I doubt 3 way SLI will be the way to go. It is simply to milk those rich kids who wants the best of best without considering hardware depreciates faster than anything else.

I will assume that there will be 3 way SLI without the need for SLI bridge. It is only a matter of a small performance hit. Whether NVIDIA will enable it really depends on how AMD markets it's CrossFire X which allows even HD 2600XT to run in tandem.

Thursday, December 13, 2007

Can Water push Yorkfield to 5 GHz?

Is 45nm An Overclocker's Dream?

Intel's Penryn family of 45 nm processors is making waves, but still hasn't had much desktop market penetration, because the firm has so far only released the expensive "Extreme" QX9650 model. As enthusiasts look forward to upcoming lower-cost versions that will give them the best "bang for the buck", a few extreme overclockers have already pushed these beyond 6 GHz using equally extreme below-ambient coolers.

As indicated in our Penryn premiere, major improvements over earlier Core 2 models include a die shrink for higher yields, and a new dielectric for reduced leakage current. The combination of these production features could eventually translate into lower pricing and lower power consumption with reduced heat. Any time we see a die shrink we expect higher clock speeds will soon follow, and the cool-running Yorkfield four-core version was capable of reaching 4 GHz on air cooling.

Lower heat is a step in the right direction when insane clock speeds are the objective, but Intel's efforts have focused instead on reducing the system's "carbon footprint" and operational cost while increasing longevity of related components. These seem like lofty goals, but we really wanted to how far we could push the performance envelope using common parts that most enthusiasts can afford.

Is More Cooling Really Needed?

While these new processors might run "cool" on nothing more than a heat sink and fan, extreme cooling has always been the "trick" to reaching the highest overclocked speeds. High overclocks require relatively large voltage increases for greater signal strength, which can overwhelm the inherent efficiency of any integrated circuit design. Excess voltage creates heat, heat causes instability, and adding even more voltage to overcome that instability only further increases heat until the component reaches its breaking point. The only way to break this never-ending loop of voltage and heat is to increase cooling capacity.

Extreme overclockers have used wild cooling methods - including dry ice and even liquid nitrogen - but none of these is practical for daily use. The next step towards "daily use" practicality has come from cascade-type phase change (refrigeration) units, where the primary cooler is itself cooled by a second cooler, but these are expensive, complicated, and require skills beyond those of most builders.

Thus, the majority of enthusiasts won't use anything more elaborate than a large water cooler. We've gotten such good results from Swiftech's Apogee GT water block, MCP-655b pump, and three 120mm radiators, that the kit is now mounted into one of our test benches.

Is More Cooling Really Needed?

Read the full Story Here

Nintendo Sells a DS Every Five Seconds

This fun piece of trivia does more than make you the life of the party when playing "trivial pursuit". It indicates just how popular the Nintendo DS really is. Nintendo claims that they have sold over six million Nintendo DS units in America this year, citing internal estimates. There also seems to be no slowdown of sales in sight. The Nintendo DS is extremely popular to both the casual and hardcore gamer, and Nintendo fully intends to propel that momentum into 2008.

To further increase your trivial pursuit scores, these recent sales figures indicate that the sale of a Nintendo DS happens more than...
  • A bad car accident (those happen once every 14 seconds)
  • A man making a sexual advance on a woman (that happens once every seven seconds)
  • Two or more script kiddies having a "flame war" on internet message boards over the previously mentioned console (which happens once every six seconds).

Wednesday, December 12, 2007

8800 GTX drops to € 369

Scared of 8800 GTS 512MB

The Geforce 8800 GTX has dropped to €369, as the new 8800 GTS 512MB G92-based cards are almost as fast as this year-old competitor.

This is a very interesting price, as the 8800 GTX has never been lower priced, and clearly Nvidia wants to get rid of the leftovers to make some space for next month’s announcement of the D8E dual PCB/chip card.

The Geforce 8800 GTX will also be a better bet for higher resolutions with FSAA and Anisotropic filtering enabled, as it has 3a 84-bit memory controller and faster memory, while the new 8800 GTS 512MB has 256-bit memory controller and slightly slower memory.

Nehalem quad core TDP is 130W

Intel's next generation native quad core with eight threads aims for 130W TDP (Thermal Design Power) and it is unlikely that Intel can do better than quad core at 45 nanometer. This is a bit better TDP than Core 2 Extreme 9770 that has 136W TDP, or 9775 that will have even higher TDP, closer to 150W.

Pre-announced eight core sixteen thread Nehalem CPU is doomed to wait for the 32 nanometer shrink as it would simply be too big and too hot for 45nm process.

Bloomfield should have a total 8MB cache memory and it fits in a brand new socket, LGA1366.

The CPU is still scheduled for Q4 2008 announcement and we are quite sure that Intel plans to have a big demonstration at IDF in Fall 2008. Until that point 3.2GHz Yorkfield at 45 nm should be the fastest one, but we’ve seen Intel changing its mind far too many times; and it can easily decide to increase the speed of Core 2 Quad, if it feels necessary.

AMD's bet for 2008 is 45nm

Much improved K10.5

If you watch Smallville you know what 33.1 is, and if you don't, then K10.5 is the new codename for AMD's K10 manufactured at 45 nanometer.
Anyway, the primary goal that AMD has for 2008 is to switch to 45 nanometers as soon as possible. The new K10.5 will produce a faster cache latency, as well as improved imc, and these were exactly the weak points of the K10.
Both the B2 and B3 revision of the K10 would benefit a lot from a faster cache latency and an improved imc. AMD hopes to give the Yorkfield processors a run for their money with this improved core.
Let’s hope to see K10.5 around Computex, at the latest.

AMD’s surprise is a new 2GB FireGL

Since initial launch of the FireGL products a few months back we have been constantly bombarded with the ultimate question – when are you going to do a head to head of the monster memory cards at the top of the scale.

At the ultra High End position this is were really the serious players come into force and items of this calibre have to be spot on, first go! There can not be one slight point of error at this level of the playing field. Last month we clearly stipulated that there was a clear and present danger to Nvidia due to the aggressive pricing and performance that the FireGL cards had brought to the fore. What does todays head to head at the Ultra High End Cards bring; well time will tell as you go through things! What we can say is that it is very good indeed and will have many of you out there wondering what next is to come? Time, Ladies and Gentlemen will tell.

Read the full story Here

FPS Command & Conquer: Tiberium Revealed

Electronic Arts is taking another stab at the FPS market with Command & Conquer. In the upcoming Command & Conquer: Tiberium FPS you'll play as Ricardo Vega, a GDI field commander dragged out of retirement to stop the latest Nod master plan 11 years after the third Tiberium war. The squad-based action will ship on PC, PlayStation 3, and Xbox 360. Unfortunately, at this stage of development, that's all the information available. Final release date is also unknown.

IBM Working on 32nm Processors

IBM has revealed that it is working on a new 32nm processor manufacturing technique which could make the production of chips more accessible to smaller companies, as well as yielding performance gains. The process works in a similar way to the high-k/metal gate technique that is currently used by both IBM and Intel for their 45nm chips, which replaces some of the silicon in transistors with more efficient materials to pack more components into a single chip. IBM’s new technique uses a modified version of this process known as high-k/gate-first to develop its 32nm chips, which works by focusing on the most advanced components first. Big Blue has already demonstrated a working example of the 32nm technology with prototype static RAM chips, and mass production is expected to begin during the second half of 2009.

AMD RS780 Chipset to Launch in January

Information coming from DigiTimes, claims that AMD will launch its next-generation performance/mainstream RS780 IGP and value RS740 chipsets in January next year, while in the second quarter, the company will launch the high-end RS780D and mainstream RS780C. RS780 motherboards will be priced around US$90-120, while RS740 motherboards will come for around US$40-60. The RS780 supports AM2+ processors, HyperTransport 3.0 and PCI Express 2.0, while the IGP will support DirectX 10 and UVD (Unified Video Decoder) 2.0 for H.264 and VC-1 decoding. The IGP also supports DisplayPort, HDMI, DVI and HDCP. Hybrid CrossFireX technology, which is a similar to NVIDIA's Hybrid SLI, allowing the IGP to work together with discrete graphics cards for extra performance, will be presented as well.

Tuesday, December 11, 2007

Asustek, MSI and Gigabyte see motherboard shipments drop in November

Due to the trend of notebooks replacing desktop PCs, first-tier motherboard makers except Elitegroup Computer Systems (ECS) saw their November motherboard shipments drop on month, according to sources at the makers.

Asustek Computer announced November consolidated revenues of NT$75.29 billion (US$2.33 billion), an increase of 11.2% on month, but a drop of 9.9% on year. The company's accumulated consolidated revenues for the first 11 months of 2007 year were NT$666.1 billion, a boost of 44.5% compared to the same period in the previous year.

Affected by its subsidiary ASRock's strategy change, Asustek shipped 5.1 million motherboards in November, a slight drop of 0.2% on year and 7% on month, while accumulated shipments for the first 11 months were only 54.55 million units, making the company's annual goal of shipping 63 million motherboards difficult to achieve.

The company also shipped 1.15 million graphics cards and 900,000-950,000 notebooks in November.

Micro-Star International (MSI) announced November non-consolidated revenues of NT$8.65 billion, a drop of 14.3% on month, but a boost of 14.7% on year. Accumulated revenues for the first 11 months were NT$83.71 billion, an increase of 22.77% compared to the same period in 2006.

The company shipped around 1.4-1.5 million motherboards and 900,000-1,000,000 graphics cards in November, a drop of 12% and 24% on month, respectively.

Gigabyte Technology's November consolidated revenues were NT$4.41 billion, a drop of 2.84% on month, while the accumulated consolidated revenues for the first 11 months of the year were NT$47.63 billion.

The company shipped 1.7 million motherboards in November, a slight drop of 2.8% on month, while it shipped 340,000 graphics cards.

ECS' November consolidated revenues were NT$8.99 billion, a drop of 7% on month and 60.05% on year. Accumulated consolidated revenues for the first 11 months were NT$84.87 billion, an increase of 32% on year.

The company shipped 2.45 million motherboards in November, an increase of 7.8% on month and 9.88% on year.