No, this wasn’t a typo, GPU manufacturer Albatron Retrotechnology is actually releasing the Nvidia 8400, 8500, and 8600 for the PCI, not PCI-Express architecture. What kind of performance can we expect from these “phat” cards riding the skinny bus?
Click the jump to find out, and why even power users might want one.
Rambus, the technology company turned responsible for RDRAM has filed suit against Nvidia claiming that they violated 17 of its memory patents. Rambus’ lawsuit alleges that at least six of Nvidia's product lines infringe the Rambus patents including chip sets, graphics processors and applications processors. They ask for an injunction that would prevent Nvidia from selling the products as well as damages.
Tom Lavelle, senior vice president and general counsel at Rambus was quoted as saying, "For more than six years, we have diligently attempted to negotiate a licensing agreement with Nvidia, but our good faith efforts have been to no avail," he adds, "We are left with no other recourse than litigation to protect and seek fair compensation for the use of our patented inventions,".
Can you smell the bull? I better get my rubber boots it’s getting thick. His own statement shows this to be an attempt to force Nvidia to cough up something to make Rambus go away and they haven't been interested. Nvidia has yet to comment on the suit.
Cost cutting must top Nvidia’s priority list after it lowered its financial outlook for Q2, 2009 and announced $150-200 million product replacement and repair expenses. It plans to cut production costs by making the shift from 65nm to 55nm manufacturing process by the end of the current quarter, according to a Commercial Times report. All of its upcoming GPUs that are expected to be out after August including G94b, G96b and G98b will utilize 55 nm processes. Although the transition will lower production costs by 20%, Nvidia will need to do more than that if it has to wrest some momentum from its resurgent rival AMD.
When it rains, it pours, and Nvidia could use a good downpour to put out the flames. Perhaps literally. Just last week Dave Murphy reported Nvidia was setting aside $150 to $200 million to cover warranty and repair costs associated with an "abnormal failure rate" in its mobile graphics cards, news of which sent Nvidia stock spiraling downward. Now there's speculation that the failures might not be limited to just a specific batch of notebook GPUs.
Rumor, news, and review site The Inquirer is saying that "all the G84 and G86 parts are bad. Period. No exceptions." That includes both mobile and desktop parts. According to The Inq, both use the same application-specific integrated circuit (ASIC), and both ASICs are plagued by a heat related problem originating from an un-named substrate or bumping material. Because of this, The Inq surmises more failures are iminent. But are they?
Find out what Nvidia has to say about the failures after the jump.
Tomshardware.com is reporting that Eran Badit editor-in-chief of ngohq.com has had some success running Nvidia’s CUDA platform and PhysX drivers on a Radeon video card. Apparently adding Radeon support to CUDA was not a big deal, but adding Radeon support for CUDA at the driver level is more challenging.
Badit says he needs support from ATI to finish out Radeon’s support for CUDA, but ATI has been slow to answer him, taking several days to reply. Surprisingly, Nvidia has been much more helpful and opened access to their Developer Relations and is providing assistance, including access to documentation, SDKs, hardware and actual engineers.
Nvidia’s official position is now that it doesn’t mind PhysX running on the Radeon an interesting change from when Justin Kerr reported that Nvidia wanted to license PhysX support to ATI pennies a GPU. This looks like additional pressure from Nvidia to make it’s platform dominant over ATI and Intel’s planned platforms. Third party implementation of CUDA on the Radeon is sure to rattle ATI’s cage.
Tom’s Hardware pinged ATI on the issue, but hadn’t heard anything back as of yet. The longer we wait for a competing platform from ATI, will only help Nvidia’s platform capture more market share. ATI looks to have an uphill battle against the already established
If you're a subscriber to Maximum PC magazine, turn to page 8 in this month's issue (and for everyone else, hit the subscription link) and read Gordon Mah Ung's take on Intel and Nvidia's Secret War. Gordon discusses the issues preventing users from being able to run SLI on an Intel chipset, and what roadblocks might be in place for future Nehalem support on upcoming Nvidia chipsets. In other words, you might end up having to choose a side. Sound familiar?
Now there's talk of Nvidia want to support Intel's Atom processor, and whether or not you care about the low-cost PC and MID market, it might be in your best interest if the two sides can come to an agreement. But can they? Earlier in the year Nvidia and VIA entered into an alliance, and speculation suggests it was forged to compete against Intel's Atom. Now it appears Nvidia's intention all along may have been to gain a bargaining chip to convince Intel to let its Atom processor support Nvidia's MCP73 IGP chipset. If Intel agrees, DigiTimes reports Nvidia will then terminate its alliance with VIA and its Nano processor. And while VIA might not be too pleased with the idea (rebound relationships never work out anyway), an agreement over licensing terms in the low-cost PC market might open the door to better communication in the mid- to high-end desktop sectors.
ATI’s resurrection as a serious contender for the top spot in the GPU industry has much to do with vastly improved technology, cheap mid-level offerings and its ability to meet demand. On the other hand, Nvidia is faced with delays and defects that are beginning to have a bearing on its financial condition. The graphics chip manufacturer has announced that it expects the repair and replacement of certain defected products to cost it $150-200 million.
It has also been forced to lower its financial outlook for Q2, 2009 due to delays and unexpected price cuts. It now expects earnings for Q2 to be between $875 million and $950 million, which would be a steep decline from Q1 earnings of $1.15 billion. If we construe talk of “unexpected price cuts” to be Nvidia’s implicit admission that it didn’t really see the AMD/ATI onslaught coming, we know exactly what is wrong with the company. Complacency!
It looks like stock prices aren't the only thing falling over at Nvidia. Competition continues to heat up in the latest round of GPU leapfrog, and reports flanking from across the web over the weekend were claiming Nvidia would issue price cuts to its add-in board partners (AIBs). The GTX 260 was to be the beneficiary of a $30 chop, while the high-end GTX 280 was to get slashed by $90. Even the 9800 GTX was to receive a $17 snip. Scouring Newegg's selection, it appears the reports were right on the money.
Thanks to Nvidia's annoying Manufacturer Advertised Pricing (MAP) policy, you'll need to do a bit of extra clicking to see the new pricing structure. But assuming your index finger is up to the task, you'll spot several GTX 280 cards selling for $499, and even lower after rebate. GTX 260 videocards have dropped down to $329, and a small handful of 9800 GTX cards have dipped below the $200 mark.
Recently in both the print and online versions of Maximum PC we looked at Nvidia’s CUDA API and what a GP-GPU future might look like. The one wild card in this equitation is the other big player in the graphics card market, ATI. Will ATI play nice by supporting CUDA and licensing PhysX? Or will it go its own way, a result which may end up killing both companies initiatives.
Nvidia shares dropped by a fourth today after the company announced it was setting aside a one-time hit of $150 to $200 million dollars to cover warranty and repair costs associated with an "abnormal failure rate" on its mobile graphics cards. The exact sources of the increased GPU problems are unknown at this time, although Nvidia believes the cards' increased thermal issues stem from weaker manufacturing and packing materials.