Nvidia's problems achieving good yields on its 130nm chips are entirely of its own making and blame should not be laid at the door of its main foundry partner, TSMC.

So says Pacific Crest Securities analyst Michael McConnell in a note the company's customers, according to a Semiconductor Business News story.

McConnell also says that Nvidia has pulled its finger out and solved the problem behind the low yields, and that everything will be back to normal by the end of the month.

Nvidia admitted there were problems with its 130nm chip yields last month when it warned that its gross margins for its most recently completed quarter would be lower than anticipated - a direct result of those yield issues.

Specifically, it blamed "higher than anticipated product costs attributed to the 0.13 micron semiconductor process technology". Nvidia claimed it will still meet its sales forecast, so its income will be effectively reduced.

Nvidia didn't explicitly blame TSMC, but there have long been claims that the company's 130nm plans have been hindered by alleged difficulties the foundry has experienced developing its 130nm process. Indeed, some observers have claimed that's one reason why Nvidia signed up IBM as a foundry partner after such a long solo relationship with TSMC.

However, ATI has always claimed it's never had any problems with TSMC's 130nm process, and now McConnell has stated that the trouble with yields is "Nvidia-specific", and not a TSMC issue.

Following the debacle of the much-delayed 130nm GeForce FX 5800 release, Nvidia admitted that moving to 130nm hadn't been easy for the company.

But it's figured the technology out at last, says McConnell: "Nvidia has informed its add-in board partners that yield problems on the NV35 and NV31 will be remedied by the end of August," he writes in his report. ®