Tom's Hardware published a harsh review of the chip July 31, the day the processor was released. Problems with the chip's design, which could lead programs to stall, prompted Intel today to recall its fastest processor.
Kyle Bennett, who publishes rival site HardOCP, said he also had problems with the review chip but held back publishing a negative review in case it was a glitch with his individual chip.
"We didn't want to shed a bad light on Intel if this were an isolated problem," Bennett said today.
However, after seeing the review on Tom's Hardware, he sent a letter to the site's publisher, Thomas Pabst, supporting its conclusions. Bennett said the chip he was given failed to compile a kernel of the Linux operating system, stalling at different points each time the test was run.
"When I saw the heat he was catching, it wasn't fair to keep silent," Bennett said.
The two sites, along with a third review site, AnandTech, began working with Intel on the bug. The chip giant eventually flew an engineer to Texas to meet with Bennett and test the chips used by the review sites.
Intel representatives said they appreciated the assistance they got from the review sites.
"They've been very helpful," said Intel spokesman George Alfs.
Intel said it initially found problems with the chip only when operated outside the recommended temperature range. Over the weekend, the company found problems even when the chip was run within specifications and decided to recall the chip.
The situation marks a rare case of cooperation between Tom's Hardware and HardOCP, Bennett said.
"We've butted heads," he said. "(Tom) knows I wouldn't go out of my way to back him up."