Word has it that NVIDIA, the world’s largest discrete GPU manufacturer, is preparing to soft-launch a 55nm revision of the G200 core which powers the GTX260 and GTX280 GPUs. Unfortunately for NVIDIA, rumor has it that the die shrink is going to arrive behind schedule with no improvements to the thermal envelope or power consumption.
The new revision, officially known as G200-xxx-B2, is expected to debut in a slate of new GPUs including a new dual-GPU solution based on a B2 GTX260. TheInq says that the problem with these otherwise grand plans is evidenced by the use of both a heatspreader and a stiffener for the upcoming GPUs.
TheInquirer points out that these devices indicate a thermally-troubled design. Especially in the case of the IHS, the cooling assembly’s ability to relieve heat is reduced by the added thermal barriers of TIM and the IHS’ alloy.
The problem with a heat spreader is that it introduces two additional thermal barriers, the paste under the lid and the lid itself, to the cooling of the silicon. Each one makes cooling incrementally less efficient, not to mention material and assembly costs. You don’t do this unless you have to.
They go on to say that the IHS is only needed when there’s a strong probability that customers will crack their CPU/GPU core as a result of an unprotected core.
If you are wondering why every modern CPU out there has one, the answer is simple, so ham-handed monkeys like most DIY people don’t crack the die when they clamp the heatsink on. Think AMD K8 here. CPU makers think the cost of a spreader, and the reduction in performance it brings, is worth the protection it gives.
TheInq reckons that the IHS is a gamble to keep a GPU with poor thermal control alive. “They can’t control hot spots. The lid is a heat spreader, and it helps keep chips with poor hot spot control alive and working,” they said. TheInquirer believes that this evidence is truly damning. “When you see a heat spreader on a part that comes assembled, it is a pretty sure sign something is wrong thermally, it simply is not worth the cost and performance drop otherwise,” they said.
Inside sources to TheInq also revealed that NVIDIA may have used the wrong production method when shelling out production to the well-known foundry TSMC. It is alleged that NVIDIA used incorrect DFM (design for manufacturing) rules which specify the dos and don’ts of IC design while preparing to make smaller revisions in the future. It is postulated that NVIDIA’s 65nm DFM required components and tools that do not make for a clean transition to 55nm production, explaining why NVIDIA is six months behind schedule on 55nm G200 parts.
While all of this remains entirely unsubstantiated, TheInquirer has had a storied history in correctly spelling out disaster long before other sites make it to the party. We hope that NVIDIA isn’t a day late and a dollar short on what may be a paper launch of the winter refresh, but something tells us that those spritely lads from the rainy country may be right.
Articles RSS