August 04, 2008 eASIC promise: 5X less power than FPGAs
By
Kenton
Williston
Today eASIC announced its Nextreme2 line of 45 nm structured ASICs* that promises to breathe new life into ASICs. Nextreme2 has many impressive features, but its main selling point is that it burns up to 5X less power than comparable FPGAs. That's a lot less power!
FPGAs have a reputation for being power hogs. Some of this reputation is undeserved. In some applications, FPGAs are actually burn less power than the alternatives. For example, FPGAs can burn less power than DSPs in high-performance signal-processing apps. However, FPGAs burn much more power than ASICs. That's where eASIC comes in.
FPGAs have several inherent features that increase power consumption. For example, FPGA contain SRAM that controls interconnects and look-up tables (LUTs). Because ASICs are hard-wired, they don't have this SRAM. Thus, FPGAs burn more power than equivalent ASICs.
eASIC says it achieves ASIC-like power by eliminating these power-hungry features. The 90 nm Nextreme product eliminated the SRAM for interconnects. Nextreme2 takes things further by eliminating the SRAM for LUTs as well. Nextreme also saves power by:
Moving to a low-power 45 nm process
Offering a three voltage options, include a low-voltage 1.0 V product
Adding via power gating, which permanently removes power from unused logic
Providing clock gating, which temporarily remove the clock from idle logic
Like all structured ASICs, Nextreme2 has its disadvantages, such as a 6-week turn-around for hardware spins. (This is only a disadvantage in comparison to FPGAs. Compared to traditional ASICs, this turn-around is very short.) However, Nextreme2's low power consumption makes it worth a close look.
*eASIC likes to call its offering a "zero mask-charge ASIC," presumably because the term structured ASIC has been associated with business failure in recent years. (See LSI, Fujitsu, NEC, Oki and Toshiba, for example.) Nonetheless, the products are clearly structured ASICs (which are also sometimes called platform ASICs). Comment on this blog entry
July 28, 2008 What would you do with free floating-point?
By
Kenton
Williston
Last week TI announced the C674x, a new DSP family that offers floating-point performance for about the same price and power as fixed-point DSPs. This is a big deal: Historically, floating-point processors were much more expensive and power-hungry than their fixed-point counterparts.
The cost difference has become less of an issue in recent years—both TI and ADI have released $5 floating-point DSPs, for example—but power has remained a stubborn problem. Even the most efficient floating-point DSPs burned over twice as much power as comparable fixed-point DSPs. Thus, the floating-point features were not "free." Designers still had to choose between the superior precision and dynamic range of floating point and the lower power of fixed point. In some cases, this was a difficult decision.
The C674x changes the equation. Its power consumption is nearly identical to that of an equivalent fixed-point C64x DSP. As such, the C674x is the first mainstream DSP with truly "free" floating point. (Atmel's mAgic DSP also offers low power floating point, but this DSP is not widely used.) It will be interesting to see how this changes DSP products. In PC's, floating-point went from a high-end bonus to a standard feature. Will we see the same evolution in DSP?
More importantly, what would you do differently with "free" floating point? Would you speed up development time? Add features to your products? Toss your ideas into the comments section—I look forward to hearing them!
July 21, 2008 Global warming revisited
By
Kenton
Williston
Global warming revisited
Here's what you had to say about global warming, and some more detail on the subject.
Several readers wrote in questioning the idea that humans are causing global warming. My thanks to everyone who shared their views. Global warming is one of the biggest challenges facing humanity, so it is absolutely vital for us to discuss the subject.
Here is what you said, and how I responded.
Many scientists reject global warming. It is wrong to stifle this debate. We should not accept global warming on blind faith.
The idea that global warming is under debate is simply incorrect. No scientific body of national or international standing rejects this idea—not one! (See Wikipedia and the update below.) I welcome scientific debate on climate change, but we should not pretend that there is a debate when none exists.
I recognize that individual scientists question global warming, but scientific debate requires more than an opposing opinion. For in idea to have scientific value, it must be backed by data and methodologies that can pass a peer review. To date, not one criticism of global warming has passed this test. For example, a study of 928 peer-reviewed papers found exactly zero papers disagreeing with the consensus on global warming. (See the Oreskes study.) Thus, I find it very difficult to see how one could characterize the global warming as "under debate."
In short, I do not think engineers should accept global warming blindly. However, we should acknowledge that scientists have reached a consensus on this issue.
Scientific consensus does not equal proof. Scientific ideas have been proven wrong in the past.
This argument is completely pointless. Any scientific idea might be disproven in the future. Even the law of gravity might be proven wrong.* Without a time machine, we cannot tell which ideas will be rejected. It is absurd to selectively ignore one idea on the basis that it might be disproven in the future.
*This was a test to see if you were paying attention! Newton's law of gravity was already proven wrong. However, it is still in use because it is accurate enough in most cases. This demonstrates both (1) that any scientific idea can be proven wrong and (2) that scientific models do no need to be perfect in order to be useful.
Climate science models are untrustworthy. They are unable to explain past climate events.
Not true! Climate models have been used to reconstruct the climate of the last 100 years with a high degree of accuracy. The ability to reconstruct the past demonstrates that the models are trustworthy.
Current models have known shortcomings, but the models do not need to be perfect in order to be useful. Short-term weather forecasts are also imperfect, yet we rely on them on a regular basis. Nobody expects clear skies when the forecast calls for rain, for example. Why anyone takes a different attitude towards global warming is beyond me.
Several readers claimed that climate science cannot explain the Little Ice Age or the following warming. This is not true—both events are well understood. The Little Ice Age and the following warming were caused by variations in solar and volcanic activity. Solar and volcanic activity will vary in the future, but such short-term variations will not change the long-term warming trends caused by greenhouse gasses.
Efforts to curb global warming might have unintended consequences. We should wait to act until we know more about the mechanisms of global warming.
I appreciate the desire not to leap into ill-advised action. However, most actions we could take to curtail global warming carry economic and health benefits. As one example, replacing incandescent bulbs with compact florescent bulbs saves money over the lifetime of the bulbs. As another example, cutting automobile emissions reduces incidences of asthma, cancer, etc. Thus, there is very little downside to acting now.
In contrast, there is a huge downside to waiting for perfect information. Scientists warn that we are nearing a tipping point where a lack of action will lead to devastating and irreversible consequences. With so much at stake, it does not make sense to wait.
Global warming is a money-making scam. Critics of global warming cannot get funding.
This is a bizarre claim. The science of global warming is open for all to see. It is published at the IPCC site as well as in any number of science journals, university web sites, etc. How could anyone pull off a con job when all of the data is out in the open?
This idea is even more preposterous when you consider the size of the industries involved. The industries that could benefit from global warming (such as solar panel manufacturers) tend to be small. The industries that might be hurt (such as oil companies and car manufacturers) are huge. If scientist were selling out to the highest bidder, one would expect them to oppose global warming, not support it.
Finally, the idea that warming deniers cannot get funding is laughable. Until recently, for example, Exxon funded warming skeptics.
Global warming is a leftist conspiracy. Critics of global warming are being silenced by political pressure.
This idea is ridiculous. Republicans controlled the US presidency, Congress, and Senate, for most of the last eight years. Despite this, the US government's acceptance of global warming has increased. How then is it possible that global warming is a left-wing conspiracy?
Furthermore, global warming is a concern to people across the political spectrum. I can easily name major Republicans who accept global warming. This list includes President Bush, Senator John McCain, and Governor Arnold Schwarzenegger. Even evangelical pastors are working to halt warming.
Update:
Several readers pointed me to articles claiming that the American Physical Society (APS) no longer accepts global warming. These articles are wrong. The APS home page makes the following statement:
The American Physical Society reaffirms the following position on climate change, adopted by its governing body, the APS Council, on November 18, 2007:
"Emissions of greenhouse gases from human activities are changing the atmosphere in ways that affect the Earth's climate."
An article at odds with this statement recently appeared in an online newsletter of the APS Forum on Physics and Society, one of 39 units of APS. The header of this newsletter carries the statement that "Opinions expressed are those of the authors alone and do not necessarily reflect the views of the APS or of the Forum." This newsletter is not a journal of the APS and it is not peer reviewed. Comment on this blog entry
July 13, 2008 Amazing Google phone app
By
Kenton
Williston
A couple of German engineers have come up with a Google phone application that is the coolest use of mapping data I have ever seen. Their innovative app places Google Earth data on top of live video from your camera phone, giving you a cyborg-like ability to identify anything you look at. Watch the video and be amazed! (The good stuff starts at 00:50.)
Today eASIC announced its 