There is a lot of truth in what you said. However, reviewers (well at least primary reviewers) get to be listed as program committee members. And PC members generally get more consideration as future PC and general chairs. As a PC chair, I would always pass along info to the next PC chairs about which reviewers were slackers in terms of writing short or useless reviews, failing to submit reviews on time, using an inordinately high number of secondary reviewers, or scoring papers significantly differently than other reviewers for the same paper. These slackers (sometimes) acquire a reputation as bad reviewers, although conferences still want recognizable names on the PC.

The American sentiment that its allies should pay more for military defense is only half the sentiment. The other unspoken and uncomfortable truth is that the US (well, mostly Trump) expects to retain full hegemony in the "alliance." However, the more other countries increase their military commitment, the more their independence from the US will grow.

Trump only knows bully tactics, but as with TACO tariffs, his assumption that the bullied will never fight back is sometimes wrong.

Hmm, a French school puts out a ranking that is dominated by European countries and that elevated Singapore, where it fortuitously has a campus, to the top spot.

What does this mean? Nothing. An arbitrary ranking from a school that most of us have never heard of.

Wow, that really sounds a lot like arguments from the Sudetenland. And that's not an exaggeration. Sort of eerie. Almost everything you said sounds like it came straight from the Russian PR machine.

Some countries collect money via tariffs on imports. Some do so via fines. Both are largely arbitrary in terms of monetary amounts and application, usually have domestic political motivations, and have no avenue for appeal. The only recourse is to exit economic activity with that country or to convince one's home country to retaliate in a way that causes the other country to back down. If this case proceeds further in India, I wouldn't be surprised to see Trump get involved. There's no way that India could collect this massive fine without massive retaliation from the US.

What does "technically perform" mean? Depending on that definition, the report could be significant or completely irrelevant. How well is a task done, what is the probability of failure, and what is the impact of the task (in terms of safety and significance)? For example, Tesla FSD could be labeled as technically performant. However, there are scenarios with a more than insignificant probability of failure. That's why FSD is legally Level 2, even though it is technically performant more than 99% of the time.

The report doesn't even come close to providing insight into these pivotal questions. Furthermore, the O*NET database that forms the basis of the framework for skills evaluation contains short descriptions that are open to wide interpretation, which likely means that the conclusions of the report are heavily dependent on the evaluation abilities and biases of the researchers.

The US can easily wipe out its entire debit overnight. Just declare a default on the entire debt. Easy-peasy.

The problem is that wiping away that debt necessarily means taking that money away from the debt holders. Erasing around a $1 trillion in assets each from Japan, China, and the UK would prompt a global economic crisis that would necessarily significantly affect the US.

The other problem is that once the debt is erased, the US will no longer be able to find foreign buyers of its future debt. That's the real problem.

Also, most of US debt is held by Americans. That includes $7 billion by Social Security and other $4 trillion by the Fed. Also, another $20 trillion by American companies and individuals.

But the real problem is that the US government would no longer be able to borrow money by selling treasuries. I guess that's one way of immediately balancing the budget and completely torching the US economy.

Or just lay down tripwires a few hundred feet from the real cables. Something that will trigger an alarm. Maybe something that traces the offending anchor line back to the boat and does something like explode a glitter bomb, play a recording of very stern warning message, attach a homing device, etc.

An alternative view is that China's cheap imports benefit mainly the American masses. The 1% can afford to pay extra to buy from anywhere, but the bottom 80% cannot. However, since the US economy is largely consumer driven, the bottom 80% must be kept alive and buying in order to support the economy (particularly the stock market) for the 1% to maintain and grow their assets.

Russia invaded Ukraine to grab land. Trump negotiates a deal to give Russia the land it wanted all along. In exchange, Ukraine gets a promise of "peace for our time." Dictators that invade neighboring countries for conquest can always be trusted to honor these agreements. We'll see if Russia beats the 11 months after Chamberlain's appeasement before completely trashing the "agreement."

I'm guessing the answer is yes. The need for heat assistance implies high coercivity at anything close to room temperature. HAMR write temperatures go beyond 400C, far above room temperatures.

Will TPU v7 finally be able to challenge the dominance of Nvidia GPUs in the data center? Some projections have the TPU garnering 5-10% of the data center processor market. That sounds dubious to me. If TPU is a viable alternative that is likely to be significantly cheaper than Nvidia, why would its market share be capped at 5-10%?

The entire world and every Nvidia customer is begging for an alternative. If TPU is truly viable, then it should immediately challenge Nvidia more forcefully. TPU can be somewhat slower, more power hungry, and even the same price and still immediately compete strongly with Nvidia because all customers want to end the Nvidia hegemony. Supply constraints are understandable, but the demand for TPUs should equal Nvidia GPUs if customers view TPU as equally viable. But that's the real question and Google's real problem. Do customers view TPU as equally viable? The world has a ton of competitors that are begging to be viewed as equally viable, from AMD to Chinese companies, but they are obviously not viewed as equally viable.

Nvidia has continued with graphics R&D, but it has slowed down pushing the boundaries of gaming products. Why? Well, their current strategy has seen their market share increase over the last few years. For all of AMD's fan support, its apparent technical prowess hasn't translated into sales. Some of that is due to the realities of the gaming market. Compared to CPUs or data center GPUs, the gaming market is tiny. Furthermore, even at AMD, gaming is a minor target, way behind CPUs and the push for data center GPUs. And AMD has to support this 3rd priority using fewer employees than Nvidia has.

Nvidia is pushing forward with their Grace/Vera CPUs. What they are pushing is the non-necessity of an x86 CPU.

What??? TSMC has their own chip designs?

It's true that TSMC can stop selling chips to Nvidia, Apple, AMD, and anyone else at any time. Of course, their stock would plummet and severely punish the executives who would dare to do that. The only events that would cause TSMC to not sell ever more to Nvidia would be a Chinese invasion or the rise of another customer that is willing to pay more than Nvidia.

TSMC is a Taiwanese company that is quite friendly with the Taiwanese CEO of Nvidia. That's helping both to increase their partnership. The only reason TSMC doesn't sell more to Nvidia right now is the slow pace of ramping up CoWoS. The wafers are there, but the packaging is not.

Since the advent of GPGPU 20 years ago, Nvidia has had the foresight to look beyond just graphics. That early focus on parallel computing allowed the GPU to adapt to the particular needs of parallel processing. By far the most challenging problem in parallel computing is data movement. GPUs look straightforward and simple until one looks at the memory subsystem, which has evolved into something that is definitely not straightforward and simple but which is able to handle the varied and demanding bandwidth requirements of a wide range of parallel applications, from AI to scientific simulations. This is the key reason why ASICs are still a generation or more behind GPUs in terms of hardware. Perhaps when AI models settle down and aren't quickly evolving, maybe then, some smart architect will design an ASIC memory subsystem that is specialized for that AI model. But we aren't there yet, so ASICs will continue to trail GPUs in the meanwhile.

This general principle should be obvious. For embedded systems and other applications that never change, ASICs will always beat general-purpose processors in speed, power, and cost. However, ASICs are generally worse for applications for which they were not designed. Yes, TPUs are designed for AI models. However, they were designed for yesterday's models and will always be trailing until the models stop evolving.

This is the reason why TPUs not only have worse MLPerf performance than Nvidia GPUs but also fail to even post numbers across all the benchmarks.