Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?
Slashdot Deals: Cyber Monday Sale! Courses ranging from coding to project management - all eLearning deals 25% off with coupon code "CYBERMONDAY25". ×

Comment Re:"Incorrect" MPG numbers (Score 1) 166

I'm not sure if they actually cheated on the US MPG numbers. The advertised MPG US are quite a bit lower than the european ones. This could be a result of a different test cycle but also a result of not being able to cheat in the US. In Europe they did stuff like mixing diesel fuel into the oil supply and increasing the tire pressure beyond the allowed range. This increases MPG but will cause the engine to break early and would be dangerous on real roads.

Comment Re:Uh... no (Score 0) 393

There are good reasons to be critical of the H1B program, but the issues of people without college degrees are not caused by H1Bs, after all there are less than 100k H1Bs and on other hand there are 121M citizens 25 years and over with no college degree. Cancel all H1Bs and these people would still have problems to find good jobs. These issues are caused by moving almost all manufacturing offshore. At the same time a lot more people went to university and businesses can usually find someone with a degree, even without H1Bs.

Comment Re:Placebos work! (Score 1) 287

More expensive placebos actually work better. (And placebos administered using a syringe also work better than pills.) At the same time many real medications are abused as placebos, e.g: when GPs prescribes an antibiotic to a patient that most likely has a viral infection without a additional bacterial super infection that is effectively a placebo but with real side effects. A "homeopathy" based placebo might work very well for some patients and in most cases will be very cheap, even if not as cheap as other sugar pills.

Comment Re:Placebos work! (Score 1) 287

Because a minor illness might still be a huge issue for the patient from his or her subjective perspective. A placebo can help a lot in these circumstances and can sometimes be a better alternative to not doing anything or prescribing medication that works only slightly better than than a placebo and has real side effects that can sometimes be worse than the wanted effects.

Comment Re:Placebos work! (Score 1) 287

Placebos work, so why shouldn't GPs be allowed to prescribe them? .

This is not an uncommon argument, even among physicians. But there's a simple rebuttal, in my view: Giving a placebo conflicts with the patient's right to be informed.

Physicians prescribing homeopathic sugar pills can fully inform their patients that no good study ever showed that these pills worked better than a placebo. Many patient will still accept them. Also GP's would not prescribe them because they think they are useless, but prescribe them because they work well as a placebo and a placebo can sometimes be the medication with the best profile of wanted effects vs. side effects. Placebo's are evidence-based medicine. The placebo effect is extremely well documented and studied.

But even if the patient is not informed, things are not that simple. Patients deserve the best possible treatment (which sometimes might be a placebo or might have extremely rare but very scary side effects that will cause patients to stop taking their medication), at the same time patient deserved to be fully informed. Sometimes it is just not possible to achieve both goals. No matter how GPs act in such a situation, they will always fail in some respect. Not properly informing patients can sometimes be the smaller ethical issue.

Comment Placebos work! (Score 0) 287

Placebos work, so why shouldn't GPs be allowed to prescribe them? I think it is much better for public health if GPs are allowed use placebos such as homeopathy, than if people are avoiding GPs and are using people that are not allowed to prescribe real medicine when beneficial. For minor illnesses without an effective evidence based treatment it is perfectly fine to prescribe a placebo. It is also fine to prescribe a placebo in addition to conventional treatment, if the conventional treatment is not effective enough.

Comment Re:A better idea (Score 5, Interesting) 284

The real prevailing wage is hard to check. Companies will just not mention some of the special skills of that person and then they can hire a very skilled person for more than 110% "prevailing wage", when they are really paying 80% prevailing wage. Or they are paying 110% prevailing wage but expecting 200% working hours.

I think a much simpler solution would be to change the random lottery to a list that is ordered by wage and give the H1B only to people on the top of that list. That would make it hard to abuse H1Bs to drive down wages and give priority to the people that would likely really contribute the US economy. It could also potentially drive up wages for us workers: If companies are required to offer 200k per year to a foreigner with a certain skillset to guarantee him a H1B, then us workers with the same skillset will also notice what their skills are worth and will demand higher wages. And if the lowest wage that still qualified for a H1B is too low, then you just reduce the number of H1Bs.

Comment USB3.0 - DVI/HDMI Adapters (Score 1) 197

As the content is likely mostly static: What about a single PC with many USB3.0 -> HDMI adapters + USB 3.0 Hubs? Sure, refresh rate will likely go down to something like 10 Hz because of bandwidth limitation but that should fine for your kind of content and driving all screens from the same PC could be very useful for administration.

Comment Re:CPU Architecture (Score 1) 311

The problem is a stall in one core can stop the other. This doesn't happen if the cores have their own instruction pipelines. It's a critical resource that's shared between the two cores of a module.

Nope, this will not happen. They are not completely stupid. I think you do not get how modern Out-of-Order CPUs such as Bulldozer are designed: There is not a single big instruction pipeline, that is either running or stalling, but many different pipelines with many waiting spots where stalled instructions can wait and non-stalled instructions can overtake stalled instructions.
A instruction cache miss will also not stall the instruction fetch and decoder pipelines: Bulldozer will just try to fetch something from the other core in the meanwhile or even prefetches instruction blocks that are predicted to be required soon by branch prediction.

Comment Re:CPU Architecture (Score 1) 311

Unless one core has a cache miss and needs to wait for the data to be fetched from main memory.
Mean while the instruction pipeline that feeds the other core in the module gets stalled, meaning nothing executes in either core.

Something that can always happen. Stalls because of cache misses are happening in all high performance CPUs. This is completely unrelated to modules vs. cores. Stalls even help the shared FPUs approach: If one core is stalled, than the shared FPU is just as good as a non-shared FPU for the other core.

Comment Re:Compare to NVIDIA (Score 1) 311

They've got nothing on NVIDIA, who advertise the GTX 980 as having 2048 "cores", when by any standard definition it only has 16 (or if you're really generous, you could maybe argue it has 64, but that's pushing it). They count every lane of their vector unit as a separate core. By that standard, AMD (and Intel) should multiply all their core counts by 8, since each AVX unit can do 8 int or float operations at once.

I think 64 cores in 16 modules would be a fair assessment of a GTX980. Each SMM contains 4 register files and dedicated execution units, 4 instruction fetch and scheduling units. Not too many things are shared within each SMM: mostly caches and shared memory.

Comment CPU Architecture (Score 1) 311

"The suit claims that Bulldozer's design means its cores cannot work independently, and as a result, cannot perform eight instructions simultaneously and independently."
If the suit really makes this claim it is easy for AMD to defend, because an AMD Bulldozer with 4 modules / 8 cores can actually execute 8 independent floating point instructions per cycle.

The two cores in each module share the floating point units, but each module contains 2 independent 128-bit FMAC units. Floating point throughput could be even higher, if both cores would have their own 2 128-bit FMACs, but would likely be lower if AMD decided to go with a single 128-bit FMAC per core. Sharing the floating point units allows higher execution speeds within the same area, as in most cases only one core is currently executing FPU instructions and that core can reach a higher speed with 2 execution units vs. one.

Leveraging always beats prototyping.