Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Intel Publishes Microcode Patches, No Benchmarking or Comparison Allowed

perens.com

Post reply on HN
Print view View original

471–480 of 499 posts

Re: Intel Publishes Microcode Patches, No Benchmarking or Comparison Allowed

#471

Post by Zelizz »

binomialxenon wrote: post #440
tomaskafka wrote: post #362

Earlier quoted context omitted.

I think Apple laptop CPUs will be like USB-C. First 'what's the point, nothing is compatible', year later 'everyone is doing it and it's actually pretty cool to attach everything including power with a single cheap dongle'.

I think it would take far longer than a year for a critical mass of OS X software to become ARM-compatible, but it could be viable in the long run. I'd like for ARM laptops to become more popular though, as a Debian user nearly all the software I use is already there.

Apple already got a huge number of developers to transition their apps to 64-bit ARM with the iPhone, I feel like if they made it simple enough they could get a critical mass on the desktop just as quickly.

Re: Intel Publishes Microcode Patches, No Benchmarking or Comparison Allowed

#472

Post by noobermin »

kllrnohj wrote: post #464
repolfx wrote: post #383

Earlier quoted context omitted.

They aren't. Spectre works fine on AMD. They avoided Meltdown only because their cores are less optimised. Meltdown isn't Intel specific - Apple cores were also affected. It's clear that AMD weren't doing anything special w.r.t. side channel attacks. They were just further behind in the optimisation race and as a consequence, were less hit.

That is a horrifically biased & wrong summary. AMD enforce privilege checks at access time rather than at retirement time. Whether or not this is due to "lack of optimization" or "good security engineering" nobody knows. But your claims that this was purely the result of "less optimised [sic]" cores is nonsense. You have zero evidence whatsoever that that was the case vs. AMD just having superior engineering on this…

Minor tidbit, but "optimised" is correct British spelling, so the disdainful "[sic]" is not needed.

Re: Intel Publishes Microcode Patches, No Benchmarking or Comparison Allowed

#473

Post by chrischen »

Bartweiss wrote: post #389
carlmr wrote: post #80

Earlier quoted context omitted.

>they've nothing significant to show for all that time and money spent when they were raking it in without a serious competitor. Big companies rarely innovate without competition around.

I don't think this is necessarily true for big hardware companies. If you're selling a subscription to a best-in-market service, there's not a lot of motive to innovate, agreed. Maybe you'd try a new product or a premium variant, but there's no reason to sink effort into advances that won't let you expand userbase or raise prices. But for Intel? Before AMD got going, Intel's biggest competition was itself 9/18 months…

This is a great argument to be against the recent trends of companies moving to subscription models just for the heck of it.

Re: Intel Publishes Microcode Patches, No Benchmarking or Comparison Allowed

#474

Post by newman314 »

onli wrote: post #351
johnchristopher wrote: post #229

Earlier quoted context omitted.

> If you loathe Intel after all the lies they have been telling and marketing speaks, you should buy AMD. > If you love Intel still after all, you should still buy AMD, teach them a painful lesson to wake them up. But how do I choose which AMD CPU I need ? Back in my youth p4 and athlon were easy to compare (freq., IPS and a modifier because AMD) but now I can't even tell the differences between any i5/3/7 and when I…

The right answer to this is to look into benchmarks. It now works again to compare Intel and AMD clocks, but only of the current generation, and then there is core count and motherboard prices to consider, so on. A project of mine is a hardware recommender, it also includes a meta-benchmark. I collect published benchmarks and build a globally sorted order of processors out of it. https://www.pc-kombo.com/benchmark/ga…

Can you add the ability to sort based on price/perf?

Also, the existing bar graph is unclear to me. What does 10/10 mean?

Re: Intel Publishes Microcode Patches, No Benchmarking or Comparison Allowed

#475

Post by vbezhenar »

cm2187 wrote: post #167
TheRealPomax wrote: post #61

Earlier quoted context omitted.

It works for Oracle (it is famously illegal to publish benchmarks of DB2 vs other engines), I'm sure intel can make it work for them thanks to Oracle's court case(s).

But you can't buy an oracle license in a shop around the corner. It is going to be hard for intel to enforce it.

You don't need to buy anything, just download from their website.

Re: Intel Publishes Microcode Patches, No Benchmarking or Comparison Allowed

#476

Post by vbezhenar »

AstralStorm wrote: post #118
amluto wrote: post #13

I can think of two theories: 1. It's a mistake. Someone in legal got carried away. 2. The performance of the L1TF mitigation is so awful that someone at Intel thought it would be a good idea to try to keep the performance secret. (Which leads to option 2b. The performance of the L1TF mitigation is so awful that somemone at Intel is afraid that Intel could be sued as a result, and they want to mitigate that risk.) I w…

A hotshot lawyer could sue them for this anticompetitive clause for big $$$. (it disallows comparing the product against the competitors)

Sue Oracle.

Re: Intel Publishes Microcode Patches, No Benchmarking or Comparison Allowed

#477

Post by celrod »

Symmetry wrote: post #403
gnulinux wrote: post #309

Earlier quoted context omitted.

Well, AMD currently has better CPUs for high-performance computing, so OP's users would benefit from this public service announcement, if anything.

I thought it was the opposite of that? That is, AMD chips are currently great for most workloads but Intel still has a definite edge in working with big vectors of floating point numbers of the sort you usually encounter in HPC. Mostly because their core's vector units are 512 instead of 256 bits wide.

Skylake-X and certain Xeons have avx-512 (which includes two 512 bit fma units). The rest only have 256 bit wide vectors, like Ryzen. But they still have the advantage of two 256 bit fma units, while Ryzen instead relies on two 128 bit fma units, meaning the fma instructions critical to matrix operations are faster on Intel.

I think the idea for HPC though is that you want to offload these highly vectorizable operations to a GPU. Or maybe you're doing a lot of Monte Carlo that it is hard to vectorize.

I really do like avx-512 though. If you're writing a lot of your own code, and that code involves many small-scale operations and has control flow (like in many Monte Carlo simulations), it's a lot easier than mucking with a GPU. If you're using gcc though, be sure to add `-mprefer-vector-width=512`, otherwise it will default to 256 bit vectors (clang and Julia use 512 by default).

Re: Intel Publishes Microcode Patches, No Benchmarking or Comparison Allowed

#478

Post by Nition »

dwild wrote: post #346
Nition wrote: post #9

They better be wary of the Streisand Effect here. Banning benchmarking just makes it seem like the performance hit is going to be serious, which makes everyone even more curious.

Could be a "reverse" Streisand Effect. Any slowdown would have been pretty public, the microcode update is still a big deal and the expectation weren't clear, I'm pretty sure the results were worst than the average expectation. Now though, the expectation are much higher, people expect it to be pretty high and maybe the results are now actually lower than theses expectations. So yeah maybe more publicity but maybe no…

That's brilliant, but probably a little too clever to be the reality.

Re: Intel Publishes Microcode Patches, No Benchmarking or Comparison Allowed

#479

Post by mcv »

gnulinux wrote: post #311
mcv wrote: post #263

Earlier quoted context omitted.

> Doesn't this show that it is time for someone to set up some kind of "ScienceLeaks" website Good idea, but I doubt it's necessary here. I'm fairly sure this sort of EULA clause is unenforceable in many jurisdictions. Run the benchmarks in a country where they are legal.

If that's the case would it be ok to publish these results in US-based websites such as Phoronix, even if they're done by other country's citizens in that other country?

If Phoronix never agreed to the license, then I can't see how that wouldn't be legal. And otherwise it's going to be a good year for Europe-based benchmarkers.

Re: Intel Publishes Microcode Patches, No Benchmarking or Comparison Allowed

#480

Post by Bartweiss »

chrischen wrote: post #473
Bartweiss wrote: post #389

Earlier quoted context omitted.

I don't think this is necessarily true for big hardware companies. If you're selling a subscription to a best-in-market service, there's not a lot of motive to innovate, agreed. Maybe you'd try a new product or a premium variant, but there's no reason to sink effort into advances that won't let you expand userbase or raise prices. But for Intel? Before AMD got going, Intel's biggest competition was itself 9/18 months…

This is a great argument to be against the recent trends of companies moving to subscription models just for the heck of it.

Definitely.

The common complaint about subscription models is obviously good: if the company folds you have nothing, instead of an unsupported product. But it neglects the other issue, which is that companies intentionally cut off the possibility of "good enough" to guarantee revenue.

I don't think its an accident that products like Microsoft Office went to subscriptions around the time it becomes very hard to imagine a new feature actually worth buying for.

Post reply on HN