User talk:Rilak

PA-RISC pages vs. OpenPA

Hi Rilak.

I'm reading with growing fascination your articles on the PA-RISC processors, currently the new page on PA-8000. What bothers me, however, is -- I'm the author of OpenPA.net -- that the PA-8000 page and its content are, well, quite similar to the OpenPA CPU page. This includes even most of the same references for the single PA-8x00 processors that I used.

What's the point of this? Rephrasing the information I compiled on that page?

And OpenPA isn't even mentioned anywhere.

I find it very odd you have mostly the same information, according to almost the same references as the CPU sections on OpenPA.

What's that supposed to be?

And how am I supposed to feel, after compiling this information for years, and now seeing very similar information with very similar structure with almost identical references on Wikipedia?

This is not the right way to go. —Preceding unsigned comment added by 85.179.24.236 (talk) 07:57, 1 May 2009 (UTC)

Hello! Thank you for notifying me of your concerns. Firstly, I do not own the article, per Wikipedia policy (WP:OWN). In regards to your concerns, my contributions are independent works, and were not derived from OpenPA.net. You claim that the content is quite similar, yet you did not elaborate as to how. It would be very helpful if you did, as I can identify problems more quickly.

Lets do a quick test for similarity. Let us consider the descriptions of the PA-8700 at OpenPA, and at Wikipedia:

Your text:

"The PA-8700 is basically an enhanced and revamped PA-8500 core with some slight modifications. As all PA-8x00 CPUs before, it logically still is very close to the original PA-8000 core from 1997. All subsequent new CPUs from HP were based on this design and added several features and some slight modifications to it while retaining the basic PA-RISC version 2.0 core. The PA-8700 enhanced the on-chip L1 caches and the TLB significantly while switching to a new CMOS-process helped boosting the clock-frequency. The chip was at its time one of the largest available commercial CPUs and one of the first to be manufactured in a SOI (Silicon On Insulator) process. The PA-8700 was manufactured by IBM, in contrast to the PA-8500 and PA-8600, which were fabbed by Intel, after HP gave up its processor fabs long time ago."

My contributions:

"The PA-8700 (PCX-W2), code-named Piranha, is a further development of the PA-8600. Introduced in August 2001, it operated at 625 to 750 MHz. Improvements were the implementation of data prefetching, a quasi-LRU replacement policy for the data cache, and a larger 44-bit physical address space to address 16 TB of physical memory. The PA-8700 also has larger instruction and data caches, increased in capacity by 50% to 0.75 MB and 1.5 MB, respectively. The PA-8700 was fabricated by IBM Microelectronics in a 0.18 µm silicon on insulator (SOI) CMOS process with seven levels of copper interconnect and low-K dielectric."

I do not see any similarity. There are multiple significant differences, for example, the texts disagree as to which microprocessor the PA-8700 was derived from, yours claim it was the 8500, mine claims it was the 8600. You omit the introduction date and clock frequency, I omit the commentary about similarity to the 8000, etc. If I do happen do cover the same fact (the larger caches for example), may I point out that most works about the PA-8700 cover it too.

I am of course, not cherry picking examples, I will review the entire article in the near future. In the mean time, if you know of obvious similarities (word for word, obvious paraphrasing), please do not hesitate to leave a message on my talk page pointing it out. Copyright violation is a serious matter that I view seriously.

Once again, thank you for bringing this to my attention. I hope that we stay in contact until this is resolved and that it be as quickly as possible.

Yours sincerely, Rilak (talk) 08:52, 1 May 2009 (UTC)

Hi, thanks for your prompt reply. Much appreciated. To make this clear up front: I did not mean to imply that you violated my or other copyright while writing this article. I would have said so if I thought it would've been the case.

So, back to my point, or rather, what I was trying to get at. I watched the page work in progress over the weeks and was unsure what to make of it.

I do not really understand the point of making pages based on mostly the same references which contain very similar information to those on other, existing pages/sites. What this looked like to me was gleaning the references from the existing (my) pages, using these references and the pre-existing art as, well, "reference"/structure, and from there, making an own page.

Again, I did not mean to imply you copied anything over from me or somewhere else. I did not like the process (as I imagined/perceived it) and I certainly did not like the outcome -- there was a pre-existing resource on that information, that resource was free/open and I was certainly open to suggestions/additions/collaboration. Who gains from this diffusion of information and content?

To your arguments: I do not want to get into any cherrypicking either. First, you're quoting from the relevant page from the date of January, why is that? Then, the information which you say is missing is included in the lists before and after the relevant paragraph -- the section in question is actually longer than the paragraph in itself. But, as you said, this is cherrypicking and not what I was trying to get at.

To wrap this up, as I'm unsure a Wiki page is the proper place for direct communication: I do not see how/if this can be resolved. I do not like the outcome, a) because it affects me and the resources I put into this, but also b) because I do not like the concept generally to duplicate information which is in more specific locations to resources with a much broader (but shallower) scope. I'd certainly happy if you want to continue to do this by mail, if you want. A contact address is given on the aforementioned site.

br —Preceding unsigned comment added by 85.179.40.203 (talk) 18:28, 2 May 2009 (UTC)

Sorry: I said you used an old version of my page. I just realized all other versions but one (from May) of the page in question are also from January. However you cite a specially old one (from 5th or 6th Jan). Not too important though.

SpaceBalls

Your SpaceBalls reference is to a brand name. A generic name would be better.

Robert.Harker (talk) 18:30, 6 March 2009 (UTC)

- vs. –

Either's fine with me :-)... Apparently WP automatically maps – back to "-" in wikilinks anyway.--NapoliRoma (talk) 05:27, 10 March 2009 (UTC)

Intel Atom - why don't we understand each other ;)

"No it isn't right..." Could you elaborate on that please.

Regards HenkeB (talk) 08:02, 28 March 2009 (UTC)

Absolutely. I was half-way through writing a message at the Atom talk page. All the relevant literature that I have seen (papers, books, articles) all describe modern x86 as decoding or translating x86 instructions into micro-ops. The number of micro-ops produced by the decoder depends on the complexity of the instruction. This is described in many ways (such as "this instruction produces n micro-ops") except for the term "divide" because nothing is divided into smaller bits.

Perhaps a good analogy is that if you have a BASIC "instruction" and it is compiled into machine code, is the BASIC instruction divided into smaller bits, or does it get interpreted as a sequence of lower-level instructions?

By interpreted, I suppose you mean translated, represented by, or similar. To be very precise: the program text (or tokenized format) is interpreted as that particular BASIC statement (not as lower-level instructions). This triggers the generation of a sequence of simpler instructions with the semantics (or meaning) of that BASIC statement divided among them. (An interpreter would, of course, instead perform a direct execution of a similar sequence.) HenkeB (talk) 10:51, 3 April 2009 (UTC)

Yes, I suppose so. A bad choice of words! :) Rilak (talk) 09:30, 5 April 2009 (UTC)

I must admit that my edits did not stress this enough. Perhaps I wasn't reading the article correctly, and I only noticed it recently. I think it has been fixed in the latest round of edits.

I realize that my edits must seem like edit warring and I must apologise if so. Regards, Rilak (talk) 08:20, 28 March 2009 (UTC)

Sorry for my late reply, I wasn't expecting an answer (on that retorical question). Regarding litterature, I belive any encyclopedia should avoid simple copying of ready-made wordings; instead, every formulation should go through the minds of (ideally) the initiated experts writing it (which is not original research btw).

I see your point, nothing physical, like instruction encodings, is divided, that absolutely true. However, if you try a slighly more abstract view, having the word instruction denote the semantics (the actions), instead of the physical representation, you should see that divide becomes much more natural. Also, as you know, one of the main points here is that the Atom does not divide most instructions into simpler tasks but merely translates them 1:1 into easily handled fixed-length encodings (that are still much larger than the "cache-efficient" x86-format).

I should have tried to emphazise this view as soon as I saw your comment: How is "divide" more descriptive? The opcode is separated from the function field? The register fields are split into half? Anyway, thanks for your tone in this reply, I appreciate it. Regards, HenkeB (talk) 10:51, 3 April 2009 (UTC)

Well, from an abstract point of view, I suppose "divide" does work.

In regards to the formulation of prose, I think using established terms has its benefits. One of them is that they are well defined in many reputable publications, so there is little likelihood that the prose would stray from the intended meaning. Too many times have I read articles from other encyclopedias and publications (including ones considered to be reliable sources by Wikipedia editors and cited in articles) where the author has decided to simplify a concept only to end up failing spectacularly. Of course, I am not saying that this is the case with the article in question.

Using established terms also maintains consistency with the rest of the world, so that Wikipedia are interoperable with the rest of the world. Another benefit is that it avoids confusion. Sometimes when something is described with different terminology, it is because it didn't quite fit the definition of the established terms. If we were to use different terms, it may be interpreted as such.

Despite all of the above, I don't feel strongly about the exact wording either way. Regards, Rilak (talk) 09:30, 5 April 2009 (UTC)

Yes, if a particular term is indeed well established and stable over time, we should definitely use it, but supported by explanatory prose that appeals to a previously uninformed reader's intuitive understanding. Many terms has several more or less reasonable meanings that changes over time and fashion, single common words given an overly specific usage in some narrow context, which then starts to compete with the original meaning. Naively copying such "established" terms tends to become a simple listing of buzzwords, something that's already far to common on WP. Instead, we should emphasize genuine understanding of fundamental concepts.

I agree that "simplified descriptions" sometimes end up strange or plain wrong. However, most topics may be described from several different points of view, and at various levels of abstraction and detail, so, sometimes, when the author is "failing spectacularly", it may perhaps be the reader that is a little to rigid to grasp it that way (no offense intended!). People think differently, so, when dealing with complicated subjects, a couple of alternative formulations is often a good thing (provided the article remains neat and well structured); it helps us serve a diverse audience better.

Naturally, I'm not expecting you to agree on everything above, instead, we probably have to agree to disagree. Regards, HenkeB (talk) 14:13, 7 April 2009 (UTC)

PowerPC

Regarding a PowerPC 620 article. Yes, If one/we/you can gather enough for one stand-alone article I'm supporting that. The reason to merge all 6xx articles into one was that there was too little for stand-alone articles. I'm happy that you are engaging the POWER3 article too. POWER4 and POWER5 is in pretty bad shape too.. We could perhaps do something there too. -- Henriok (talk) 16:33, 13 April 2009 (UTC)