Jump to content

Talk:Cray

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia


"Joseph Patrick O'Brien Jr."... actual Cray Inventor?

[edit]

This quote (below) from the intro paragraph of the History section seems especially pertinent. If the original architecture for the Cray supercomputer came from an outside inventor we should be informed about this mystery genius. Who is Joseph Patrick O'Brien Jr., where is he now, what was his ownership stake in Cray (did he ever benefit from the use of his designs?)? Am I the only one who thinks this is important information? We have all heard of Seymour Cray... what of the guy who did the low-level designs upon which his computers seem to be based?

"One of the contributing factors to his leaving CDC was that he obtained a design from a student at the Control Data Institute of Technology in Los Angeles, California. This design increased the speed of the fastest known computer (at the time) tenfold. The design was submitted by Joseph Patrick O'Brien Jr. to Control Data's Designing and Engineering Headquarters in Minneapolis, Minnesota in 1971. The schematic presentation was in Boolean algebra, as well as the component level. The circuit (now public domain) has remained the fastest to this day."

--Randall Lee Reetz —Preceding unsigned comment added by 69.232.238.168 (talk) 18:23, 21 September 2007 (UTC)[reply]

I see no reason to keep this paragraph. An E-mail address to a person claiming to have letters is not a reliable source. If the circuit truly is in the public domain, then at the very least, there should be evidence of it. I'm unable to find any such evidence. 76.204.88.63 20:32, 9 November 2007 (UTC)[reply]

For information regarding this subject, the following Wikipedia Talk link pretty much covers it.

[1]

--Jopedia (talk) 09:16, 4 January 2008 (UTC)[reply]

SX-6 picture irrelevant/"un-historic"?

[edit]

Why is there a picture of SX-6. It's a NEC system marketed by Cray in the US. It definitively is not an original Cray product and does not reflect historic achievements of Cray. You should at least put a picture of X-1. --Anonymous #1

Sarbanes-Oxley

[edit]

I'm not sure the section "Cray Inc. and Sarbanes Oxley Compliance: Common Example of the Hard-Hitting SOX" belongs in the Cray article. I think it belongs in an article about Sarbanes Oxley Compliance. --Anonymous #2

I think this is exactly where it belongs! I quit Cray just shortly after the hulabaloo over the SBO non-compliance. They are in terrible shape and this is just points out their managment problems. --Anonymous #3

Intro text unclear re S. Cray's startups

[edit]

The first paragraph is confusing. Cray start the first company in 1972? Did he leave it in 1976? I mean, it sounds like he started both companies but only one is explicited mentioned as a company he started and well - its just confusing. By reading it I have no idea what really happened. --Anonymous #4

split suggestion for Computers section

[edit]

A list of Cray computers needs to be on a seperete article, as they were created by multiple companies (Cray Research, Cray Computer Corp., and Cray Inc.), so to have them listed on the article for Cray Inc. (this article) is innacurate IMO.

Slark 08:28, 16 February 2006 (UTC)[reply]

On the other hand, this article currently covers both CRI and Cray Inc, and the Cray-3 originated at CRI, leaving the Cray-4 as the only model not strictly related to CRI or Cray Inc. Letdorf 13:40, 16 February 2006 (UTC)[reply]
I'm not intending to be a part of this discussion, but just a note on capitalization: You've blanked Cray Computers, which used to be a redirect to Cray Inc.. If you intend to split the page, make sure it is to Cray Computers, or if not, then ask an admin to speedy the page. Thanks =) Kareeser|Talk! 05:22, 19 February 2006 (UTC)[reply]
sorry about blanking the Cray Computers redirect, me bad :( I changed the template suggestion to Cray supercomputers - is this more appropriate? Slark 03:42, 20 February 2006 (UTC)[reply]
If you want to split that section, I'd recommend list of Cray supercomputers as the article title. —Mulad (talk) 01:28, 23 March 2006 (UTC)[reply]
I changed the suggested section split title to list of Cray supercomputers. Slark 08:53, 22 April 2006 (UTC)[reply]
Oppose. The list isn't all that long, and this IS an an all-encompassing Cray article. Maybe if someone dug up the details for each supercomputer (or series, I guess), it would warrant a split. --CCFreak2K 22:12, 19 July 2006 (UTC)[reply]

I deleted the split suggestion, as the consensus is to keep the section here Slark 03:34, 29 December 2006 (UTC)[reply]

As a former Cray Research employee who has done his homework, there should be a split between Cray RESEARCH, Inc. and the Cray Inc. They are two distinct entities.

I also note many omissions and actual details of our organization at that time. I have had a intimate relationship with the the politics and resultant behavior of the main parties at that time. The details would flesh out the whole CRAY RESEARCH, INC. story.

For proof, check the ExCray.com site. I'm listed there. Arthur Blackwell, a member of Dr. Chen's Design Team.

Aeb1barfo (talk) 03:49, 6 July 2012 (UTC)[reply]

[edit]

Sorry if it's a newby quesion but how do I edit the External link on the Cray Inc. page from Fred Gannet's Cray Supercomputer FAQ to Fred Gannett's Cray Supercomputer FAQ , gotta have tthatt exttra t ? Gannett 10:26, 26 March 2006 (UTC)[reply]

Sorry about that, fixed it for you now. Letdorf 21:58, 26 March 2006 (UTC)[reply]

Cray I faster or slower than ILLIAC IV?

[edit]

This article on Cray states that "The Cray-1 was a major success when it was released, faster than all computers at the time except for the ILLIAC IV." But the linked article on ILLIAC IV states that "(ILLIAC IV) was finally ready for operation in 1976, after a decade of development that was now massively late, massively over budget, and outperformed by existing commercial machines like the Cray-1." These two articles seem to contradict each other. —The preceding unsigned comment was added by 71.50.113.68 (talkcontribs) 21:17, 25 April 2006 (UTC{{{3}}})

Not only that, but the ILLIAC article contradicts itself. It says at the top of the page that the Cray-1 was faster, but says at the bottom of the page that the Illiac began operation in 1976, the same year the Cray-1 was released with roughly the same performance. --Blainster 08:25, 26 April 2006 (UTC)[reply]
Having written the majority of both articles, let me explain. The ILLIAC actually started running in 1974, although at the time it was really just the processor and was very unreliable. NASA spent a considerable amount of time trying to get it to the point where people could use it for real-world programs, which was complete by 1976, when it started running commercial loads.
In theory, the ILLIAC could boast about 250 MFLOPS, about the same as the absolute best performance of the Cray. However, the Cray made it much easier to actually reach that level of performance -- remember, the Cray was the fastest scalar computer too, it wasn't just good at vectors. Overall, a program running on both was much more likely to run faster on the Cray, but that said, for the kind of stuff NASA was running on the ILLIAC, it would likely outperform the Cray.
I admit its somewhat complex, which is why I left most of the discussion out of the articles. Maury 21:13, 18 July 2006 (UTC)[reply]
The qualitative technical problem is called Amdahl's Law. But to understand the problem qualitatively is to use an analogy from Amdahl's colleague Fred Brooks, another IBM 360 colleague, which came to describe his separate but related "law", Brooks Law. The analogy is that 9 women can't have a baby in 1 month; so what makes a 64-processor machine faster than a single processor machine? The answer is "only certain kinds of decomposable problems." Speed was made slower for memory lacks and having to go to disk. 198.123.50.52 (talk) 18:13, 16 May 2008 (UTC)[reply]

Financial troubles no longer relevant

[edit]

It's my contention that the section on Cray's financial and compliance issues of 2005 are no longer interesting as part of a wikipedia entry. Unless issues of stock price and listing compliance are timely, or represent an imminent threat to the continued operation of the company, or listing on a stock exchange, I don't think the company's wikipedia page is the best place to document the ups and downs of a stock. I plan to remove this section unless there is a strong objection. —The preceding unsigned comment was added by Forestlaughing (talkcontribs) 03:48, 18 April 2007 (UTC).[reply]

I think it was a good change. Jdavis79 04:41, 23 April 2007 (UTC)[reply]

E10K vs. "traditional Intel systems" ?

[edit]

From the article:

the extremely successful Enterprise 10000 range of servers [...]continue to be sold to this day, and only recently have traditional Intel-based systems started to approach [their] performance

A couple of problems here. First, I wouldn't say the E10K is being sold today, other than perhaps some kind of niche resale market. You could say its successors are still being sold today, perhaps.

Second: "traditional Intel system" ? The implication here is that "computers" has always equalled "Intel systems" and that anything else in the history of computing is an oddity. If the intent is to say that a current PC is as fast as an E10K, that's a bit of a stretch. If it means to compare to top-line Intel-based supercomputers, that's not what I'd call "traditional", unless again it means to say that all HPC systems are Intel-based.

So I'm not sure how to fix this. There's a point to be made here ("SGI gave away what became an enormously successful product line for Sun"), and there needs to be an explanation as to why this is so, but as written this is way off the mark.--NapoliRoma 04:16, 6 August 2007 (UTC)[reply]

I agree that this is a problem. In the general case far too many computing-related articles are written as though the author is unaware of any non-PC technology, in this particular case I think it needs to be emphasised that Sun had 8-20 way SMP systems while most PC-derivatives stopped at 2x. Another relevant point is that there's a whole lot of accumulated cruft in the PC architecture that quite simply isn't Intel technology- when the PC first came out Intel had a multi-master bus and were working on additional coprocessors. MarkMLl 11:06, 6 August 2007 (UTC)[reply]

Case of beer accompanying new delivery

[edit]

Do we want to add a comment under the trivia section that anytime a new supercomputer is delivered and installed on site for Cray, that the engineers will also bring a case of (can't remember the name) beer. I don't know of any other supercomping company that does this, which makes it unique. -- Joelhacker 06:16, 31 October 2007 (UTC)[reply]

That was a case of LEINIES brewed in Chippewa Falls, you know, where we brewed the CRAYS......

Art, a Cray-on...Aeb1barfo (talk) 04:01, 22 December 2007 (UTC) —Preceding unsigned comment added by Aeb1barfo (talkcontribs) 03:55, 22 December 2007 (UTC)[reply]


I see a LOT of mis-information about the Cray Research years here. I knew the juicy bits about the breakup, who caused it, SSI and bits about the death of Seymour....Aeb1barfo (talk) 04:01, 22 December 2007 (UTC)[reply]

Future Products

[edit]

I'm a little cloudy on the correlation between code names and released products. I believe that "Baker" is the current XT5, and "Black Widow" is the current XT5h. Is that correct? It seems that Marble will be the successor to the current XT5, and Granite will be the successor to the XMT. I notice that the roadmap slide makes it look like the XT5h product line just fades out without a successor. Does this mean Cray will be abandoning vector processors in favor of the Tera architecture? Also, for Cray's x86 line (XT5), are they locked into Opteron, or will they also be able to support Xeon? Given all the troubles that AMD has had recently (Barcelona delays and bugs, and push-out of the HPC products Shanghai and Bulldozer), it may hurt Cray that they are locked into Opteron. I don't see how Cray could have the resources to support a proprietary interconnect for both Opteron and Xeon. So I think Cray is hosed by being locked into Opteron. Any thougths? Westwind273 (talk) 16:31, 28 February 2008 (UTC)[reply]

Apple

[edit]

I removed the Apple/Cray item because I can't find a good source for it. Yes, the story is told. Yes, Apple used Crays and Cray Research used Apples. But no one seems to know whether Seymour Cray actually said what he's claimed to have said. WhatamIdoing (talk) 02:15, 29 April 2008 (UTC)[reply]

In writing, it was confirmed decades back in a Cray User Group proceedings. 198.123.50.52 (talk) 18:04, 16 May 2008 (UTC)[reply]

Propose rename to Cray Inc

[edit]

Propose rename this article to Cray Inc (since Cray Research redirects here) and make Cray into a disambiguation page. Could be the first step to moving Seymour's biography to beef up Seymour Cray. Any objections to the rename ? Rod57 (talk) 21:50, 26 October 2008 (UTC)[reply]

I'm guessing the disambiguation exception to the "no legal status in title" rule in WP:NCCORP would apply here? But renaming this article would imply that it can't be considered the primary topic (WP:PRIMARYTOPIC) for the title "Cray". Is this the case? Letdorf (talk) 23:30, 26 October 2008 (UTC).[reply]
IMO, Yes and yes. Rod57 (talk) 23:40, 19 November 2008 (UTC)[reply]

$

[edit]

Recent edits have added a space between "$" and n. Why? Rilak (talk) 14:25, 22 May 2009 (UTC)[reply]

Probably best to ask Electron9 (talk · contribs), I think. Letdorf (talk) 16:28, 22 May 2009 (UTC).[reply]

ECL vs. CMOS

[edit]

I believe the biggest factor that made the Cray machines obsolete was not competition with parallelism (Thinking Machines wasn't really that successful for very long), but the basic physics as integrated circuits shrank. CMOS logic was fairly slow in the late 1970s and early 1980s, and Cray used a completely different and faster system of ECL (emitter coupled logic), based on flowing and switching current, rather than charging complementary capacitors. In a talk at Caltech in 1979, Carver Mead predicted that CMOS would overtake ECL in speed once the circuits shrank to a sufficiently small size. Years later, when the clock speeds of micro processors began to exceed 100 MHz, that pulled the rug out from under the large ECL-based super computers. DonPMitchell (talk) 01:17, 10 June 2009 (UTC)[reply]

Some old page history

[edit]

Some old edits that used to be at the title "Cray" are now at Talk:Cray/Old edits. Graham87 13:18, 10 August 2009 (UTC)[reply]

Founded date

[edit]

The infobox currently states that the company was founded in 1972 as Cray Research, Inc. This is incorrect, since the current company is the same business entity as Tera Computer Company, which was founded in 1987, as stated on the Tera Wikipedia page, and on the Bloomberg and Yahoo! websites. True, the Cray website's history page says that "Cray Inc. builds upon a rich history that extends back to 1972", but that's not a statement of when the company was founded.

Does anyone have a reason to retain 1972 in the infobox? If not, I'll correct the year by changing it to 1987.

- Dotyoyo (talk) 08:11, 25 November 2011 (UTC)[reply]

Well, this article is entitled simply "Cray" and (currently) covers all the businesses associated with Seymour Cray that used the name; including CRI, CCC, Cray as an SGI subsidiary, and Cray Inc. Since the oldest of these companies was founded in 1972, this seems to me to be an appropriate date to have in the infobox. There is a separate Tera Computer Company article covering this company before the Cray merger. Regards,Letdorf (talk) 10:46, 25 November 2011 (UTC).[reply]
I think the root of the problem is that this article conflates "Cray" the brand with "Cray" the company known as Cray Inc. The top of the article states "This article is about the corporation.", but as you say, it covers several corporations. The current company infobox is misleading, in that it implies that there was a company founded in 1972 that is now called Cray Inc. If the article covers several companies, then perhaps it should either (a) have an infobox that lists the relevant information for each of those companies, or (b) have no company infobox at all. Are there any other options that don't mislead the reader? - Dotyoyo (talk) 19:32, 25 November 2011 (UTC)[reply]
Yes, I guess we could have separate infoboxes in the appropriate sections, if the relevant details can be found. I've edited the hatnote to clarify the subject of the article. Regards, Letdorf (talk) 21:32, 25 November 2011 (UTC).[reply]

What about the Cray XC Machines

[edit]

As of 2014, several Cray XC machines are listed among the world's most powerful computers, yet the XC30 and XC40 are not mentioned anywhere in this article. Perhaps an update is needed? Marchino61 (talk) 00:00, 12 November 2014 (UTC)[reply]

Significant Events of Cray

[edit]

The article has too much trivia and really lacks an outline the important events in the Cray history.

1. The Cray 1 was fast because of the vector loading from memory. The ECL ram was of secondary importance. The vector coding is a significant restriction on the types of jobs that can be run. Supers have always had some sort of restrictions on the type of coding allowed. The word vector is only mentioned a few times in the article.

2. The Cray 2 was a dog at 100MFl per cpu. Steve Chen's YMP, supposedly a minor project, was 3X faster (300MFl per cpu) and half the price. Seymour Cray left and started the rival. Shame? Guilt? Jealousy? Self-sacrifice? Who knows.

3. The T90 was not bad, but too slow compared to the emerging Japanese machines.

4. The T3D was a total dog. The aggregate disk i/o speed was slower than a PC. It took 30mins to do a core dump. (I had a few dumps so I know that for a fact.) The T3E was too little too late. IBM's MPP using RS6k cpus was also a dog. The MPP experiment sealed Cray's fate. They started the J series of poor-mans' supers to try and get revenue going.

5. The J machines were $3million badly spent. By this time the Xenon PC cpu was not too far behind. The NEC machines were light years ahead, and highly scalable, from desk-side, to requiring their own building.27.33.21.199 (talk) 00:30, 30 September 2015 (UTC)[reply]

Cray unclear

[edit]

In the section on Cray Research (CR) and CCC at least 3 times it says 'Cray' without indicating if it refers to Seymour, CR or CCC. Whole section needs clarification of 'cray' and more dates. - Rod57 (talk) 08:29, 3 January 2016 (UTC)[reply]

Move discussion in progress

[edit]

There is a move discussion in progress on Talk:Cray (disambiguation) which affects this page. Please participate on that page and not in this talk page section. Thank you. —RMCD bot 18:04, 29 January 2021 (UTC)[reply]