Jump to content

Talk:PARAM

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

15 September 2011: Article Cleanup

[edit]

The following changes were made:

  • Unreferenced material removed
  • Citations checked
  • New material added
  • Everything related to GARUDA grid computing removed. GARUDA has a separate page.
  • 1 Picture added
  • New technical information added
  • New Citations added
  • Article divided into subsections

Requesting peer review. Kenfyre 09:04, 15 September 2011 (UTC)[reply]

Article needs technical detail

[edit]

The article needs more architectural detail and detail about software beyond the OS. E.g, base CPU, memory, cache, registers, and compilers and applications. 143.232.210.38 (talk) 06:28, 20 July 2010 (UTC)[reply]

The best place to get that would be straight from the source (C-DAC). Keep in mind though that these specs are for the previous version of PARAM (Padma) and not the current Yuva cluster. Vedant (talk) 19:07, 20 July 2010 (UTC)[reply]

just vogas — Preceding unsigned comment added by 180.234.78.136 (talk) 19:01, 30 May 2011 (UTC)[reply]

Export

[edit]

At least one of PARAM 8000 was exported to Russia and been installed in Keldysh Institute of Applied Mathematics in Moscow. But I can not find any prove of that. The Holm (talk) 04:18, 15 December 2015 (UTC)[reply]

A prototype produced was benchmarked at the 1990 Zurich Supercomputering Show. It surpassed most other systems, placing India second after US.

[edit]

Its not clear what the Zurich Supercomputering Show is, I can't find any record of such benchmarks and the best estimate for the system I can find was around 1 GFLOP which would have been good but unexceptional at the time. Additionally the most powerful system at the time was the Fujitsu VP2600/10 which was Japanese not american.©Geni (talk) 21:02, 18 September 2019 (UTC)[reply]

I re-added the section because it is sourced. However I do understand your concerns. From reading the source article again, I have clarified the statement to be "at the show" as that what it suggests. I have also added a note as I suspect (after googling) that the show was the "CONPAR 90 — VAPP IV Joint International Conference on Vector and Parallel Processing" which occurred in Zurich. There is a Springer book on the proceedings here: https://link.springer.com/book/10.1007%2F3-540-53065-7 . I suspect that a the time the generic terms around supercomputing hadn't been fully settled. I hope this is acceptable. - Master Of Ninja (talk) 09:56, 14 September 2021 (UTC)[reply]
The problem is its making claims we know are false. At around 1GFLOP its slower than the Japanese HITAC S-820 /70 and /80 models one of which was shipped in 1987. This is also the period where japan overtook the US with the Fujitsu VP2000. I suspect the issue is that the various authors telling the story of the plucky Indian team defying the US have sidestepped the Japanese issue because it complicates things.©Geni (talk) 15:02, 14 September 2021 (UTC)[reply]
Note that I did edit that it was "at the show" - can you prove the subtlety in this statement is false? I am not a supercomputer expert, but can read through the sources without straying into original research. I started editing these articles as there was a lack of citations in them. I have not looked into the S-820, but even looking at the linked wikipedia article, there are no references to cite to backup these claims. This is not to say they are true, but again where's the proof? I also note the VP2000 is minimally cited. I did edit the article to suggest that the computer was the 2nd fastest "at the show", and perhaps a further edit to say "the creators claim". The cited article did claim that the Japanese team in effect could not get their machine working at the same show. - Master Of Ninja (talk) 21:07, 14 September 2021 (UTC)[reply]
I've switched from publicly demonstrated to ran at the show. Until we can find another source on the bench-marking thing and preferably what was going in with super computers in general in 1990 its probably the best we can get.©Geni (talk) 23:31, 14 September 2021 (UTC)[reply]