Re: More Alpha AXP Performance Numbers

Article View: comp.arch

Article #8019

#8019

From: Jason W Solinsky
Date: Fri, 13 Nov 1992 14:09

35 lines
2967 bytes

In article <1992Nov10.204139.5148@ryn.mro4.dec.com>, bhandarkar@wrksys.enet.dec.com (Dileep Bhandarkar) writes:
|> This is in response to e-mail requests for more benchmark data.
|> 
|>     Alpha AXP OpenVMS (version 1.0) SYSTEMS PERFORMANCE DATA 
|> 
|>  |---------------------------------------------------------------------------|
|>  |METRIC              | 3000/400  | 3000/500 |   4000   | 7000      | 10000  |
|>  |====================|===========|==========|==========|===========|========|
|>  |SPECint89           |   65.8    |   73.5   |   83.7   |   95.1    | 104.5  |
|>  |SPECfp89            |  150.6    |  169.9   |  188.4   |  244.2    | 268.6  |
|>  |SPECmark89          |  108.1    |  121.5   |  136.2   |  167.4    | 184.1  |
|>  |--------------------|------------------------------------------------------|
|>  |SPECint92           |   65.3    |   74.3   |   83.5   |   96.6    | 106.5  |
|>  |SPECfp92            |  112.2    |  126.0   |  143.1   |  182.1    | 200.4  |
|>  |--------------------|------------------------------------------------------|
|>  |Dhrystone(DMIPS)V1.1|  129.9    |  146.7   |  158.8   |  177.3    | 194.5  |
|>  |--------------------|------------------------------------------------------|
|>  |Linpack 100x100     |   26.4    |   30.2   |   36.3   |   38.6    |  42.5  |
|>  |       1000x1000    |   70.8    |   79.9   |   86.4   |  102.1    | 111.6  |
|>  |--------------------|------------------------------------------------------|
|>  |Perfect BM Suite    |   18.1    |   20.4   |   22.9   |   26.0    |  28.6  |
|>  | Mflops(Geom. Mean) |           |          |          |           |        |
|>  |--------------------|------------------------------------------------------|
|>  |DN&R Labs           |           |          |          |           |        |
|>  |CPU2    (MVUPS)     |  159.1    |  179.7   |  193.2   |  217.9    | 239.3  |
|>  |--------------------|------------------------------------------------------|
|>  |Livermore Loop Mflop|   18.7    |   21.3   |   22.9   |   25.6    |  28.1  |
|>  |--------------------|------------------------------------------------------|

Based on this data, it is extraordinarily unclear why somebody would be willing
to go for the 10000 box when the 3000 delivers better than half the performance
at a TINY fraction of the price. Also, why aren't there any multi-processing
configurations available?

Jason W. Solinsky

Technical Details

Message-ID: <1992Nov13.140907.2304@athena.mit.edu>

Path: arc1-ca.pugleaf.net!sparky!uunet!noc.near.net!hri.com!snorkelwacker.mit.edu!bloom-picayune.mit.edu!athena.mit.edu!solman

References: <1992Nov10.204139.5148@ryn.mro4.dec.com>

Show Full Headers

Complete Article Headers

Path: arc1-ca.pugleaf.net!sparky!uunet!noc.near.net!hri.com!snorkelwacker.mit.edu!bloom-picayune.mit.edu!athena.mit.edu!solman
From: solman@athena.mit.edu (Jason W Solinsky)
Newsgroups: comp.arch
Subject: Re: More Alpha AXP Performance Numbers
Message-ID: <1992Nov13.140907.2304@athena.mit.edu>
Date: 13 Nov 92 14:09:07 GMT
References: <1992Nov10.204139.5148@ryn.mro4.dec.com>
Sender: news@athena.mit.edu (News system)
Organization: Massachusetts Institute of Technology
Lines: 35
Nntp-Posting-Host: m4-035-18.mit.edu

Reply to Parent Message