Steve,

Thanks for the results... now you've really complicated the results tables,
by adding VM versions! ;-) I'll add these ASAP.

Did the GC performance improve at all? Some people may be willing to
sacrifice a few FPS if the pauses for GC are minimal.

Daniel Selman
Tornado Labs Ltd.

Email:   [EMAIL PROTECTED]
Web:     http://www.tornadolabs.com
Phone:   +44 (0131) 343 2513
Fax:     +44 07070 800 483



-----Original Message-----
From: Stephen E. Schweibinz [mailto:[EMAIL PROTECTED]]
Sent: 03 May 1999 15:16
To: [EMAIL PROTECTED]
Cc: [EMAIL PROTECTED]
Subject: Re: [java3d] Hotspot and Java3D



>Guillaume Bedard wrote:
>>
>> Hi all!
>>
>> I have tried the new HotSpot VM with our application which makes use
>> of Java3D, and for an unknown reason, it is noticeably slower than
>> the classic VM in all situations.  I must admit that I was hoping
>> a totally different result.
>>
>> If someone has more details or similar results, please answer.
>
>I noticed this too, when I was using the pre-release version. I didn't
>want to say anything because you couldn't talk about performance in
>public, but I guess that's changed now.
>
>Maybe a member of the J3D team can comment on this ? I'm sure they have
>tested Hotspot and noticed it ...
>


I found similar results.  HotSpot didn't seem to
speed up Java3D.  Here are my scores on the tornado
labs benchmark, before and after HotSpot.

Steve Schweibinz


CPU     Dual PII OverDrive
Speed   333 x 2
OS      Win NT SP3
Video   STB Velocity 4400, PCI, nVidia Riva TNT (16Mb), v. 1.01
RAM     320 MB

------------------------------------------------------------------
512x512 Resolution

        FPS without HotSpot     FPS with HotSpot

Test 1          35.97                   32.66
Test 2          177.93                  177.94
Test 3          31.68                   31.68

------------------------------------------------------------------
256x256 Resolution

        FPS without HotSpot     FPS with HotSpot

Test 1          38.08                   34.39
Test 2          188.67                  188.68
Test 3          32.01                   31.07


/*******************************************************

   Stephen E. Schweibinz
   Planning Support Systems Group, MCP candidate, MIT
   (617) 679-0360
   [EMAIL PROTECTED]
   http://www.vcp.com

*******************************************************/



=====================================================================
To subscribe/unsubscribe, send mail to [EMAIL PROTECTED]
Java 3D Home Page: http://java.sun.com/products/java-media/3D/

Reply via email to